Development of a concise synthesis of (-)-oseltamivir (Tamiflu)

Article abstract of DOI:10.1002/chem.201003454

We report a full account of our work towards the development of an eight-step synthesis of anti-influenza drug (-)-oseltamivir (Tamiflu) from commercially available starting materials. The final synthetic route proceeds with an overall yield of 30 %. Key transformations include a novel palladium-catalyzed asymmetric allylic alkylation reaction (Pd-AAA) as well as a rhodium-catalyzed chemo-, regio-, and stereoselective aziridination reaction.

Full text of DOI:10.1002/chem.201003454

DOI: 10.1002/chem.201003454
Development of a Concise Synthesis of (ꢀ)-Oseltamivir (Tamiflu)
Barry M. Trost*[a] and Ting Zhang[b]
Abstract: We report a full account of our work towards the development of an
eight-step synthesis of anti-influenza drug (ꢀ)-oseltamivir (Tamiflu) from commer-
cially available starting materials. The final synthetic route proceeds with an over-
all yield of 30%. Key transformations include a novel palladium-catalyzed asym-
metric allylic alkylation reaction (Pd-AAA) as well as a rhodium-catalyzed
chemo-, regio-, and stereoselective aziridination reaction.
Keywords: heterogeneous catalysis ·
palladium catalysis rhodium
catalysis · Tamiflu · total synthesis
·
Introduction
sive attention from the chemical community.[4] Our group,
among many others, have contributed to the development of
alternative chemical synthetic routes and completed an
eight-step azide-free synthesis of (ꢀ)-oseltamivir in 2008,
which represents the shortest synthesis to date.[4i] This article
elaborates in detail our development of the aforementioned
synthetic route.
Tamiflu (oseltamivir phosphate, 1, see below), developed by
Gilead Sciences, is an orally active neuraminidase inhibitor,
which had been widely used for the treatment of H5N1 in-
fluenza as well as a recent outbreak of H1N1 flu.[1] In order
to be effective in treating infected patients, Tamiflu must be
administered to influenza patients no later than 36–48 h
after the manifestation of the symptoms. With a daily dose
of 150 mg per patient, sufficient stockpiles of Tamiflu are
necessary in times of a flu pandemic, as it is both a therapy
and a preventive agent.
Results and Discussion
Retrosynthetic analysis: In our first-generation retrosynthet-
ic analysis, the alkyl chain of (ꢀ)-oseltamivir 2 would be in-
stalled by etherification of the allylic alchohol 3, which
could come from isomerization of epoxide 4. The expoxide
would in turn result from an allyl amine 5. To install the
second amino group, aziridination was proposed starting
from alkene intermediate 7. Finally, a novel Pd-AAA reac-
tion is envisioned, where a nitrogen-centered nucleophile is
to be used to directly open racemic cis-lactone 8 and to set
the requisite stereochemistry for the entire synthesis.
(Scheme 1)
Such high public demands raise the challenge of providing
large quantities of tamiflu to meet the worldwide need.[2]
Currently, tamiflu is marketed by Roche and is prepared via
a semisynthetic approach starting from (ꢀ)-shikimic acid,
which is still a limited resource given the massive demand.[3]
Therefore, the development of alternative synthetic ap-
proaches that start from simple materials has drawn exten-
Pd-AAA opening of lactone: Commercially available lac-
tone 8 was chosen as the starting material because of its six-
membered carbon backbone, the cis-disubstituted stereo-
chemistry, and its potential to undergo a deracemization.[5]
Scheme 2 illustrates how a Pd-AAA deracemization process
differs from the normal nucleophilic opening of lactone 8:
Instead of a normal nucleophilic attack at the carbonyl
carbon C4 to generate ring-opened product 9, a Pd-AAA
process results in nucleophilic attack at the ether carbon C1,
generating an alternative ring-opened product 10, forming a
new bond on the six-membered ring.
When a chiral ligand is used, this Pd-AAA opening of lac-
tone process also generates enantiomeric excess. Thus, lac-
tone 8 is not only opened by the incoming nucleophile, but
also deracemized. The deracemization mechanism is shown
in Scheme 3. A chiral palladium complex reacts with a race-
mic mixture of 8 to form a pseudo-meso p-allyl intermediate
[a] Prof. B. M. Trost
Department of Chemistry, Stanford University
Stanford, CA 94305-5080 (USA)
Fax : (+1)650-725-0002
E-mail: bmtrost@stanford.edu
[b] Dr. T. Zhang
Department of Medicinal Chemistry, Merck Research Laboratories
PO Box 2000, RY800-B310 Rahway, NJ 07065 (USA)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201003454.
3630
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3630 – 3643

