Cyclic guanidines as dual 5-HT5A/5-HT7 receptor ligands: Optimising brain penetration

Article abstract of DOI:10.1016/j.bmcl.2007.10.078

The optimisation of molecular properties within a series of 2-amino dihydroquinazoline 5-HT_5A/5-HT₇ receptor ligands resulted in a significantly improved brain-to-plasma ratio, enhancing the pharmacological utility of these compounds

Full text of DOI:10.1016/j.bmcl.2007.10.078

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 262–266
Cyclic guanidines as dual 5-HT5A/5-HT7 receptor
ligands: Optimising brain penetration
Jens-Uwe Peters,a,* Thomas Lubbers,a Alexander Alanine,a Sabine Kolczewski,a
¨
Francesca Blascob, and Lucinda Stewardc
aDiscovery Chemistry, Pharma Division, F. Hoffmann-La Roche Ltd, CH-4070 Basel, Switzerland
bSafety Technical Services, F. Hoffmann-La Roche Ltd, CH-4070 Basel, Switzerland
cPreclinical CNS Research, F. Hoffmann-La Roche Ltd, CH-4070 Basel, Switzerland
Received 4 September 2007; revised 22 October 2007; accepted 25 October 2007
Available online 30 October 2007
Abstract—The optimisation of molecular properties within a series of 2-amino dihydroquinazoline 5-HT5A/5-HT7 receptor ligands
resulted in a significantly improved brain-to-plasma ratio, enhancing the pharmacological utility of these compounds. By modulat-
ing the lipophilicity and pKa, a 20-fold increase in brain-to-plasma ratio could be achieved, leading to micromolar brain concentra-
tions after oral administration. The enantiomers of one representative of this series of improved compounds were separated, and the
configuration of the eutomer was determined by X-ray crystallography.
Ó 2007 Elsevier Ltd. All rights reserved.
5-Hydroxytryptamine (5-HT) receptor (R) subtypes
have proven to be tractable targets for drug discovery,
leading to well-established medicines for the treatment,
for example, of anxiety (buspirone, 5-HT1AR partial
agonist),1a migraine (‘triptans’, 5-HT1DR agonists),1a
emesis (ondansetron, 5-HT3R antagonist),1b or GI tract
motility disorders (cisapride, 5-HT4R agonist).1c The
more recently discovered 5-HT6 and 5-HT7Rs have been
the subject of extensive preclinical research.1d,e In con-
trast, the 5-HT5A subtype has been less well studied, de-
spite its expression in the cerebral cortex, hippocampus,
amygdale, thalamus, hypothalamus, basal ganglia, brain
stem, and cerebellum,2 and an association of the 5-HT5A
receptor polymorphism with schizophrenia,3 pointing to
a potential role in the modulation of mood disorders.4a,b
Recently disclosed, selective 5-HT5A receptor antago-
nists demonstrate antipsychotic-like potential in animal
models.5 These findings have generated an increased
interest in the 5-HT5AR as a potential drug target.
type 5-HT5AR antagonists with high affinity and high
functional activity, however, with no selectivity over
the 5-HT7R (see preceding publication). Compound 1a
penetrates into the brain after oral administration, with
a brain-to-plasma ratio of ꢀ0.2, which is modest for a
compound targeting the CNS. Selected molecular prop-
erties of 1a are summarised in Table 1, further pharma-
cokinetic properties are given in Table 5.
With the goal of developing a pharmacological tool for
behavioural studies, we aimed to further improve the
brain penetration. We speculated that this might be pos-
sible by adjusting the physicochemical properties (e.g.,
pKa, logD) of these compounds more towards the opti-
mal range for CNS drugs (pKa < 10,6 logD ꢀ 27). Espe-
cially lipophilicity is known to be an important
parameter governing brain penetration,8 so we first at-
tempted to improve the brain-to-plasma ratio of our
compounds by increasing lipophilicity. We decided to
explore the modulation of lipophilicity by attaching a
variety of substituents onto the exocyclic nitrogen.
Compound 2 (Table 2) was chosen as a model com-
pound due to its straightforward synthetic access. Com-
pounds 2b–2i were prepared to explore the tolerance of
substituents. A significant loss of 5-HT5AR affinity was
observed for all compounds except 2d. Also, the stabili-
ties in rat liver microsome incubations were very
low (except 2i), in sharp contrast to the unsubstituted
In a search for novel 5-HT5AR ligands, we identified 1a
as the best representative of a novel class of guanidine-
Keywords: 5-HT; GPCR; Guanidine; DMPK; Molecular properties.
