A new semisynthesis of paclitaxel from Baccatin III

Article abstract of DOI:10.1021/np990040k

A new method for the semisynthesis of paclitaxel (Taxol) from baccatin III via a dioxo-oxathiazolidine intermediate is reported.

Full text of DOI:10.1021/np990040k

1068
J . Nat. Prod. 1999, 62, 1068-1071
A New Sem isyn th esis of P a clita xel fr om Ba cca tin III
Erkan Baloglu and David G. I. Kingston*
Department of Chemistry, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061-0212
Received February 3, 1999
A new method for the semisynthesis of paclitaxel (Taxol) from baccatin III via a dioxo-oxathiazolidine
intermediate is reported.
Sch em e 1a
The important anticancer drug paclitaxel (Taxol) (1)1 was
originally isolated in low yield from the bark of the Pacific
Yew, Taxus brevifolia Nutt. (Taxaceae).2 Because of its
clinical importance, and because extraction on a large scale
from yew bark was perceived as a threat to the environ-
ment, a great deal of effort has gone into finding alternate
sources of this important substance. The major method
used at the present time for the commercial preparation
of paclitaxel is a semisynthetic route from a protected form
of baccatin III (2)3 which can readily be prepared from 10-
deacetyl baccatin III (10-DAB, 3), which itself can be
a
Key: (a) SOCl2, Et3N, PhH, 0-5 °C; 7, 68%; 8, 14%; 9, 0-3%; (b) NaIO4,
RuCl3, CCl4, CH3CN, H2O, 91%; (c) LiOH, H2O, THF, 92%.
III proceeds poorly and gives lower yields of paclitaxel,
sometimes coupled with epimerization at the 2′ position.8
We selected an oxathiazolidine derivative of the pacli-
taxel side chain for investigation, based on reports that this
ring system is stable to basic conditions but can be
hydrolyzed under acidic conditions.9 Treatment of the
methyl ester of the paclitaxel side chain (6) with thionyl
chloride in benzene gave a mixture of the two isomeric
2-oxo-1,2,3-oxathiazolidines 7 and 8 together with small
amounts of oxazoline 9, previously prepared as part of an
alternative protected side chain derivative.10 The use of
solvents more polar than benzene gave higher yields or
even exclusive formation of oxazoline 911 (Scheme 1).
isolated in significant quantities (1 g/Kg) from the renew-
able needles of the European Yew, Taxus baccata L.
(Taxaceae).4 Since the yield of 10-DAB is about 10 times
greater than the amount of paclitaxel that can be isolated
from the bark of T. brevifolia, its conversion to paclitaxel
is an attractive option.
A paclitaxel synthesis from 10-DAB requires the prepa-
ration of the N-benzoyl-â-phenylisoserine paclitaxel side
chain (4),5,6 its protection and coupling to a suitably
protected baccatin III derivative, and deprotection to the
final product. Several methods of coupling the paclitaxel
side chain to protected baccatin III have also been devel-
oped,5 but not all of them give high yields, and so we
decided to develop a new protected form of the paclitaxel
side chain. The more successful coupling methods have
relied on the formation of some sort of heterocyclic deriva-
tive of 4 to “tie back” the R-OH and â-NHCOPh groups and
thus reduce their steric bulk.7 It has been reported that if
this kind of linkage is not used, then acylation of baccatin
The isomeric nature of the two products 7 and 8 was
proved by oxidation of both compounds with ruthenium
chloride and periodate to the same 2,2-dioxo-1,2,3-oxathia-
zolidine derivative 10 (Scheme 1). The geometry of the
isomeric products was assigned by NMR spectrometry. It
is reported that in the case of a five-membered heterocyclic
ring the sulfoxide bond (SdO) deshields the ring substit-
uents that are cis to it.9 The chemical shift of H4 is further
downfield when it is cis to the SdO bond, and further
upfield when it is trans. The same trend applies to H5 as
well. Thus, the major product was assigned as 7, and the
secondary product as 8. Interestingly, the coupling constant
of the H4 and H5 protons of the major product (7) was found
to be 8.8 Hz and that of the corresponding protons of the
* To whom enquiries should be addressed: Tel.: (540) 231 6570. Fax:
(540) 231 7702. E-Mail: dkingston@vt.edu.
