4-Fluorosulfonylpiperidines: Selective 5-HT2A ligands for the treatment of insomnia

Article abstract of DOI:10.1016/j.bmcl.2005.05.104

Incorporation of fluorine at the 4-position of an existing series of sulfonyl piperidine 5-HT_2A antagonists gave compounds with increased selectivity over the IKr potassium channel. This work led to the identification of 3b, a compound that gave no increase in QT_c in the anesthetized dog up to plasma levels as high as 148 μM. Furthermore, 3b has been shown to increase slow-wave sleep bout duration and to decrease the number of awakenings in rats, indicating the potential utility of 5-HT_2A antagonists in the treatment of insomnia.

Full text of DOI:10.1016/j.bmcl.2005.05.104

Bioorganic & Medicinal Chemistry Letters 15 (2005) 3665–3669
4-Fluorosulfonylpiperidines: Selective 5-HT2A ligands
for the treatment of insomnia
L. Rebecca Fish,a Myra T. Gilligan,a Alexander C. Humphries,a,* Magnus Ivarsson,a
Tammy Ladduwahetty,a,* Kevin J. Merchant,a Desmond OÕConnor,a Smita Patel,a
Elisabeth Philipps,a Hugo M. Vargas,b Peter H. Hutsona and Angus M. MacLeoda
aThe Neuroscience Research Centre, Merck Sharp and Dohme, Terlings Park, Harlow, Essex CM20 2QR, UK
bDepartment of Pharmacological Sciences, Safety Assessment, West Point, PA 19486, USA
Received 31 March 2005; revised 9 May 2005; accepted 26 May 2005
Available online 1 July 2005
Abstract—Incorporation of fluorine at the 4-position of an existing series of sulfonyl piperidine 5-HT2A antagonists gave com-
pounds with increased selectivity over the IKr potassium channel. This work led to the identification of 3b, a compound that gave
no increase in QTc in the anesthetized dog up to plasma levels as high as 148 lM. Furthermore, 3b has been shown to increase slow-
wave sleep bout duration and to decrease the number of awakenings in rats, indicating the potential utility of 5-HT2A antagonists in
the treatment of insomnia.
ꢀ 2005 Elsevier Ltd. All rights reserved.
Chronic insomnia affects about 10% of the global popu-
lation, with between 30% and 50% of people affected by
poor sleep (of any duration or severity) at any time.1
Insomnia is associated with significant impairments in
physical and mental well-being, it can have a major neg-
ative impact on quality of life, and leads to a higher inci-
dence of psychiatric disturbances including depression
and anxiety.2 The leading prescription drugs for the
treatment of insomnia, including Ambien,ꢁ are sedative
hypnotics that act at the benzodiazepine site on the
GABA-A ion channel. Unresolved issues with these
drugs include abuse potential3 and the risk of fractures
in elderly patients, due to an increased incidence of
falls.4 Hence, there is a need for new therapeutics that
will provide improvements in quality of sleep without
the adverse effects of existing treatments.
both humans and rats.6 Furthermore, in clinical trials,
several agents that antagonize 5-HT2A receptor activity,
including ritanserin,7 nefazodone,8 and trazodone,9 have
been shown to reduce sleep disturbance in patients who
suffer from a primary disorder other than sleep distur-
bance, for example, schizophrenia, depression, and anx-
iety. While these agents have potent activity at several
G-protein-coupled receptors (GPCRs), preliminary
(unpublished) information indicates that the selective
5-HT2A antagonist M100907 (Fig. 1) increases SWS
Serotonin (5-hydroxytryptamine, 5-HT) was one of the
first neurotransmitters associated with the regulation
of the sleep-wake cycle.5 It is known that ritanserin, a
5-HT2A antagonist, increases the amount of slow-wave
sleep (SWS), one of the components of deep sleep, in
*
Corresponding authors. Tel.: +44 1279 440239; fax: +44 1279 440390
(A.C.H.); Tel.: +44 1279 440420; fax: +44 1279 440390 (T.L.); e-mail
addresses: tamara_ladduwahetty@merck.com; alexander_humphries
@merck.com
Figure 1. Selected 5-HT2A antagonists.
