A pyridazine series of α2/α3 subtype selective GABAA agonists for the treatment of anxiety

Article abstract of DOI:10.1021/jm051144x

The development of a series of GABA_A α2/α3 subtype selective pyridazine based benzodiazepine site agonists as anxiolytic agents with reduced sedative/ataxic potential is described, including the discovery of 16, a remarkably α3-selective compou

Full text of DOI:10.1021/jm051144x

2600
J. Med. Chem. 2006, 49, 2600-2610
A Pyridazine Series of r2/r3 Subtype Selective GABAA Agonists for the Treatment of Anxiety
Richard T. Lewis,* Wesley P. Blackaby, Timothy Blackburn, Andrew S. R. Jennings, Andrew Pike, Rowan A. Wilson,
David J. Hallett, Susan M. Cook, Pushpinder Ferris, George R. Marshall, David S. Reynolds, Wayne F. A. Sheppard,
Alison J. Smith, Bindi Sohal, Joanna Stanley, Spencer J. Tye, Keith A. Wafford, and John R. Atack
The Neuroscience Research Centre, Merck Sharp & Dohme Research Laboratories, Terlings Park, Eastwick Road,
Harlow, Essex, CM20 2QR, United Kingdom
ReceiVed NoVember 15, 2005
The development of a series of GABAA R2/R3 subtype selective pyridazine based benzodiazepine site agonists
as anxiolytic agents with reduced sedative/ataxic potential is described, including the discovery of 16, a
remarkably R3-selective compound ideal for in vivo study. These ligands are antagonists at the R1 subtype,
with good CNS penetration and receptor occupancy, and excellent oral bioavailability.
Introduction
The question remained as to whether it was possible to
discover subtype selective BZ site ligands, and whether it would
thus be possible to separate the anxiolytic from the sedative
properties. Elegant studies in transgenic mice,2 and in vivo
studies with subtype selective ligands generated in our labora-
tories,3 have now shown that receptors containing R1 seem to
be strongly linked with the sedative and ataxic properties of
BZ ligands, while R2 may be linked with at least part of their
anxiolytic profile. The role of R3 in anxiolysis has recently been
established.3d Here we report the culmination of our endeavors
to develop a series of R2 or R3 subtype selective BZ ligands
based on a pyridazine motif, as anxiolytic agents with reduced
ataxic potential, including the discovery of 16, a remarkably
R3 selective compound ideal for in vivo study.
γ-Aminobutyric acid (GABA), the major inhibitory neu-
rotransmitter in mammalian brain, regulates chloride ion flux
into neuronal cells by ligand gating GABAA receptors. This
family of membrane spanning ion channels serve as arbiters of
neuronal inhibitory tone and thereby play a crucial role in the
regulation of neurotransmission in the CNS. The benzodiazepine
class of anxiolytic, sedative, hypnotic, and myorelaxant thera-
peutic agents are known to exert their pharmacological effects
through allosteric modulation of the effects of endogenous
GABA on the conductance of the GABAA receptor. Since these
agents have no significant effect upon the ion channel in the
absence of GABA binding at its modulatory site, their mech-
anism of action ensures a safer pharmaceutical profile in
comparison with barbiturates, for example, which may activate
the channel directly. The myorelaxant/ataxic, sedative/hypnotic,
and memory impairing properties of benzodiazepines are
considered of clinical benefit when they are utilized in anes-
thesia, particularly for minor surgical procedures. However these
same properties are considered a significant liability when these
agents are used in the treatment of anxiety; such side effects
may make it difficult for the patient to function normally while
on medication and may even be considered as potentially
dangerous for those who work in a mechanized environment.
The GABAA channel is composed of a pentameric assembly
of protein subunits, and 19 subtypes of these subunits are
known1a (R1-6, â1-3, γ1-3, δ, ꢀ, π, θ, and F1-3), which may
theoretically coassemble to form a plethora of structurally
different but functionally competent ion pores. Speculation that
these might reasonably be expected to possess different proper-
ties and distribution in various brain regions is now supported
by a growing body of evidence. Fortunately, in studying the
interaction with benzodiazepine (BZ) ligands, one need not
consider the multitude of possible receptor subtype combina-
tions, since only those which contain a γ2 or γ31c subunit in
conjunction with R1, 2, 3, or 5 seem to bind this class of agents
with significant affinity. The benzodiazepine binding pocket
appears to be located at the interface of γ and R subunits, with
residues on both proteins making significant contributions to
the active site. Since the GABA binding site requires contribu-
tions from both an R and â subunit, we therefore need only
consider channels composed of R, â, and γ subtypes1 for our
purposes.
Results and Discussion
In our extensive search for subtype selective ligands at the
BZ site, we have observed that it is generally not possible to
obtain more than 10-fold binding selectivity for R2/R3 over R1.4
While binding selectivity may be difficult to achieve via the
BZ site, our observations indicate that functional selectivity is
attainable. In this work, the selectivity described is functional
in origin, i.e., selective efficacy. Because the BZ site is an
allosteric modulatory site (vide supra), compounds which bind
to it may elicit a range of different effects. Those which cause
an enhancement of the potency of GABA and hence increased
channel Cl- conductance are termed agonists (more properly
known as positive allosteric modulators), while those which
reduce the potency of GABA are referred to as inverse agonists
(negative allosteric modulators). Compounds which bind at the
BZ site but which do not influence the properties of GABA on
the channel are commonly termed antagonists, and since there
is generally considered to be no endogenous ligand for the BZ
binding site, such compounds have no functional effects, either
in vitro or in vivo. We have therefore sought compounds with
an antagonist profile at R1, and varying degrees of agonism at
R2 and R3. A role for R5-containing subtypes in learning and
memory impairment has been suggested, so low efficacy at this
subtype is preferred.5 Widely prescribed BZ ligands such as
diazepam and chlordiazepoxide (CDZ) are full agonists at all
these subtypes and cause maximal potentiation of GABA. The
compounds described below have, in many cases, higher affinity
for the BZ binding site, but are less efficacious at modulating
the effects of GABA; they are therefore termed partial agonists,
and their efficacy is reported as a fraction relative to the response
achievable with chlordiazepoxide under the same test conditions.
* To whom correspondence should be addressed. Tel +44 (0)1279
440000; Fax +44 (0) 1279 440390; e-mail: richard_t_lewis@btinternet.com.
10.1021/jm051144x CCC: $33.50 © 2006 American Chemical Society
Published on Web 03/22/2006

R2/R3 Subtype SelectiVe GABAA Agonists
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2601
Figure 1. Consideration of structural similarity of pyridazine lead with known7 imidazopyrimidine and imidazotriazine BZ site selective agonists.
Table 1. Binding Affinity and Efficacy at GABAA Receptor Subtypes
The pyridazine series of benzodiazepine site ligands 1
described herein originated from 4, identified by screening of
the Merck sample collection, and preliminary exploratory
structure-activity relationships (SAR) for this lead have been
reported.6 We were attracted by the results of these early studies,
which hinted that R2 or R3 subtype selective compounds might
be obtained. In addition we recognized certain structural
similarities with imidazotriazine 2 and imidazopyrimidine 3
series of subtype selective GABAA ligands which we have
extensively explored at Merck.7 It was initially postulated that
the adjacent nitrogens of the pyridazine core-structure might
effectively mimic the disposition of the nitrogen atoms at
positions 4 and 5 of imidazotriazine 2 (or the corresponding 1
and 8 positions of 3), and this influenced the direction in which
we chose to develop the SAR of this series (Figure 1).
Gratifyingly, the extensive SAR which had been mapped out
for the biaryl appendage in the imidazotriazine and imidazopy-
rimidine work7 proved to be directly transferable into the new
series and conferred similar effects upon selectivity of efficacy
modulation with respect to R unit subtype when appended at
the 5-position of the pyridazine core. We selected the two
fluoropyridyl moieties (1, R ) H or F) indicated in Table 1 for
several compelling reasons. Generally they both afforded
selective efficacy in their own right in the desired sense (but
differed in that the 3,5-difluoropyridine conferred lower efficacy
at all subtypes while retaining the selectivity window between
them), offered structural diversity from our other series, and
conveyed good pharmacokinetic (PK) properties (vide infra).
Unsurprisingly, the SAR of the 3-substituent, which crudely
maps to the 3-position of the imidazotriazine (or 7-position in
the imidazopyrimidine) proved to be significantly different from
these other series, presumably at least partially as a consequence
of alteration of the geometry and size of the heterocyclic
scaffold.8 This was regarded as advantageous in providing a
structurally distinct alternative series. A pendant aryl moiety
emerged as the 3-substituent of choice, conferring enhancement
of affinity. The receptor was also quite tolerant to a range of
substituents on this appendage, although obtrusive polar or
sterically demanding substituents at the para position tended to
be detrimental. Selected compounds of interest for their selective
efficacy profile are listed in Table 1. All show good (but
relatively unselective) binding affinity for R1, 3, and 5 subtypes.9
Selective efficacy could be modulated by appropriate choice
of substitution pattern on this moiety. Thus substitution at one
or both ortho positions proved to be beneficial both for affinity
and opening the window of selective R3 agonism over R1,
particularly when the flanking groups possessed lone pair
electron density to buttress the adjacent pyridazine nitrogen lone
pair. This is suggestive of an orthogonal disposition of these
two rings being optimal from an R3 selective efficacy standpoint.
