Bicyclic heteroarylpiperazines as selective brain penetrant 5-HT 6 receptor antagonists

Article abstract of DOI:10.1016/j.bmcl.2005.06.107

Starting from the potent and selective but poorly brain penetrant 5-HT ₆ receptor antagonist SB-271046, a successful strategy for improving brain penetration was adopted involving conformational constraint with concomitant reduction in hydrogen bond count. This provided a series of bicyclic heteroarylpiperazines with high 5-HT₆ receptor affinity. 5-Chloroindole 699929 combined high 5-HT₆ receptor affinity with excellent brain penetration and also had good oral bioavailability in both rat and dog.

Full text of DOI:10.1016/j.bmcl.2005.06.107

Bioorganic & Medicinal Chemistry Letters 15 (2005) 4867–4871
Bicyclic heteroarylpiperazines as selective brain penetrant
5-HT6 receptor antagonists
Mahmood Ahmed,a Michael A. Briggs,a Steven M. Bromidge,b Tania Buck,a
Lorraine Campbell,c Nigel J. Deeks,a Ashley Garner,c Laurie Gordon,c
Dieter W. Hamprecht,b Vicky Holland,c Christopher N. Johnson,a,*
Andrew D. Medhurst,a Darren J. Mitchell,a Stephen F. Moss,a Jenifer Powles,c
Jon T. Seal,a Tania O. Stean,a Geoffrey Stemp,a Mervyn Thompson,a
Brenda Trail,a Neil Upton,a Kim Winbornc and David R. Wittya
aNeurology and GI Centre of Excellence for Drug Discovery, GlaxoSmithKline, New Frontiers Science Park,
Third Avenue, Harlow, Essex CM19 5AW, UK
bPsychiatry Centre of Excellence for Drug Discovery, GlaxoSmithKline, New Frontiers Science Park,
Third Avenue, Harlow, Essex CM19 5AW, UK
cDiscovery Research, GlaxoSmithKline, New Frontiers Science Park, Third Avenue, Harlow, Essex CM19 5AW, UK
Received 4 April 2005; revised 4 April 2005; accepted 14 June 2005
Available online 6 September 2005
Abstract—Starting from the potent and selective but poorly brain penetrant 5-HT6 receptor antagonist SB-271046, a successful
strategy for improving brain penetration was adopted involving conformational constraint with concomitant reduction in hydrogen
bond count. This provided a series of bicyclic heteroarylpiperazines with high 5-HT6 receptor affinity. 5-Chloroindole 699929
combined high 5-HT6 receptor affinity with excellent brain penetration and also had good oral bioavailability in both rat and dog.
Ó 2005 Elsevier Ltd. All rights reserved.
5-HT6 receptor mRNA is almost exclusively expressed
within the brain, and many CNS drugs have high affinity
for the 5-HT6 receptor. Hence, there has been consider-
able interest in elucidating the role of the 5-HT6 receptor
in CNS disorders through the development of selective
agents. Several classes of agonists and antagonists have
been disclosed in recent years,1 including the highly
potent and selective 5-HT6 receptor antagonist
SB-271046 1.2 Following the discovery that 5-HT6
receptor antagonists have a beneficial effect on memory
consolidation in animal models of cognitive enhance-
ment such as the Morris water maze and Novel Object
Recognition paradigms,3,4 a possible role has been sug-
gested for 5-HT6 receptor antagonists in the treatment
of learning and memory disorders.5 The proposed
involvement of 5-HT6 receptor antagonism in memory
consolidation is further supported by the effect of 1 on
neuronal cell adhesion molecule (NCAM) polysialyla-
tion in rat brain.6 NCAM polysialylation is a process
that contributes to learning-associated dendritic remod-
elling in the adult central nervous system. Acute and
chronic administration of SB-271046 has been shown
to increase the frequency of NCAM polysialylated neu-
rons activated in the entorhinal and perirhinal cortex, in
response to water maze spatial learning.7 These data im-
ply that 5-HT6 receptor antagonists may have a benefi-
cial effect on synaptic plasticity, a property which may
underpin their broad spectrum of activity in preclinical
models of cognition enhancement.