0960-894X/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.05.104

3666
L. R. Fish et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3665–3669
and total sleep time, as well as reducing time spent
awake after sleep onset in elderly subjects.10
Hence, deprotonation of sulfones 1a and 1b with
NaHMDS, followed by reaction with N-fluoro com-
pound 4, successfully afforded the desired 4-fluor-
osulfonylpiperidines 3a and 3b, respectively, albeit in
low yield (Scheme 1).
Previous work from our laboratories11 identified a series
of sulfonylpiperidines of general structure 1, with high
affinity for the 5-HT2A receptor and good selectivity
over other GPCR targets, including 5-HT2C and D2.
However, in common with many related structures,12
compounds of this class have moderate affinity for the
hERG (IKr) potassium channel. It is well known that
IKr blockade can lead to prolongation of the QTc inter-
val in vivo, which in turn can lead to the episodes of
Torsade de Pointe that may result in death. Compound
1a gave rise to a significant increase in QTc interval after
i.v. infusion to an anesthetized ferret.11 However, we
found that the acetophenone 2 has reduced affinity for
the IKr channel and subsequently demonstrated that 2
did not significantly increase the QTc interval in ferret.
Some improvement was made by changing both the
base, to n-butyllithium, and the electrophilic fluorine
source, to N-fluorobenzenesulfonamide 5. However, giv-
en that the yields of fluorinated products 3c–e were still
low, we switched to using the N-Boc intermediate 6
(Scheme 2). In this case, deprotonation with n-butyllithi-
um, followed by N-fluorobenzenesulfonamide quench,
afforded the a-fluorosulfone 7 in good yield.15 The syn-
thesis of the core was then completed by removal of the
Boc-protecting group and alkylation using standard
protocols.
To further vary the 4-aryl substituent, we again used the
intrinsic reactivity of the system, this time by nucleophil-
One possible reason that acetophenone 2 has reduced
affinity for the IKr channel is its lower basicity com-
pared to the parent phenethyl compound 1a (Table 1).
However, a drawback with acetophenones such as 2 is
that compounds of this class chemically degrade in solu-
tion, particularly in polar solvents such as methanol and
DMSO. Hence, we sought other means of reducing the
basicity of the sulfonylpiperidines. Fluorine is common-
ly incorporated into biologically active compounds since
it gives rise to minimal steric impact, whilst its electro-
negativity influences pKa.13 Moreover, it is known that
the presence of a fluorine at the 4-position of a piperi-
dine ring can substantially reduce its pKa.14 We there-
fore set out to synthesize a-fluorosulfones, of general
structure 3, as an alternative approach to reduce the
pKa of the piperidine ring.
To gain rapid access to this class of compounds, we
sought to harness the intrinsic reactivity of the
alkylsulfonyl moiety and incorporate a fluorine atom
by an electrophilic quench onto an alkylsulfonyl anion,
thus allowing us to use available advanced fragments.
Scheme 1. Reagents and conditions: (a) 1a or 1b, NaHMDS, THF,
ꢀ78 ꢂC to room temperature, then ꢀ78 ꢂC, 4, to room temperature,
25% (3a), 18% (3b); (b) 1c–e, nBuLi, THF, ꢀ78 ꢂC, then 5, to room
temperature, 25–35% (3c–e).
Table 1. Binding affinities, in vivo occupancy data, and pKa values
h5-HT2A Ki/nMa
r5-HT2A Ki/nMb
h5-HT2C Ki/nMa
hD2 Ki/nMa
hIKr Ki/nMa
In vivo occupancyc
pKa
d
Compound
M100907e
0.31
0.33
0.42
2.4
0.10
0.18
13
25
1300
>1500
>1500
>1500
>1000
313
1100
1000
710
1ae
1be
2e
7.3
7.2
6.3
92
0.51
0.63
130
49
6800
2446
5561
1525
398
3a
3b
3c
3d
3e
3f
0.68
0.39
0.30
0.32
0.64
2.7
176
49
98 5%
6.3
791
266
14
398
1491
844
>4000
>4000
>1000
863
397
1.33
0.30
1.6
2702
9971
>9000
>9000
80 4%
49 8%f
37 8%
62 7%
6.4
3g
3h
3i
1.0
2.0
>2000
>2000
>2000
0.66
a h5-HT2A, h5-HT2C, hD2, and hIKr were determined as described in Ref. 11 (n = 2).
b r5-HT2A data were determined as described in Ref. 16 (n = 2).
c Occupancy is quoted at 30 min after dosing at 10 mg/kg po as a mean SEM (%), measured using the protocol described in Ref. 17.
d pKa values were determined using a Sirius SGA apparatus.
e Data from Ref. 11.
f At 3 mg/kg.