FULL PAPER
HNACHTNUTRGENNG(U Cbz)2 has also shown
promising results in related Pd-
AAA reactions.[10,11] However,
as shown in Table 1, no reaction
was observed in the Pd-AAA
opening lactone reaction at var-
ious temperatures (Table 1,
entry 9, 10, and 11). Another
known Pd-AAA nucleophile,
NaN
proceed with or without phase-
transfer reagent THAB
(Table 1, entry 12 and 13).
Phthalimide, either in salt
(CHO)2,[12] also failed to
ACHTUNGTRENNUNG
Scheme 1. Retrosynthetic analysis. PG=protecting group.
form[13] or neutral form,[14] has shown to function as good
nucleophiles in AAA reactions. First, sodium phthalimide
was tested but showed zero reactivity in Pd-AAA reaction
with 8. Considering the charge repulsion between the nega-
tively-charged carboxylate leaving group and the incoming
sodium phthalimide may have prevented the addition
(Table 1, entry 14),[15] neutral phthalimide was tested as the
nucleophile with base as an additive to increase the nucleo-
philicity. However, no reaction occurred with either buf-
fered system (CsOAc/HOAc) or with the addition of excess
base (Table 1, entry 15 and 16).
Scheme 2. Different product outcome between normal nucleophilic open-
ing of lactone and Pd-AAA deracemizing lactone.
As nitrogen-centered nucleophiles were not effective for
the desired Pd-AAA reaction for lactone 8, a subsequent
strategy involving ring-opening of the lactone prior to the
Pd-AAA was envisioned. Lactone 8 was transformed in two
steps into the ring-opened 15. Interestingly, in contrast to
the reactions with lactone 8, the Pd-AAA reaction with 15
proceeded excellently with phthalimide as the nucleophile
in the presence of Cs2CO3, affording 13 in high yield and
enantiomeric excess (Scheme 4).
A rationale for the different reactivities between 8 and 15
is provided in Scheme 4. Upon oxidative insertion of the Pd0
Scheme 3. Mechanism of Pd-AAA deracemizing lactone.
with the attached carboxylate as a leaving group. Chiral
ligand biased nucleophilic attack yields an enantiomerically
enriched product 10.
Our group had demonstrated that stabilized carbon nucle-
ophiles such as nitromethane and malonate can deracemize
the lactone with high enantioselectivity (Table 1, entry 1 and
2).[6,7] Using similar strategies, several stabilized nitrogen nu-
cleophiles that were successfully used in previous Pd-AAA
reactions were examined to deracemize lactone 8 and install
the first amino group at C1. Literature known compound
[8,9]
HN
G
was examined as the initial nucleophile. Using
standard Pd-AAA conditions, no reaction was observed
(Table 1, entry 3). We hypothesized that HN(Boc)2 was not
sufficiently nucleophilic, therefore, variety of bases
AHCTUNGTRENNUNG
a
(Table 1, entry 4, 5 and 6) was tested, however, still no reac-
tion was observed. Then more reactive Pd-source, [{Pd-
ACHTUNGTRENNUNG
Scheme 4. Pd-AAA with phthalimide. MEIM = N-Methylimidazole.
Chem. Eur. J. 2011, 17, 3630 – 3643
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3631

B. M. Trost and T. Zhang
Table 1. Opening of lactone with imide nucleophiles.
Entry
Nuc
Conditions
Results[a]
[6]
1
2
3
4
5
6
7
8
CH3NO2
2 mol% [Pd
2.5 mol% [{Pd
G
74%, 99% ee
57%, 85% ee
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
malonate[7]
G
[8,9]
HN
HN
HN
HN
HN
HN
HN
HN
HN
N
2.5 mol% [Pd
2.5 mol% [Pd
2.5 mol% [Pd
2.5 mol% [Pd
2.5 mol% [{Pd
2.5 mol% [{Pd
2.5 mol% [{Pd
2.5 mol% [{Pd
2.5 mol% [{Pd
2 mol% [{Pd
2 mol% [{Pd
2.5 mol% [{Pd
2.5 mol% [{Pd
2ACHTUNGTRENNUNG
C
R
G
R
E
[10,11]
9
E
10
11
12
13
14
15
16
G
R
[12]
NaN
NaN
U
N
E
NaPhth[13c]
HPhth[14]
HPhth
A
2 mol% [{Pd
[a] NR=no reaction. [b] THAB=tetrahexylammonium bromide. [c] NaPhth=sodium phthalimide.
catalyst into substrate 15, an intermediate involving a
pseudo-meso PdII p-allyl complex bearing an attached car-
boxylate ester is formed (16). The chiral ligand biases the
nucleophilic attack by the deprotonated phthalimide and
leads to enantiomerically enriched 13. However, when using
lactone 8 as the substrate, it generates an intermediate 12
with an attached carboxylate anion. Presumably, the ap-
proach of an anionic phthalimide towards the Pd p-allyl
complex is disfavored by repulsion between the negative
charges on the nitrogen atom and the carboxylate group. In
addition, it is also possible that the intramolecular reaction
with the carboxylate is favored over the intermolecular reac-
tion with the phthalimide, therefore the reverse reaction of
12 back to 8 is faster than the formation of 13. We hypothe-
sized that, to overcome the repulsion and the intramolecular
Scheme 5. Opening of lactone with TMS-phthalimide.
trapping of 12, the carboxylate anion of 12 must be trapped
prior to the attack of the phthalimide without diminishing
the nucleophilicity of the phthalimide.
boxylate was in situ converted to the ethyl ester affording
the desired compound 13. The initial conditions afforded 13
in 24% yield and 96% ee (Scheme 6). This approach instal-
led both the amino group and the ester functionality in one
step and established the desired stereochemistry for the
entire synthesis. (R,R)-Ligand was randomly chosen at this
point and was later confirmed to lead to the desired abso-
lute stereochemistry.
Further optimization of this reaction is shown in Table 2.
Our goal was to improve the yield without using large
excess of reagents. Therefore, higher reaction concentration,
temperature, and various esterification conditions were ex-
amined (Table 2, entry1–3). The enantioselectivity was opti-
mal at 0.2m. Unfortunately, the yield was still unacceptable.
TMS-phthalimide was stored and handled in the glove box.
Despite this precaution and the use of rigorous drying tech-
niques, the TMS-phthalimide was found to undergo compet-
itive hydrolysis to the unreactive phthalimide during the
Based upon the above analysis, we proposed using a tri-
methylsilyl (TMS) group, whose oxophilicity would result in
a selective trapping of the carboxylate oxygen over the ni-
trogen (Scheme 5). Considering the simple addition of exter-
nal silylating reagent did not facilitate the reaction (Table 1,
entry 6), we considered bringing in the TMS group attached
to the nitrogen nucleophile, wherein silyl transfer should
capture the carboxylate and simultaneously reveal the imide
nucleophile (17). Therefore, we chose commercially avail-
able TMS-phthalimide. Utilizing another advantage of the
TMS group, that is, its susceptibility to acid hydrolysis,
esterification could be carried out in the same pot, forming
the carboxylic ethyl ester.
To our delight and in stark contrast to phthalimide itself,
with [{PdACHTUNGTRENNUNG(C3H5)Cl}2] and ligand (R,R)-11 as the catalyst, and
TMS-phthalimide as the nucleophile, the racemic lactone 8
was opened and desymmetrized, and the resulted TMS-car-
3632
www.chemeurj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3630 – 3643