*
Corresponding author. Tel.: +41 61 6882636; e-mail: jens-uwe.
peters@roche.com
Present address: Novartis Institute for BioMedical Research, CH-
4002 Basel, Switzerland.
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.10.078

J.-U. Peters et al. / Bioorg. Med. Chem. Lett. 18 (2008) 262–266
263
Table 1. Properties of the dual 5-HT5A/5-HT7R ligand, 1a
this trend was less pronounced with human microsomes.
We found that the attachment of small, lipophilic and
electron-withdrawing fluoro-ethyl substituents was a
superior way to improve logD, by increasing not only
the intrinsic lipophilicity, logP, but also by lowering
the high basicity (pKa) of the guanidine core (2o, 2p).
The reduction of pKa leads to a smaller pH-dependent
term in Eq. 111 and thus contributes to the increase of
the measured lipophilicity, logD.
h5-HT5A pKi = 8.3
35S]GTPcS h5-HT5A pA2 = 8.5
Cl
[
NH
NH2
h5-HT7 pKi = 7.8
logD7.4 = 0
pKa = 9.9
MAB (rat)a = 65%
MAB (human)b = 100%
N
1a
a Maximal achievable bioavailability estimated from incubations with
rat liver microsomes.
b Human liver microsomes.
log D ¼ log P ꢁ logð1 þ 10ðpHꢁpK ÞÞ
ð1Þ
a
Whereas the monofluoro-ethyl (2n), and the trifluoro
ethyl, substituent (2p) led to a ꢀ10-fold drop in 5-
HT5AR affinity, the bis-fluoro-ethyl substituent (2o)
gave an only slight decrease in affinity, a retained micro-
somal stability, a reduced basicity, and an improved
lipophilicity.
compounds 1a and 2a. Additionally, the measured lipo-
philicity did not exceed logD = 1.3, despite the lipophilic
nature of substituents. The latter fact might be explained
by the high basicity of the compounds (pKa = 9.9–10.7),
leading to a very low fraction of unionised compound
under the near-neutral conditions (pH = 7.4) of the
logD measurement.9
The introduction of the difluoroethyl side chain into
compounds with optimised aromatic substituents (see
preceding publication) led to highly active compounds
1b, 3–7 (Table 4) with improved physicochemical prop-
erties. The lipophilicity was especially increased in com-
pounds carrying a lipophilic, electron-withdrawing
aromatic chloro substituent, which further lowers the
basicity of the guanidine core (1b, 3, 6, 7), whereas com-
pounds with 8-alkoxy substituents retain high basicities
In a more systematic investigation, an attachment of
lipophilic alkyl chains to the 2-amino function did not
lead to higher logD values, as would be expected from
ClogP calculations10 predicting a lipophilicity increase
of ꢀ0.5log units per methylene unit (2j–m, Table 3).
Moreover, the stabilities of the compounds in rat liver
microsomes deteriorated with increasing chain lengths;
Table 2. Exploration of substituents R
NH
R
N
N
H
2
Compound
R=
H
logD7.4
MABa (rat, %)
MABb (human, %)
5-HT5A Ki (nM)
2ac
2b
ꢁ1.4
75
87
55
38
241
O
O
0.1
1.5
2c
2d
ꢁ0.3
2.1
1.4
41
23
176
37
O
1.0
O
O
2e
2f
2g
2h
2i
ꢁ0.6
ꢁ0.1
ꢁ0.8
1.3
3.7
5.3
5.4
1.3
46
100
71
163
279
96
N
O
O
O
66
5.1
83
124
122
ꢁ0.6
a Maximal achievable bioavailability estimated from incubations with rat liver microsomes.
b Human liver microsomes.
c As HI salt.

264
J.-U. Peters et al. / Bioorg. Med. Chem. Lett. 18 (2008) 262–266
Table 3. The low lipophilicity can be improved by substituents which lower the guanidine basicity
NH
R
N
N
H
2
Compound
R=
logD7.4
pKa
MABa (rat, %)
MABb (human, %)
5-HT5A Ki (nM)
2ac
2j
H
Me
ꢁ1.4
ꢁ0.6
ꢁ0.9
ꢁ0.8
ꢁ0.3
10.5
10.6
10.8
10.7
10.7
75
77
68
58
17
87
100
93
38
103
196
153
189
397
87
2kc
2l
Et
Pr
82
74
2m
2n
2o
2p
Bu
CH2–CH2F
CH2–CHF2
CH2–CF3
Not determined
ꢁ0.2
9.7
9.2
78
62
100
93
0.5
488
a Maximal achievable bioavailability estimated from incubations with rat liver microsomes.
b Human liver microsomes.
c As HI salt.