We favored a bias toward an electron deficient aromatic moiety
from an early stage in this work because we were aware of the
intrinsically high turnover of the earlier compounds6 in liver
microsomal preparations, which we routinely employed as a
convenient surrogate marker of in vivo stability. The strategy
a Ki values for benzodiazepine sites on stably transfected GABAA
receptors Rxâ3γ2 (x ) 1, 2, 3, or 5). Inhibition curves were carried out
using receptors labeled with [3H]Ro15-1788 at a concentration of ap-
proximately twice the Kd. Ki values were calculated according to the Cheng-
Prussof equation. Data shown are mean values for three to six determina-
tions.3 Efficacy measured at GABAA receptors stably expressed in L(tk-)
cells using whole cell patch clamp recording and represents the effect of
the test compound on the current produced by an EC20-equivalent of
GABA.3 b Efficacy is expressed relative to the full agonist chlordiazepoxide
(CDZ) (relative equivalent efficacy ) 1.0), except where a percentage value
is shown, which represents a direct percent modulation of the EC20 control.
Values are the mean from at least four independent experiments. Negative
values reflect a reduction in the GABA EC20 induced chloride current and
indicate inverse agonist responses; these are reported as a percentage
modulation since expression of inverse agonism relative to an agonist
standard is inappropriate.3
utilized to optimize metabolic stability in the pyridazine series
built upon the extensive knowledge of the metabolic fate of
several other GABA ligands from these laboratories. The two
pyridazine-5-biaryl substituents chosen from the start in opti-

2602 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Lewis et al.
Table 2. Liver Microsomal Turnover, Selected Pharmacokinetic Parameters, and Rat In Vivo Receptor Occupancy Data
a Microsomal turnover data were obtained by incubation of the test compound (and diazepam as a control compound) at 1 µM concentration with liver
microsomes pooled from a number of individuals, at a protein concentration of 0.5 mg/mL for 0.25 h. Concentration of test compound remaining after
incubation was determined by LC/MS/MS analysis. Turnover values are the mean of three determinations, with a standard deviation of </) 5% for the
values quoted.3 b Determined in six animals, three dosed iv and three po. c Rat in vivo receptor occupancy was measured with [3H]Ro15-1788 according
to the published protocol3 and is the mean of three to six independent determinations.12
mizing this series were well-known to us as very resistant to
microsomal metabolism from our study of the imidazotriazine
series.7d Any remaining metabolic weak spots were therefore
likely to reside in the 3-arylpyridazine moiety. Pyridazine
N-oxidation was a strong suspect, which might be ameliorated
by appending an electron-withdrawing aromatic moiety at the
3-position. The 31% turnover of 8 in dog microsomal incubation
was reduced to just 2%, for its pyridazine-2-oxide derivative
5a, (and 10% for 5b Figure 2). More strongly electron-
withdrawing 3-substituents, such as difluorophenyls, 9 and 10

R2/R3 Subtype SelectiVe GABAA Agonists
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2603
Figure 2. Synthesized pyridazine N-oxides 5 are significantly more
resistant to liver microsomal turnover in vitro than is parent pyridazine
8.
Figure 4. Circumventing the principal route of metabolism in rat.
Pyridine N-oxide 20 is the major characterized metabolite of 14 in vitro.
The 2-pyridyl isomer 16 is resistant to microsomal turnover in vitro.
properties in man. It is possessed of an interesting selective
efficacy profile, being a clean antagonist at R1 and R5, and
weak-to-moderate partial agonist at both R2 and R3 subtypes
(Figure 5). The compound occupies the BZ receptor very
effectively in rat after po dosing,11 with an Occ50 of 0.22 mpk12
(milligrams per kilogram), at a plasma concentration of 38 nM.
In the rat plus-maze11 it was anxiolytic at a minimum effective
dose of 0.3 mpk po, corresponding to 56% receptor occupancy12
(Figure 6). Furthermore, compound 14 was fully effective in
the squirrel monkey conditioned emotional response (CER)
paradigm, a more stringent test of anxiolytic potential.11 In the
presence of a stimulus associated with shock (conditioned
stimulus (CS)), the response rate at 1 mpk po was 60% of that
observed in the absence of the CS (compared with a 10%
response rate in vehicle treated control animals), a significant
anxiolytic response. At doses between 1 and 10 mpk (the highest
studied, which corresponded to an 80% release rate) no overt
signs of ataxia or sedation were observed. In contrast, a similarly
therapeutic dose of diazepam (1 mpk, resulting in a 68% release
rate) causes obvious behavioral disruption in this assay. In the
mouse rotarod assay,11 a measure of ataxia, the compound had
no significant effect at 10 mpk po (corresponding to 97%
receptor occupancy12) in either the absence or presence of a
subthreshold dose of ethanol. It is therefore apparent that the
anxiolytic properties may indeed be separated from the ataxic
effects in an appropriately subtype selective compound, and that
its effects are not as prone to potentiation by ethanol as
traditional unselective BZ full agonist anxiolytic agents. As such,
it will be of significant interest to observe its properties in man13
where a more accurate gauge of side effect profile, including
sedative and cognitive effects, may be made. As a clean R5
antagonist, one might expect less memory disruption or cogni-
tive impairment to be evident.
Compound 15 is the most potent of the R2/R3 selective
agonists discovered in this series. Structurally it differs from
14 in two respects. Fluorine and nitrogen have been transposed
in the 3-substituent leading to a much improved rat PK profile
(Table 2), with low clearance translating to a 10 h half-life in
this species. The second point of variation is deletion of a
fluorine (R ) H, Table 1) on the other pyridine substituent,
which results in significantly increased R2/R3 efficacy (∼0.2
elevation at both subtypes, cf. close analogue 16, Figure 5).
Gratifyingly, it is also devoid of efficacy at R1, but has agonism
at R5 subtypes. The compound is a potent anxiolytic in the rat
plus-maze at 0.1 mpk po, which corresponds to 57% receptor
occupancy (Figure 6, Occ50 0.09 mpk). It showed no significant
effect at 3 mpk po in the mouse rotarod ataxia assay, with no
significant impairment in the presence of ethanol at doses which
gave 90% receptor occupancy.
Figure 3. Illustration of the conceptually pseudo-symmetric nature of
ligand structure resulting from consideration of nitrogen/fluorine
interchangeability.
had excellent affinity and interesting efficacy profiles (Table
1), but microsomal turnover, although much improved over
earlier compounds, was still considered to be too high,
particularly in dog (Table 2).
The second conceptual approach, which afforded a solution
to this problem, arose from a consideration of the pseudo-
symmetric nature of the molecular structure (Figure 3). In our
exploration of the SAR of several series of BZ ligands, we have
encountered a reasonably general observation that a ring nitrogen
may be successfully replaced with a pendant fluoro substituent
or vice versa. The 5-appended fluorophenyl-fluoropyridine biaryl
moiety is metabolically stable, and the arylpyridazine motif is
superficially quite similar to it when nitrogen/fluorine transposi-
tion is factored in. This concept is most apparent in the
comparison of compound 10 with 18, as shown in Figure 3. If
the enzymes responsible for the metabolism recognize only a
portion of the entire molecule, i.e. a biaryl motif, then it might
conceivably be possible to render both biaryl moieties similarly
stable. This was borne out in practice. Hence, replacement of
the difluoroaryl of 10 with 3-fluoropyridine 18 gave improved
microsomal stability in rat, dog, and human (Table 2), with
turnover reduced from around 20% in rat and dog for 10 to
less than 5% for 18. Further exploration of this SAR led to
discovery of a selection of compounds with good microsomal
stability (Table 2) and very interesting efficacy profiles (Table
1). It is interesting to compare, for example, the efficacy profiles
of compounds 9 and 12; 10 and 18; and 14, 16, and 17.
The somewhat poorer rat PK profile of compounds bearing
a 4-pyridyl motif was traced to a propensity for pyridine N-oxide
formation (20, Figure 4) in this species, confirmed in the case
of 14 by comparison of its metabolite profile with an authentic
standard.10 Relocation of the pyridine nitrogen to the ortho
position, as in 16, circumvented this route and accounts for the
exceptional stability and half-life of this compound in rat. This
metabolic pathway does not appear to be quite so predominant
in the other species studied. While compounds which are highly
turned over by liver microsomes are very likely to have poor
PK, low turnover is not necessarily indicative of good PK. It is
clear from these studies that for this series, the rat microsomal
turnover is not particularly well correlated with in vivo stability.
It served a very useful function in assisting the optimization
process, but should not be relied upon in isolation.
Compound 16, differing from 15 only by the presence of a
5-fluoro substituent (R ) F, Table 1) has an intriguing profile,
remarkable in that its only significant efficacy is through the
R3 subtype, being effectively functionally silent at R1, 2, and
5 (Figure 5). In addition, with low intrinsic clearance in rat,
Compound 14 has acceptable PK properties in three safety
species, and with low intrinsic clearance in dog, monkey and
human liver microsomes, is predicted to have good PK

2604 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Lewis et al.
