Synthetic study on the unique dimeric arylpiperazine: Access to the minor contaminant of aripiprazole

Article abstract of DOI:10.1016/S0960-894X(02)00857-0

The dimeric derivative of Aripiprazole was synthesized via the two notable synthetic technologies as a key step: (1) efficient aldehyde bis-arylation by Bi(OTf)₃ and (2) facile Wynberg amination at room temperature. The synthesis has established the structural identity with the minor contaminant sometimes present in Aripiprazole.

Full text of DOI:10.1016/S0960-894X(02)00857-0

Bioorganic & Medicinal Chemistry Letters 13 (2003) 65–68
Synthetic Study on the Unique Dimeric Arylpiperazine:
Access to the Minor Contaminant of Aripiprazole
Yasuhiro Torisawa,* Koichi Shinhama, Takao Nishi and Jun-ichi Minamikawa
Process Research Laboratory, Second Tokushima Factory, Otsuka Pharmaceutical Co., Ltd.,
Kawauchi-cho, Tokushima, 771-0182, Japan
Received 19 August 2002; accepted 3 October 2002
Abstract—The dimeric derivative of Aripiprazole was synthesized via the two notable synthetic technologies as a key step: (1) effi-
cient aldehyde bis-arylation by Bi(OTf)3 and (2) facile Wynberg amination at room temperature. The synthesis has established the
structural identity with the minor contaminant sometimes present in Aripiprazole.
# 2002 Elsevier Science Ltd. All rights reserved.
Arylpiperazine skeleton has been widely recognized as
an important and essential pharmacophore in con-
temporary medicinal chemistry, especially in close rela-
tion to some serotonin receptor ligand or other CNS
modulators. Thus, many structural variants of arylpi-
perazines have been prepared in order to attain specific
biological profile. Trends of such medicinal research
mostly focused on the specified ring-substituted deriva-
tives of arylpiperazines.
Our synthesis is based on the two new technologies
developed for the improvement of the classical aldehyde
bis-arylation reaction, as well as uncommon piperazine
introduction reaction by the lithium amide species.
These tactics are briefly summarized in Figure 1.
Thus, aldehyde bis-arylation was first examined with the
substrates (A and B). After that, arylpiperazine intro-
duction was investigated thoroughly with the substrates
which bears 1,1-diarylethane skeleton (C) as detailed in
Scheme 1.
Very few examples of dimeric or polymeric arylpiper-
azines have so far emerged as a promising candidate for
drug substance. As part of our process research on
antipsychotic agent Aripiprazole (1),1 a trace amount of
dimeric derivative was detected in a raw bulk material,
causing problem in further characterization and pur-
ification.
For the precise characterization, a careful inspection by
NMR (1H and 13C) indicated the bis-arylpiperazine
structure (2), in which two moles of Aripiprazole was
connected by 1,1-diarylethane bridge as shown in Fig-
ure 1. To identify the proposed structure (2) and further
to detect the origin of this contaminant, we started the
chemical synthesis of 2. We report herein the whole
story regarding the synthesis of 2 along with some
speculations regarding the most probable origin of 2.
*Corresponding author. Tel.: +81-88-665-2126; fax: +81-88-637-
1144; e-mail: torisawa@fact.otsuka.co.jp
Figure 1. Dimeric arylpiperazines.
0960-894X/03/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(02)00857-0