Although 1 had good oral bioavailability and showed
activity in centrally mediated animal models of cogni-
tion, its brain–blood ratio in rat was low (0.05:1) and
the compound was shown to be a substrate for the
P-glycoprotein (P-gp) efflux pump. This combination
of low brain penetration in rat and P-gp substrate liabil-
ity was considered to give a high risk for poor brain
penetration in human. Increased brain penetration and
reduced P-gp liability were, therefore, key requirements
for a back-up drug candidate.
*
Corresponding
Christopher_N_Johnson@gsk.com
author.
Tel.:
+44
1279
627735; e-mail:
0960-894X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.06.107

4868
M. Ahmed et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4867–4871
Table 2. 5-HT6 receptor binding affinity (pKi): alternative core
templatesa
To identify compounds with improved brain–blood
ratio, a strategy was adopted involving conformational
constraint of the SB-271046 template with resultant
removal of an acidic NH group as in 2. In addition, a
rat ex vivo binding assay8 was used as a key early stage
filter to identify compounds with inherently good oral
exposure and high brain–blood ratio. This article de-
scribes the optimisation of a series of bicyclic heteroaryl-
piperazines related to 2.
Compoundb
Template
R
pKi
26
3-Cl
9.5
27
28
H
H
7.6
8.1
Novel compounds 5–22, 26–29, 32 and 35 (Tables 1
and 2) were prepared as described in Schemes 1–4.
Compounds 5–10 were prepared in two steps from
Boc-piperazinylindole 3a by reaction with the appro-
priate arylsulfonyl chloride under basic conditions,
followed by deprotection (Scheme 1). Introduction of
substituents into the indole moiety could be carried
out on either 3a or on the arylsulfonated derivative
4a.9 Thus, chlorination of 3a with N-chlorosuccini-
mide in 1,4-dioxane at room temperature gave a
mixture of 3-, 5- and 7-chloro derivatives 3b–d, from
which the major component 3c was separated by chro-
29
32
35
3-Cl
H
8.6
8.9
9.3
Table 1. 5-HT6 receptor binding affinity (pKi): 4-piperazinyl indolesa
H
a All pKi values represent the mean of at least three experiments, each
within 0.3 of the mean. In this assay, SB-271046 1 gave pKi 8.9.
b All new compounds gave satisfactory analytical and/or mass spectral
data.16
Compoundb
Ar
R
H
pKi
5
8.5
6
7
Ph
H
9.5
9.4
9.6
8.7
9.0
9.5
8.6
9.5
9.2
9.4
8.3
8.8
9.0
8.2
9.3
8.1
7.6
C6H4(2-Cl)
C6H4(3-Cl)
C6H4(4-Cl)
2-Pyridyl
C6H4(3-Cl)
C6H4(3-Cl)
C6H4(3-Cl)
C6H4(3-Cl)
C6H4(3-Cl)
C6H4(3-Cl)
Ph
H
8
H
9
H
10
11
12
13
14
15
16
17
18
19
20
21
22
H
3-Cl
5-Cl
7-Cl
3-Me
2-Me
5-CN
5-Cl
5-Cl
5-Cl
5-Cl
5-Cl
5-Cl
Scheme 1. Reagents and conditions: (i) 1. KOBut, THF, rt, 2 h;
2. ArSO2Cl, rt, 15 h. (ii) CF3CO2H, CH2Cl2, rt, 4 h.
C6H4(2-Cl)
C6H4(4-Cl)
2-Pyridyl
1-Naphthyl
2-Naphthyl
matography. The mixture of 3-chloro and 7-chloro
isomers was treated with the appropriate arylsulfonyl
chloride as described above and the resultant mixture
of 4b and 4d was separated by chromatography. A
3-methyl substituent was introduced by treatment of
a All pKi values represent the mean of at least three experiments, each
within 0.3 of the mean. In this assay, SB-271046 1 gave pKi 8.9.