Figure 5. Efficacy profiles of 14, 15 and 16 at human recombinant GABAA R(n)â3γ2receptors. Footnote: Efficacy profiles of 14, 15, and 16 at
human recombinant GABAA receptors comprising â3, γ2, and either an R1, R2, R3, or R5 subunits as measured using whole-cell patch clamp
electrophysiology. A potentiation of the current produced by an EC20-equivalent of GABA represents benzodiazepine site agonism whereas an
attenuation (negative modulation) corresponds to inverse agonism. Values shown are mean ( SEM (n ) 4-6/data point).
Figure 6. Anxiolytic effects (elevated plus maze), and benzodiazepine binding site occupancy of 14, 15, and 16 following oral dosing to rats.
Footnote: Rats were dosed with either vehicle (0.5% methyl cellulose), pyridazine 14, 15, or 16 or, as a positive control, the nonselective full
agonist chlordiazepoxide (CDP, 5 mg/kg ip). Thirty min later animals were placed on the elevated plus maze and the time spent on the open arms
was recorded as an index of anxiety. A significant increase in time on the open arms relative to vehicle treated animals (*, p < 0.05) represents
anxiolytic-like activity. Following completion of the plus-maze trial, a subgroup of rats were taken and occupancy measured using an in vivo
[3H]Ro 15-1788 binding assay. Compounds 14, 15, and 16 all produced dose-dependent occupancy with respective ED50 values of 0.22, 0.09, and
0.52 mg/kg. Values shown are mean ( SEM (n ) 16-18/group, plus-maze experiment, and n ) 5-10/group, occupancy assay).
dog, and human liver hepatocytes, it is also predicted to have
good PK properties in man. In the rat plus-maze it was anxiolytic
at a minimum effective dose of 1 mpk p.o, corresponding to
71% receptor occupancy12 (Figure 6, Occ50 0.52 mpk). The
compound is apparently less potent in this assay than 14,
presumably because its efficacy is now mediated predominantly
through the R3 subtype, which is of relatively low abundance
in the CNS. This result is of particular interest because it offers
further evidence of a role for the R3 subtype in anxiolysis.3d
Other properties and benefits pertaining to this efficacy profile
await exploration.
Chemistry
An efficient five-step synthesis of key chloropyridazine
intermediates 26a,b was developed (Scheme 1), which pro-
ceeded with an overall yield of 40% 26b to 50% 26a based
upon the starting halopyridine. Thus coupling of 2-fluorophe-
nylboronic acid with the requisite halopyridine proceeded

R2/R3 Subtype SelectiVe GABAA Agonists
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2605
Scheme 1. Synthesis of Key Coupling Partner Chloropyridazine 26a
a Reagents and conditions: (i) Pd2(dba)3, t-Bu3P, KF, THF, 55 °C; (ii) CH3CN, H2SO4; (iii) (a) Pd(dppf)Cl2, KOAc, dioxane, 80 °C, (b) HCl; iv) Pd(dppf)Cl2,
Na2CO3, dioxane, 85 °C; (v) POCl3, 75 °C.
Scheme 2. Routes to Selectively Metallated 3,5-difluoropyridinea
a Reagents and conditions: (i) n-BuLi, TMEDA, diethyl ether, -78 °C; (ii) TMSCl; (iii) ZnCl2; (iv) Me3SnCl; (v) n-BuLi, THF, -78 °C.
smoothly under the Fu conditions.14 Addition of a small quantity
of water was found to significantly enhance the rate of this
reaction, presumably by rendering the potassium fluoride more
soluble and therefore facilitating boron-ate formation. Bromi-
nation of 22 with dimethylhydantoin dibromide in the presence
of acid15 afforded bromide 23. Some bis-bromination was
observed if the reaction was run for a longer period, but could
be avoided by careful control of reaction time. Palladium-
catalyzed borylation16 was efficient. It was found that although
the boronate ester could be isolated, a base extraction followed
by acidic hydrolysis afforded clean boronic acid without
recourse to chromatography, a process amenable to large scale
preparation of 24. Suzuki coupling with 5-chloropyridazin-3-
one,6 followed by brief exposure to hot phosphorus oxychloride
afforded 26a or 26b. All these transformations could be
performed at scale to access significant quantities of material.
The remaining 3-aryl appendage could be introduced into 26
under a variety of different coupling protocols depending on
availability of precursors. On a small scale, the pyridines 11 to
19, and also compound 17, were prepared by Stille cross-
coupling of the corresponding stannanes, or by Negishi coupling
of the arylzinc, both species being readily accessible from the
corresponding lithiated pyridine (Scheme 2). The Negishi
protocol is preferable on a larger scale, as the final compound
is easier to render free from heavy metal contamination. The
regioselective deprotonation of functionalized pyridines has been
the subject of some excellent studies.17 3,5-Difluoropyridine 27
may be cleanly deprotonated at the 4-position to afford
thermodynamically more stable anion 28 which may be readily
transformed into the appropriate Stille or Negishi coupling
partner as indicated in Scheme 2. Silylation of 28, followed by
a second deprotonation, allows access to 32 or 33 in a controlled
manner. This procedure could be performed either stepwise with
isolation of compound 30, which is to be preferred on a larger
scale; alternatively 28 could be transformed into 32 in one pot
without isolation of intermediate 30. (The kinetic deprotonation
of 27 at the 2-position under the Queguiner conditions was
difficult to control and led to some scrambling of regiochem-
istry.) The silane was readily cleaved by warming with 5 N
aqueous HCl after coupling 32 to 26. For the corresponding
Negishi reaction of 33, cleavage of the silane occurred under
the coupling conditions, and for a large scale synthesis of 16
was completed by including a KF treatment in the workup. The
Stille coupling could also be performed in the reverse sense,
via in situ preparation of stannane 36 (Scheme 3), and was the
method of choice for compounds 10 and 17.

2606 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Lewis et al.
Scheme 3. Coupling Strategies forAssembly of the Completed
the strongest peaks from the mass spectra are reported.) Spectra
were recorded between 100 and 800 Da. Accurate mass data was
obtained on a Micromass QTof micro. Elemental analyses for
carbon, hydrogen, and nitrogen were performed by Butterworth
Laboratories, 54-56 Waldegrave Road, Teddington, U.K. Analytical
thin-layer chromatography (TLC) was conducted on precoated silica
gel 60 F254 plates (Merck). Visualization of the plates was
accomplished by using UV light and/or iodine and/or aqueous
potassium permanganate solution. Chromatography was conducted
on silica gel 60, 220-440 mesh (Fluka) under low positive pressure.
Solutions were evaporated on a Bu¨chi rotary evaporator at reduced
pressure. Microwave reactions were performed using a Smith
Synthesizer, supplied by Personal Chemistry, Uppsala, Sweden. All
starting materials were obtained from commercial sources and used
as received unless otherwise indicated. Purity for all target
compounds was greater than 98% as assessed by NMR and HPLC.
3,5-Difluoro-2-(2-fluorophenyl)pyridine (22a). 2-Fluoroben-
zeneboronic acid (25.0 g, 178 mmol), 2-bromo-3,5-difluoropyridine
(34.65 g, 178.6 mmol), dry THF (360 mL), potassium fluoride
(34.25 g, 0.589 mol), and water (30 mL) were combined in a 2 L
flask, and the mixture was degassed by vacuum/N2 cycles. Tris-
(dibenzylideneacetone)dipalladium(0) (2.66 g, 2.92 mmol) and tri-
tert-butylphosphine (11.2 mL of a 0.33 M solution in hexanes, 3.7
mmol) were added. The reaction was heated at 55 °C for 24 h under
nitrogen and then allowed to cool to ambient temperature. The
organic phase was decanted and the residue rinsed with ethyl
acetate. The combined organics were evaporated, and the residue
was partitioned between ethyl acetate (400 mL) and water (200
mL). The organic phase was separated, dried (MgSO4), filtered,
and evaporated. The residue crystallized from hot isohexane to give
22a as a white solid (33.64 g, 90%). δH (500 MHz, CDCl3): 7.18
(1 H, t, J ) 9.2 Hz), 7.28 (1 H, dd, J ) 0.9, 7.5 Hz), 7.32 (1 H,
m), 7.45 (1 H, m), 7.56 (1 H, ddd, J ) 1.7, 7.5, 9.2 Hz), 8.47 (1
H, d, J ) 2.3 Hz); m/z (ES+) 210 (MH+).