66
Y. Torisawa et al. / Bioorg. Med. Chem. Lett. 13 (2003) 65–68
alcohol). (3) Reduction of the bromide by NaBH3CN
furnished the requisite 1,1-diarylethane (6) (80%). The
overall yield from 4 to 6 was at first around 60% using
crystallization in each step. Improvements of reagent
combination of each step have been attained success-
fully, and the whole story of this progress has already
been detailed in another article.2b
Piperazine Introduction
The most important task after aldehyde bis-arylation
was an introduction of two piperazine units into 6.
There were two notable (complementary) pathways for
the conversion of anisole derivative into arylpiperazines:
Buchwald amination and Wynberg amination. We
made intensive study along this line until our develop-
ment of useful modifications as summarized below.
Buchwald amination
The direct catalytic amination of aromatic halides has
become popular since 1995, when Buchwald and Hart-
wig independently reported clever reagent systems by
the choice of bulky phosphine in combination with a
strong base. There were many follow-up of this amina-
tion technology including some applications to arylpi-
perazine synthesis in medicinal chemistry research.
Like others, we reported some successful applications in
the closely related substrates (9).3 In order to apply this
protocol to the present substrate, we examined the
reaction of the triflate (7), which was stable and easily
obtainable from 6 by the standard transformations (de-
methylation by BBr3 in CH2Cl2 followed by the treat-
ment of Tf2O and pyridine).
Scheme 1. C-Arylation tactics.
Aldehyde Bis-arylation
Direct reaction of phenylpiperazines (A) with acetalde-
hyde and its derivatives (i.e., A into C) simply led to the
recovery of SM. This made us find a good reaction
partners and catalysts for aldehyde bis-arylation.
Among the reaction candidates shown in Scheme 1, we
found that 2,3-dichloroanisole (3) was a good substrate
(in reactivity and availability), which gave 2,2-diaryl-
acetate (4) in nearly quantitatively, via the reaction with
ethyl glyoxylate polymer by the aid of a catalytic metal
triflate [Sc(OTf)3, Bi(OTf)3].2 We did not detect any de-
methylation reaction, which was often occurred by the
use of a classical Lewis acid catalyst such as AlCl3.
In contrast to our expectations, however, the attempted
Buchwald amination of 7 with N-Boc piperazine gave
no desired arylpiperazines even under our optimized
conditions (i.e., 9 into 10).3,4
Instead, the reaction of 7 turned out to be very slow and
we only isolated the phenol by-products (derived from
hydrolysis of the triflate) along with a large amount of
starting material (7).
Furthermore, the key para para dimeric structure was
secured on the basis of cosy and noesy spectra as shown.
Predominant formation of the one isomer (4) was cru-
cial for further elaboration to 2.
This unsuccessful outcome of Buchwald catalytic ami-
nation prompted us to investigate the feasibility of
Wynberg amination, which required very harsh condi-
tions employing anisole and lithium amide under long h
of reflux.5
As other reaction candidates, amine derivatives such as
2,3-dichloroaniline and 2,3-dichlorophenylpiperazines
(structure: A) were very sluggish in this glyoxylate bis-
arylation reaction. Use of more selective catalysts
including Clay and Zeorite were ineffective.
Modified Wynberg amination
Wynberg has reported a protocol for amination, which
enabled direct aromatic substitution reaction of anisole
by the lithium amide in 1993. There were however very
few application of this technology, partly because of the
harsh reaction conditions and long reaction time with
refluxing. In the original reported procedures, reactions
usually required long h of reflux (>10 h) after tedious
formation of lithium amide.5
Conversion of 4 into the desired 1,1-diarylethane (6)
was carried out in a straightforward way as shown in
Scheme 1. (1) Reduction of 4 by ZrCl4/NaBH4 afforded
the alcohol (nearly quantitative). (2) Bromination of the
alcohol by Ph3P/CBr4 gave the bromide (5) (80% from

