On the photoisomerization of the benzisothiazole portion of ziprasidone

Article abstract of DOI:10.1016/S0040-4039(03)00028-5

Photostability challenge of ziprasidone in solution shows that the benzisothiazole moiety undergoes isomerization to the corresponding benzthiazole. A model compound, 3-piperazinyl-1,2-benzisothiazole, also undergoes this photoisomerization. Identification of the products has been confirmed by synthesis of the proposed molecules.

Full text of DOI:10.1016/S0040-4039(03)00028-5

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 44 (2003) 1559–1561
On the photoisomerization of the benzisothiazole portion of
ziprasidone
a,
b
b
a
Thomas R. Sharp, * Kyle R. Leeman, D. Eugene Bryant and George J. Horan
a
Analytical R&D Department, Pfizer Global Research & Development, Groton, CT 06340, USA
Chemical R&D Department, Pfizer Global Research & Development, Groton, CT 06340, USA
b
Received 19 November 2002; revised 2 January 2003; accepted 3 January 2003
Abstract—Photostability challenge of ziprasidone in solution shows that the benzisothiazole moiety undergoes isomerization to the
corresponding benzthiazole. A model compound, 3-piperazinyl-1,2-benzisothiazole, also undergoes this photoisomerization.
Identification of the products has been confirmed by synthesis of the proposed molecules. © 2003 Elsevier Science Ltd. All rights
reserved.
Photostability challenge of ziprasidone (1), an antipsy-
1,2
chotic drug substance recently approved by the US
Food and Drug Administration, induces the compound
to undergo a clean conversion to a single, chromato-
graphically distinct compound. Electrospray liquid
chromatography–mass
spectrometry
(LC–MS)
3
experiments indicate identical mass spectra for ziprasi-
done and the new product. Similar photochallenge
4
experiments with the two synthetic precursors of
ziprasidone indicate that the benzisothiazole portion
undergoes the conversion, not the oxindole portion.
Comparison with reference standards of 2, the benzthi-
azole analog of ziprasidone, and of 5, the benzthiazole
analog of a synthetic precursor, confirm the photochal-
lenge products as benzthiazole isomers.
The partial chromatograms shown in Figure 1 illustrate
the clean conversion of ziprasidone to its photochal-
lenge product. Comparison with the electrospray mass
spectrum of ziprasidone, taken from the same LC–MS
run, shows a spectrum (Fig. 2) identical to that of
ziprasdone, complete with the isotope pattern associ-
+
ated with the m/z 413 [M+H] indicative of the pres-
ence of
a
single chlorine atom. MS–MS
collision-induced decomposition product ion spectra of
ziprasidone and the photochallenge product are very
similar.
Figure 1. UV–vis (top) and total ion current (bottom) chro-
matograms of photochallenged ziprasidone (30.3 min) and its
photoproduct (28.2 min).
The simplest rationalization for this observation is a
photo-induced isomerization. Similar photochallenge
experiments were conducted with 3 and 4, both syn-
*
Corresponding author. Tel.: 860-441-5932; fax: 860-441-0437;
e-mail: thomas r sharp@groton.pfizer.com
–
–
0
040-4039/03/$ - see front matter © 2003 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(03)00028-5

1
560
T. R. Sharp et al. / Tetrahedron Letters 44 (2003) 1559–1561
Figure 2. ESP mass spectrum of the 28.2 min chromatographic peak of Figure 1. This component is produced by light challenge
of ziprasidone.
thetic precursors of ziprasidone. Oxindole 4 remained
unchanged, but benzisothiazole–piperazine structure 3
showed a clean conversion to a chromatographically
distinct species, similar to the conversion of ziprasi-
done. LC–MS analysis showed that the product from
photochallenge of 3 also showed a 219 dalton relative
molecular mass identical to that of the starting mate-
rial. MS–MS product ion spectra of both compounds
were very similar. Since neither 3 nor its photochallenge
product produced a large signal response by electro5-
spray in the mass spectrometer, GC–MS experiments
were also performed to support the assignment of iden-
tical relative molecular mass and structural similarity.
The GC–MS run showed only two chromatographic
peaks, the necessary 219 dalton molecular mass for
both peaks, and spectra which differed only slightly in
relative abundances of the fragments. Both electron
ionization-induced fragmentation (by GC–MS) of 3
and its isomer, and collision-induced fragmentation (by
LC–MS) of ziprasidone and its isomer could be justified
with respect to literature discussions on these types of
intermediate. Nucleophilic displacement is proposed to
occur at either the azirine carbon or the nitrogen by the
sulfur (Scheme 1), assisted by development of an aro-
matic pi-bonding system orthogonal to the azirine ring
plane. The sulfur would have access to either carbon or
nitrogen, producing the benzthiazole in one instance
and the benzisothiazole in the other. The literature is
unclear on the experimental basis for this mechanism. It
may only be Gilchrist’s extrapolation. Some indication
has been given that the conversion of the azirine to the
benzisothiazole has a thermal dependence. The litera-
ture, however, is also unclear about the extent of
thermal and/or light dependence of this step. A signifi-
cant thermal dependence may be an explanation for
why the extent of isomerization seems to show a ther-
mal dependence. In attempts to photoconvert ziprasi-
done on a large scale to the benzthiazole, elevated
temperatures (ca. 60°C) could be shown to push the
conversion further than at room temperature.
6
structures.
A search of the literature indicates that 5 has been
7
previously prepared. An appropriate test for identity
of the photoisomerization hypothesis—to compare the
HPLC and GC retention time behavior and mass spec-
tral properties—was performed with an authentic sam-
ple of 5, and indicated that the photochallenge product
of 3 was indeed 5. In addition, confirmation that
ziprasidone indeed photoisomerizes to the benzthiazole8
analog was achieved by comparison with authentic 2.
Although examples of photoisomerizations of
9
10
phenyloxazoles and benzoxazoles have been reported
in the literature, we find no reports of photoisomeriza-
11
tion of benzisothiazoles. Gilchrist discusses a mecha-
nism for photoisomerization involving an azirine
Scheme 1.