10.1021/np990040k CCC: $18.00
© 1999 American Chemical Society and American Society of Pharmacognosy
Published on Web 05/21/1999

Notes
J ournal of Natural Products, 1999, Vol. 62, No. 7 1069
Sch em e 3a
Sch em e 2a
a
Key: (a) DCC, DMAP, PhCH3, 75 °C, 20%.
Sch em e 4
a
Key: (a) DCC, DMAP, PhCH3, 75 °C, 30%; (b) HCl, EtOH.
secondary product (8) was found to be 2.4 Hz. It is
surprising that these values are so different, but it is
assumed that the difference is due to subtle differences in
the conformations of the five-membered rings in 7 and 8.
During these studies the major product (7) was found to
be sparingly soluble in ethanol. Thus, it was possible to
isolate 7 in pure form, without the need of further purifica-
tion, by washing the crude product with ethanol after
workup. The secondary product (8) was not very stable and
was hydrolyzed to starting material even on a TLC plate.
Sch em e 5
The next step was the hydrolysis of the methyl ester 7
to the corresponding carboxylic acid. Although several
hydrolysis conditions were tried, in all cases the sulfur
linkage was broken before the methyl ester was hydrolyzed,
contrary to literature reports that oxathiazolidines are
stable under basic conditions.9 We thus investigated the
use of the 2,2-dioxo-1,2,3-oxathiazolidine derivative 10,
which could readily be prepared by oxidation of 7 or of a
mixture of 7 and 8. Compound 10 was readily hydrolyzed
to the corresponding carboxylic acid 11 in the presence of
lithium hydroxide, and the stage was set for the coupling
of 11 with 7-(triethylsilyl)baccatin III (5) to give paclitaxel
(Scheme 2).
In an attempt to prepare the 2,2-dioxo-1,2,3-oxathiazo-
lidine 10 in one step from the paclitaxel side chain methyl
ester, ester 6 was treated with sulfuryl chloride and DMAP
in CH2Cl2. Again, the results were surprising, since the
cis-oxazoline 14 was the observed product (Scheme 5). The
same product was also obtained when the reaction condi-
tions were changed to sulfuryl chloride/pyridine or sulfuryl
chloride/triethylamine/benzene. The structure of product
14 was assigned with the help of NMR and mass spectral
data and by comparison with the trans-oxazoline 9. In
particular, the coupling constant of the H4 and H5 protons
(J ) 10.8 Hz) was consistent with that obtained by
molecular modeling and Karplus correlation12 for the cis
isomer. Formation of 14 presumably occurs by esterification
of the 2′-OH group with sulfuryl chloride, followed by
backside attack of the amide carbonyl group with displace-
ment of the chlorosulfate group. The cis-oxazoline 14 was
then converted to its corresponding acid 15 in the presence
of lithium hydroxide.
Coupling of 11 with 7-(triethylsilyl)baccatin III (5) was
carried out in the presence of dicyclohexylcarbodiimide
(DCC) and 4-(dimethylamino)pyridine (DMAP). Surpris-
ingly the product was not the expected dioxothiazolidine
derivative, but instead the previously prepared oxazoline
derivative 12. The structure of 12 was assigned by com-
parison of its 1H NMR spectrum with the corresponding
spectrum of the literature compound.10 The unexpected
product 12 was then converted to paclitaxel (1) by the
published procedure (Scheme 2).10
This surprising result prompted a short mechanistic
study of the coupling reaction. Compound 10 was subjected
to the coupling conditions in the absence of 7-(triethylsilyl)-
baccatin III. Conversion of the 2,2-dioxo-1,2,3-oxathiazo-
lidine derivative 10 to the trans-oxazoline 9 was observed
(Scheme 3), indicating that conversion of 11 to its oxazoline
analogue can occur in the absence of baccatin III. Presum-
ably attack of a nucleophile on the C-2 position of 10 gives
the intermediate 13 by an SN2 reaction, resulting in
inversion of configuration at C-2. This is then followed by
intramolecular attack of the lone pair electrons of the
carbonyl oxygen of the N-benzoyl group with concomitant
elimination of sulfur trioxide to give the trans oxazoline 9
with overall retention of configuration (Scheme 4).