Y. Torisawa et al. / Bioorg. Med. Chem. Lett. 13 (2003) 65–68
67
To circumvent these inherent problems of the original
Transformation to 2
protocol, and to find a clue for congested arylpiperazine
synthesis, we conducted some model experiments with
careful modification in solvent system and in the way of
addition of alkyllitium reagents. Very simple modifi-
cations did work very well as summarized below.
To obtain our target molecule (2), de-methylation of 12
was investigated next. Two protocols were available for
this purpose: 1-chloroethyl chloroformate (ACE-Cl)
and trichroloethyl chloroformate (TROC-Cl). Success-
ful one-pot de-methylation was possible by the choice of
ACE-Cl in EDC followed by refluxing in MeOH as
summarized in Scheme 3. Simple evaporation of all
volatile materials under reduced pressure afforded the
hydrochloride (13) as a brown caramel, which was pur-
ified by precipitation in EtOH to afford the solid
hydrochloride (13) with good purity. Partial deprotec-
tion product (mono N-Me derivative of 12) or other side
products were removed in the filtrate. The solid material
(13) was then subjected to the next N-alkylation in a
usual manner as shown in Scheme 3.
In our model study, we found that the reaction of 2,3-
dichloroanisole (3) and N-methylpiperazine (11) was
feasible at room temperature during the slow addition
of a solution of n-BuLi (3 M) to a stirred mixture of 3
and 11 in DME (not THF). This operation did simplify
the time-consuming protocol originally reported by
Wynberg.5 Reactions with other polar solvents (DMF,
DMA, NMP, DMSO) were unsuccessful which accom-
panied substantial formation of unwanted side products
derived from the reaction of solvent itself. As originally
reported by Wynberg, ethereal solvents (DME, THF)
were essential for this transformation under Ar atmo-
sphere. The yields were usually higher in DME (75–
80%) than in THF (50–55%).
Thus, the two segments (13 and 14) were refluxed in
H2O–IPA in the presence of K2CO3 for 6 h. After
removing IPA, resulting mixture was diluted and
extracted with CH2Cl2. Crude product was further pur-
ified by SiO2 column chromatography (AcOEt–MeOH–
Et3N) to give nearly pure product 2 (40%) as a major
product. The improved result was obtained by the
addition of PTC catalyst (TBAI) or by performing this
reaction in DMF–NaH composition (65–70%).6 The
product was crystallized from EtOH–AcOEt to furnish
the solid material that was super-imposable with the
authentic sample isolated from the raw Aripiprazole in
1H NMR, and 13C NMR, as shown below.
In the real substrate (6), our modified protocol worked
quite well, affording the desired bis-arylpiperazine (12)
by the slow addition (exothermic) of excess n-BuLi (2.5
to 3.5 equiv to 6) to the stirred solution of 6 and 11 (3.5
to 4 equiv to 6) in DME. After stirring for ca. 2h at rt,
NMR analysis of the crude extracts indicated ca. 85%
yield of the desired bis-arylpiperazine (9).
As indicated in Table 1 in Scheme 2, key to the repro-
ducible result was a slight excess piperazine and n-BuLi
and the rate of addition of BuLi so as to keep the mix-
ture below 55 ꢀC under Ar.
Dimerization of Aripiprazole
We also attempted this reaction with different piper-
azines and bases. As a more convenient base, we
attempted to use LiOR or other alkoxides (MOR), but
almost no reactions were observed in DME at ambient
temperature. Only a trace of product was formed after
heating. Attempted reactions with other piperazine
derivatives such as N-benzylpiperazine and N-Boc
piperazine were sluggish and found to be impractical for
our synthesis. As far as we surveyed with various ani-
sole derivatives, reaction of N-methylpiperazine in
DME as solvent gave the best result as shown above.
After having obtained the authentic samples of 2 as well
as its precursor (13), we searched for their origin in Ari-
piprazole production. First of all, we attempted the direct
reaction of Aripiprazole with acetaldehyde (or equiva-
lents) to obtain 2 (or other dimeric products). Although
we carried out many reaction conditions but none of the
dimeric structure related 2 was obtained. Instead, we
could isolate another type of dimer (15), under very
harsh reaction conditions, whose structure was tenta-
tively assigned as shown in the following Scheme 4.
Scheme 2.
Scheme 3.

