A concise synthesis of (-)-oseltamivir

Article abstract of DOI:10.1002/anie.200800282

(Chemical Equation Presented) Tackling the supply problem: A short and efficient synthesis of (-)-oseltamivir has been developed which requires eight steps from commercially available starting materialand proceeds with an overall yield of 30%. Key transformations include a novel palladium-catalyzed asymmetric allylic alkylation reaction (Pd-AAA, see scheme) as well as a chemo-, regio-, and stereoselective aziridination reaction. Phth=phthaloyl.

Full text of DOI:10.1002/anie.200800282

Angewandte
Chemie
DOI: 10.1002/anie.200800282
Drug Synthesis
A Concise Synthesis of (À)-Oseltamivir**
Barry M. Trost* and Ting Zhang
Since its first isolation from a human patient a decade ago, the
avian flu virus H5N1 has caused pandemic life-threatening
influenza outbreaks in multiple areas around the world.[1]
Efforts to find an orally active neuraminidase inhibitor were
fruitful upon the discovery of Tamiflu ((À)-oseltamivir
phosphate, 1·H3PO4) by Gilead Sciences.[2] However, the
daily dose of 150 mg per patient requires large quantities of
the drug to meet the global need, especially that from the
developing world.[3] Currently, tamiflu is marketed by Roche
and is prepared by a semisynthetic approach starting from
(À)-shikimic acid, which is still a limited resource given the
Scheme 1. Retrosynthetic analysis of (À)-oseltamivir. PG=protecting
massive demand.[4] Therefore, the development of alternative
synthetic approaches, which start from simple materials, has
drawn extensive attention from the chemical community.
Despite numerous reports on the development of chemical
syntheses of tamiflu from simple starting materials,[5] to date,
enhanced efficiency in its synthesis remains a challenge. We
took on the challenge and herein report our success in the
development of a concise and efficient synthesis of (À)-
oseltamivir (1).
group.
Commercially available lactone 5 was chosen as the
starting material on account of its six-membered carbon
backbone, its relative stereochemistry, and its potential to
deracemize. A palladium-catalyzed AAA process was pro-
posed to open the lactone and thus install an amino-group
equivalent onto the ring enantioselectively. Different from
common lactone-opening procedures, where nucleophiles
would usually attack C7, the Pd-AAA process would result
in a nucleophilic addition at C2 or C4 (Scheme 2). Uniquely
for 5, with the tethered carboxylate as a leaving group, the
requisite carboxylic ester moiety could also be revealed by the
Pd-AAA reaction. A proof of principle for this approach was
obtained from our previous work with carbon-centered
nucleophiles.[7] Surprisingly and in contrast to that work, the
use of imides as nucleophiles (for example, NHBoc2, NHCbz2,
NH(CHO)2, and phthalimide; Boc = tert-butoxycarbonyl,
Cbz = benzyloxycarbonyl), with or without the addition of
bases, failed to give any Pd-AAA reaction product. Revisiting
the reaction mechanism gave rise to a few conclusions and
postulates:[7a] a) the full or partial deprotonation of the
nitrogen-centered nucleophile was necessary to increase its
nucleophilicity; b) the approach of an anionic nucleophile N
towards intermediate E could induce repulsion between the
negative charge on the nitrogen atom and that on the
carboxylate group thus inhibiting the nucleophilic addition
(Scheme 2); c) the conversion of intermediate E back into
starting material 5 is dominant over the nucleophilic attack of
the imide anion. Therefore, to overcome the repulsion, the
As shown by our retrosynthetic analysis (Scheme 1), we
envisioned the installation of the alkoxy group towards the
end of the synthetic sequence, with a direct opening of
aziridine 2 as the most efficient option.[2,4,5] In our approach, a
direct but unprecedented chemo-, regio-, and stereoselective
aziridination of diene 3 was envisioned. Furthermore, diene 3
could be quickly obtained from intermediate 4 by using
sulfoxide elimination chemistry that was developed by our
research group.[6] Finally, a novel palladium-catalyzed asym-
metric allylic alkylation (Pd-AAA) reaction in which a
nitrogen-centered nucleophile, was used to directly open
racemic cis-lactone 5 was proposed to set the requisite
stereochemistry for the entire synthesis.
[*] Prof. B. M. Trost, T. Zhang
Department of Chemistry
Stanford University
Stanford, CA 94305 (USA)
Fax: (+1)650-725-0002
E-mail: bmtrost@stanford.edu
Homepage: http://www.stanford.edu/group/bmtrost/
[**] We thank the National Institutes of Health (GM 033049) and the
National Science Foundation for their generous support of our
programs. T.Z. thanks Novartis for a graduate fellowship. Mass
spectra were provided, in part, by the Mass Spectrometry Regional
Center of the University of California, San Francisco. Palladium salts
were generously supplied by Johnson-Matthey. We are indebted to
Dr. Kiran Guthikonda, David N. Zalatan, and Prof. Justin Du Bois for
valuable discussions and donation of chemicals for the aziridina-
tion step. We thank Gilead Sciences for a reference sample of
Tamiflu.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Scheme 2. Palladium-catalyzed asymmetric allylic alkylation. L=ligand.