2-(5-Bromo-2-fluorophenyl)-3,5-difluoropyridine (23a). To
22a (33.6 g, 160.76 mmol) in acetonitrile (330 mL) was added
dibromodimethylhydantoin (1.5 equiv, 68.85 g, 240.5 mmol). The
mixture was cooled to 4 °C, and concentrated sulfuric acid (51.9
mL, 6 equiv, 95.4 g, 0.974 mol) was added slowly with efficient
stirring to maintain the internal temperature below 20 °C. On
completion of addition, the cooling bath was removed and the
mixture stirred a further 10 min at RT. The vessel was then
transferred to a preheated oil bath at 55 °C and the mixture heated
at 55 °C for 2 h. Most of the solvent was then stripped at reduced
pressure, the residue diluted with water, and 4 N aq NaOH (310
mL) added, and the mixture extracted with isohexane. The organic
phase was washed with water, separated, dried (MgSO4), filtered,
and evaporated and the residue chromatographed on silica, eluent
45% CH2Cl2 in isohexane, to give 23a as a white solid (32.34 g,
111.9 mmol, 70%). δH (500 MHz, CDCl3): 7.07 (1 H, t, J ) 9.2
Hz), 7.33 (1 H, ddd, J ) 2.4, 7.7, 8.9 Hz), 7.55 (1 H, ddd, J ) 2.7,
4.7, 8.9 Hz), 7.72 (1 H, dd, J ) 2.6, 6.2 Hz), 8.47 (1 H, d, J ) 2.4
Hz); m/z (ES+) 288/290 (MH+).
Ligandsa
a Reagents and conditions: (i) ArB(OH)2, Pd(Ph3P)4, Na2CO3, dioxane,
(85 °C thermal or 150 °C microwave); (ii) ArSnR3, LiCl, CuI, Pd(Ph3P)4,
dioxane, 100 °C; (iii) Me3SnSnMe3, Pd(Ph3P)4, LiCl, dioxane, 100 °C; (iv)
ArBr, Pd(Ph3P)4, LiCl, dioxane, 100 °C.
Alternatively, with the availability of bromopyridine 21a,18
the extreme rapidity of lithium-halogen exchange at low
temperature may be exploited to give access to 2-lithio species
34, which may be cleanly trapped at low temperature to afford
Negishi and Stille partners 35 or 37. Controlled deprotonation
of 3-fluoropyridine at either 2 or 4-positions, proceeds smoothly
according to literature precedent,17 to give access to the required
stannanes for compounds 11, 12, 18, and 19. Compounds 6, 7,
8, and 9, were similarly assembled from the commercial boronic
acids by Suzuki coupling (Scheme 3).
Conclusions
In summary, we have designed a series of benzodiazepine
site ligands with an array of attractive functional selective
efficacy profiles. In particular, compounds 14, 15, and 16 stand
out as antagonists at the R1 subtype, with good CNS penetration
and receptor occupancy, and excellent oral bioavailability. It
will be interesting to see how these subtype selective compounds
perform, not only with regard to the sedative, myorelaxant,
memory impairment, and anxiolytic properties associated with
current clinically utilized unselective agents, but also whether
they offer advantage in reducing such liabilities as potential to
induce tolerance and dependence, abuse liability, and potentia-
tion with ethanol. These compounds are well suited to examine
the hypothesis that BZ ligands devoid of R1 efficacy will be
effective treatments for anxiety disorders with fewer debilitating
side effects; hopefully this will lead to more effective treatments
for a range of anxiety conditions. Furthermore, a comparison
of 14 with 15 may permit an examination of the role of R5
subtypes in the memory and cognitive impairment liabilities,
while the remarkably selective 16 affords the opportunity of
examining the pharmacology of the R3 subtype in isolation.
Together these compounds are useful tools for advancing
understanding of how subtly GABA fulfils its role as the major
inhibitory neurotransmitter in a host of neuronal pathways in
mammalian brain.
[3-(3,5-Difluoropyridin-2-yl)-4-fluorophenyl]boronic Acid (24a).
To a solution of 23a (31.34 g, 108.4 mmol) and bis(neopentyl-
glycolato)diboron (29.6 g, 131 mmol) in dry dioxane (230 mL)
was added potassium acetate (6.13 g, 230.4 mmol), and the mixture
was degassed for 1 h. Dichloro[1,1′-bis(diphenylphosphino)fer-
rocene]palladium(II) DCM adduct (2.02 g, 2.5 mmol) was added
and the mixture degassed for a further 15 min and then heated at
80 °C for 24 h. The reaction was cooled to ambient temperature, 1
M aq NaOH (1060 mL) was added, and the mixture was stirred at
RT for 1 h in air. Ethyl acetate in isohexane (30%) (250 mL) was
added, the mixture was stirred a further 10 min, and then the phases
were separated. The aqueous layer was washed with 30% ethyl
acetate in isohexane, and the combined organics were washed with
water (200 mL). The combined aqueous layers were filtered through
a fine filter paper and then taken to pH 4 by addition of 10 N aq
HCl (∼108 mL). The mixture was allowed to get warm during the
acidification and was stirred at 40 °C for 0.5 h to complete boronate
hydrolysis. The mixture was then aged at RT for 2 h and the solid
Experimental Section
Experimental Details for Biological Assays. Detailed experi-
mental protocols for the in vitro and in vivo experiments have been
published in ref 3 and references therein.
General Chemical Experimental Details. 1H NMR spectra were
recorded on Bruker AMX500 or DPX400 spectrometers. Chemical
shifts are reported in parts per million (δ) downfield from
tetramethylsilane as internal standard and coupling constants in
hertz. Melting points were determined on a Leica Galen III hot
stage apparatus and are uncorrected. Mass spectra were recorded
on a Waters Micromass ZMD2000 or Micromass ZQ instrument
operating in electrospray (ES) mode, as indicated. (Note that only

R2/R3 Subtype SelectiVe GABAA Agonists
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2607
collected by filtration, washed with water, and then dried in vacuo
at 65 °C to afford 24a as a white solid (26.5 g, 105 mmol, 97%).
δH (500 MHz, DMSO-d6): 7.33 (1 H, dd, J ) 8.5, 10.5 Hz), 7.97
(1 H, m), 8.01 (1 H, dd, J ) 1.5, 7.8 Hz), 8.10 (1 H, m), 8.20 (2
H, s), 8.70 (1 H, d, J ) 2.3 Hz); m/z (ES+) 254 (MH+).
was thoroughly degassed. Dichloro[1,1′-bis(diphenylphosphino)-
ferrocene]palladium(II) DCM adduct (0.416 g, 0.515 mmol) was
added and the mixture degassed and then heated at 80 °C for 18 h
under N2. The reaction was cooled to RT, 1 M aq NaOH (210 mL)
was added, and the mixture was stirred at RT for 0.5 h in air. Diethyl
ether (120 mL) was added, the mixture was stirred a further 10
min, and then the phases were separated. The aqueous layer was
filtered through a fine filter paper and then taken to pH 4 by addition
of 2 N aq HCl (∼117 mL). The mixture was allowed to get warm
during the acidification and was stirred at 45 °C for 0.5 h to
complete boronate hydrolysis. The mixture was then aged at RT
for 2 h, and the solid was collected by filtration, washed with water,
and then dried in vacuo at 60 °C to afford 24b as a cream solid
(5.0 g, 94%). δH (500 MHz, DMSO-d6): 7.32 (1 H, dd, J ) 8.4,
10.5 Hz), 7.58 (1 H, m), 7.87 (1 H, m), 7.97 (1 H, m), 8.05 (1 H,
dd, J ) 1.5, 7.9 Hz), 8.20 (2 H, s), 8.59 (1 H, m); m/z (ES+) 236
(MH+).
5-[3-(3-Fluoropyridin-2-yl)-4-fluorophenyl]-2H-pyridazin-3-
one (25b). To 24b (5.0 g, 21.15 mmol), 5-chloropyridazin-3(2H)-
one6 (3.54 g, 27.08 mmol), and dioxane (88 mL) was added 2 M
aqueous sodium carbonate (41 mL). The mixture was degassed by
vacuum/N2 cycles, and then dichloro[1,1′-bis(diphenylphosphino)-
ferrocene]palladium (II) DCM adduct (1.0 g, 1.23 mmol) was added
and the mixture heated at 85 °C for 20 h under N2. Most of the
dioxane was stripped at reduced pressure, water (150 mL) was
added, and the mixture was stirred at 50 °C for 0.5 h and then
allowed to stand at RT for 2 h. The solid was collected by filtration
washed with water, cold ethanol, and then diethyl ether. The filter
cake was dried in vacuo at 60 °C to give 25b, a white solid (5.078
g, 84%). δH (500 MHz, DMSO-d6): 7.19 (1 H, s), 7.54 (1 H, t, J
) 9.2 Hz), 7.62 (1 H, m), 7.92 (1 H, t, J ) 9.0 Hz), 8.04 (2 H, m),
8.33 (1 H, d, J ) 1.9 Hz), 8.62 (1 H, d, J ) 4.3 Hz), 13.14 (1 H,
s); m/z (ES+) 286 (MH+).
3-Chloro-5-[3-(3-fluoropyridin-2-yl)-4-fluorophenyl]py-
ridazine (26b). To phosphorus oxychloride (100 mL) preheated at
70-80 °C was added 25b (5.078 g, 17.81 mmol) portionwise with
stirring, and the solution was heated for 0.5 h after becoming
homogeneous. Excess phosphorus oxychloride was stripped at
reduced pressure and the residue azeotroped with toluene. The dark
oil was treated with ice (100 g) with external ice cooling. The
mixture was adjusted to pH 6 with saturated NaHCO3 solution and
the resulting solid collected by filtration, washed with water, and
then dried in vacuo at 60 °C (4.66 g, 86%). δH (500 MHz, DMSO-
d6): 7.63 (2 H, m), 7.94 (1 H, m), 8.22 (1 H, m), 8.26 (1 H, dd, J
) 2.4, 6.7 Hz), 8.36 (1 H, d, J ) 1.9 Hz), 8.63 (1 H, m), 9.73 (1
H, d, J ) 1.9 Hz); m/z (ES+) 304/306 (MH+).