L. R. Fish et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3665–3669
3667
ence in ion channel activity compared to 1a. However,
it became clear that IKr channel affinity is more sensitive
to incorporation of the fluorine atom on the piperidine
ring in instances when there is also substitution on the
phenylsulfone moiety. Thus, the 4-fluorophenyl com-
pound 3b shows significantly less IKr affinity compared
with its parent, 1b. Comparison of 3b with the 3- and 2-
fluoro-substituted compounds (3c,d) demonstrates that
there is also a degree of positional dependence in activity
at the IKr channel. Finally, the significance of the nature
of the 4-substituent is further illustrated by compounds
3e–i, with chlorine (3e) leading to increased IKr channel
affinity, whereas reduced affinity was obtained with the
azetidine (3g) and polar heterocycles, such as pyrazole
(3h) and triazole (3i).
To prioritize the compounds for further profiling, in
vivo CNS activity of the selective compounds, 3b and
3f–i, was measured using a competitive 5-HT2A receptor
occupancy assay, after oral dosing in rat (Table 1).17 On
the basis of superior pharmacodynamic activity, 3b was
selected for further evaluation.
Scheme 2. Reagents and conditions: (a) nBuLi, THF, ꢀ78 ꢂC, then 5,
to room temperature, 76%; (b) 11 N HCl (aq), EtOH, 80 ꢂC, >95%;
(c) 2,4-difluorophenylethyl bromide, K2CO3, MeCN, 85 ꢂC, 84%;
(d) NaCN, DMSO, 80 ꢂC, 72%; (e) azetidine, DMSO, 150 ꢂC, 10 min
(Smithꢁ Personal Synthesiser), >95%; (f) pyrazole or 1,2,3-triazole,
Cs2CO3, DMSO, 150 ꢂC, 10 min (Smithꢁ Personal Synthesiser),
70–80%.
Compound 3b has good bioavailability (F 51%) in
rat and, despite having a high i.v. clearance (Clp
69 mL/min, Vd 11 L/kg), the compound has a moder-
ate half-life (t1/2 3.8 h) in this species.18
We assessed the cardiovascular impact of 3b using rising
dose studies in anesthetized dogs (n = 3; Fig. 2A).19
Hence, at cumulative doses of 3, 10, and 30 mg/kg (each
dose infused i.v. over 30 min), there was no observed
QTc prolongation or significant change in blood pres-
sure and heart rate. The maximum plasma levels reached
in this study were 148 lM. In contrast, M100907 gave
significant changes in QTc interval at plasma levels as
low as 0.49 lM in dogs (n = 2) following intravenous
infusion at increasing dose levels (1, 3, and 10 mg/kg;
Fig. 2B). The completely benign profile of 3b in the
dog cardiovascular study is intriguing, considering that
the difference in IKr binding between 3b and M100907
is only ca. fivefold. Clearly, factors such as plasma pro-
ic displacement of the fluorine atom of the 4-fluorophe-
nylsulfone moiety (Scheme 2). Hence, 3b was reacted
with NaCN, azoles (as Cs salts), or cyclic amines, in
DMSO, either thermally or by fixed temperature micro-
wave irradiation, to give the desired products 3f–i in
high yield.
These novel compounds were screened for activity at the
5-HT2A receptor, together with related GPCRs (5-HT2C
and D2) and the hIKr channel (Table 1). Initial results
with 3a were disappointing in that there is little differ-
Figure 2. Cardiovascular effects of 3b (A) and M100907 (B) in anesthetized dogs.