Angew. Chem. Int. Ed. 2008, 47, 3759 –3761
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3759

Communications
carboxylate anion must be trapped as it is formed without
diminishing the nucleophilicity of the nitrogen-centered
nucleophile. On the basis of the above analysis, we proposed
using a trimethylsilyl (TMS) group, whose oxophilicity would
result in a selective trapping of the carboxylate oxygen atom
over the nitrogen atom. Furthermore, the TMS group could
alternatively be introduced attached to the nitrogen-centered
nucleophile, wherein silyl transfer should capture the carbox-
ylate moiety while simultaneously revealing the imide anion.
Consequently, we chose commercially available TMS-phthal-
imide as the nitrogen-centered nucleophile. Utilizing another
advantage of introducing the TMS group, that is, its suscept-
ibility to acid hydrolysis, esterification could be carried out in
the same pot, thus revealing the carboxylic ethyl ester
functionality. To our delight, and in stark contrast to using
phthalimide itself, the use of [(h3-C3H5PdCl)2]and the Trost
ligand (R,R)-13 as the catalyst, and with TMS-phthalimide as
the nucleophile, the racemic lactone 5 was opened and
desymmetrized. The resulting TMS-carboxylate was con-
verted into the ethyl ester with ethanol in situ to afford 6 in
84% yield and 98% ee in one pot (Scheme 3).
Sulfenylation of 6 with PhSSO2Ph and KHMDS as the
base gave an approximate 1:1 diasteromeric mixture of a-
thioester 7 in 94% yield.[6] Oxidation of this diastereomeric
mixture with mCPBA resulted in the corresponding sulfoxide
intermediate, which underwent thermal elimination in situ to
give diene 8[8] as well as a small amount of the corresponding
1,4-diene. DBU was used as an additive to slightly increase
the regioselectivity in the elimination process. After chroma-
tography, 8 was obtained as a 10:1 regioisomeric mixture in
85% yield, and was carried through as a mixture to the next
step.
An unexpected difficulty arose in the selective aziridina-
tion step. Copper catalysts, despite their popularity in
common aziridination reactions,[9] showed virtually no selec-
tivity between the a,b double bond and the g,d double bond of
8, despite the extensive screening of ligands. In this case,
changing Cu to a larger metal such as Ag or Au resulted in low
reactivity, although only one isomer, the desired g,d-aziridine
9, was obtatined.[10] Recently Guthikonda, Du Bois, and co-
workers reported a Rh-catalyzed intermolecular aziridination
reaction, which highlights the generation of a nitrene species
in situ and with the olefin being the limiting reagent.[11] When
their exact reagents and reaction conditions were applied to
our substrate 8, only the ring-opened product of the presumed
vinylaziridine was obtained.[12] Thus, finding a stablizing yet
easy to remove protecting group on the aziridine was
important. Aryl sulfonyl groups were shown to be problem-
atic: tosyl and benzenesulfonyl groups were difficult to
remove later in the synthesis, whereas p-nitrobenzenesul-
fonyl-protected aziridine was unstable and readily decom-
posed either after a prolonged reaction time or during
chromatography. Finally, the 2-(trimethylsilyl)ethanesulfonyl
(SES) group was found to be the best choice. Further
optimization of the reaction showed that [Rh2(esp)2](bis-
[rhodium(a,a,a’,a’-tetramethyl-1,3-benzenedipropionic
Scheme 3. Synthesis of (À)-oseltamivir. Reagents and conditions:
a) 2.5 mol% [(h3-C3H5PdCl)2], 7.5 mol% (R,R)-13, 1.5 equiv trimethyl-
silylphthalimide, THF, 408C, then TsOH·H2O, EtOH, reflux, 84%,
98% ee; b) 1.5 equiv KHMDS, 1.8 equiv PhSSO2Ph, THF, À788C to RT,
94%; c) 1 equiv mCPBA, 2 equiv NaHCO3, 08C, then 1 equiv DBU,
608C, toluene, 85%; d) 2 mol% 14, 1.1 equiv SESNH2, 1.3 equiv PhI-
(O2CCMe3)2, 2.3 equiv MgO, PhCl, 08C to RT, 86%; e) 1.5 equiv
BF3·Et2O, 3-pentanol, 758C, 65%; f) 2 equiv DMAP, 20 equiv pyridine,
Ac2O, MW, 1508C, 1 h, 84%; g) 2 equiv TBAF, THF, RT, 95%;
h) 5 equiv NH2NH2, EtOH, 688C, quant. DBU=1,8-diazabicyclo-
[5.4.0]undec-7-ene, DMAP=4-dimethylaminopyridine, HMDS=
1,1,1,3,3,3-hexamethyldisilazane, mCPBA=meta-chloroperbenzoic
acid, MW=microwave, Phth=phthaloyl, SES=2-(trimethylsilyl)etha-
nesulfonyl, TBAF=tetra-n-butylammonium fluoride, Ts=para-toluene-
sulfonyl.
