The design and preparation of metabolically protected new arylpiperazine 5-HT1A ligands

Article abstract of DOI:10.1016/j.bmcl.2004.01.045

New arylpiperazines related to buspirone, gepirone and NAN-190 were designed and screened in silico for their 5-HT_1A affinity and potential sites of metabolism by human cytochrome P450 (CYP3A4). Modifications to these structures were assessed in silico for their influence on both 5HT _1A affinity and metabolism. Selected new molecules were synthesized and purified in a parallel chemistry approach to determine structure activity relationships (SARs). The resulting molecules were assessed in vitro for their 5HT_1A affinity and half-life in a heterologously expressed human CYP3A4 assay. Molecular features responsible for 5-HT_1A affinity and CYP3A4 stability are described.

Full text of DOI:10.1016/j.bmcl.2004.01.045

Bioorganic & Medicinal Chemistry Letters 14 (2004) 1709–1712
The design and preparation of metabolically protected new
arylpiperazine 5-HT1A ligands
Manish Tandon, Mary-Margaret O’Donnell, Alex Porte, David Vensel, Donglai Yang,
Rocio Palma, Alan Beresford and MarkA. Ashwell *
ArQule Inc., 19 Presidential Way, Woburn, MA 01801, USA
Received 8 December 2003; revised 16 January 2004; accepted 19 January 2004
Abstract—New arylpiperazines related to buspirone, gepirone and NAN-190 were designed and screened in silico for their 5-HT1A
affinity and potential sites of metabolism by human cytochrome P450 (CYP3A4). Modifications to these structures were assessed in
silico for their influence on both 5HT1A affinity and metabolism. Selected new molecules were synthesized and purified in a parallel
chemistry approach to determine structure activity relationships (SARs). The resulting molecules were assessed in vitro for their
5HT1A affinity and half-life in a heterologously expressed human CYP3A4 assay. Molecular features responsible for 5-HT1A affinity
and CYP3A4 stability are described.
# 2004 Elsevier Ltd. All rights reserved.
The cytochrome P450 3A4 (CYP3A4) has a broad range
of substrate specificity and is responsible for metaboliz-
ing nearly 50% of marketed drugs.1 It is therefore a
significant challenge in the pursuit of orally active small
molecule therapeutics. Due to CYP3A4s broad sub-
strate specificity, predicting which site on a drug is sus-
ceptible to metabolism by this enzyme has proved
difficult. We initiated a library design project incorpor-
ating our predictive global site lability CYP3A4 model
to aid in the identification of new 5-HT1A ligands with
increased stability towards CYP3A4 metabolism. Bus-
pirone, 12 is a known drug, which is metabolized in vivo
primarily by CYP3A4. Therefore it was selected as the
structural manifold on which to demonstrate a parallel
design and synthesis process.
initial findings for (a) and (b) and provide a summary of
the SAR. We discuss the structural changes made in
relationship to the measured 5-HT1A affinity and CYP
3A4 stability.
The CYP3A4 metabolism model estimates regioselec-
tivity by calculating the relative stabilities of free radi-
cals using semi-empirical quantum calculations.3 The
model is generally applicable as it is not derived from
limited structure–activity information and requires only
a structure as input. The resulting free energy estimates
are combined with steric accessibility and global orien-
tation terms to rank-order regiolabilities. The total sus-
ceptibility to metabolism is categorized relative to the
competing decoupling reaction resulting in the form-
ation of water: sites that have predicted rates of meta-
bolism much greater than that of water formation are
classed as highly labile, while those that have much
lower rates are classed as stable.
Our objective was 3-fold; (a) maintain or improve in
vitro 5HT1A affinity compared to 1 (Table 1, IC50=25
nM, upper limit for new designs=250 nM), (b) increase
the stability to CYP3A4 metabolism 2-fold or greater
compared to 1 (Table 1, t1/2=4.6 min) and (c) to pro-
vide additional SAR data to refine our predictive mod-
els to further increase their effectiveness in designing
additional molecules with improved 5HT1A activity and
CYP3A4 stability. Herein we report the results of our
Although the number of predicted labile sites in a
molecule may not correlate directly to the rate of meta-
bolism, the relative number of labile sites in a molecule
corresponds to the number of modifications required to
improve its stability.
The visual representation of the metabolic landscape
(Fig. 1) assigns the different sites of metabolism as
labile, moderately labile, moderately stable and stable.
* Corresponding author. Tel.: +1-781-994-0598; fax: +1-781-994-
0690; e-mail: mashwell@arqule.com
0960-894X/$ - see front matter # 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.01.045