3668
L. R. Fish et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3665–3669
Figure 3. Effect of 3b on sleep patterns in rat.
Eds.; W.B. Saunders Company: Philadelphia, 2000, pp
615–623.
2. Ohayon, M. M. Sleep Med. Rev. 2002, 6, 97.
3. Hajak, G.; Muller, W. E.; Wittchen, H. U.; Pittrow, D.;
Kirch, W. Addiction 2003, 98, 1371.
tein binding and the functional IKr activities of these
two compounds may have further influenced the out-
come in this assay.
To assess the effects of 3b on the sleep patterns in rats, we
developed an EEG telemetry assay in this species.20
Hence, rats were dosed orally with either 3b at 10 mg/kg
or vehicle (0.5% methyl cellulose) during the first 15 min
of the light phase in a cross-over design. EEG and EMG
recordings were then collected for the whole of the light
phase (12 h) and at every 12 s, the data were scored as
either awake, SWS, or rapid eye movement (REM) sleep.
This assay demonstrated that 3b significantly decreases
the number of awakenings (Fig. 3) and increases SWS
bout duration. In contrast, 3b showed no significant effect
on REM bout duration.
4. Wang, P. S.; Bohn, R. L.; Glynn, R. J.; Mogun, H.;
Avorn, J. J. Am. Geriatr. Soc. 2001, 49, 1685.
5. Jones, B. E. In Principles and Practise of Sleep Medicine;
Kryger, M. H., Roth, T., Dement, W. C., Eds.; W.B.
Saunders Company: Philadelphia, 2000, pp 134–154.
6. (a) Kantor, S.; Jakus, R.; Bodizs, R.; Halasz, P.; Bagdy,
G. Brain Res. 2002, 943, 105–111; (b) Viola, A. U.;
Brandenberger, G.; Toussaint, M.; Bouhours, P.; Macher,
J. P.; Luthringer, R. Clin. Neurophysiol. 2002, 113, 429.
7. Da Roza-Davis, J. M.; Sharpley, A. L.; Cowen, P. J.
Psychopharmacol. 1992, 108, 387.
8. Rush, A. J.; Armitage, R.; Gillin, J. C.; Yonkers, K. A.;
Winokur, A.; Moldofsky, H.; Vogel, G. W.; Kaplita, S. B.;
Fleming, J. B.; Montplaisir, J.; Erman, M. K.; Albala, B.
J.; McQuade, R. D. Biol. Psychiatry 1998, 44, 3.
9. Schwartz, T.; Nihalani, N.; Virk, S.; Jindal, S.; Costello,
A.; Muldoon, R.; Azhar, N.; Hussein, J.; Tirmazi, S. Int.
J. Psychiatr. Nurs. Res. 2004, 10, 1146.
10. Markham, R. J. Aventis 2003 Press and Analyst Confer-
ence, London, UK, February, 2004.
11. Fletcher, S. R.; Burkamp, F.; Blurton, P.; Cheng, S. K. F.;
Clarkson, R.; OÕConnor, D.; Spinks, D.; Tudge, M.; van
Niel, M. B.; Patel, S.; Chapman, K.; Marwood, R.;
Shepheard, S.; Bentley, G.; Cook, G. P.; Bristow, L. J.;
Castro, J. L.; Hutson, P. H.; MacLeod, A. M. J. Med.
Chem. 2002, 45, 492.
In conclusion, we have described our work on a novel
series of high affinity, selective 4-fluorosulfonylpiperi-
dine 5-HT2A antagonists and examined the effect of fluo-
rine incorporation on affinity for the IKr channel. We
found that modulating the pKa of the piperidine nitro-
gen alone does not significantly affect IKr affinity in vi-
tro. However, incorporation of the fluorine on the
piperidine ring together with substitution at the 4-posi-
tion of the phenyl ring, as in 3b, does significantly reduce
this ion channel activity. In contrast to the flagship com-
pound in this class, M100907, 3b shows no cardiovascu-
lar effects in dogs. Furthermore, 3b was assessed in rat
sleep studies and shown to significantly increase SWS
bout duration and decrease the number of awakenings,
thus supporting the use of 5-HT2A antagonists as a po-
tential treatment for insomnia.