Table 4. Combination of aromatic substitution patterns leading to high affinity, with a difluoroethyl substituent leading to improved logD, pKa
Cl
Cl
NH
Cl
F
NH
NH
N
N
H
F
F
F
O
N
N
N
N
H
H
F
F
1b
3
4
Cl
O
NH
Cl
NH
NH
F
F
N
N
F
N
N
H
N
N
H
F
O
H
F
Cl
F
6
7
5
Compound
logD7.4
pKa
Ki (nM)
5-HT5A
5-HT1A
5-HT1D
5-HT2C
5-HT7
1ba
3a,b
(S)-3a,c
(R)-3
4a
1.5
2.7
2.7
2.7
ꢁ0.5
0.1
1.5
3.0
8.9
8.5
8.5
8.5
10.2
10.2
8.9
7.8
10.1
5.7
85
1367
630
401
162
121
238
1658
2145
233
675
33.3
23.3
7.7
124
75
1.8
233
504
28.2
22.2
22.2
33.4
612
53
>104
747
1704
38.9
4.2
5a
240
348
87
82
6a
7a
1810
>104
17.6
366.3
a >30-fold selectivity over 5-HT2A, 5-HT3, 5-HT6.
b hH1 Ki = 18.3 nM.
c hH1 Ki = 2.6 nM.
and low lipophilicities (4, 5). Compound 1b and 3 com-
bined good physicochemical properties with a high 5-
HT5AR affinity. However, 3 carries the undesired 5,6-di-
chloro substitution pattern which we associated with
high hH1R affinity (see preceding publication).
0.8–4 h after an oral dose of 6.6 mg/kg. However, this
improvement was achieved at the expense of selectivity,
as 1b now had significant 5-HT1A affinity (Ki: 85 nM,
Table 4). Thus we had obtained two structurally related
compounds 1a and 1b, the former having the better
selectivity profile, the latter with substantially improved
brain penetration.
Consequently, 1b emerged as the best compound and
was evaluated in a pharmacokinetic experiment in mice
(Table 5). As anticipated, 1b shows an improved brain-
to-plasma ratio as compared to 1a (20-fold increase),
with otherwise similar pharmacokinetic properties,
resulting in brain concentrations greater than 1 lM over
The increased lipophilicity and relatively low basicity al-
lowed a separation of the racemic 3 into enantiomers by
chiral HPLC. An X-ray crystal structure was determined
for the eutomer of 3, which allowed the assignment of

J.-U. Peters et al. / Bioorg. Med. Chem. Lett. 18 (2008) 262–266
Table 5. Pharmacokinetic properties of 1a and 1b (mice)
265
Compound
CL (ml/min/kg)
Vss (l/kg)
t1/2 (h)
Cmax (ng/ml)
Tmax (h)
F (%)
b/p
PBa (%)
1a
1b
30b
31b
9.0b
3.6b
4.0b
1.4b
660c
740d
0.75c
0.25d
59c
36d
ꢀ0.2c
84
92
ꢀ4d
Compound 1b has an improved brain/plasma ratio (b/p), presumably due to its higher lipophilicity.
a Protein binding, determined in vitro.
b 5.0 mg/kg iv.
c 10.0 mg/kg po.
d 6.6 mg/kg po.
the (S)-configuration for this compound (Fig. 1). More
polar compounds could not be separated by HPLC be-
cause of pronounced peak tailing and overlap. However,
we were later able to circumvent this limitation by sep-
aration of the racemic thiourea precursors (e.g., 11,
Scheme 2) into their enantiomers, and the synthesis of
enantiomerically pure target compounds from these
intermediates. It was found that for all enantiomeric
pairs, the 5-HT5AR affinity resides principally in one
of the enantiomers; the eudismic ratios were always
greater than 100-fold (e.g., (S)-3, (R)-3, Table 4). In
analogy to 3, the configuration of the eutomers of 1a
and 1b was assigned also as (S). The pharmacological
profiles of (S)-1a and (S)-1b are summarised in Table 6.
Table 6. In vitro pharmacological profile of the two best compounds,
(S)-1a and (S)-1b
Cl
Cl
NH
NH
F
N
N
N
NH2
H
F
(S)-1a
(S)-1b
(S)-1aa
(S)-1ba
Ki (nM)
Selectivity
Ki (nM)
Selectivity
h5-HT5A
r5-HT5A
h5-HT1A
h5-HT1D
h5-HT2A
h5-HT2C
h5-HT3
1.6
1.3
6.8
2.9
95.2
58-Fold
669-Fold
157-Fold
34-Fold
648-Fold
627-Fold
4-Fold
56.0
8-Fold
1097.9
257.0
55.0
1437.4
491.0
351.7
N.D.