Synthesis of (ꢀ)-Oseltamivir (Tamiflu)
FULL PAPER
it is generated in situ and subsequently trapped by the olefin
species via an ene reaction.[17] However, under literature
conditions the ene reaction with 13 failed to proceed, even
when heated to 808C (Scheme 7). Starting material 13 was
recovered in both cases. We hypothesized that the alkene of
13 is not electron-rich enough to react with 17.
Scheme 6. Initial results of the opening of lactone with TMS-phthalimide.
AAA reaction. This undesired
side reaction explained the low
conversion. To rectify this ob-
servation, we first decided to
add 1 equiv external silylating
reagent BSA to maintain the
effective concentration of TMS-
Scheme 7. Attempted second amination via ene reaction.
phthalimide in the reaction.
Satisfyingly, such method led to
full conversion in the Pd-AAA
step, however, the following esterification became problem-
atic with BSA present (Table 2, entry 4). Finally, we decided
to use excess amount of TMS-phthalimide. As little as
1.5 equiv of the nucleophile was sufficient enough to lead to
full conversion to 13 with an isolated yield of 80% (Table 2,
entry 5 and 6). The yield improved further when scaled up
(Table 2, entry 7).
Alternatively, the second amino group could be installed
via a reaction between olefin 13 and a highly reactive
AcNH+ equivalent.[18] A straightforward way to generate
AcNH+ equivalence is reacting N-haloacetamide such as 19
with a silver salt, such as AgBF4. A non-nucleophilic base,
2,6-di-tert-butylpyridine, was added to deprotonate the allyl-
ic proton of intermediate 21 thus leading to the formation
of allylic amine 20. Unfortu-
nately, although the AgBr pre-
cipitation was observed under
Table 2. Optimization of opening of lactone with TMS-phthalimide.
the reaction conditions, none of
the desired amination product
was detected (Scheme 8).
Acknowledging the challenge
of installing the second amino
Entry
X
[m] Acid
Results
1
2
3
4
1.1 0.1 conc. H2SO4, RT
24%, 96% ee
32%, 97% ee
47%, 94% ee
full conv. in Pd-AAA; low conv. in hydrolysis
step
80%, 97% ee (0.3 mmmol scale)
80%, 98% ee (3 mmol scale)
84%, 98% ee (20 mmol scale)
group on an electron deficient
double bond via ene or acyl
cation strategy, we decided to
investigate aziridination reac-
tions as originally planned, as
they are more developed ami-
nation reactions for electron
deficient double bonds.[19] Grat-
ifyingly, under standard aziridi-
1
1
1
0.2 conc. H2SO4, 408C
0.3 10 mol% TsOH·H2O, 40 8C
0.2 1 equiv BSA, 10 mol% TsOH·H2O,
408C
5
6
7
2
0.2 10 mol% TsOH·H2O, 40 8C
1.5 0.2 10 mol% TsOH·H2O, 40 8C
1.5 0.2 10 mol% TsOH·H2O, 40 8C
In summary, the first amino group was successfully instal-
led via a novel Pd-AAA deracemizing lactone reaction with
a unique N-centered nucleophile in high yield and ee, and in
the same pot the desired ethyl ester was formed. This reac-
tion set up the absolute stereochemistry for the entire syn-
thesis.
nation conditions, the desired aziridine 22 was formed in
69–74% yield as a single diastereoisomer (Scheme 9). One
major drawback of the reaction was the requirement of
large excesses of PhI=NTs and high catalyst loading. Previ-
ously, our group had reported that CuACTHNUTRGNEUNG(IPr)Cl is an effective
aziridination catalyst.[20] However, in this case, it led to no
reaction for substrate 13.
Installation of the second amino group: With 13 in hand, we
turned to install the second amino group onto the ring.
Before following our synthetic plan of utilizing aziridination
reaction to form aziridine 6, we made an attempt of access-
ing allylic amine 5 directly from 13. Initially, ene reaction
was examined. It is known that an ene-type reaction be-
tween an alkene and a nitroso compound affords an allylic
amine.[16] It is proposed that the reaction occurs via a six-
membered transition state (Scheme 7). Nitroso compound
17 is highly reactive and thus cannot be isolated. Therefore,
In summary, the second amine was installed via a copper-
catalyzed diastereoselective aziridination in good yield. The
next step in the total synthesis of oseltamivir was regioselec-
tive ring opening of the aziridine 22.
Further conversions: In order to obtain the desired allylic
amine 23 from aziridine 22, an elimination reaction was pro-
posed (Table 3). In the literature, allyl amines can be ac-
cessed from aziridines utilizing either R2NLi or RLi to pro-
mote the desired eliminative rearrangement.[21,22] However,
Chem. Eur. J. 2011, 17, 3630 – 3643
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3633