12. Cavalli, A.; Poluzzi, E.; De Ponti, F.; Recanatini, M. J.
Med. Chem. 2002, 45, 3844.
13. Welch, J. T.; Eswarakrishnan, S. In Fluorine in bioorganic
chemistry; John Wiley & Sons: New York, 1991, pp 1–6.
14. Van Niel, M. B.; Collins, I.; Beer, M. S.; Broughton, H.
B.; Cheng, S. K.; Goodacre, S. C.; Heald, A.; Locker, K.
L.; MacLeod, A. M.; Morrison, D.; Moyes, C. R.;
OÕConnor, D.; Pike, A.; Rowley, M.; Russsell, M. G.
N.; Sohal, B.; Stanton, J. A.; Thomas, S.; Verrier, H.;
Watt, A.; Castro, J. L. J. Med. Chem. 1999, 42, 2087.
15. In a typical example, n-butyllithium (1.6 M in isohexanes;
13 mL, 20.8 mmol) was added dropwise to a stirred
solution of N-BOC 4-(4-fluorophenylsulfonyl)piperidine
(6 g, 17.4 mmol) in THF (70 mL) at ꢀ78 ꢂC. After 1 h, a
solution of N-fluorobis(phenylsulfonyl)amine (6.04 g, 19
mmol) in THF (17 mL) was added dropwise and the
mixture was brought to ambient temperature, stirred for
1 h, quenched by the addition of water (1 mL), and then
partitioned between saturated aqueous NH4Cl (80; mL)
and EtOAc (80 mL). The organic fraction was washed
with brine (30 mL), dried over Na2SO4, filtered, and
Acknowledgments
We thank Paul Scott-Stevens for in vivo pharmacokinetic
work, Richard Woltmann and Kevin Fitzgerald for
carrying out the dog CV studies, and Steve Thomas
and Helen Atherton for carrying out pKa determinations.
References and notes
1. Zorick, F. J.; Walsh, J. K. In Principles and Practise of
Sleep Medicine; Kryger, M. H., Roth, T., Dement, W. C.,

L. R. Fish et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3665–3669
3669
concentrated in vacuo. Purification by column chroma-
tography (silica, 20% EtOAc/isohexane) afforded N-BOC
4-fluoro-4-(4-fluorophenylsulfonyl)piperidine as an off-
white solid (4.8 g, 76%): dH (360 MHz, CDCl3) 1.46 (9H,
s), 1.80–1.95 (2H, m), 2.05–2.30 (2H, m), 2.90–3.00 (2H,
m), 4.12–4.22 (2H, m), 7.28 (2H, t, J = 8.5 Hz), 7.93–7.96
(2H, m). m/z (ES+) 261 [(M-BOC)H]+.
ꢁ1.67 mm, prefilled with heparinised saline; 20 U/mL)
were implanted into both the right and the left femoral
vein and one femoral artery. One venous catheter was used
for drug administration and the contralateral for blood
sampling. Arterial blood pressure was recorded from the
catheterized femoral artery with a disposable-type pres-
sure transducer. Heart rate was derived from pulsatile
arterial pressure or the electrocardiogram (ECG). ECG
activity from Leads II and V was captured simultaneously
from subcutaneous limb and precordial needle electrodes,
respectively. After surgery, the animals were allowed to
stabilize for a minimum of 45 min prior to collecting
pretreatment control data. A computerized data acquisi-
tion system (CARecorder; DISS, Inc.) was used to collect
and analyze raw input signals. Mean arterial pressure
(mmHg), heart rate (beats/min), and cardiac intervals (QT,
QTc, PR, and QRS; s) were recorded continuously
throughout the study and averaged at 1 min intervals.
Baseline values (pretreatment) were taken at time 0 min
(control) and compared to values measured at 5 min
16. r5-HT2A affinities were measured by the same protocol as
described for h5-HT2A receptor binding in Ref. 11, except
HEK cells stably expressing rat 5-HT2A receptor were
used.