Coupling of the cis oxazoline 15 with 7-triethylsilylbac-
catin III 5 gave the same trans-oxazoline ester 12 as did
the trans-oxazoline 910 or the dioxo-oxathiazolidine 10. The
2′-carbon thus undergoes epimerization during coupling to
give the more stable trans oxazoline product. This finding
means that a practical paclitaxel synthesis can be achieved
from a mixture of oxazolines epimeric at the 2′-position,
much as was previously reported for oxazolidine deriva-
tives.13
In summary, the chemistry described provides an alter-
nate to the previously reported routes for the semisynthesis
of paclitaxel. In addition, the transformation of the pacli-
taxel side chain to the trans-oxazoline 9 via the dioxo-
oxathiazolidine derivative 10 and to the cis-oxazoline 14

1070 J ournal of Natural Products, 1999, Vol. 62, No. 7
Notes
3-N-Ben zoyl-4-p h en yl-(4S,5R)-2,2-d ioxo-1,2,3-oxa th ia -
zolid in e Meth yl Ester (10). To a stirred solution of the
oxathiazolidine (7 or 8) in CH3CN/CCl4/H2O (1:1:2) solvent
system at room temperature was added an excess amount of
NaIO4 and a catalytic amount of RuCl3. The reaction mixture
was stirred for 45 min. and then filtered through sand/Celite/
silica gel. After workup in the usual way with EtOAc, the crude
product was isolated via PTLC (40% EtOAc/hexanes) in 91%
by treatment with sulfuryl chloride provides routes to both
oxazoline diastereomers from the same starting material,
and the epimerization of oxazoline 14 during coupling with
7-triethylsilylbaccatin III allows the use of either oxazoline
isomer for paclitaxel synthesis.
Exp er im en ta l Section
1
yield. The same product was obtained from both 7 and 8. H
Gen er a l Exp er im en ta l P r oced u r es. Chemicals were
obtained from Aldrich Chemical Co. and were used without
further purification, unless otherwise noted. Thionyl chloride
was obtained from Acros Chemical Co. All anhydrous reactions
were performed in oven-dried glassware under argon. Tet-
rahydrofuran (THF) was distilled over sodium/benzophenone,
dichloromethane was distilled over calcium hydride, and
toluene was distilled over sodium prior to use. All reactions
were monitored by E. Merck analytical thin-layer chromatog-
raphy (TLC) plates (silica gel 60 GF, aluminum back) and
analyzed with 254 nm UV light and/or vanillin/sulfuric acid
spray and/or iodine vapor. Silica gel for column chromatogra-
phy was purchased from E. Merck (230-400 mesh). Prepara-
tive thin-layer chromatography (PTLC) plates (silica gel 60
NMR (CDCl3, 400 MHz δ 7.38-7.83 (m, 10H), 6.03 (d, 1H, J
) 8.4 Hz), 5.16 (d, 1H, J ) 8.4 Hz), 3.87 (s, 3H); 13C NMR
(CDCl3, 400 MHz) δ 166.49, 164.73, 135.07, 133.41, 132.47,
129.56, 129.28, 128.71, 128.46, 127.25, 78.38, 62.98, 53.71;
HRFABMS m/z 361.0699 (calcd for C17H15NO6S, 361.0620);
LRFABMS (M + H)+ m/z 362.0.
3-N-Ben zoyl-4-p h en yl-(4S,5R)-2,2-d ioxo-1,2,3-oxa th ia -
zolid in e Ca r boxylic Acid (11). To a stirred solution of 40
mg (0.11 mmol) starting material (10) in THF (2 mL) was
added 500 µL of water, and the mixture was stirred for 5 min.