1710
M. Tandon et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1709–1712
zine or with substituted phenyl rings. The labile site in
Region 2 was addressed by incorporation of (1) a steric
block(alpha methyl or beta gem dimethyl), (2) an elec-
tronic blockwith a beta OH, (3) a combination of a
methyl group and OH on the beta carbon. The study
was extended to include two additional Region 3 var-
iants, the 4,4-dimethylpiperidinone of gepirone and the
phthalimide of NAN-190 with both straight chain and
alpha methyl variants.8
Figure 1. Metabolic landscape of buspirane
CYP3A4 model (most labile sites shown).
1 as predicted by
All straight chain and alpha methyl analogues were
prepared as shown in Scheme 1.9
Furthermore this analysis provides relative proportions
of each metabolite likely to be formed by CYP3A4.
Focused libraries containing only the 3,3-tetra-
methyleneglutarimide in Region 3 were prepared by
reacting aryl piperazines with the dimesylate of 2,2-
dimethylbutane-1,4-diol 2 to give the cyclic quarternary
ammonium salts 3. Which upon treatment with 3,3-tet-
ramethyleneglutarimide gave the final products as
shown in Scheme 2.10
The predicted metabolic landscape for 1 (Fig. 1, only
labile sites depicted) is in good agreement with the
reported experimental data.4 Specifically, it shows a
propensity (labile) for aromatic hydroxylation at C5 of
the pyrimidine ring and oxidation on the carbon alpha
to the piperazine ring (labile), which leads to N-dealkyl-
ation producing 1-pyrimidinyl-piperazine. Both of these
sites are metabolized in vivo. Our design strategy
assumed that a reduction in the number of labile sites in
the molecule would result in concomitant increase in the
stability towards the CYP3A4.
Access to the analogues modified with both a methyl
and a hydroxyl group on the beta carbon was achieved
by reacting the epoxides 4 with aryl piperazines as
described in Scheme 3.
Our synthetic strategy was devised to provide straight-
forward access to molecules modified at the primary
sites of metabolism. The most labile sites were addressed
first as these were expected to have the most dramatic
effect on the overall stability of the molecule.
The scaffold was divided into three regions (Fig. 1) and
new compounds were designed around these. The pro-
tonatable arylpiperazine recognition element, crucial for
5-HT1A affinity, was maintained in all designs.5
Substituents R1, R2 and R3, in Region 2 were filtered to
reject any combination which (a) rendered the basic N
on the arylpiperazine non-basic, as predicted by ACD
pkA predictor6 and (b) did not fit the Catalyst pharma-
cophore model. This pharmacophore was constructed
from literature data and was consistent with the estab-
lished SAR requiring the presence of a basic arylpiper-
Scheme 1. (a) Aryl piperazine, K2CO3, DMF, 80 ꢀC.
azine nitrogen, an aromatic hydrophobe and
a
hydrogen bond acceptor. The pharmacophore preferred
an extended conformation in agreement with published
CoMFA models.7 The pharmacophore provided a
powerful and convenient predictive tool with which to
apply the known SAR.
In an iterative virtual design process, structural modifi-
cations in each region were made and the effect on pre-
dicted CYP3A4 was assessed. If the design was accepted
the two remaining regions were also subjected to virtual
library enumeration. The prioritized virtual arrays were
then assessed in both the CYP3A4 and the 5-HT1A
models. A similar process was employed to address
multiple sites in parallel.
Scheme 2. (a) Aryl piperazines, DMF, decant.; (b) 3,3-tetramethyl-
eneglutarimide, K2CO3, DMF, 100 ꢀC.
The labile site of Region 1 was addressed by; (1) block-
ing the oxidation site with a fluorine atom, (2) replacing
the pyrimidine ring with either a 2-pyridine or 2-pyra-
Scheme 3. (a) Aryl piperazine, LiClO4, CH3CN, 80 ꢀC.