(CF3CONH)4]for the aziridination reactions of 8 with
PhI(OAc)2 and SESNH2 as the nitrene source. Moreover, by
changing the oxidant from PhI(OAc)2 to PhI(O2CCMe3)2 an
even better conversion was achieved. Most satisfyingly, g,d-
aziridine 9 was observed as the only product in all cases. In the
end, with chlorobenzene as the solvent,[14] PhI(O2CCMe3)2
and SESNH2 as the nitrene source, [Rh2(esp)2]as the catalyst,
and MgO as the base and dessicant, the single isomer 9 was
obtained in 86% yield from the diene mixture 8.[15]
acid)], 14),[13] a new catalyst for C H insertion reactions
The remaining transformations were straightforward.
Opening aziridine 9 with BF3·Et2O in 3-pentanol gave 10 in
65% yield.[16] Acylation of 10 in Ac2O with DMAP and
À
reported by Du Bois and co-workers, gave better conversion
compared to [Rh2(O2CCPh3)4], [Rh2(O2CCMe3)4], and [Rh2-
3760 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 3759 –3761

Angewandte
Chemie
Bunt, PhD Thesis, Stanford University (US), 1996; c) B. M.
Trost, J.-P. Surivet, Angew. Chem. 2000, 112, 3252 – 3254; Angew.
Chem. Int. Ed. 2000, 39, 3122 – 3124.
pyridine was accelerated by a microwave reactor and yielded
11 (84%) at 1508C after 1 hour.[17] Once acylated, the SES
protecting group could be easily removed by treatment with
TBAF to afford 12 in 95% yield.[17] Finally, heating 12 in
ethanolic hydrazine yielded 1 in quantitative yield.
In summary, we have successfully synthesized (À)-oselta-
mivir. The synthesis features a novel palladium-catalyzed
deracemizing lactone reaction with a nitrogen-centered
nucleophile as well as an unprecedented Rh-catalyzed
chemo-, regio-, and stereoselective direct aziridination reac-
tion on an electron-deficient conjugated diene system. This
azide-free synthesis required just eight steps from commer-
cially available starting materials and proceeded with an
overall yield of 30%,[18] which to our knowledge represents
the shortest synthesis to date.
[8]A similar diene intermediate, bearing an NH-Boc group rather
than an N-phthaloyl group, was reported by Corey and co-
workers (Ref. [5a]). In their approach, the synthesis of the diene
required eight steps, whereas in our approach, 8 was prepared in
only three steps. From the conjugate diene, they applied a two-
step haloamination–SN2 displacement sequence to form the
desired aziridine while we developed a direct aziridination
process.
[9]For a recent review on metal-catalyzed aziridination reactions,
see: J. A. Halfen, Curr. Org. Chem. 2005, 9, 657 – 669.
[10]a) For a recent report on Ag-catalyzed aziridination reactions,
see: Y. Cui, C. He, J. Am. Chem. Soc. 2003, 125, 16202 – 16203;
for a recent report on Au-catalyzed aziridination reactions, see:
Z. Li, X. Ding, C. He, J. Org. Chem. 2006, 71, 5876 – 5880.
[11]a) K. Guthikonda, J. Du Bois, J. Am. Chem. Soc. 2002, 124,
13672 – 13673; b) K. Guthikonda, P. M. Wehn, B. J. Caliando, J.
Du Bois, Tetrahedron 2006, 62, 11331 – 11342.
Received: January 19, 2008
Published online: April 10, 2008
[12]See the Supporting Information for details.
[13]a) C. G. Espino, K. W. Fiori, M. Kim, J. Du Bois, J. Am. Chem.
Soc. 2004, 126, 15378 – 15379; b) K. W. Fiori, J. Du Bois, J. Am.
Chem. Soc. 2007, 129, 562 – 568.
[14]Optimization studies showed that instead of dry chlorobenzene,
dry isopropyl acetate (a non-halogenated solvent) could give the
same result under otherwise identical reaction conditions, but
Keywords: aziridination · lactones · palladium · rhodium ·
total synthesis
.