Representative Suzuki Coupling. 5-[3-(3,5-Difluoropyridin-
2-yl)-4-fluorophenyl]-3-(2-fluorophenyl)pyridazine (8). To 26a
(0.2 g, 0.62 mmol) and 2-fluorophenylboronic acid (0.131 g, 0.93
mmol) in dioxane (3 mL) was added sodium carbonate solution (2
N, 1 mL) followed by tetrakistriphenylphosphinepalladium(0) (36
mg; 31 µmol), and the mixture was heated at 150 °C in a Smith
microwave reactor for 600 s. The reaction was diluted with dioxane
(20 mL), separated, and evaporated onto silica gel. The crude
product was chromatographed on silica eluting on a gradient with
50% to 100% ethyl acetate in isohexane to give 8 as a white solid
(149 mg, 63% yield). δH (500 MHz,CDCl3): 7.20-7.24 (1 H, m),
7.34-7.41 (3 H, m), 7.47-7.53 (1 H, m), 7.80-7.84 (1 H, m),
7.96 (1 H, dd, J ) 2.3, 6.7 Hz), 8.13 (1 H, t, J ) 2.0 Hz), 8.19-
8.24 (1 H, m), 8.52 (1 H, d, J ) 2.3 Hz), 9.44 (1 H, d, J ) 2.0
Hz); m/z (ES+) 382 (MH+). Acc Mass (C21H11F4N3) requires:
382.0962, found: 382.0950 Da.
5-[3-(3,5-Difluoropyridin-2-yl)-4-fluorophenyl]-2H-pyridazin-
3-one (25a). To 24a (26.5 g, 105 mmol), 5-chloropyridazin-3(2H)-
one6 (16.6 g, 127 mmol), and dioxane (500 mL) was added 2 M
aqueous sodium carbonate (140 mL). The mixture was degassed
by vacuum/N2 cycles, and then dichloro[1,1′-bis(diphenylphosphi-
no)ferrocene]palladium (II) DCM adduct (3.0 g, 3.69 mmol) was
added and the mixture heated at 85 °C for 22 h under N2. Water
(500 mL) was added while still hot and the mixture stirred with
cooling in an ice bath at 4 °C for 0.75 h. The solid was collected
by filtration washed with water until filtrate colorless, air-dried,
and then washed with isohexane. The filter cake was dried in vacuo
at 60 °C to give 25a a white solid (26.6 g, 83%). δH (500 MHz,
DMSO-d6): 7.17 (1 H, s), 7.55 (1 H, t, J ) 9.5 Hz), 8.03 (2 H, m),
8.16 (1 H, t, J ) 8.3 Hz), 8.32 (1 H, s), 8.72 (1 H, s), 13.14 (1 H,
br s).
3-Chloro-5-[3-(3,5-difluoropyridin-2-yl)-4-fluorophenyl]py-
ridazine (26a). To phosphorus oxychloride (300 mL) preheated at
70-80 °C was added 25a (13 g, 45.6 mmol) portionwise with
stirring, and the solution was heated for 0.5 h after becoming
homogeneous. Excess phosphorus oxychloride was stripped at
reduced pressure and the residue azeotroped with toluene. The dark
oil was treated with ice (300 g) with external ice cooling. The
mixture was adjusted to pH 6 with saturated NaHCO3 solution and
extracted with EtOAc (2 × 300 mL). The combined organics were
dried (MgSO4), filtered, and evaporated to give a beige solid, which
was recrystallized from hot EtOAc/isohexane (14.4 g, 98%). δH
(500 MHz, DMSO-d6): 7.62 (1 H, t, J ) 9.2 Hz), 8.2 (3 H, m),
8.35 (1 H, d, J ) 1.9 Hz), 8.74 (1 H, d, J ) 2.1 Hz), 9.72 (1 H, d,
J ) 1.8 Hz); m/z (ES+) 322/324 (MH+).
3-Fluoro-2-(2-fluorophenyl)pyridine (22b). 2-Fluorobenzenebo-
ronic acid (5.32 g, 38 mmol), 2-chloro-3-fluoropyridine (5.0 g, 38
mmol), dry THF (100 mL), potassium fluoride (11.2 g, 0.193 mol),
and water (3.0 mL) were combined, and the mixture was degassed
by vacuum/N2 cycles. Tris(dibenzylideneacetone)dipalladium(0)
(0.776 g, 0.852 mmol) and tri-tert-butylphosphine (4.90 mL of a
0.33 M solution in hexanes, 1.6 mmol) were added. The reaction
was heated at 55 °C for 24 h under nitrogen and then allowed to
cool to ambient temperature. The organic phase was decanted and
residue rinsed with ethyl acetate. The combined organics were
evaporated, and the residue was chromatographed on silica, eluent
CH2Cl2, to give 22b as a yellow oil (6.26 g, 86%). δH (500 MHz,
CDCl3): 7.18 (1 H, t, J ) 9.1 Hz), 7.28 (1 H, t, J ) 7.6 Hz), 7.34
(1 H, m), 7.44 (1 H, m), 7.50 (1 H, t, J ) 8.8 Hz), 7.61 (1 H, t, J
) 7.2 Hz), 8.54 (1 H, s); m/z (ES+) 192 (MH+).
2-(5-Bromo-2-fluorophenyl)-3-fluoropyridine (23b). To 22b
(6.26 g, 32.79 mmol) in acetonitrile (67 mL) was added dibro-
modimethylhydantoin (1.5 equiv, 14.04 g, 49 mmol). The mixture
was cooled to 4 °C, and concentrated sulfuric acid (10.58 mL, 6
equiv, 0.198 mol) was added slowly with efficient stirring to
maintain the internal temperature below 20 °C. On completion of
addition, the cooling bath was removed and the mixture stirred a
further 10 min at RT. The vessel was then transferred to a preheated
oil bath at 55 °C and the mixture heated at 55 °C for 2 h. Most of
the solvent was then stripped at reduced pressure and the residue
neutralized by addition of 4 N aq NaOH (65 mL) with cooling in
an ice water bath. The product was extracted with ethyl acetate,
the organic phase was washed with water, separated, dried (MgSO4),
filtered, and evaporated, and the residue was chromatographed on
silica, eluent 50% CH2Cl2 in isohexane to give 23b as a white solid
(6.12 g, 69%). δH (500 MHz, CDCl3): 7.08 (1 H, t, J ) 9.1 Hz),
7.37 (1 H, m), 7.53 (2 H, m), 7.76 (1 H, dd, J ) 2.5, 6.2 Hz), 8.55
(1 H, dt, J ) 1.4, 4.6 Hz); m/z (ES+) 270/272 (MH+).
5-[3-(3,5-Difluoropyridin-2-yl)-4-fluorophenyl]-3-(4-fluorophe-
nyl)pyridazine (6). Prepared in the same manner as 8 by Suzuki
reaction. δΗ (500 MHz, CDCl3): 7.42 (2 H, t, J ) 9.0 Hz), 7.62 (1
H, dd, J ) 8.6, 0.8 Hz), 8.17-8.22 (1 H, m), 8.27-8.39 (4 H, m),
8.55 (1 H, d, J ) 2.3 Hz), 8.75 (1 H, d, J ) 2.3 Hz), 9.66 (1 H, d,
J ) 2.3 Hz); m/z (ES+) 382 (MH+). Anal. (C21H11F4N3. 0.5(H2O))
C, H, N.
[3-(3-Fluoropyridin-2-yl)-4-fluorophenyl]boronic Acid (24b).
To a solution of 23b (6.12 g, 22.58 mmol), and bis(neopentylgly-
colato)diboron (6.17 g, 27.3 mmol) in dry dioxane (50 mL) was
added potassium acetate (4.70 g, 176.6 mmol), and the mixture
5-[4-Fluoro-3-(3-fluoropyridin-2-yl)phenyl]-3-(2-fluorophenyl)-
pyridazine (7). Prepared in the same manner as 8 via Suzuki

2608 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Lewis et al.
reaction. δΗ (500 MHz, CDCl3): 7.41-7.46 (2 H, m), 7.53-7.66
(3 H, m), 7.91-8.02 (2 H, m), 8.20-8.27 (2 H, m), 8.37 (1 H, t,
J ) 2.0 Hz), 8.63-8.64 (1 H, m), 9.72 (1 H, d, J ) 2.3 Hz); m/z
(ES+) 364 (MH+). Acc Mass (C21H12F3N3) requires: 364.1062,
found: 364.1060 Da.