LiOH (3 equiv, 8 mg, 0.33 mmol) was then introduced, and
the reaction mixture was stirred for 45 min. When the TLC
showed the disappearance of starting material, the reaction
mixture was diluted with EtOAc, acidified with dilute HCl,
and the usual workup procedure was performed. The crude
product was applied on a PTLC plate (10% MeOH/CH2Cl2) in
1
GF) were purchased from Analtech. H and 13C NMR spectra
were obtained in CDCl3 or CD3OD on Varian Unity 400
spectrometer (operating at 399.951 MHz for 1H and 100.578
MHz for 13C) and were assigned by comparison of chemical
shifts and coupling constants with those of related compounds.
Chemical shifts were reported as δ values relative to tetra-
methylsilane (TMS) as internal reference, and coupling con-
stants were reported in Hz. FAB mass spectra were obtained
at the Nebraska Center for Mass Spectrometry, University of
Nebraska, and CI mass spectra were obtained in the Depart-
ment of Chemistry, Virginia Polytechnic Institute and State
University.
The phrase “worked-up in the usual way” refers to diluting
the reaction mixture with excess organic solvent, washing with
water and brine, drying over anhydrous sodium sulfate, and
evaporating the solvent in vacuo unless otherwise noted. The
methyl ester of the paclitaxel side chain and 7-(triethylsilyl)-
baccatin III14 were prepared following the procedures that are
reported in the literature, and NMR data of these compounds
were identical to those in the literature.
3-N-Ben zoyl-4-p h en yl-(4S,5R)-2-oxo-1,2,3-oxa th ia zoli-
d in e m eth yl ester s (7 a n d 8) a n d (4S,5R)-2,4-Dip h en yl-
5-(m eth oxyca r bon yl)-2-oxa zolin e (9). To a stirred solution
of (2S,3R)-N-benzoyl-3-phenylisoserine methyl ester (100 mg,
0.33 mmol) in anhydrous benzene (4 mL) under argon was
added triethylamine (5 equiv, 0.2 mL), and the mixture was
stirred for 5 min at room temperature. The reaction mixture
was then cooled to 3 °C where thionyl chloride (4 equiv, 1.336
mmol, 0.1 mL in 0.2 mL of benzene) diluted in benzene was
introduced dropwise over 15 min. TLC immediately showed
the formation of three new compounds, along with the starting
material; one of the products was the major product (7), while
one was of intermediate amount (8) and the third (9) was
either minor or absent, depending on the exact conditions.
After workup in the usual way with EtOAc and saturated
aqueous sodium bicarbonate, EtOH was added to the crude
product to yield the major ethanol-insoluble product 7 in pure
form (68%). The rest of the crude product was subjected to
PTLC purification (40% EtOAc/hexanes) to yield 8 (14%) and
9 (0-3%).
1
order to give the desired acid in 92% yield. H NMR (CDCl3,
400 MHz) δ 7.38-7.79 (m, 10H), 7.14 (bs, 1H), 6.03 (d, J )
8.0 Hz), 5.18 (d, 1H, J ) 8.0 Hz); 13C NMR (CDCl3, 400 MHz)
δ 167.60, 166.79, 135.06, 133.49, 132.31, 129.62, 129.29,
128.75, 128.43, 127.31, 78.16, 62.89.