b All new compounds gave satisfactory analytical and/or mass spectral
data.16

M. Ahmed et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4867–4871
4869
hexane/tetrahydrofuran at À78 °C followed by
quenching with iodomethane. 5-Cyano 4g was ob-
tained by treatment of 4a with chlorosulfonyl isocya-
nate in acetonitrile at 0 °C followed by addition of
triethylamine. Conversion of indole-substituted deriva-
tives 3b–e, 4f and 4g to final products 11–22 was car-
ried out using methods analogous to those used for
the synthesis of compounds 5–10.9
To prepare 7-piperazinyl indole isomer 26, routes were
initially investigated involving palladium-catalysed
reaction of a suitable 7-bromoindole derivative with
N-Boc-piperazine.10 This approach proved unsuccessful
whether or not the indole NH was protected, for exam-
ple, with a Boc group. A far more successful strategy in-
volved use of the Bartoli indole synthesis,11 involving
reaction of an aromatic nitro compound with an alkenyl
Grignard reagent (Scheme 2). We reasoned that a piper-
azine group, suitably protected on the distal nitrogen,
should serve as an appropriate large moiety to force
the nitro group out of plane, reportedly a prerequisite
for a successful outcome.12 Nitro derivative 23a, ob-
tained by reaction of 2-fluoronitrobenzene with N-Boc-
piperazine, was treated with a solution of 3.2 molar
equivalents of vinylmagnesium bromide in tetrahydrofu-
ran at À40 °C to afford 7-piperazinyl indole 24a in 11%
yield.13 The 3-arylsulfonyl group was introduced by
reaction with the appropriate aryl disulfide under basic
conditions, followed by oxidation with 3-chloroperben-
zoic acid to give 25a. Acid-mediated deprotection gave
desired target 26. Corresponding N-methyl indole 29
was obtained from 25a by alkylation with dimethyl
sulfate under basic conditions prior to removal of the
Boc-protecting group. The success of this approach
prompted us to investigate the Bartoli synthesis for the
preparation of azaindoles 27 and 28. Nitropyridines
23b and 23c (obtained by reaction of the appropriate flu-
oro compound with N-Boc-piperazine) were successfully
treated with vinylmagnesium bromide as described
above to afford desired azaindoles 24b and 24c in 17%
and 51% yield, respectively, and these were converted
into final products 27 and 28 as shown in Scheme 2.
Scheme 2. Reagents and conditions: (i) CH2@CHMgBr, THF,
À40 °C, 0.7 h. (ii) 1. NaH, DMF, rt; 2. [(R)C6H4S]2, rt, 2 h. (iii)
3-chloroperbenzoic acid, CH2Cl2, rt, 18 h. (iv) HCl, 1,4-dioxane, rt,
2 h. (v) Me2SO4, K2CO3, acetone, EtOH, reflux, 18 h.
Scheme 3. Reagents and conditions: (i) ClCH2SO2Ph, KOBut, DMF,
rt, 2 h. (ii) concd H2SO4, AcOH, 60 °C, 10 h. (iii) H2, Pd/C, DMF,
EtOH, rt, 18 h. (iv) MeN(CH2CH2Cl)2 Æ HCl, Na2CO3, BuOH, 120 °C,
48 h. (v) 1. MeCH(Cl)OCOCl, ClCH2CH2Cl, 70 °C, 1 h; 2. MeOH,
reflux, 1 h.
Benzofuran 32 was prepared as shown in Scheme 3.