B. M. Trost and T. Zhang
acidic indium triflate, at an ele-
vated temperature, elimination
product 23 was observed for
the first time, although irrepro-
ducible, in 23% yield (Table 3,
entry 11). The desired product
23 coeluted with another prod-
uct, which is believed to be 25.
In addition to 23 and 25, the re-
action also afforded 30% yield
of 24. We hypothesized that
water was acting as the nucleo-
phile for the opening of aziri-
Scheme 8. Attempted amination with acyl cation equivalent.
dine 22 to form 24, in which
case, addition of molecular
sieves might prevent this unde-
sired reaction. However, when 3 ꢁ MS was added, no reac-
tion was observed after 24 h (Table 3, entry 12).
Taking the good result from Table 3, entry 10, we tried to
convert 24 into 23 (Scheme 10). Various one-step dehydra-
tion conditions were investigated, however, no desired prod-
uct was observed. Therefore, a stepwise procedure was pur-
sued. Interestingly, after transforming 24 to its correspond-
ing triflate, DBU induced elimination afforded aziridine 22
instead of the desired allylic amine 23, presumably due to
adjacent group participation.
Scheme 9. Copper-catalyzed aziridination.
in our case, it would lead to the undesired regioisomer due
to chelating effects of such bases. Nonetheless, when we did
perform the reaction with LDA, it simply led to decomposi-
tion even at ꢀ788C (Table 3, entry 1). Additionally, nucleo-
philic bases would attack the ester functionality, which
should also be avoided. Therefore, our initial studies focused
on non-nucleophilic, non-coordinating bases and E2 type
elimination conditions to promote the formation of 23.
When 1 equiv of potassium tert-butoxide (KOtBu) was used,
no desired product was observed (Table 3, entry 2). Adding
excess amount of potassium tert-butoxide resulted in messy
results, where attack of the ester group by the base was ob-
served (Table 3, entry 3). Superbase LIDAKOR[21,23] was
used with the intention to minimize the nucleophilic attack
at the ester. However, either no reaction was observed in
hexanes or decomposition occurred in THF (Table 3, entry 4
and 5). Changing to organic base DBU gave no reaction
(Table 3, entry 6). Next, instead of only using base to pro-
mote the elimination, Lewis acid-non-nucleophilic base
combinations were used to generate a “push–pull” effect.
However, still no reaction was observed (Table 3, entry 7
and 8). In order to force the reaction under the “push-pull”
conditions to occur, chlorobenzene was used as the solvent
in order to increase the substrate solubility as well as the re-
action temperature (Table 3, entry 9). Oxophilic lanthanide
Scheme 10. Attempted dehydration of alcohol 24.
At the same time we were studying the one-step E2 rear-
rangement, we also started investigating a two-step se-
quence to access 23 as a backup plan (Scheme 11). Because
of the observed adjacent group participation indicated in
Scheme 10, we proposed to open the aziridine utilizing a nu-
cleophile that would not participate in the intramolecular
SN2 reactions. Therefore, PhSeꢀ was chosen to generate in-
termediate 26, followed by oxidative thermal elimination to
afford the allylic amine 23. One major challenge of this
strategy was the regioselectivity of the aziridine-opening
step. Electronically, the equatorial attack to afford desired
product 26 is favored (Scheme 11, path a; similar result has
been observed in Table 3, entry 10). However, stereoelec-
tronically, the axial attack of the selenide to afford the unde-
sired regioisomer 27 is favored (Scheme 11, path b), leading
holmium triflate (HoACHTUNTRGNEUNG(OTf)3) was examined as a Lewis acid
to activate the sulfonyl group on the aziridine. However, no
reaction occurred even at 1008C. As lanthanides did not
assist the aziridine opening, main group Lewis acids were
studied. Gratifyingly, AlACTHNUTRGNE(UNG OTf)3 gave nearly quantitative
yield of aziridine opening product 24, which could potential-
ly lead to allylic amine 23 via a dehydration reaction
(Table 3, entry 10). Delightfully, by switching to less Lewis
3634
www.chemeurj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3630 – 3643

Synthesis of (ꢀ)-Oseltamivir (Tamiflu)
FULL PAPER
Table 3. Attempted elimination conditions.
Entry Conditions
Notes
1
2
3
4
5
6
7
8
2 equiv LDA, THF, ꢀ788C
decomposition to baseline by TLC
no desired product
1 equiv KOtBu, THF, 0!ꢀ408C
5 equiv KOtBu, THF, RT
messy, tBuOꢀ attacked ester
NR (substrate not dissolving)
decomposition to baseline by TLC
NR
1 equiv LIDAKOR,[a] hexanes, 08C!RT
1 equiv LIDAKOR, THF, ꢀ788C
1.5 equiv DBU, toluene, reflux
50 mol% In
20 mol% In
U
NR
no desired reaction; In
the reaction.
ACHTUNGRTEN(NNUG OTf)3 reacted with proton sponge to form insoluble solids in
9
HoACHTUNGTRENNUNG
10
Al
N
11
12
In
In
ACHTUNGTRENNUNG(OTf)3, 2,6-di-tert-butylpyridine, chlorobenzene, 1008C NMR yield: 23% 23, 30% 24, 30% 25
(OTf)3, 2,6-di-tert-butylpyridine, chlorobenzene, 3 ꢁ
NR, recovered starting material
MS, 1008C
[a] LIDAKOR=LDA + tBuOK. [b] NR=no reaction. [c] NMR yields were determined by using 1,3,5-trimethoxybenzene as an internal standard.
to allylic amine 28. A mixture of regioisomers may also be
generated due to the competition between the two pathways
(as already shown in Table 3, entry 11). With this in mind,
we started to look into the reactivity and the selectivity of
the aziridine opening reaction.
As shown in Scheme 12, the aziridine opening reaction
was carried out in ethanol, however, only 10% conversion
was observed; this was presumably due to the low solubility
of 22 in ethanol. The conversion increased to 62% when
THF was added as a co-solvent to dissolve the starting ma-
terial. When TFA was added to activate the Ts-aziridine, full
conversion was achieved with a 90% isolated yield for the
opening of aziridine. Unfortunately, the subsequent elimina-
tion did not occur under the standard oxidative conditions.
Later it was discovered, the reaction yielded 27 rather than
the desired 26, which explained the difficulty of elimination
due to lack of aligned proton. This intriguing result indicates
that the axial attack (Scheme 11, path b) predominates
under those reaction conditions, suggesting that the torsional
strain in the transition state is dictating the regioselectivity
when a nucleophile is used, leading to the formation of un-
desired 27 (Scheme 11 and 12).
Revised synthetic plan: In observation of the difficulties in
the direct E2-type elimination from 22, we decided to revise
Scheme 11. Planned two-step sequence to form allylic amine 23.
our synthetic plan. As shown in Table 3, AlACTHGNUTRENNU(G OTf)3 and In-
Chem. Eur. J. 2011, 17, 3630 – 3643
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3635