17. The assay was carried out as follows: drug compound
(10 mg/kg; 1 mg/mL in 0.5% methylcellulose), vehicle, or 2
(to determine non-specific binding, 10 mg/kg; 1 mg/mL in
0.5% methylcellulose) was dosed to male CD1 rats at a
volume of 1 mL/kg, p.o. After 5 min (for test compounds/
vehicle) or 1.5 h (for 2), 6 lCi [3H] radioligand was
administered by tail vein injection. After a further
30 min, rats were culled by stun and decapitation and
the brain was removed. The frontal cortex was dissected,
weighed, and homogenized in 50 volumes of ice-cold
buffer (50 mM Tris, pH 7.4, HCl) using a Teflon homog-
enizer (10–15 strokes, 1000 rpm). Aliquots of homogenate
(1 mL) were filtered over Whatmann GF/B filters, which
were presoaked in 0.1% PEI. The filters were washed with
10 mL ice-cold buffer and filters placed in vials. Scintilla-
tion fluid was added and radioactivity was counted using a
beta liquid scintillation counter. Occupancy at the 5-HT2A
receptor was calculated using the following equation:
Total binding (TB) = Vehicle counts ꢀ Non-specific bind-
ing counts; Compound binding (CB) = Compound
counts ꢀ Non-specific binding counts; Occupancy
(%) = 100 ꢀ [100 · (CB/TB)].
intervals during each 30 min i.v. infusion of vehicle or test
p
agent. BazettÕs formula [QTC = QT/ (R ꢀ R)] was used to
correct the QT interval. Vehicle or test agent were infused
in a rising dose paradigm and each dose was delivered in a
volume of 10 mL over 30 min using a programmable
syringe infusion pump. Cumulative doses of 1, 3, and
10 mg/kg of each agent were obtained by infusing doses of
1, 2, and 7 mg/kg i.v., respectively, and cumulative doses
of 3, 10, and 30 mg/kg by infusing doses of 3, 7, and
20 mg/kg, respectively. Whole blood samples (2 mL) were
taken at the 10, 20, and 29 min time points of each
infusion for determination of plasma drug levels.
20. Animals were dosed with either drug or vehicle at light
onset in a cross-over design. Recording of EEG and EMG
signals began immediately after dosing, and EEG and
EMG data were continuously sampled with Data Sciences
International hardware and software. The vigilance states
were automatically scored with software developed in
house. The automatic analysis is based on the changes of
EEG and EMG activity that defines the different vigilance
states in rat—WAKE is characterized by desynchronized,
low-amplitude EEG activity with increased EMG activity;
NREM displays synchronized, high-amplitude EEG activ-
ity accompanied by low-muscle activity; REM shows
desynchronized low-amplitude EEG activity with low or
absent EMG activity (Timo-Iaria, C.; Negrao, N.; Sch-
midek, W. R.; Hoshino, K.; Lobato de Menezes, C. E.;
Leme da Rocha, T. Physiol. Behav., 1970, 5, 1057–1061).
After rejection of epochs with artifacts, every 12 s epoch
was assigned to one of the three vigilance states: WAKE,
NREM, or REM.
18. Dosed 1 mg/kg i.v. to rat as a 1 mg/mL solution in 3:1
PEG300:water (n = 1). Dosed 5 mg/kg p.o. to rat (n = 1) as
a 1 mg/mL suspension in 0.5% methylcellulose.
19. Conditioned, male or female mixed-breed or beagle dogs
(5–14 kg) were anesthetized with sodium pentobarbital
(35 mg/kg i.v.) following approximately 18 h of food
deprivation. Anesthesia was maintained during the study
with
a constant infusion of sodium pentobarbital
(6 mg/kg/10 mL/h). Normal body temperature (ꢁ37 ꢂC)
was maintained with circulating-water warming pad and
infrared lamp monitored through a controller unit with an
esophageal thermistor. Each dog had an endotracheal
tube inserted and was ventilated with ambient room using
a large animal ventilator (tidal volume, ꢁ140–160 mL;
breaths/min, ꢁ15–19; %inspiration/expiration ratio,
40:60). Animals were vagotomized bilaterally, then the
femoral arteries and veins isolated through incisions in the
right and left inguinal regions. Polyethylene catheters (ID