Nitroaldehyde 30 was treated with the anion
derived from chloromethylphenyl sulfone to give
dihydrobenzofuran 31 in moderate yield, the reaction
proceeding via rearrangement of an initially formed
epoxide.14 Dehydration of 31 followed by catalytic
hydrogenation gave the 7-aminobenzofuran and
subsequent piperazine ring formation with bis-(2-chlo-
roethyl)methylamine hydrochloride gave, after demeth-
ylation with 1-chloroethylchloroformate, the desired
NH-piperazine 32. The low overall yield of this sequence
was primarily due to low recovery from the piperazine
ring forming stage, though this step was not optimised.
Benzothiophene 35 was prepared as shown in Scheme 4.
7-Bromo-3-hydroxybenzothiophene 33 was condensed
with thiophenol under acid conditions and then oxidised
with monomagnesium peroxyphthalate to give sulfone
34. Displacement of the 7-bromo substituent by N-
Boc-piperazine was then carried out using palladium
catalysis and a final deprotection gave the desired target
Scheme 4. Reagents and conditions: (i) PhSH, p-TsOH, C6H6, 80 °C,
16 h. (ii) monomagnesium peroxyphthalate, CH2Cl2, rt, 16 h. (iii)
N-Boc-piperazine, Pd(OAc)2, BINAP, Cs2CO3, 1,4-dioxane, 100 °C,
16 h. (iv) HCl, 1,4-dioxane, rt, 2 h.
3a with sodium hydride in tetrahydrofuran followed
by chlorotriisopropylsilane to give the 1-triisopropylsi-
lyl indole, treatment of the latter with Eschenmoserꢀs
salt in acetonitrile and subsequent catalytic hydroge-
nation over palladium on charcoal to give 3e, after
chromatography. A 2-methyl substituent was intro-
duced by treatment of 4a with tert-butyllithium in

4870
M. Ahmed et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4867–4871
35. All final compounds were purified by chromatogra-
phy and isolated as hydrochloride salts.
Compound 12 (699929) was selected for further profil-
ing. The native tissue 5-HT6 affinity of 12 in rat striatum
(pKi 7.9) and human caudate (pKi 8.1) was lower in
comparison with affinity at the human recombinant
receptor (pKi 8.6). The compound was shown to be
>100-fold selective across a range of 50 receptors and
ion channels. In a cAMP accumulation assay using hu-
man cloned 5-HT6 receptors expressed in HeLa cells, 12
was an antagonist with apparent pKb 7.2. The pharma-
cokinetic profile of 12 was further investigated and the
compound was found to have 49% oral bioavailability
in rat and 90% in dog, together with a long plasma half
life (6 and 17 h, respectively). The P-gp substrate liabil-
ity for 12 was assessed using MDCK cell monolayers
expressing human multi-drug resistance (MDR) pro-
tein.17 In this assay, 12 was shown to have reduced
P-gp liability in comparison to SB-271046 1.
All compounds were evaluated by displacement of
[3H]LSD from human cloned 5-HT6 receptors expressed
in HeLa cells with data expressed as pKi values (Tables 1
and 2).
Conformational constraint as described above led to a
series of sulfonylated piperazinyl indoles 5–10 with high
affinity for the 5-HT6 receptor (Table 1). In all, over 130
Ar groups were investigated of which 96 had pKi > 8,
with 3-chlorophenyl 8 giving the highest affinity (pKi
9.6). A selection of these was profiled in a steady-state
CNS penetration assay in the rat15 and, encouragingly,
brain–blood ratios between 2:1 and 4:1 were obtained.
However, all compounds evaluated suffered from high
in vivo blood clearance due to poor metabolic stability.
Substitution of the indolyl moiety was investigated for
Ar = 3-chlorophenyl to block potential sites of metabo-
lism. A range of substitution patterns was tolerated with
respect to 5-HT6 receptor affinity. Introduction of elec-
tron-withdrawing groups at C-5 reduced 5-HT6 affinity
to some extent (12, 16) but improved selectivity against
other 5-HT receptors to >100-fold. In a rat steady-state
CNS penetration assay, 5-chloro derivative 12 had sig-
nificantly reduced blood clearance (CLb 44 mL/min/
kg) compared to the C-5 unsubstituted compound and,
importantly, maintained a 3:1 brain–blood ratio. Varia-
tion of the arylsulfonyl group was investigated in the 5-
chloroindole series and a selection is shown in Table 1.