B. M. Trost and T. Zhang
DMSO led to no improvement
in the yield of 29 (Table 4,
entry 4). Addition of desiccant
Na2SO4 into the reaction in-
creased the yield of 29 to 58%
(Table 4, entry 5). Finally, while
keeping the catalyst concentra-
tion unchanged, increasing the
substrate concentration, and
therefore lowering the water
concentration in the reaction
resulted in 54% yield of 29
(Table 4, entry 6).
Having ketone 29 in hand,
enone 30 was successfully
formed by IBX oxidation
(Scheme 14).[24] Fortunately, by
Scheme 12. Opening of aziridine 22 with selenide.
Table 4. Oxidative opening of aziridine.
ACHTUNGTRENNUNG(OTf)3 assisted the ring-opening of aziridine 22 with small
nucleophiles such as water to afford the desired regioisomer
24. We decided to take advantage of this result. With a care-
ful choice of the nucleophile employed to open 22 and a
slight alteration of our synthetic plan, the target molecule 2
may be accessed with equal efficiency as the first generation
synthetic plan. As shown in Scheme 13, oxidative opening of
aziridine 22 with DMSO may be followed by further oxida-
tion to the enone 30. Subsequent selective reduction would
afford allylic alcohol 31. Finally, an etherification reaction
would complete the synthesis of oseltamivir.
Entry
1
Conditions
Results
50 mol% In
0.1m DMSO, 808C
(OTf)3,
no reaction
2
3
4
5
6
50 mol% In
0.1m DMSO, MW, 2008C, 30 min
50 mol% In(OTf)3 (wet),
0.1m DMSO, 1208C
50 mol% In(OTf)3 (dry),
0.1m DMSO (distilled), 1208C
50 mol% In(OTf)3 (dry),
0.1m DMSO, 1 equiv Na2SO4, 1208C
5 mol% In(OTf)3 (dry),
1m in DMSO, 1208C
A
aromatization
E
46% 29, 50% 24
43% 29, 42% 24
58% 29, 34% 24
54% 29, 22% 24
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
Scheme 13. Revised synthetic plan.
Oxidative opening of aziridine 22 was examined and the
results are summarized in Table 4. At 808C with InACTHNUTRGNEUNG(OTf)3
Scheme 14. Further oxidation to enone.
and DMSO, no reaction occurred (Table 4, entry 1). Increas-
ing the reaction temperature to 2008C and applying a micro-
wave reactor resulted in aromatization of the starting mate-
rial (Table 4, entry 2). Further screening showed that at
1208C the desired ketone 29 could be obtained in moderate
yields together with a significant amount of alcohol 24
(Table 4, entry 3). Water was considered to be the source of
the formation of 24. Therefore, several methods were ap-
plied in order to minimize waterꢂs presence in the reaction.
doubling the equivalence of IBX, alcohol 24 could be direct-
ly oxidized to enone 30, albeit in a modest yield.
Since the ring opening of aziridine 22 and subsequent oxi-
dation both took place in DMSO, we investigated a one-pot
procedure for the two steps (Scheme 15). By adding all re-
agents at once, enone 30 was formed in a modest 30% yield
and was accompanied by 49% yield of the unoxidized alco-
hol 24. When IBX was added after the opening of aziridine
Utilization of ultra dry InACTHNUTRGNE(NUG OTf)3 as well as freshly distilled
3636
www.chemeurj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3630 – 3643

Synthesis of (ꢀ)-Oseltamivir (Tamiflu)
FULL PAPER
pentanol as the nucleophile. Such method should lead to 2
in the same number of steps as the planned mesylation-dis-
placement protocol. Encouragingly, in the presence of
BF3·Et2O, reaction between 34 and 3-pentanol gave a 26%
yield of the desired product 32, along with a substantial
amount of hydrated product 35 (Scheme 18). Attempted re-
ductive etherification of 35 to form 32 gave only 9%
yield.[25]
In addition to the low yield in the alkylation step, later it
was found the removal of the tosyl group on 32 was rather
difficult. Both direct deprotection of tosyl[26] and the acyla-
Scheme 15. One-step procedure for transforming aziridine to enone.
but still in the same pot, the
yield of 30 can be improved to
52%.
Moving forward in the syn-
thesis, enone 30 was reduced to
allylic alcohol 31 as a single ste-
reoisomer
in
76%
yield
(Scheme 16). The relative ste-
reochemistry was determined
by comparison of the NMR
spectra with the other isomer
35 obtained later in the synthe-
sis.
Scheme 18. Opening of vinyl aziridine with 3-pentanol.
tion[27] of tosylamine to facilitate the tosyl removal failed. In
addition to the deprotection problem, this synthetic route
suffered several low yielding steps after the Pd-AAA step
(Scheme 19). Therefore, we decided to investigate an alter-
native synthetic plan to address these problems.
Scheme 16. Reduction of enone 30 to allylic alcohol 31.
Final synthetic plan: As shown in Scheme 19, from the aziri-
dine 22 to the vinyl aziridine 34, many oxidation stage
changes were made only to install one double bond as a net
result. It was against our initial goal of developing an effi-
cient synthesis. Therefore, a major revision of the previous
synthetic route is needed to access the vinyl aziridine 34
from the amino ester intermediate 13 more efficiently.
To avoid the multiple oxidation stage changes involved in
Further transformation of 31 to 2 showed interesting re-
sults (Scheme 17): The planned mesylation of the allylic al-
cohol 31 in order to install the alkyl chain via an SN2 dis-
placement led to the formation of aziridine 34 rather than
the mesylate, despite the cis-configuration of the sulfon-
amide and hydroxy groups.
At this point, we decided to take advantage of the forma-
tion of 34 and install the 3-pentoxy side chain via a Lewis
acid catalyzed opening of vinyl aziridine reaction with 3-
ꢀ
installing the C C double bond in 34, we proposed to first
take advantage of the existing double bond in 13 to intro-
duce a second double bond in conjugation (38), and then in-
stall the aziridine by a selective aziridination reaction. In the
mean time, we sought alternative protecting groups, in the
hope of improving the yields of the aziridination and open-
ing of the aziridine, as well as easy deprotection. According-
ly, a final revision of our synthetic plan is shown in
Scheme 20.
First, we attempted to oxidatively form 38 directly from
13. Similar strategy to form a,b-unsaturated system from
carbonyl compound has been reported for ketones and was
also demonstrated in our synthetic study in Scheme 14.[24]
However, under similar or more rigorous conditions, the for-
mation of 38 was not observed (Scheme 21).
Scheme 17. Attempted installation of alkyl fragment.
Chem. Eur. J. 2011, 17, 3630 – 3643
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3637