Acyla tion of 7-Tr ieth ylsilylba cca tin III w ith Acid (11)
a n d Isola tion of P a clita xel Der iva tive 12. To a stirred
emulsion of the acid (11) in anhydrous toluene was added 1
equiv of DMAP and 4 equiv of DCC, and the mixture stirred
under argon for 5 min. 7-(Triethylsilyl)baccatin III (0.25 equiv)
was then introduced, and the reaction mixture was warmed
to 75 °C and stirred overnight. TLC showed the formation of
a new product. The workup was performed in the usual way
with EtOAc after the filtration of the reaction mixture through
Celite. The product (12) was isolated via PTLC (40% EtOAc/
hexane) in 30% yield. 1H NMR (CDCl3, 400 MHz) δ 7.37-8.23
(m, 15H), 6.42 (s, 1H, H-10), 6.18 (t, 1H, H-13), 5.68 (d, 1H, J
) 7.2 Hz, H-3′), 5.60 (d, 1H, J ) 6.8 Hz, H-2), 4.93-4.96 (m,
2H, H-2′,5), 4.50 (dd, 1H, H-7), 4.29 (d, 1H, J ) 8.4 Hz, H-20),
4.14 (d, 1H, J ) 8.4 Hz, H-20), 3.84 (d, 1H, J ) 6.8 Hz, H-3),
2.54 (m, 1H, H-6), 2.23-2.40 (m, 2H, H2-14), 2.16 (s, 3H, Me,
4-Ac), 2.08 (s, 3H, Me,10-Ac), 2.06 (t, 1H, H-6), 1.68 (s, 3H,
Me-18), 1.58 (s, 3H, Me-19), 1.24 (s, 3H, Me-17), 1.19 (s, 3H,
Me-16), 0.92 (t, 9H, 3 × Me, 7-TES), 0.56 (q, 6H, 3xCH2,
7-TES); HRFABMS (M + H)+ m/z 950.4166 (calcd for C53H64
NO13Si, 950.4147).
-
Con ver sion of 3-N-Ben zoyl-4-p h en yl-(4S,5R)-2,2-d ioxo-
1,2,3-oxa th ia zolid in e Meth yl Ester (10) to (4S,5R)-2,4-
Diph en yl-5-(m eth oxycar bon yl)-2-oxazolin e (9). To a stirred
emulsion of 10 mg of compound 10 in anhydrous toluene was
added 1 equiv of DMAP and 4 equiv of DCC, and the mixture
was stirred under argon for 5 min. TLC showed the formation
of a new product. The workup was performed in the usual way
with EtOAc after the filtration of the reaction mixture through
Celite. The product (9) was isolated via PTLC (40%EtOAc/
hexane) in 20% yield. Spectral data were identical to those
recorded.
(4S ,5S )-2,4-Dip h e n yl-5-(m e t h oxyca r b on yl)-2-oxa zo-
lin e (14). To a stirred solution of (2S,3R)-N-benzoyl-3-phe-
nylisoserine methyl ester (25 mg, 0.08 mmol) in CH2Cl2 was
added DMAP (7 equiv). The reaction mixture was then cooled
to 0 °C, and sulfuryl chloride (3 equiv) was introduced
dropwise. The product 14 was isolated in 65% yield after
workup in the usual way and purification via PTLC. 1H NMR
(CDCl3, 400 MHz) δ 7.22-8.11 (m, 10H), 5.74 (d, 1H, J ) 10.8
Hz), 5.38 (d, 1H, J ) 10.8 Hz), 3.198 (s, 3H); 13C NMR (CDCl3,
400 MHz) δ 168.45, 164.73, 136.86, 131.92, 128.66, 128.41,
128.11, 128.06, 127.70, 126.69, 81.04, 73.45, 51.54; CIMS (M
+ H)+ m/z 282.
Ma jor p r od u ct (7): 1H NMR (CDCl3, 400 MHz) δ 7.35-
7.74 (m, 10H), 5.69 (d, 1H, J ) 8.8 Hz), 5.66 (d, 1H, J ) 8.8
Hz), 3.84 (s, 3H); 13C NMR (CDCl3, 400 MHz) δ 167.65, 166.69,
135.517, 134.045, 132.71, 129.05, 128.91, 128.76, 127.99, 86.14,
63.40, 53.34; HRFABMS m/z 345.0746 (calcd for C17H15NO5S,
345.0671); LRFABMS (M + H)+ m/z 346.
1
In ter m ed ia te p r od u ct (8): H NMR (CDCl3, 400 MHz) δ
7.35-7.67 (m, 10H), 6.224 (d, 1H, J ) 2.4 Hz), 5.273 (d, 1H, J
) 2.4 Hz), 3.89 (s, 3H).