4476.2
24.8
210-Fold
72-Fold
52-Fold
Throughout these studies, numerous compounds were
tested in functional assays and were found to be antag-
onists at the human 5-HT5AR, with functional activities
correlating with their binding constants. For instance,
1b was found to be an antagonist in a [35S]GTPcS assay,
with a pA2 value of 7.41 (pKi 8.2). The binding affinities
for the rat 5-HT5AR were found to be similar to the
affinities for the human receptor (e.g., Table 6).
1062.3
1028.4
6.0
h5-HT6
h5-HT7
655-Fold
4-Fold
a Affinities of R-enantiomers, h5-HT5A: Ki ([R]-1a) = 415 nM; Ki ([R]-
1b) = 1472 nM; N.D., not determined.
the cyclic thioureas (e.g., 11, Scheme 2) as their synthetic
precursors are less polar and were conveniently sepa-
rated. The conversion of the enantiomerically pure thio-
ureas, for example, (S)-11 into the dihydroquinazolines
according to Scheme 2, provided the enantiomerically
pure final compounds, for example, (S)-1b. Racemisa-
tion during this conversion was not found to take place
by analytical chiral HPLC.
All compounds were obtained in analogy to the methods
described in the preceding publication. For instance,
2-aminoacetophenone (8) was converted via 9 into
isothiourea 10; nucleophilic replacement of the methyl-
sulfanyl leaving group by amines then afforded dihydro-
quinazolines 2 (Scheme 1).
Compound 3 could be separated into enantiomers by
preparative, chiral HPLC (chiralpak AD, ethanol/ hep-
tane), whereas 1a, 1b, 4 and 5 could not be separated
by this method because of their high polarity. However,
In summary, the introduction of a difluoroethyl side
chain improved the physicochemical properties of a no-
a
NH
b
O
N
H
S
NH2
8
9
c
NH
NH
R
N
N
N
S
H
10
2
Scheme 1. Preparation of dihydro-quinazolinylamines, 2. Reagents
and conditions: (a) NaBH4, EtOH, 65 °C overnight; then KSCN, HCl,
3 h, 65 °C, 31%; (b) MeI (3 equiv), acetone, o.n. rt, 67%; (c) R–NH2,
CH3CN, 80 °C, or microwave.
Figure 1. An X-ray crystal structure determination of the eutomer of 3
established its (S)-configuration.

266
J.-U. Peters et al. / Bioorg. Med. Chem. Lett. 18 (2008) 262–266
Curr. Drug Targets CNS Neurol. Disord. 2004, 3, 27; (c)
Cl
Cl
Cl
Bockaert, J.; Claeysen, S.; Compan, V.; Dumuis, A. Curr.
Drug Targets CNS Neurol. Disord. 2004, 3, 39; (d)
Woolley, M. L.; Marsden, C. A.; Fone, K. C. F. Curr.
Drug Targets CNS Neurol. Disord. 2004, 3, 59; (e)
Thomas, D. R.; Hagan, J. J. Curr. Drug Targets CNS
Neurol. Disord. 2004, 3, 81.
a
b
NH
NH
N
H
S
N
H
S
11
(S)-11
Cl
2. Oliver, K. R.; Kinsey, A. M.; Wainwright, A.; Siri-
nathsinghji, D. J. S. Brain Res. 2000, 867, 131.
3. (a) Iwata, N.; Ozaki, N.; Inada, T.; Goldman, D. Mol.
Psychiatry 2001, 6, 217; (b) Arias, B.; Collier, D. A.;
Gasto, C.; Pintor, L.; Gutierrez, B.; Valles, V.; Fananas,
L. Neurosci. Lett. 2001, 303, 111; (c) Birkett, J. T.; Arranz,
M. J.; Munro, J.; Osbourn, S.; Kerwin, R. W.; Collier, D.
A. NeuroReport 2000, 11, 2017.
NH
c
NH
F
N
N
N
S
H
F
(S)-12
(S)-1b
Scheme 2. Preparation of (S)-1b. Reagents and conditions: (a) chiral
HPLC, separation of enantiomers; (b) MeI (3 equiv), acetone, weekend
rt, 87%; (b) H2N–CH2–CHF2, CH3CN, 2d, 80 °C, 46%.