B. M. Trost and T. Zhang
this case, it led to satisfying sul-
fenylation of 13 (Scheme 23).
KHMDS was found to be the
optimal base after careful
screening. Sulfenylation of 13
with PhSSO2Ph and KHMDS
yielded an approximately 1:1
diastereomeric mixture of b-ke-
tosulfide 39 in 94% yield
(Scheme 23). The sulfide was
oxidized to the sulfoxide and a
subsequent thermal elimination
occurred upon refluxing in tolu-
ene. The elimination afforded
diene 38[29] and 40 in 5.8:1 ratio.
Lower temperature gave higher
ratio of 38 with a lower conver-
sion. Further investigation re-
vealed that one diastereomer of
Scheme 19. Overview of the early synthetic route.
Scheme 23. Sulfenylation-elimination.
Scheme 20. Final synthetic plan. PG=protecting group.
39 spontaneously eliminated after oxidation to give a 11:1
regioselectivity favoring 38 (Scheme 24). Unfortunately, the
other diastereomer of 39, upon oxidation, was stable up to
room temperature. When heated to 608C, a 2.7:1 ratio of
Scheme 21. Attempted oxidative installation of conjugation.
Hence, a stepwise strategy was undertaken. Starting with
22, a-selenation followed by oxidative thermal elimination
should afford diene 38. However, the a-selenation only af-
forded about 10% NMR yield of 37 (Scheme 22).
Scheme 24. Difference in elimination of the two a-thioester isomers.
38:40 was obtained. After optimization, DBU was found ef-
fective in increasing the regioselectivity in the elimination
step (Scheme 25). After chromatography, 38 and 40 was ob-
tained as a 10:1 regioisomeric mixture in 85% yield
(Scheme 25). This mixture was carried on to the next step.
With diene 38 in hand, we would like to install the second
amino group via a selective aziridination reaction. Consider-
ing the difficulty of removing the tosyl group in the previous
synthetic route, this time we wanted to carefully choose a
protecting group on the nitrogen.
Scheme 22. a-Selenation.
Alternatively,
a highly reactive sulfenylating reagent
phenyl benzenethiosulfonate (PhSSO2Ph) discovered by our
group has been successfully utilized in a-sulfenylation of ke-
tones and esters.[28] Compared with PhSeCl, the reactions
with it are cleaner, less odoriferous and much less toxic. In
3638
www.chemeurj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3630 – 3643

Synthesis of (ꢀ)-Oseltamivir (Tamiflu)
FULL PAPER
Delightfully, diphenylsulfilimine, a good nucleophile for
1,4-conjugate addition,[32] reacted with model compound 42
via 1,6-conjugate addition to yield the desired aziridine
product 43 as a single regioisomer (Scheme 27). Unfortu-
nately, the same conditions on real system 38 led to exclu-
sive addition to the carbonyl group of the phthalimide func-
tionality.
Scheme 25. Optimized elimiation conditions.
Ideally, we wanted to bring in the acyl-protected amine
directly as it is present in the final product. But considering
the difficulties of performing a metal catalyzed aziridination
of acyl nitrene, a 1,6-conjugate addition-elimination process
was proposed instead to install an acyl aziridine on the
double bond distal to the ester.[30,31] Unfortunately, utilizing
O-acetyl acetamide as nucleophile, and copper thiophene-
carboxylate (CuTC) as the catalyst, no reaction occurred
with reaction temperatures up to 408C (Scheme 26).
Scheme 27. 1,6-Conjugate addition on model compound. TMG=1,1,3,3-
tetramethylguanidine.
Considering the difficulties in
conjugate 1,6-addition, a selec-
tive aziridination was proposed
using copper as catalyst.[33]
First, we tried to find a good
protecting group on the aziri-
dine.
4-Nitrobenzenesulfonyl
group is known for its ease of
removal. However, it led to de-
Scheme 26. Planned 1,6-conjugate addition.
composition in the aziridination
step. A more stabilized, yet rel-
In order to develop conditions for the conjugate addition,
a model compound 42 was studied under a variety of Lewis
Acid catalyzed 1,4-conjugate addition reaction conditions
(Table 5, entry 1). No reaction was observed under all condi-
tions. Since Lewis acidic conjugate addition didn’t work, we
turned our attention to basic conditions. Commonly used
conjugate addition reagents lithium anion and cuprates gave
no reaction with 42 (Table 5, entry 2 and 3). Using strongly
basic sodium hydride along with heating still led to no reac-
tion (Table 5, entry 4). Lastly, suspecting the a-protons of
the acyl group may be deprotonated under basic conditions,
we changed the acyl amine to a Boc protected amine, and
switched the acetate leaving group to a more active phos-
phate leaving group. However, such changes did not give
improved results (Table 5, entry 5).
atively easy-to-remove protecting group, benzenesulfonyl
group, was then quickly examined. Unfortunately, standard
benzenesulfonyl group removal condition led to decomposi-
tion. Therefore, finding a sulfonyl group which deprotected
under mild conditions was necessary for the synthesis. 2-
(Trimethylsilyl)ethanesulfonyl (SES) group was known for
its ease of removal by fluoride sources and thus was chosen
to be the protecting group on the aziridine. A variety of con-
ditions and catalysts were examined for the aziridination of
38 with SESNH2 (Table 6).
An unexpected difficulty arose in the regio- and stereose-
lective aziridination of 38. Extensive conditions were
screened; selected examples are shown in Table 6. Copper
catalysts, which were commonly used in aziridination reac-
tions, showed virtually no selectivity between the a,b- and
the g,d-double bonds of 38 and
formed an inseparable mixture
of 44 and 45. This lack of selec-
tivity was seen throughout ex-
Table 5. Attempted 1,6-conjugate addition on model compound.[a]
tensive screening of ligands (se-
lected results: Table 6, entry 1
and 2). Additionally, changing
copper for larger metals such as
Ag or Au, resulted in low reac-
tivity though excellent regiose-
lectivity of the desired g,d-aziri-
Entry
Conditions
Results
NR[c]
NR
NR
NR
NR
1
2
3
4
5
AcNHOAc, Cs2CO3, LA,[b] CH3CN, RT!808C
nBuLi, AcNHOAc, THF, ꢀ408C!RT,
A
dine 44 was obtatined (Selected
AcNHOAc, NaH, DMF, RT!908C
results: Table 6, entry 3 and
BocNHOP(O)
E
4).[34] Guthikonda and Du Bois
[a] PG=protecting group. [b] LA=[Pd
G
E
N
ACTHNUGTERN(NUNG OTf)3, or CuACHTUNGTRENN(UNG OTf)3. [c] NR=
no reaction.
reported
a
rhodium-catalyzed
3639
Chem. Eur. J. 2011, 17, 3630 – 3643
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org