3-(2,4-Difluorophenyl)-5-[3-(3,5-difluoropyridin-2-yl)-4-fluo-
rophenyl]pyridazine (9). Prepared in the same manner as 8 by
Suzuki reaction. δΗ (400 MHz, CDCl3): 6.96-7.02 (1H, m), 7.08-
7.13 (1H, m), 7.36-7.42 (2H, m), 7.79-7.83 (1H, m), 7.95 (1H,
dd, J ) 2.3, 6.3 Hz), 8.09 (1H, t, J ) 2.2 Hz), 8.23-8.29 (1H, m),
8.52 (1H, d, J ) 2.3 Hz), 9.44 (1H, d, J ) 2.3 Hz); m/z (ES+) 400
(MH+). Anal. (C21H10F5N3. 0.4(H2O)) C, H, N.
d, J ) 2.2 Hz), 8.64 (1 H, br d, J ) 4.6 Hz), 8.76 (1 H, d, J ) 2.1
Hz), 9.80 (1 H, d, J ) 2.2 Hz); m/z (ES+) 383 (MH+). Anal.
(C20H10F4N4. 0.45(H2O)) C, H, N.
5-[3-(3,5-Difluoropyridin-2-yl)-4-fluorophenyl]-3-(2,4,6-tri-
fluorophenyl)pyridazine (17). Prepared by the same method as
10. δΗ (500 MHz, CDCl3) 6.86 (2 H, t, J ) 8.2 Hz), 7.34-7.41 (2
H, m), 7.81 (2 H, t, J ) 6.4 Hz), 7.94 (1 H, dd, J ) 2.2, 6.4 Hz),
8.51 (1 H, d, J ) 2.2 Hz), 9.47-9.57 (1 H, m).; m/z (ES+) 418
(MH+). Anal. (C21H9F6N3. 0.4(H2O)) C, H, N.
5-[4-Fluoro-3-(3-fluoropyridin-2-yl)-phenyl]-3-(3-fluoropyri-
din-2-yl)pyridazine (18). Prepared in the same manner as 14 by
Stille proceedure: δΗ (400 MHz, CDCl3) 7.35-7.48 (3 H, m),
7.55-7.61 (1 H, m), 7.63-7.68 (1 H, m), 7.84-7.89 (1 H, m),
8.05 (1 H, dd, J ) 3, 7 Hz), 8.41 (1 H, d, J ) 2 Hz), 8.58-8.62
(2 H, m), 9.54 (1 H, d, J ) 2 Hz); m/z (ES+) 365 (MH+). Acc
Mass. C20H11F3N4 requires: 365.1014, found: 365.0993 Da.
5-[3-(3,5-Difluoropyridin-2-yl)-4-fluorophenyl]-3-(3-fluoropy-
ridin-2-yl)pyridazine (19). Prepared in the same manner as 14 by
Stille proceedure: δΗ (500 MHz, CDCl3) 7.36-7.41 (2 H, m),
7.46-7.49 (1 H, m), 7.66 (1 H, t, J ) 10 Hz), 7.85-7.89 (1 H,
m), 8.01 (1 H, dd, J ) 2, 7 Hz), 8.41 (1 H, s), 8.52 (1 H, s), 8.61
(1 H, d, J ) 4 Hz), 9.53 (1 H, s); m/z (ES+) 383 (MH+). Anal.
(C20H10F4N4) C, H, N.
Representative Negishi Coupling. 5-[4-Fluoro-3-(3,5-difluo-
ropyridin-2-yl)-phenyl]-3-(3,5-difluoropyridin-2-yl)pyridazine (16).
To a stirred solution of TMEDA (7.04 g; 60.6 mmol) in dry diethyl
ether (120 mL) at -78 °C under a N2 atmosphere was added
n-butyllithium (58.6 mmol; 23.5 mL of 2.5 M in hexanes), and the
mixture was stirred for 15 min. 30 (10.59 g, 56.57 mmol) was then
added dropwise over 10 min and then the reaction aged for 1.5 h.
Zinc chloride (61 mmol, 61 mL of 1 M in diethyl ether) was added
over 15 min and the mixture stirred at -78 °C for 1 h and then
allowed to warm to -10 °C to afford 33. The chloropyridazine
26a (13.00 g, 40.41 mmol) was added as a solid, followed by
dichloro[1,1′-bis(diphenylphosphino)ferrocene]palladium(II) DCM
adduct (1.32 g, 1.62 mmol) and dry THF (150 mL). The mixture
was then heated at reflux for 6 h. The reaction was concentrated to
half volume, DMA (15 mL) was added, followed by KF (5.0 g, 86
mmol), and the reaction was heated at 60 °C for 1 h to complete
desilylation. On cooling, the reaction was treated with water (500
mL) containing EDTA (100 g) and NaOH (35 g), and the mixture
was vigorously stirred for 0.5 h and then extracted with CH2Cl2 (2
× 500 mL). The organic phase was separated, washed with water,
dried (MgSO4) filtered, and evaporated onto silica. Chromatography
on silica, gradient eluent CH2Cl2 to 40% ethyl acetate in CH2Cl2,
gave 16, which was dissolved in hot methanol (700 mL) and then
boiled with activated charcoal (5 g) for 1 h. The mixture was filtered
hot through GF/A filter paper, and the filtrate was concentrated to
100 mL and then allowed to stand at RT. The solid was collected
by filtration and dried in vacuo, to afford 16 as a pale yellow solid
(9.0 g, 56%). δH (500 MHz, DMSO-d6): 7.61-7.65 (1 H, m),
8.16-8.24 (4 H, m), 8.48 (1 H, s), 8.75 (2 H, d, J ) 12.2 Hz),
9.79 (1 H, s).; m/z (ES+) 401 (MH+). Anal. (C20H9F5N4. 0.5(H2O))
C, H, N.
Representative Reverse Stille Coupling. 3-(2,6-Difluorophe-
nyl)-5-[4-fluoro-3-(3-fluoropyridin-2-yl)-phenyl]pyridazine (10).
To a degassed mixture of 26b (0.25 g, 0.82 mmol), hexamethylditin
(0.323 g, 0.98 mmol), and lithium chloride (0.1024 g, 2.46 mmol)
in 1,4-dioxane (12 mL) was added tetrakis(triphenylphosphine)-
palladium(0) (0.048 g, 0.041 mmol), and the mixture was heated
at 100 °C for 3 h. Mass spectroscopy showed formation of stannane
36. 1-Bromo-2,6-difluorobenzene (0.3177 g, 0.187 mL, 1.65 mmol)
was added, followed by tetrakis(triphenylphosphine) palladium(0)
(0.01 g, 0.0082 mmol), and the mixture was heated at 100 °C for
6 h. The reaction was cooled to RT, filtered, and evaporated, and
the crude product was chromatographed on silica, eluent 4%
methanol-CH2Cl2, to give 10 which was recrystallized from ethyl
acetate/isohexane as a white solid (42 mg, 14%). δH (400 MHz,
CDCl3) 7.05-7.12 (2 H, m), 7.32-7.59 (4 H, m), 7.79-7.85 (2
H, m), 8.00 (1 H, dd, J ) 2.5, 6.5 Hz), 8.58-8.60 (1 H, m), 9.51
(1 H, d, J ) 2.3 Hz); m/z (ES+) 382 (MH+). Anal. (C21H11F4N3.
0.5(H2O)) C, H, N.
Representative Stille Coupling. 5-[4-Fluoro-3-(3,5-difluoro-
pyridin-2-yl)-phenyl]-3-(3,5-difluoropyridin-4-yl)pyridazine (14).
To 26a (0.2 g; 0.62 mmol) and 3,5-difluoro-4-trimethylstannylpy-
ridine (0.35 g, 1.26 mmol) in dioxane (3 mL) were added copper-
(I) iodide (6 mg, 0.03 mmol) and lithium chloride (0.026 g, 0.62
mmol) followed by tetrakis(triphenylphosphine)palladium(0) (71
mg, 62 µmol), and the mixture was heated at 100 °C for 18 h. The
reaction was diluted with DCM (20 mL) and the organic layer
separated, dried over MgSO4, filtered, and evaporated. Chroma-
tography on silica eluted with 7% ethyl acetate in isohexane gave
14 as a white solid (115 mg, 46% yield). Crystallized from hot
ethyl acetate. Mp 199-201 °C. δH (500 MHz, DMSO-d6): 7.65
(1 H, m), 8.16 (1 H, ddd, J ) 2.4, 9, 9 Hz), 8.24-8.26 (2 H, m),
8.53 (1 H, br s), 8.74 (1 H, d, J ) 2.2 Hz), 8.81 (2 H, s), 9.87 (1
H, d, J ) 2.3 Hz); m/z (ES+) 401 (MH+). Anal. (C20H9F5N4) C,
H, N.
5-[4-Fluoro-3-(3-fluoropyridin-2-yl)-phenyl]-3-(3-fluoropyri-
din-4-yl)pyridazine (11). Prepared in the same manner as 14 by
Stille procedure: δH (500 MHz, CDCl3) 7.42 (2 H, m), 7.58 (1 H,
t, J ) 8.9 Hz), 7.83 (1 H, m), 8.01 (1 H, dd, J ) 2.4, 6.5 Hz), 8.22
(2 H, m), 8.60 (1 H, m), 8.64 (1 H, d J ) 5 Hz), 8.67 (1H, d, J )
2 Hz), 9.54 (1 H, d, J ) 2.0 Hz); m/z (ES+) 365 (MH+). Acc
Mass (C20H11F3N4) requires: 365.0951, found: 365.0945 Da.