Min or p r od u ct (9): 1H NMR, 13C NMR and mass spectral
data identical with literature data.11 CIMS (M + H)+ m/z 282.

Notes
(4S,5S)-2,4-Dip h en yl-2-oxa zolin eca r boxylic Acid (15).
J ournal of Natural Products, 1999, Vol. 62, No. 7 1071
no longer accurately describes the chemical substance 1, the name
paclitaxel will be used whenever the chemical substance taxol is
intended.
To a stirred solution of ester 15 (40 mg, 0.11 mmol) in EtOH
(2 mL) was added water (500 µL) and the mixture stirred for
5 min. LiOH (8 mg, 0.33 mmol, 3 equiv) was then introduced,
and the reaction mixture was stirred for 45 min. When TLC
showed the disappearance of starting material, the reaction
mixture was diluted with EtOAc, acidified with dilute HCl,
and the usual workup procedure was performed. The crude
product was purified by PTLC (10% MeOH/CH2Cl2) to give the
(2) Wani, M. C.; Taylor, H. L.; Wall, M. E.; Coggon, P.; McPhail, A. T. J .
Am. Chem. Soc. 1971, 93, 2325-2327.
(3) Della Casa de Marcano, D. P.; Halsall, T. G. J . Chem. Soc., Chem.
Commun. 1975, 365-366.
(4) Denis, J . N.; Greene, A. E.; Gue´nard, D.; Gue´ritte-Voegelein, F.;
Mangatal, L.; Potier, P. J . Am. Chem. Soc. 1988, 110, 5917-5919.
(5) Boge, T. C.; Georg, G. I. In Enantioselective Synthesis of â-Amino
Acids; J uaristi, E., Ed.; Wiley-VCH: New York, 1997; pp 1-43.
(6) Denis, J . N.; Correa, A.; Greene, A. E. J . Org. Chem. 1990, 55, 1957-
1959.
1
acid 15 in 92% yield. H NMR (CDCl3, 400 MHz) δ 7.22-8.11
(m, 10H), 5.74 (d, 1H, J ) 11.2 Hz) 5.38 (d, 1H, J ) 11.2 Hz);
(7) Wuts, P. Curr. Opin. Drug Discov. Dev. 1998, 1, 329-337.
(8) Commerc¸on, A.; Be´zard, D.; Bernard, F.; Bourzat, J . D. Tetrahedron
Lett. 1992, 33, 5185-5188.
CIMS (M + H)+ m/z 268.
(9) Deyrup, J . A.; Moyer, C. L. J . Org. Chem. 1969, 34, 175-179 and
Ack n ow led gm en t. We are grateful for financial support
for this work by grant CA 69571 to D.G.I.K. from the National
Cancer Institute and to the Nebraska Center for mass spec-
trometry for FAB mass spectra.
references therein.
(10) Kingston, D. G. I.; Chaudhary, A. G.; Gunatilaka, A. A. L.; Middleton,
M. L. Tetrahedron Lett. 1994, 35, 4483-4484.
(11) Gou, D.; Liu, Y.; Chen, C. J . Org. Chem. 1993, 58, 1287-1289.
(12) Lambert, J . B.; Shurvell, H. F.; Lightner, D. A.; Cooks, R. G. Organic
Structural Spectroscopy; Prentice Hall: Englewood Cliffs, NJ , 1998;
pp 72-73.
Refer en ces a n d Notes
(1) The name taxol was assigned to compound 1 by its discoverers.2
Several years after this assignment of the name, the name Taxol was
registered as a trademark by Bristol-Myers Squibb Pharmaceuticals
for its clinical formulation of taxol, and the generic name paclitaxel
was substituted for the compound of structure 1. To avoid infringing
on Bristol-Myers Squibb’s trademark, and because the name Taxol
(13) Denis, J .-N.; Kanazawa, A. M.; Greene, A. E. Tetrahedron Lett. 1994,
35, 105-108.
(14) Samaranayake, G.; Neidigh, K. A.; Kingston, D. G. I. J . Nat. Prod.
1993, 56, 884-898.
NP990040K