4. (a) Nelson, D. L. Curr. Drug Targets CNS Neurol. Disord.
2004, 3, 53; (b) Thomas, D. R. Pharmacol. Ther. 2006,
111, 707.
5. Data presented at the 36th Annu. Meet. Soc. Neurosci.,
October 14–18, Atlanta, 2006. (a) Garcia-Ladona, F.;
Amberg, W.; Kling, A.; Lange, U. E. W.; Hornberger, W.;
Wernet, W.; Netz, A.; Ochse, M.; Mezler, M.; Meyer, A.
H.; Hahn, A.; Hillen, H.; Beyerbach, A.; Bowmik, S.;
Schoemaker, H.; Sullivan, J. P. Abst. 33.1; (b) Drescher,
K. U.; Amberg, W.; Kling, A.; Gross, G.; Schoemaker,
H.; Sullivan, J. P.; Garcia-Ladona F. J. Abst. 33.2; (c)
Rueter, L. E.; Wicke, K.; Basso, A. M.; Jongen-Relo, A.
L.; van Gaalen, M. M.; Gross, G.; Decker, M. W.; Kling,
A.; Schoemaker, H.; Sullivan, J. P.; Amberg, W.; Garcia-
Ladona, F. J. Abst. 33.3; (d) Ebert, U.; Kling, A.;
Amberg, W.; Garcia-Ladona, F. J.; G. Gross, G.; Schoe-
maker, H.; Sullivan, J. P. Abst. 529.25; (e) Jongen-Relo,
A. L.; Bespalov, A. Y.; Rueter, L. E.; Freeman, A. S.;
Decker, M. W.; Gross, G.; Schoemaker, H.; Sullivan, J.
P.; van Gaalen, M. M.; Wicke, K. M.; Zhang, M.;
Amberg, W.; Garcia-Ladona, F. J. Abst. 529.26.
6. Many brain penetrating drugs, such as imipramine,
propranolol, tacrin and paroxetine, have basicities in the
range of pKa = 9.5–10.
7. Van de Waterbeemd, H.; Smith, D. A.; Jones, B. C. J.
Comput. Aided Mol. Des. 2001, 15, 273.
8. (a) Hitchcock, S. A.; Pennington, L. D. J. Med. Chem.
2006, 49, 7559; (b) Atkinson, F.; Cole, S.; Green, C.; Van
de Waterbeemd, H. Curr. Med. Chem. Centr. Nerv. Syst.
Agents 2002, 2, 229.
9. Measurements were performed according to an in-house
developed miniaturized method to measure the shake
flask octanol–water partition coefficient at pH 7.4
(logD).
vel series of guanidine-based 5-HT5AR antagonists and
led to a high brain-to-plasma ratio. (S)-1a and (S)-1b
emerged as the two best compounds from our studies.
(S)-1a is a high affinity 5-HT5AR antagonist with more
than 30-fold selectivity over related receptors (except
5-HT7). The structurally related 1b shows a high
brain-to-plasma ratio but (S)-1b is somewhat less selec-
tive than (S)-1a. Compounds in this series were so far
not found to be selective over the 5-HT7R and are thus
dual 5-HT5A/5-HT7R ligands. Additional structural
modifications will be required to identify 5-HT5A
R
antagonists with an improved selectivity profile, includ-
ing selectivity towards 5-HT7. Such compounds would
be valuable pharmacological tools to further elucidate
the physiological role of the 5-HT5A receptor.
Acknowledgments
The authors are grateful for the excellent technical assis-
´
tance of Philipp Ernst, Silvia Herzig, Stephane Henriot,
Stephane Kritter, Nadine Marquardt, Catherine Muller,
´
Gabrielle Py, Silja Weber and Daniel Zimmerli. The
authors would also like to thank Dr. Andre Alker for
the determination of the X-ray crystal structure of (S)-
3, and Dr. Holger Fischer, Frank Senner and Bjo¨rn
Wagner for the measurement and discussion of logD
and pKa values.
10. Meylan, W. M.; Howard, P. H. J. Pharm. Sci. 1995, 84,
83.
References and notes
11. Smith, D. A.; van de Waterbeemd, H.; Walker, D. K. In
Pharmacokinetics and Metabolism in Drug Design; Mann-
hold, R., Kubinyi, H., Timmerman, H., Eds.; Wiley-VCH:
Weinheim, 2003.
1. (a) Lanfumey, L.; Hamon, M. Curr. Drug Targets CNS
Neurol. Disord. 2004, 3, 1; (b) Costall, B.; Naylor, R. J.