5-[3-(3,5-Difluoropyridin-2-yl)-4-fluorophenyl]-3-(3-fluoropy-
ridin-4-yl)pyridazine (12). Prepared in the same manner as 14 by
Stille procedure: δH (400 MHz, CDCl3) 7.36-7.44 (2 H, m), 7.81-
7.85 (1 H, m), 7.97 (1 H, dd, J ) 2, 7 Hz), 8.21-8.24 (2 H, m),
8.52 (1 H, d, J ) 2 Hz), 8.64-8.68 (2 H, m), 9.53 (1 H, d, J ) 2
Hz); m/z (ES+) 383 (MH+). Acc Mass (C20H10F4N4) requires:
383.0914, found: 383.0960 Da.
3-(3,5-Difluoropyridin-4-yl)-5-[4-fluoro-3-(3-fluoropyridin-2-
yl)phenyl]pyridazine (13). Prepared in the same manner as 14 by
Stille procedure. Crystallized from hot 2-propanol by addition of
water: δH (500 MHz, DMSO-d6) 7.62-7.66 (2 H, m), 7.94 (1 H,
br t, J ) 9.2 Hz), 8.23-8.27 (2 H, m), 8.54 (1 H, br s), 8.63 (1 H,
br d, J ) 4.5 Hz), 8.80 (2 H, s), 9.88 (1 H, d, J ) 2.2 Hz); m/z
(ES+) 383 (MH+). Anal. (C20H10F4N4) C, H, N.
3,5-Difluoro-4-trimethylstannanylpyridine (29). To a solution
of N,N,N′,N′-tetramethylethylenediamine (3.85 g, 5 mL, 33.13
mmol) in dry ether (100 mL) at -78 °C was added n-butyllithium
(10.0 mL, 25 mmol, 2.5 M in hexanes), and the mixture was stirred
at -35 °C for 1 h and then cooled to -78 °C. A solution of 3,5-
difluoropyridine 27 (2.63 g, 22.8 mmol) in dry ether (8 mL) was
added and the mixture stirred for 2 h. Trimethyltin chloride (25
mL, 25 mmol, 1 M solution in hexane) was added and the mixture
stirred at -78 °C for 1 h and allowed to warm to RT. The reaction
was then cooled to -78 °C and quenched by the addition of water
(5 mL). After being warmed to RT, the organic phase was separated,
dried (MgSO4), filtered, and evaporated. The crude product was
purified by dry flash chromatography on silica eluted with 10%
ethyl acetate in isohexane, to give 29 as a yellow oil (1.67 g, 26%).
δΗ (400 MHz, CDCl3) 8.22 (2 H, s), 0.48 (9 H, s).
5-[4-Fluoro-3-(3-fluoropyridin-2-yl)-phenyl]-3-(3,5-difluoro-
pyridin-2-yl)pyridazine (15). Prepared in the same manner as 14
by Stille procedure. δH (500 MHz, DMSO-d6) 7.60-7.66 (2 H,
m), 7.93 (1 H, br t, J ) 9.2 Hz), 8.21-8.25 (3 H, m), 8.48 (1 H,
3,5-Difluoro-4-trimethylsilanylpyridine (30). A solution of
n-butyllithium (250 mmol, 100 mL of 2.5 M in hexanes) was added

R2/R3 Subtype SelectiVe GABAA Agonists
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2609
Containing GABAA Receptors is a Nonsedating Anxiolytic in Rodents
and Primates. J. Pharmacol. Exp. Ther. 2006, 316, 410-422, and
references therein. (c) See also ref 6. (d) Dias, R.; Sheppard, W. F.;
Fradley, R. L.; Garrett, E. M.; Stanley, J. L.; Tye, S. J.; Goodacre,
S.; Lincoln, R. J.; Cook, S. M.; Conley, R.; Hallett, D.; Humphries,
A. C.; Thompson, S. A.; Wafford, K. A.; Street, L. J.; Castro, J. L.;
Whiting, P. J.; Rosahl, T. W.; Atack, J. R.; McKernan, R. M.;
Dawson, G. R.; Reynolds, D. S. Evidence for a significant role of
R3-containing GABAA receptors in mediating the anxiolytic effects
of benzodiazepines. J. Neurosci. 2005, 25(46), 10682-10688.
(4) Unpublished results from these laboratories. This observation is based
upon examination of more than 10 chemically distinct structural
classes. The γ2 subunit (the most abundant γ subtype in CNS), which
is held invariant in these studies, makes significant contributions to
the binding pocket. Furthermore the R subunits are highly homolo-
gous, and the modest binding selectivity achievable for R2/R3 over
R1 may be attributable to a single amino acid residue difference which
is isoleucine in R3 and valine in R1 (P. Wingrove, personal
communication). For published studies on the molecular determinants
of subtype selective binding affinity, see also: Pritchett, D. B.;
Seeburg, P. H. γ-Aminobutyric acid type A receptor point mutation
increases the affinity of compounds for the benzodiazepine site. Proc.
Natl. Acad. Sci. U.S.A. 1991, 88, 1421-1425. Korpi, E. R.; Grunder,
G.; Luddens, H.; Drug interactions at GABAA receptors. Prog.
Neurobiol. 2002, 67, 113-159.
(5) Chambers, M.; Atack, J.; Carling, R.; Collinson, N.; Cook, S.;
Dawson, G.; Ferris, P.; Hobbs, S.; O’Connor, D.; Marshall, G.;
Rycroft, W.; MacLeod, A. An Orally Bioavailable, Functionally
Selective Inverse Agonist at the Benzodiazepine Site of GABAA R5
Receptors with Cognition Enhancing Properties. J. Med. Chem. 2004,
47(24), 5829-5832 and references therein.
(6) Van Niel, M. B.; Wilson, K.; Adkins, C. H.; Atack, J. R.; Castro, J.
L.; Clarke, D. E.; Fletcher, S.; Gerhard, U.; Mackey, M. M.; Malpas,
S.; Maubach, K.; Newman, R.; O’Connor, D.; Pillai, G. V.; Simpson,
P. B.; Thomas, S. R.; MacLeod, A. M. A New Pyridazine Series of
GABAA R5 ligands. J. Med. Chem. 2005, 48, 6004-6011.
via a cannula to a solution of TMEDA (34.57 g, 297.5 mmol) in
diethyl ether (400 mL) at -78 °C, and the solution was stirred for
0.5 h. A solution of 3,5-difluoropyridine (27.39 g, 238 mmol) in
diethyl ether (100 mL) was then added via a cannula, and the
resulting mixture was stirred at -78 °C for 2 h, before adding
chlorotrimethylsilane (33 mL, 262 mmol) The solution stirred for
0.5 h and then allowed to warm to RT over 2 h. The reaction was
quenched into water (500 mL), and the organic phase was separated,
washed with 10% aqueous citric acid, water, dried (MgSO4), and
evaporated. The residue was distilled at reduced pressure to give
30 as a colorless oil (bp 62 °C @ 5 mbar), (30 g, 67%). δΗ (400
MHz, CDCl3) 8.24 (2 H, s), 0.41 (9 H).
3,5-Difluoro-4-trimethylsilanyl-2-trimethylstannanylpyri-
dine (32). To a solution of 3,5-difluoropyridine (3.0 g, 26.01 mmol)
in diethyl ether (100 mL) at -78 °C was added lithium diisopro-
pylamide (13.7 mL, 27.4 mmol, 2 M solution in THF), and the
solution was stirred for 0.5 h. Chlorotrimethylsilane (2.95 g, 3.47
mL, 27.15 mmol) was added and the solution stirred for 0.5 h and
then allowed to warm to RT and stirred for 1 h. The solution was
then cooled to -78 °C, a solution of lithium diisoproylamide (13.7
mL, 27.4 mmol, 2 M solution in THF) was added, and the solution
was stirred for 0.5 h. Trimethyltin chloride (28.7 mL, 28.7 mmol,
1 M in hexanes) was added and the solution stirred for 0.5 h and
then allowed to warm to RT over 18 h. The mixture was diluted
with isohexane (100 mL) and poured into water (200 mL). The
organic phase was washed with water (3 × 200 mL) and then dried
over MgSO4, filtered, and evaporated to give an orange oil.
Purification by distillation at reduced pressure gave a pale yellow
oil (7.4 g, 81%; bp 110 °C 1.3 mmHg). δΗ (400 MHz, CDCl3)
8.42 (1 H, s), 0.39 (9 H, s), 0.38 (9 H, s).
(7) Russell, M. G. N.; Carling, R. W.; Street, L. J.; Hallett, D. J.;
Mezzogori, E.; Atack, J. R.; Cook, S. M.; Bromidge, F. A.; Wafford,
K. A.; Marshall, G. R.; Reynolds, D. S.; Stanley, J.; Lincoln, R.;
Tye, S. J.; Sheppard, W. F. A.; Sohal, B.; Pike, A.; Dominguez, M.;
Castro, J. L. Discovery of Imidazo[1,2-b][1,2,4]triazines as GABAA
R2/3 Subtype Selective Agonists for the Treatment of Anxiety. J.