[1]For information on avian flu and H5N1 virus, visit: http://
www.nature.com/avianflu/index.html.
with
a slight inconvenience during the workup; see the
[2]Contributions from Gilead Sciences: a) C. U. Kim, W. Lew,
M. A. Williams, H. Liu, L. Zhang, S. Swaminathan, N. Bischof-
berger, M. S. Chen, D. B. Mendel, C. Y. Tai, G. Laver, R. C.
Stevens, J. Am. Chem. Soc. 1997, 119, 681 – 690; b) J. C. Rohloff,
K. M. Kent, M. J. Postich, M. W. Becker, H. H. Chapman, D. E.
Kelly, W. Lew, M. S. Louie, L. R. McGee, E. J. Prisbe, L. M.
Schultze, R. H. Yu, L. Zhang, J. Org. Chem. 1998, 63, 4545 –
4550; c) N. W. Bischofberger, C. U. Kim, W. Lew, H. Liu, M. A.
Williams, US Patent 5763483, 1998.
Supporting Information for details. J. Chan, K. D. Baucom,
J. A. Murry, J. Am. Chem. Soc. 2007, 129, 14106 – 14107.
[15]See the Supporting Information for selected optimized reaction
conditions for aziridination.
[16]The modest yield of 10 resulted from the formation of two by-
products, 10a and 10b (the stereochemistry of 10b has not been
verified), which are separable from 10 and from each other. It
[3]a) For a recent review on the supply problem and syntheses of
oseltamivir, see: V. Farina, J. D. Brown, Angew. Chem. 2006, 118,
7488 – 7492; Angew. Chem. Int. Ed. 2006, 45, 7330 – 7334; b) for a
recent microreview on synthetic strategies for oseltamivir, see:
M. Shibasaki, M. Kanai, Eur. J. Org. Chem. 2008, DOI: 10.10027/
ejoc:200800033.
[4]For a review that includes the extensive studies at Roche, see: S.
Abrecht, P. Harrington, H. Iding, M. Karpf, R. Trussardi, B.
Wirz, U. Zutter, Chimia 2004, 58, 621 – 629.
turned out to be extremely difficult to convert 10a into 10
through alkylation reactions. However, converting 10a into 9
was successful. An unoptimized reaction condition is shown
below. DIAD = diisopropylazodicarboxylate.
[5]a) Y.-Y. Yeung, S. Hong, E. J. Corey, J. Am. Chem. Soc. 2006,
128, 6310 – 6311; b) Y. Fukuta, T. Mita, N. Fukuda, M. Kanai, M.
Shibasaki, J. Am. Chem. Soc. 2006, 128, 6312 – 6313; c) X. Cong,
Z.-J. Yao, J. Org. Chem. 2006, 71, 5365 – 5368; d) T. Mita, N.
Fukuda, F. X. Roca, M. Kanai, M. Shibasaki, Org. Lett. 2007, 9,
259 – 262; e) K. Yamatsugu, S. Kamijo, Y. Suto, M. Kanai, M.
Shibasaki, Tetrahedron Lett. 2007, 48, 1403 – 1406; f) K. M.
Bromfield, H. Graden, D. P. Hagberg, T. Olsson, N. Kann,
Chem. Commun. 2007, 3183 – 3185; g) N. Satoh, T. Akiba, S.
Yokoshima, T. Fukuyama. Angew. Chem. 2007, 119, 5836–5838;
Angew. Chem. Int. Ed. 2007, 46, 5734 – 5736; Angew. Chem. Int.
Ed. 2007, 46, 5734 – 5736; h) J.-J. Shie, J.-M. Fang, S.-Y. Wang, K.-
C. Tsai, Y.-S. E. Cheng, A.-S. Yang, S.-C. Hsiao, C.-Y. Su, C.-H.
Wong, J. Am. Chem. Soc. 2007, 129, 11892 – 11893.
[17]At room temperature 11 existed as an approximately 4:1
rotameric mixture in CDCl3. NMR spectroscopy studies in
C6D6 showed the two sets of rotameric signals of 11 to gradually
coalesce as the temperature increased and became one set of
resonances upon reaching 808C. The same effect was observed
for 12. See the Supporting Information for details.
[6]B. M. Trost, G. S. Massiot, J. Am. Chem. Soc. 1977, 99, 4405 –
4412.
[7]For mechanistic discussions on the Pd-AAA opening of lactones
and its success with carbon nucleophiles, see: a) B. M. Trost,
R. C. Bunt, J. Am. Chem. Soc. 1996, 118, 235 – 236; b) R. C.
[18]The overall yield would be higher if 10a were to be recycled as
indicated in Ref. [16].
Angew. Chem. Int. Ed. 2008, 47, 3759 –3761
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3761