B. M. Trost and T. Zhang
Table 6. Selected conditions for the aziridination of diene 38.
Entry
Catalyst
Conditions
Results[b]
1
2
3
4
5
6
7
8
10 mol% [Cu
10 mol% [Cu
10 mol% [Ag
10 mol% [Au
U
PhI=O; 3 ꢁ MS; CH3CN; RT
PhI=O; 3 ꢁ MS; CH3CN; 08C
PhI=O; 3 ꢁ MS; CH3CN; 08C!RT
84%; 44/45 1.7:1
ACHTUNGTRENNUNG
C
E
ACHTUNGTRENNUNG
R
PhI
PhI
PhI
(OAc)2; MgO; C6H5Cl; 08C!RT
PhI
(OAc)2; MgO; C6H5Cl; 08C!RT
PhI
(OAc)2; MgO; C6H5Cl; 08C!RT
PhI
(OAc)2; MgO; C6H5Cl; 08C!RT
PhI
(OPiv)2; MgO; C6H5Cl; 08C!RT
PhI
(OPiv)2; MgO; isopropyl acetate; 08C!RT
(OAc)2; 4 ꢁ MS; CH3CN; 80–508C
ACHTUNGTRENNUNG
2 mol% [Rh
5 mol% [Rh
5 mol% [Rh
5 mol% [Rh
2 mol% [Rh
2 mol% [Rh
2 mol% [Rh
N
A
ACHTUNGTRENNUNG
R
ACHTUNGTRENNUNG
E
ACHTUNGTRENNUNG
G
ACHTUNGTRENNUNG
9
10
11
N
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
A
ACHTUNGTRENNUNG
1
[a] The stereochemistry of 45 is undetermined. Compound 44 and 45 are inseparable on column. [b] Yields in parentheses were determined by H NMR
with 1,3,5-trimethoxybezene as an internal standard. [c] IPr=1,3-bis(2,6-diisopropylphenyl)imidazol-2-ylidene. [d] tBu3tPy=4,4’,4’’-tri-tert-butyl-2,2’;6’,2’’-
terpyridine. [e] Conversion was accompanied by decomposition, resulting in irreproducible results. [f] Dry isopropyl acetate (IPAC), a non-halogenated
solvent, gave the same result as dry chlorobenzene.
intermolecular aziridination reaction, highlighting in situ
generation of a nitrene species. In these reactions the olefin
is the limiting reagent.[35] Screening of Rh-catalysts revealed
ꢀ
erated the formation of 46 relative to 47 (Table 7, entry 3).
Weaker Lewis acids, such as In(OTf)3 or Cu(OTf)2 also
G
ACHTUNGTRENNUNG
slowed down the formation of 46, and thus increased the
yield of the undesired 47 (Table 7, entry 4 and 5). Entry 1–5
in Table 6 were performed when 44 still contained 45 as an
impurity and yields were calculated based on the assay of
44. Later, with pure 44 in hand, a third product with an un-
confirmed structure 48 was isolated (Table 7, entry 6–12).
Therefore 46, 47 and 48 nicely account for all the mass bal-
ance of this reaction. Lowering the substrate concentration
in order to increase the concentration of the nucleophile 3-
pentanol did not improve the yield (Table 7, entry 6 and 7).
Adding molecular sieves to minimize waterꢂs presence also
did not reduce the formation of 47 (Table 7, entry 8). Prefor-
mation of the 3-pentyl borate and subjection of it to the re-
action conditions led to no conversion (Table 7, entry 9).
However, thoroughly drying 44 slightly improved the yield
of 46 (Table 7, entry 10). Adding 1 equiv N,O-bis[trimethyl-
silyl]acetamide (BSA) to chemically remove any residual or
newly generated water afforded a lower yield of 46 (Table 7,
entry 11). Eventually, after premixing BSA with dry 44 fol-
lowed by full removal of all volatiles, the aziridine-opening
gave a 65% isolated yield of 46 (Table 7, entry 12).
Du Boisꢂs C H insertion catalyst, [Rh2ACTHUNTRGNE(UGN esp)2] (bis-
ACHTUNGTRENNUNG[rhodium(a,a,a’,a’-tetramethyl-1,3-benzenedipropionic
acid)]),[36] gave higher conversions than [Rh
[Rh2A(O2CCMe3)4] or [Rh2A(CF3CONH)4] (Table 6, entry 5–8).
2ACHTUNGRTNE(NUNG O2CCPh3)4],
C
CHTUNGTRENNUNG
The catalyst loading could be lowered to 2 mol% without
decay in yield (Table 6, entry 9). Moreover, changing the ox-
idant from PhIACHTUNGTRENNUNG(OAc)2 to PhIACHTUNTGREN(NUNG OPiv)2 afforded even better
conversion to 44 (Table 6, entry 10).[37] The environmentally
benign solvent isopropyl acetate could be used, replacing
chlorobenzene, and leading to similar yield and selectivity,
however, with a slightly inconvenient workup. Most satisfy-
ingly, g,d-aziridine 44 was observed as the exclusive product
in all Rh-catalyzed cases. Finally, in chlorobenzene, using
PhIACHTUNGTRENNUNG(O2CCMe3)2 and SESNH2, 2 mol% Rh2ACHTUTGNREN(NUGN esp)2, and MgO
as the base and desiccant, an 86% isolated yield of a single
isomer 44 was obtained from the diene mixture 38
(Scheme 28). Replacing chlorobenzene with isopropyl ace-
tate gave the same result (entry 11).
As the formation of 47 could not be avoided, conditions
to convert 47 to 46 were studied. Unfortunately, various at-
tempts to directly convert 47 to 46 failed. However, under
Mitsunobu conditions, we were able to convert compound
47 back to 44 which would allow the recycling of 44
(Scheme 29).
Scheme 28. Optimized aziridination condition.
With 44 in hand, similar opening of aziridine as shown in
Scheme 18 was performed with good regio- and diastereose-
lectivity (Table 7). Similar to tosyl aziridine, the alcohol 47
The next step in the synthesis of 2 was to deprotect the
SES group. However, direct removal of the SES group was
unsuccessful under a variety of literature conditions. There-
fore, we decided to acylate the SES-amino group prior to
the deprotection. Optimization of the acylation step is sum-
marized in Table 8. Under standard acylation conditions, no
reaction occurred at room temperature even with large ex-
was
a major byproduct in this ring-opening reaction
(Table 7, entry 1). At lower temperatures, 46 was formed
more slowly than 47, resulting in less 46 formation (Table 7,
entry 2). Running the reaction at higher temperature accel-
3640
www.chemeurj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3630 – 3643