Conformational constraint of the SB-271046 template
with concomitant reduction of hydrogen bond count
proved to be a successful strategy for obtaining com-
pounds with improved brain penetration. The resulting
series of bicyclic heteroarylpiperazines had high affinity
for the 5-HT6 receptor and compounds such as 12 and
29 had improved potency in an ex vivo binding assay
compared to SB-271046 1. From this series, 5-chloro-
4-piperazinyl indole 12 (699929) was identified, with a
brain–blood ratio of 3:1 in rat and good oral bioavail-
ability in two species, and was selected for further
progression.
For Ar = substituted phenyl, an ortho substituent, for
example, 2-Cl 18, was most beneficial for 5-HT6 affinity,
while a para substituent reduced affinity, for example, 4-
Cl 19. 2-Pyridylsulfonyl 20 gave the highest affinity in
this series and also achieved >1000-fold selectivity over
other 5-HT receptors. However, this compound gave
poor exposure following oral dosing in the rat. Bicyclic
arylsulfonyl groups were poorly tolerated in this series,
for example, 1- and 2-naphthyl (21 and 22, respectively).
An isomeric 7-piperazinyl indole series together with
alternative 6,5-bicyclic replacements for the indole unit
were investigated (Table 2).
Acknowledgment
The authors thank Kerry-Ann Garwood for assistance
in the preparation of the manuscript.
References and notes
1. Glennon, R. A. J. Med. Chem. 2003, 46, 2795.
2. Bromidge, S. M.; Brown, A. M.; Clarke, S. E.; Dodgson,
K.; Gager, T.; Grassam, H. L.; Jeffrey, P. M.; Joiner, G.
F.; King, F. D.; Middlemiss, D. N.; Moss, S. F.; Newman,
H.; Riley, G. J.; Routledge, C.; Wyman, P. J. Med. Chem.
1999, 42, 202.
3. Rogers, D. C.; Hagan, J. J. Psychopharmacology 2001,
158, 114.
4. King, M. V.; Sleight, A. J.; Woolley, M. L.; Topham, I.
A.; Marsden, C. A.; Fone, K. C. F. Neuropharmacology
2004, 47, 195.
Isomeric 7-piperazinyl 3-arylsulfonyl indole 26 retained
good affinity (pKi 9.5), which was reduced 10-fold by
indole N-methylation 29. The corresponding 6- and
4-azaindoles had markedly lower 5-HT6 affinity (27
and 28, respectively). Replacement of indole by benzofu-
ran 32 and benzothiophene 35 was, by contrast, well
tolerated but these compounds had low in vitro meta-
bolic stability as determined using rat and human liver
microsomes. In the rat steady-state CNS penetration as-
say, 7-piperazinyl indoles 26 and 29 had moderate blood
clearance (41 and 34 mL/min/kg, respectively). Howev-
er, the brain–blood ratio of N-methyl 29 was significant-
ly higher than that of NH derivative 26 (2.6:1 vs 0.7:1),
reflecting the increased number of H-bond donors pres-
ent in 26. In the rat ex vivo binding assay,8 12 and 29
had ED50 3 and 5 mg/kg po, respectively. In comparison
SB-271046 was significantly less potent, with ED50
11 mg/kg.
5. Reavill, C.; Rogers, D. C. Curr. Opin. Investig. Drugs
2001, 2, 104.
6. Regan, C. M.; Foley, A. G.; Murphy, K. J.; Hirst, W.
D.; Gallagher, H. C.; Hagan, J. J.; Upton, N.; Walsh, F.
S. Abstracts of Papers, Society for Neuroscience 33rd
Annual Meeting, New Orleans, USA, Nov 8–12, 2003;
835.21.