Med. Chem. 2006, 49(4), 1235-1238. (b) Carling, R. W.; Madin,
A.; Guiblin, A.; Russell, M. G. N.; Moore, K. W.; Mitchinson, A.;
Sohal, B.; Pike, A.; Cook, S. M.; Ragan, I.C.; McKernan, R. M.;
Quirk, K.; Ferris, P.; Marshall, G. R.; Thompson, S. A.; Wafford,
K. A.; Dawson, G. R.; Atack, J. R.; Harrison, T.; Castro, J. L.; Street,
L. J.; 7-(1,1-Dimethylethyl)-6-(2-ethyl-2H-1,2,4-triazol-3-ylmethoxy)-
3-(2-fluorophenyl)-1,2,4-triazolo[4,3-b]pyridazine: A Functionally
Selective GABAA R2/R3-Subtype Selective Agonist which Exhibits
Potent Anxiolytic Activity but is Not Sedating in Animal Models. J.
Med. Chem 2005, 48, 7089-7092. (c) Goodacre, S. G.; Street, L. J.;
Hallett, D. J.; Crawforth, J. M.; Kelly, S.; Owens, A. P.; Blackaby,
W. P.; Lewis, R. T.; Stanley, J.; Smith, A. J.; Ferris, P.; Sohal, B.;
Cook, S. M.; Pike, A.; Teall, M.; Wafford, K. A.; Marshall, G. R.;
Castro, J. L.; Atack, J. R.; Imidazo[1,2-a]pyrimidines as Functionally
Selective and Orally Bioavailable GABAA R2/R3 Binding Site
Agonists for the Treatment of Anxiety Disorders. J. Med. Chem.
2006, 49(1), 35-38. (d) Jennings, A. S. R.; Lewis, R. T.; Russell,
M. G. N.; Hallett, D. J.; Crawforth, J.; Street, L. J.; Castro, J. L.;
Atack, J. R.; Cook, S. M.; Lincoln, R.; Stanley, J.; Smith, A. J.;
Reynolds, D. S.; Sohal, B.; Pike, A.; Marshall, G. R.; Wafford, K.
A. Imidazo[1,2-b][1,2,4]triazines as R2/R3 subtype selective GABAA
agonists for the treatment of anxiety. Bioorg. Med. Chem. Lett. 2006,
16(6), 1477-1480.
(8) Unpublished results from these laboratories. For example, the
hydroxydimethyl moiety of 2, a key feature of the imidazotriazine
series, was not well tolerated as a 3-substituent in the pyridazine
series: Ar ) (CH3)2COH, R) H Ki R3 226 nM.
(9) The R2 binding affinity was not routinely measured. It closely
parallels the R3 affinity for those compounds studied; 16: Ki R2
10.91 nM; Ki R3 8.5 nM and 17: Ki R2 4.6 nM; Ki R3 5.0 nM.
(10) A combination of HPLC-MS-MS. analysis of the material recovered
after incubation with rat hepatocytes.
(11) The protocol for these assays has been published elsewhere. See ref
3 and references therein for experimental protocols for rat in vivo
occupancy, plus-maze, and rotarod assays in the presence or absence
of coadministered ethanol, and primate CER.
(12) It is important to note that occupancy of all BZ receptor subtypes is
measured11 by the test compound’s ability to displace tritiated Ro
15-1788, but that the efficacy derives only from occupancy of those
native receptors containing R2 or R3 for this compound.
Acknowledgment. We thank Sarah Kelly for the synthesis
of compound 30, and Drs. Angus M MacLeod, Leslie J.
Street, Jose´ L. Castro, and Ruth M. Mckernan for useful
discussions.
Supporting Information Available: Table S1 contains analyti-
cal or accurate mass data for compounds 6-19. This material is
available free of charge via the Internet at http://pubs.acs.org.
References
(1) Simon, J.; Wakimoto, H.; Fujita, N.; Lalande, M.; Barnard, E. A.
Analysis of the set of GABAA Receptor Genes in the Human
Genome. J Biol. Chem. 2004, 279(40), 41422-41435. (b) Atack, J.
R. The benzodiazepine binding site of GABAA receptors as a target
for the development of novel anxiolytics. Expert Opin. InVestig.
Drugs 2005, 14(5), 601-618. See also: Whiting, P. J. The GABA-A
receptor gene family: New opportunities for drug development. Curr.
Opin. Drug DiscoVery DeV. 2003, 6(5), 648-657, and references
therein. (c) The γ3 subtype is of low abundance in mammalian brain.
In this work only the γ2 subtype combination has been studied in
vitro. Knoflach, F.; Rhyner, Th.; Villa, M.; Kellenberger, S.; Drescher,
U.; Malherbe, P.; Sigel, E.; Mohler, H. The γ3-subunit of the
GABAA-receptor confers sensitivity to benzodiazepine receptor
ligands. FEBS Lett. 1991, 293(1-2), 191-194; see also: Baer, K.;
Essrich, C.; Benson, J.; Benke, D.; Bluethmann, H.; Fritschy, J.-M.;
Luscher, B. Postsynaptic clustering of γ-aminobutyric acid type A
receptors by the γ3 subunit in vivo. Proc. Natl. Acad. Sci. U.S.A.
1999, 96(22), 12860-12865.
(2) Low, K.; Crestani, F.; Keist, R.; Benke, D.; Brunig, I.; Benson, J.
A.; Fritschy, J. M.; Rulicke, T.; Bluethmann, H.; Mohler, H.;
Rudolph, U. Molecular and neuronal substrate for the selective
attenuation of anxiety. Science 2000, 290(5489), 131-134, and
references therein.
(3) 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.; 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. Nature Neurosci. 2000, 3(6),
587-592, and references therein. (b) Atack, J. R.; Wafford, K. A.;
Tye, S. J.; Cook, S. M.; Sohal, B.; Pike, A.; Sur, C.; Melillo, D.;
Bristow, L.; Bromidge, F.; Ragan, C. I.; Kerby, J.; Street, L. J.;
Carling, R. W.; Castro, J. L.; Whiting, P. J.; Dawson, G. R.;
McKernan, R. M. TP A023, An Agonist Selective for R2- and R3-

2610 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Lewis et al.
(13) Compound 14 possesses no significant off target activities at 1000
times its effective anxiolytic plasma concentration, as assessed in a
‘MDS Pharma screen where it was assayed against a panel of 170
enzyme, G-protein-coupled receptor, ion channel, and amino acid
transporter receptors. MDS Pharma Services, Taiwan Ltd., Pharma-
cology Laboratories, 158 Li-Teh Road, Peitou, Taipei, Taiwan 112,
R. O. C.
(14) Littke, A. F.; Fu, G. C. Palladium-Catalysed Coupling Reactions of
Aryl Chlorides. Angew. Chem., Int. Ed. 2002, 41, 4176-4211, and
references therein.
(15) Eguchi, H.; Kawaguchi, H.; Yoshinaga, S.; Nishida, A.; Nishiguchi,
T.; Fujisaka, S. Halogenation using N-halogeno compounds. II. Acid-
catalyzed bromination of aromatic compounds with 1,3-dibromo-
5,5-dimethylhydantoin. Bull. Chem. Soc. Jpn. 1994, 67(7), 1918-
1921. (b) Olah, G. A.; Wang, Q.; Sandford, G.; Surya, P. G. K.
Synthetic methods and reactions 181. Iodination of deactivated
aromatics with N-iodosuccinimide in trifluoromethanesulfonic acid.
J. Org. Chem. 1993, 58(11), 3194-3195.
(16) Ishiyama, T.; Itoh, Y.; Kitano, T.; Miyaura, N. Synthesis of
arylboronates via the palladium(0)-catalyzed cross-coupling reaction
of tetra(alkoxo)diborons with aryl triflates. Tetrahedron Lett. 1997,
38(19), 3447-3450.
(17) Queguiner, G.; Marsais, F. Review on the Metallation of π-deficient
Heteroaromatic Compounds. Tetrahedron 1983, 39(12), 2009-2021.
(b) Gros, P.; Choppin, S.; Mathieu, J.; Fort, Y. Lithiation of
2-heterosubstituted Pyridines J. Org. Chem. 2002, 67, 234-237. (c)
See also Chambers, R. D.; Hall, C. W.; Hutchinson, J.; Millar, R.
W. Polyhalogenated heterocyclic compounds. Part 42. Fluorinated
nitrogen heterocycles with unusual substitution patterns. J. Chem.
Soc., Perkin Trans. 1 1998, (10), 1705-1713.
(18) 2-Bromo-3,5-difluoropyridine 21a may be prepared by the action of
AlBr3 and HBr on 2,3,5-trifluoropyridine by adaptation of the method
of Chambers et al. in ref 17c for the preparation of 2,4,6-tribromo-
3,5-difluoropyridine from pentafluoropyridine.
JM051144X