M. Tandon et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1709–1712
1711
Table 1. Region 1 and 2 SAR of the 3,3-tetramethyleneglutarimide containing analogues
Compd
R1
R2
R3
R4
5-HT1A
IC50 (mM)a
CYP 3A4
t1/2 (min)b
1
5
6
7
H
H
H
H
H
H
H
H
H
Pyrimidine
5-F Pyrimidine
p-F phenyl
2-Pyrazine
0.025
0.063
0.064
0.46
4.6
52.3
3.2
H
H
H
5.5
8
9
H
H
H
H
Me
H
OH
OH
OH
OH
OH
H
H
H
H
Me
H
H
Me
H
H
H
p-OMe Phenyl
Pyrimidine
5-F Pyrimidine
Pyrimidine
o-OMe Phenyl
Pyrimidine
Pyrimidine
2-Pyrazine
o-OMe Phenyl
p-F phenyl
p-F phenyl
Phenyl
0.93
21.1
3.8
14.8
2.9
5.1
10.2
8.6
9.6
7.8
3.8
2.4
Me
Me
H
H
H
H
H
H
H
0.004
0.046
0.71
0.002
0.84
1.46
>1.0
0.012
0.069
0.099
0.015
10
11
12
13
14
15
16
17
18
19
Me
H
H
H
H
8.0
a Binding affinity at 5-HT1A receptors labeled with [3H]-8-OH-DPAT.
b LC/MS/MS analysis of parent ion fragmentation at each time point, peakarea reported normalized to internal standard.
Analysis of modifications in Region 1 (Region 2 and 3
fixed as in 1) demonstrates that the labile site (5-position
in pyrimidine ring) plays a dominant role in determining
overall stability to CYP 3A4 metabolism11 (Table 1).
Removal of the Region 2 labile site by introduction of
beta-OH was successful in providing analogues with
increased metabolic stability. For example analogue 13
had improved stability (t1/2=10.2 min) with respect to 1
(t1/2=4.6 min). The same trend is observed with the
regioisomeric 15 and its unsubstituted parent 7. How-
ever, both 13 and 15 are considerably less potent 5-
HT1A ligands than 1. No additional improvement either
in stability or affinity is seen when both a methyl and a
–OH are combined on the same carbon atom of 1, as in
14.
When the C5 position is substituted with a fluorine
atom as in 5, 5-HT1A affinity12 is maintained (63 nM)
and the compound is more stable than 1. However,
when the Region 2 labile site is blocked with a methyl
group as in 9 (or a beta gem dimethyl 11), the stability is
not improved compared to 1. Removal of both labile
sites by introduction of an alpha methyl into 5 provid-
ing 10 decreases CYP3A4 stability relative to 5. Our
global CYP3A4 model did not accurately predict this.
In general, introduction of a methyl group alpha to the
piperazine ring reduces CYP3A4 stability compared to
the unsubstituted analogues a trend, which held true
when the Region 3 substituent was varied (compare 20
and 21 and 23 and 24 in Tables 2 and 3).
Significantly, the introduction of a beta –OH group in
Region 2 did not reduce the 5-HT1A affinity as drama-
tically when introduced into the 2-OMePh substituted
piperazines (common to many potent 5-HT1A ligands).
A comparison of the unsubstituted 12 and its beta-OH
analogue 16 shows that the more metabolically stable 16
also has good 5-HT1A affinity.
Table 2. Region 1 SAR of the 4,4-dimethylpiperidine-2,6-dione ana-
logues
Table 3. Region 1 SAR of the phthalimide containing analogues
Compd
R1
R4
5-HT1A
IC50 (mM)a
CYP 3A4
t1/2 (min)b
Compd R1
R4
5-HT1A IC50 (mM)a CYP 3A4 t1/2 (min)b
23
24
25
H
H
Me
Pyrimidine
5-F Pyrimidine
5-F Pyrimidine
0.06
0.44
0.21
26.8
20.4
18.2
20
21
22
H
H
Pyrimidine
5-F Pyrimidine
0.114
0.205
0.087
30.3
78.3
40.1
Me 5-F Pyrimidine
a Binding affinity at 5-HT1A receptors labeled with [3H]-8-OH-DPAT.
b LC/MS/MS analysis of parent ion fragmentation at each time point,
peakarea reported normalized to internal standard.
a Binding affinity at 5-HT1A receptors labeled with [3H]-8-OH-DPAT.
b LC/MS/MS analysis of parent ion fragmentation at each time point,
peakarea reported normalized to internal standard.