7. Johnson, C. N.; Roland, A.; Upton, N. Drug Discovery
Today 2004, 1, 13.
8. Rats received vehicle or 5-HT6 receptor antagonist, dosed
orally, 4 h pretreatment. Animals were then killed and
striatum was removed. Tissue was homogenised and a
binding assay was performed with the selective 5-HT6

M. Ahmed et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4867–4871
4871
receptor antagonist radioligand
[
125I]-SB-258585. An
14. Makosza, M.; Ziobrowski, T.; Kwast, A. Tetrahedron
1998, 54, 6811.
ED50 value was determined from the dose–response curve.
9. Bromidge, S. M. WO2002041889; Chem. Abstr. 2002, 136,
401783.
10. Hartwig, J. F. In Negishi, E., Ed.; Handbook of Organ-
opalladium Chemistry; Wiley: Hoboken, NJ, 2002; Vol. 1,
pp 1051–1096.
15. CNS penetration was determined at steady state in rat.
Compounds were dissolved in 2% (v/v) DMSO in 5% (w/
v) dextrose aq and administered at a constant infusion
rate over 12 h at a target dose rate of 0.3 mg free base/
kg/h. Blood samples were removed during the latter part
of the infusion to confirm steady-state blood concentra-
tions. Blood and brain samples were analysed by LC/MS/
MS.
11. Bartoli, G.; Palmieri, G.; Bosco, M.; Dalpozzo, R.
Tetrahedron Lett. 1989, 30, 2129.
12. Bosco, M.; Dalpozzo, R.; Bartoli, G.; Palmieri, G.;
Petrini, M. J. Chem. Soc. Perkin Trans. 2 1991, 657.
13. Procedure for Bartoli reaction: A solution of 23a (33.1 g,
0.108 mol) in dry THF (750 mL) under argon was cooled
to À45 °C and then treated with a solution of vinylmag-
nesium bromide (1 M, 345 mL) over 0.3 h. The mixture
was stirred at À40 °C for 0.7 h and then quenched with
saturated aqueous ammonium chloride (30 mL). The
resulting mixture was extracted with dichloromethane
(500 mL), then the organic phase was washed with water
(500 mL), dried (MgSO4), and evaporated in vacuo to give
an oil. Chromatography on silica gel with 3–8% acetone–
toluene gradient elution gave 24a (3.55 g, 11%) dH
(CDCl3) 1.50 (9H, s), 3.06 (4H, t, J = 5.0 Hz), 3.65 (4H,
t, J = 4.8 Hz), 6.56 (1H, m), 6.83 (1H, d, J = 7.5 Hz), 7.07
(1H, t, J = 7.7 Hz), 7.20 (1H, m), 7.39 (1H, d, J = 7.9 Hz),
8.28 (1H, br s).
16. 1H NMR spectra were recorded at 250 MHz in CDCl3 as
solvent. Compound 12, mp 240–241 °C (dec.) (hydrochlo-
ride salt); dH 3.4 (4H, br s), 3.6 (4H, br s), 4.8 (2H, br s),
6.9 (1H, d, J = 3.8 Hz), 7.23 (1H, d, J = 8.8 Hz), 7.52 (1H,
t, J = 4.0 Hz), 7.65–7.76 (2H, m), 7.82 (1H, d, J = 8.8 Hz),
7.90 (1H, d, J = 8.4 Hz), 7.96 (1H, t, J = 1.8 Hz).
17. Confluent MDR-MDCK monolayers on collagen-coated,
microporous polycarbonate filters were used in a test
medium of Hankꢀs balanced salt solution containing
10 mM HEPES and 15 mM glucose at pH 7.4. Apical-
to-basal and basal-to-apical permeability (PappA-B and
PappB-A, respectively) of test compounds (3 lM) was
assessed in the presence and absence of the P-gp inhibitor
GF120918A (2 lM). P-gp substrates were defined as
having PappB-A/PappA-B ratio >2, which reverted to unity
in the presence of GF120918A.