68
Y. Torisawa et al. / Bioorg. Med. Chem. Lett. 13 (2003) 65–68
ing mixture was warmed at 50–60 ꢀC for 5 h. TLC indi-
cated the major formation of less polar spots (N-
chloroethylformate derivatives). The mixture was fur-
ther heated to reflux after addition of MeOH (100 mL)
to form the hydrochloride (13). The dark mixture was
stirred at ambient temperature to precipitate insoluble
solid mass, which was subsequently removed by filtra-
tion. Evaporation of the filtrate afforded crude product
as brown caramel (18.5 g), which was treated with hot
EtOH to remove side products. Thus obtained solid
material 13 (12.5 g) showed enough purity (85–95%) by
HPLC analysis, which was directly subjected to the next
N-alkylation to furnish 2.
Selected data for 13. 13C NMR (DMSO-d6) d: 144.1,
134.5, 128.4; 122.9, 122.8, 115.4, 43.7, 38.9, 15.9. 1H
NMR (DMDO-d6) d: 9.50 (br, 4H), 7.20 (m, 4H), 4.78
(q like, 1H), 3.29, (m, 8H), 1.52 (d, 3H, J=7.2Hz).
Scheme 4.
In the meantime, we carried out careful inspection of
the starting material supplied by the foreign bulk
makers. This survey revealed the presence of 13 in the
bulk material of arylpiperazine segment (16).7 We thus
tentatively speculate that the origin of the dimeric con-
taminant (2) must be the presence of 13 in the starting
material (16). This means that pure Aripiprazole can be
obtained from the pure starting material 16 definitely free
from the contaminants formed during the preparation.7
Selected data for 2. 13C NMR (CDCl3) d: 172.1, 158.5,
149.1, 138.4, 138.1, 133.7, 18.5, 128.1, 125.9, 118.0,
115.5, 108.6, 102.1, 67.7, 58.1, 53.2, 51.2, 39.7, 31.0, 27.2,
24.4, 23.3, 20.0. 1H NMR (CDCl3) d: 8.19 (br, 2H), 7.08
(d, J=7.1 Hz, 2H), 6.98 (m, 4H), 6.56 (dd, J=7.1,
2.4 Hz), 6.36 (d like, 2H), 4.89 (q like, 1H), 4.00 (t like,
4H), 3.09 (m, 8H), 2.93 (m, 5H), 2.66 (m, 12H), 2.52 (m,
4H), 1.86–1.70 (m, 12H), 1.55 (3H, d, J=7.0 Hz).
In summary, we have succeeded in the synthesis of the
dimeric derivative of Aripiprazole (2) by the newly
devised synthetic technologies. The sequence shown
here demonstrated the utility of Wynberg amination for
the sterically demanding substrate such as 6, instead of
the catalytic amination developed recently. Some of the
representative procedures are summarized below. Our
synthesis also indicated the presence of the origin of the
contaminant (13) in the bulk starting material (16).
Acknowledgements
We thank Mrs. K. Yamamoto, Y. Nishioka and I. Miura
in our laboratories for their helpful assistance in isolation
of 2 and subsequent spectroscopic assignment. For the
intensive discussion about Wynberg amination reactions,
we also thank to Dr. S. Morita now in our laboratory.
References and Notes
Selected Experimental Procedures
Modified Wynberg amination
1. Oshiro, Y.; Sato, S.; Kurahashi, N.; Tanaka, T.; Kikuchi, T.;
Tottori, K.; Uwahodo, Y.; Nishi, T. J. Med. Chem. 1998, 41, 658.
2. (a) Torisawa, Y.; Nishi, T.; Minamikawa, J. Org. Process
Res. & Dev. 2001, 84. (b) Torisawa, Y.; Nishi, T.; Minami-
kawa, J. Bioorg. Med. Chem. 2002, 10, 2583. (c) Loux, C. L.;
Dubac, J. Synlett 2002, 181. (d) Carrigan, M. D.; Sarapa, D.;
Smith, R. C.; Wieland, L. C.; Mohan, R. S. J. Org. Chem.
2002, 67, 1027.
3. Torisawa, Y.; Nishi, T.; Minamikawa, J. Bioorg. Med.
Chem. Lett. 2000, 10, 2489.
4. Torisawa, Y.; Nishi, T.; Minamikawa, J.; Bioorg. Med.
Chem. 2002, 10, 4023.
To a stirred solution of 6 (31 mM)2 and N-methylpiper-
azine (11, 4.5 equiv; distilled and flushed with Ar) in dry
DME (Kanto, 100 mL) was added under cooling (25–
30 ꢀC), a solution of n-buthylithium in hexane (Kanto,
ca. 2.5 M; 50+10 mL; ca. 2.5 equiv) via syringe (exo-
thermic reaction; inner temp below 55 ꢀC) under Ar.
Resulting mixture was stirred without cooling for fur-
ther 3 h, before it was quenched by the addition of H2O
(17 mL; added by drops). The resulting mixture was
then diluted with AcOEt and stirred with Na2SO4. Fil-
tration and evaporation of the organic layer afforded
crude products, which was subjected to vacuum pump-
ing to form a caramel or half-solid material (22.5 g).
NMR analysis of this crude product was critical. Com-
parison with standard sample guaranteed that 12 was
the sole product formed with trace of 11 after intensive
vacuum pumping.
5. Hoeve, W.; Krause, C. G.; Luteyn, J. M.; Thiecke, R. G.;
Wynberg, H. J. Org. Chem. 1993, 58, 5101.
6. Morita, S.; Kitano, K.; Matsubara, J.; Ohtani, T.; Kawano,
Y.; Otsubo, K.; Uchida, M. Tetrahedron 1998, 54, 4811.
7. Careful inspection of the bulk material (16) revealed the
presence of two minor contaminants. Beside 13, we identified
the new sym-triamine derivative (ArNHCH2CH2NHCH2-
CH2NHAr), which was fortunately eliminated during the
transformation to Aripiprazole. Attempted N-alkylation of
this triamine with excess 14 did not give any N-alkylated pro-
ducts under standard conditions. Details will be published
elsewhere together with the contaminants of other arylpiper-
azines having different substituents.
1-Chloroethyl chloroformate (TCI, 25.5 mL) was added
at rt to a stirred solution of 12 obtained above (22.5 g;
ca. 30 mM) in dry EDC (100 mL) under Ar. The result-