T. R. Sharp et al. / Tetrahedron Letters 44 (2003) 1559–1561
1561
Acknowledgements
detector, equipped with split injection port and electronic
pressure control. The column was a J&W Scientific DB-1
(100% poly(dimethylsiloxane)) bonded phase fused silica
capillary column (30 m×0.25 mm×0.25 m film thickness).
Injector and detector temperatures were set, respectively,
at 250 and 280°C. Oven temperature was held for 2 min
at 100°C, then increased linearly at 5°/min to 240°C, then
held at 240°C for 10 min. Helium carrier gas flow
through the column was approx. 1.3 mL/min (45 cm/s).
Injections were made by autosampler into a split injection
port with split ratio set at 30:1. The mass selective
detector was autotuned and mass calibrated using
perfluorotributylamine (FC-43) according to instrument
manufacturer’s criteria. The detector was operated in
repetitive scan mode, recording full scan mass spectra
from m/z 40 to m/z 450 every 1.9 s.
The authors thank J. A. Lowe III and H. R. Howard
for discussions during preparation of this manuscript.
References
1
2
. Seeger, T. F.; Seymour, P. A.; Schmidt, A. W.; Zorn, S.
H.; Shulz, D. W.; Lebel, L. A.; McLean, S.; Guanowsky,
V.; Howard, H. R.; Lowe, J. A., III; Heym, J. J. Pharma-
col. Expt. Therapeutics 1995, 275, 101–113.
. Howard, H. R.; Lowe, J. A., III; Seeger, T. F.; Seymour,
P. A.; Zorn, S. H.; Maloney, P. R.; Ewing, F. E.;
Newman, M. E.; Schmidt, A. W.; Furman, J. S.;
Robinson, G. L.; Jackson, E.; Johnson, C.; Morrone, J.
J. Med. Chem. 1996, 39, 143–148.
. LC–MS analyses were carried out on a Micromass
Instruments Quattro I triple quadrupole mass spectrome-
ter, coupled to a Hewlett–Packard 1050 modular HPLC
system. Reverse phase separations were carried out on a
Waters Puresil C18 column (4.6 mm×150 mm). Over the
course of a 60 min run, the mobile phase (1 mL/min) was
linearly changed from 10% methanol and 90% aqueous
.1% trifluoroacetic acid to 100% methanol. UV–vis
detection was carried out at 229 nm. The column was
thermostated at 30°C. Column effluent was split approx.
0:1 after passing through the UV–vis detector using a
Valco T-splitter, the smaller portion being presented to
the LC–MS.
. Photochallenge of compounds in solution was conducted
in a Rayonet Model RPR-200 photochemical reaction
chamber with fluorescent lamps. Samples were dissolved
in a mixed solvent of 60% (v/v) methanol and 40%
6
. Porter, Q. N. Mass Spectrometry of Heterocyclic Com-
pounds, 2nd ed.; New York: John Wiley & Sons, 1985;
pp. 896 ff and pp. 906 ff.
3
7
. Diouf, O.; Depreux, P.; Lesieur, D.; Poupaert, J. H.;
Caignard, D. H. Eur. J. Med. Chem. 1995, 30, 715–719.
. Compound 2 was synthesized by reacting one equivalent
of 2-chlorobenzthiazole and the alkylpiperizine derivative
of 6-chlorooxindole in isopropanol. Two equivalents of
triethylamine were added over the course of a ca. 24 h
reflux. The reaction only proceeded to ca. 45% comple-
tion using simple HPLC evaluation. The reaction was
quenched by addition of five equivalents of HCl. The
product was isolated by filtration, washed with iso-
8
0
1
propanol and H O, producing approximately 11% yield
4
2
of a material greater than 95% pure by simple HPLC
evaluation. The reaction was not optimized for yield.
. Maeda, M.; Kojima, M. J. Chem. Soc., Perkin Trans. 1
9
1977, 239–247.
distilled H O, at an initial concentration of 0.2 mg/mL,
2
1
1
0. Goeth, H.; Schmid, H. Chimia 1966, 20, 148.
1. Gilchrist, T. L. Heterocyclic Chemistry; Longman Scien-
tific & Technical Publishers: UK, 1985; p. 214.
and were contained in ordinary Pyrex glass containers.
. GC–MS experiments were conducted on a Hewlett–Pack-
ard 6890 Series gas chromatograph and mass selective
5