1712
M. Tandon et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1709–1712
Introduction of either a F or OMe blockon to the 4
position of the arylpiperazine again highlights the diffi-
culty in maintaining 5-HT1A affinity while increasing
metabolic stability. Although the 4-fluorophenyl analo-
gues (6, 17 and 18) are all potent, none of them have
improved stability relative to 1 or the unsubstituted
phenyl, 19. In contrast, the alternate blocking sub-
stituent 4-OMe of 8, while increasing metabolic stabi-
lity, reduces activity to the unacceptable micromolar
level.
This workand additional analogues provided invalu-
able SAR for the construction and implementation of a
new local CYP3A4 rate model design, which will be
described in detail elsewhere.
Acknowledgements
We thankDrs. Ken Korzekwa and Stuart Russell for
their pADME modeling insight, Dr. Andrew Smellie for
his molecular modeling assistance and Dr. Sheila
DeWitt and Dr. Valery Antonenko for their support of
this project.
We were predominantly interested in modifications to
reduce CYP3A4 metabolism at those sites indicated
to be most labile by our predictive model. However, we
also wanted to explore the role of the substituent in
Region 3. It has recently been suggested that a general
strategy for avoiding CYP2D6 metabolism in com-
pounds related to 1, is to introduce bulky substituents
into Region 3.13
References and notes
1. Rendic, S. Drug Metab. Rev. 2002, 34, 83.
2. (a) Wu, Y.-H.; Rayburn, J. W.; Allen, L. E.; Ferguson,
H. C.; Kissel, J. W. J. Med. Chem. 1972, 15, 477. (b)
Tunicliff, G.; Eison, A. S.; Taylor, D. P., Eds. In Buspir-
one: Mechanismand Clinical Aspects ; Academic Press Inc,
1991.
3. (a) Jones, J. P.; Mysinger, M.; Korzekwa, K. R. Drug
Metab. Dispos. 2002, 30, 7. (b) Korzekwa, K. R.; Jones,
J. P.; Gillette, J. R. J. Am. Chem. Soc. 1990, 112, 7042.
4. (a) Jajoo, H. K.; Mayol, R. F.; LaBudde, J. A.; Blair, I. A.
Drug Metab. Dispos. 1989, 17, 634. (b) Jajoo, H. K.;
Mayol, R. F.; LaBudde, J. A.; Blair, I. A. Drug Metab.
Dispos. 1989, 17, 625. (c) Sakr, A.; Andheria, M. J. Clin.
Pharmacol. 2001, 41, 886.
5. Lopez-Rodriguez, M. L.; Ayala, D.; Benhamu, B.; Mor-
cillo, M. J.; Viso, A. Curr. Med. Chem. 2002, 9, 443.
6. PhysChem Batch software, v6.13; http://www.acdlabs.-
com/products/phys_chem_lab/pka/
7. Lopez-Rodriguez, M. L.; Rosado, M. L.; Benhamu, B.;
Morcillo, M. J.; Fernandez, E.; Schaper, K.-J. J. Med.
Chem. 1997, 40, 1648.
Our findings from this study demonstrate that these
bulky substituents also dramatically affect the metabo-
lism by CYP3A4. Overall stability is greatest when
Region 3 contains 4,4-dimethylpiperidinone, followed
by phthalimide (Table 2) and is least with 3,3-tetra-
methyleneglutarimide (Table 3), as in 1.
A design-make-test paradigm was implemented to
assess the impact of an in silico model of human
CYP3A4 metabolism. We successfully, demonstrated
the use of in silico models for 5-HT1A affinity and
CYP3A4 metabolism. Strategies were found which
maintained 5-HT1A affinity and improved in vitro
CYP3A4 stability within the limits established for the
goals of the project. Key SAR features which emerged
were: (i) the effect of steric crowding around the site of
N-dealkylation negatively affecting the CYP3A4 stabi-
lity, (ii) introduction of a hydroxyl group beta to this
site which increased metabolic stability and (iii) the
previously under appreciated effect of the Region 3
substituent in determining the rate of CYP3A4 meta-
bolism for this general class of arylpiperazine 5HT1A
ligands.
8. Glennon, R. A.; Naiman, N. A.; Pierson, M. E.; Titeler,
M.; Lyon, R. A.; Herndon, J. L.; Misenheimer, B. Drug
Dev. Res. 1989, 16, 335.
9. All new compounds were prepared in a parallel chemistry
fashion, purified by RP-HPLC and characterized by
HPLC-MS.
10. Ishizumi, K.; Kojima, A.; Antoku, F.; Ikutaro, S.; Yosh-
igi, M. Chem. Pharm. Bull. 1995, 43, 2139.
11. For a detailed description of the assay methodology see:
Saraswat, L.; Caserta, K. A.; Laws, K.; Wei, D.; Jones,
S. J.; Adedoyin, A. J. Biomolecular Screening 2003, 8, 544.
12. The method was based on a scintillation proximity assay
previously described but adapted for human 5-HT1A
binding: Kahl, S. D.; Hubbard, F. R.; Sittampalam, G.;
Zock, J. M. J. Biomolecular Screening 1997, 2, 33.
13. Kalgutkar, A. S.; Zhou, S.; Fahmi, O. A.; Taylor, T. J.
Drug Metab. Dispos. 2003, 31, 596.
In general, for close structural analogues of 1 significant
improvements in stability against CYP3A4 were always
at the expense of 5-HT1A affinity except in the case of
the known incorporation of a 5-F substituent into the
pyrimidine ring. Other potent 5-HT1A arylpiperazines
templates such as 2-OMePh had improved CYP3A4
stability and maintained affinity when substituted with a
beta OH.