Synthesis of (ꢀ)-Oseltamivir (Tamiflu)
FULL PAPER
Table 7. Opening of vinyl aziridine.
Conclusion
In summary, we have accom-
plished an azide-free eight-step
synthesis of (ꢀ)-oseltamivir
from commercially available
materials with 30% overall
yield. Key transformations in-
Entry Conditions
Yield
volve
a
novel Pd-catalyzed
1
2
3
4
5
6
1.5 equiv BF3·Et2O, 0.15m in 3-pentanol, RT
1.2 equiv BF3·Et2O, 0.15m in 3-pentanol, ꢀ208C
1.5 equiv BF3·Et2O, 0.15m in 3-pentanol, 758C
1.5 equiv In
10 mol% Cu
1.5 equiv BF3·Et2O, 0.15m in 3-pentanol, 758C
1.5 equiv BF3·Et2O, 0.05m in 3-pentanol, 708C
1.5 equiv BF3·Et2O, 0.05m in 3-pentanol, 3 ꢁ MS, 708C
31% 46, 54% 47
20% 46, 41% 47
60% 46, 22% 47
17% 46, 71% 47
87% 47
54% 46, 36% 47,
12% 48
54% 46, 34% 47,
12% 48
50% 46, 34% 47,
16% 48
No reaction
60% 46, 26% 47,
13% 48
AAA deracemization of a lac-
tone with a nitrogen nucleo-
phile and an unprecedented
Rh-catalyzed chemo-, regio-
and stereoselective direct aziri-
dination reaction on an electron
ACHTUNGTRENNUNG(OTf)3, 0.15m in 3-pentanol, 758C
A
7
8
deficient
conjugated
diene
system. The final synthetic
route for (ꢀ)-oseltamivir is
shown in Scheme 31.[4i]
9
1.5 equiv B
(OCHEt2)3, 0.15m in 3-pentanol, 758C
10
azeotropically dired with toluene, then 1.5 equiv BF3·Et2O, 0.15m in 3-pentanol,
758C
11
12
1.5 equiv BF3·Et2O, 1 equiv BSA in reaction, 0.15m in 3-pentanol, RT
25% 46, 57% 47,
7% 48,
11% starting ma-
terial
Experimental Section
Selected experimental procedures for
the two key steps, Pd-AAA deracemi-
zation of lactone and Rh-catalyzed se-
lective azirdination, are shown below.
Full experimental details for all new
compounds and (ꢀ)-oseltamivir are
pre-mixed with BSA, volatile fully removed; then 1.5 equiv BF3·Et2O, 0.15m in 3- 65% 46, 20% 47,
pentanol, 758C 15% 48
given in the Supporting Information.
Pd-AAA of deracemization lactone 8
AHCTUNGTERG(NNUN 1S,5S)-Ethyl 5-(1,3-dioxoisoindolin-2-yl)cyclohex-3-enecarboxylate (13):
All glassware was flame-dried under high vacuum. All chemicals were
flushed with N2 by evacuating under high-vacuum then refilling with N2
for 5 cycles prior to the reaction. THF was degassed using a freeze–thaw
technique under N2 for five cycles. Lactone 8 was redistilled before use
and TMS-phthalimide was stored and weighed in a glove box. To a
Scheme 29. Attempted alkylation of allylic alcohol 47.
100 mL pear shaped flask was loaded [{PdACHTNURGTNEUNG(C3H5)Cl}2] (183 mg,
0.50 mmol), (R,R)-11 (1.04 g, 1.50 mmol) and a stir bar. Degassed dry
cesses of reagents (Table 8, entry 1 and 2). At elevated tem-
perature as high as 1208C with
acetic anhydride as solvent,
Table 8. Acylation of 46.
conversion
was
observed
(Table 8, entry 3). Addition of
pyridine improved the yield
from to 64% after 3 h and 74%
after
overnight
heating
(Table 8, entry 4 and 5). Even-
tually, a microwave reactor was
used and 84% yield was ob-
tained at 1508C after 1 h
(Table 8, entry 6).
The optimized acylation con-
ditions and the final steps in the
synthesis are summarized in
Scheme 30. After acylation, the
SES group was easily removed
by TBAF. Finally, removal of
the phthaloyl with hydrazine
finished the synthesis of 2.
Entry
1
Conditions
10 equiv AcCl,
10 equiv DMAP, DMF, RT
10 equiv Ac2O,
10 equiv DMAP, Py, RT
3 equiv DMAP,
0.1 M in Ac2O, 1208C, 4 h
1 equiv DMAP,
10 equiv Py, 0.1 M in Ac2O, 1208C, 3 h
1 equiv DMAP,
10 equiv Py, 0.1 M in Ac2O, 1208C, overnight
2 equiv DMAP,
Results
sm[a] + elimination
2
3
4
5
6
sm
54% product+sm
64% product + 36% sm
74% product
84% product
20 equiv Py, 0.1 M in Ac2O, MW 1508C, 1 h
[a] sm=starting material.
Chem. Eur. J. 2011, 17, 3630 – 3643
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3641

B. M. Trost and T. Zhang
Rh-catalyzed selective azirdination
(1S,5S,6R)-ethyl 5-(1,3-dioxoisoindolin-2-yl)-7-(2-(trimethylsilyl)ethylsul-
fonyl)-7-aza-bicyclo[4.1.0]hept-2-ene-3-carboxylate (44) : All glassware
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
was flamed dried under high vacuum. MgO was activated by flame-dry
under high vacuum before use. Chlorobenzene was distilled over P2O5
and stored under inert atmosphere over 4 ꢁ MS beads. Under a positive
pressure of N2, to SESNH2 in C6H5Cl (1 mL) was added sequentially mix-
ture 38 (149 mg, 0.5 mmol), MgO (46 mg, 1.15 mmol), [Rh
2ACHTUNGTRENNUNG(esp)2]
(7.6 mg, 0.01 mmol). The reaction was cooled to 08C. PhI(OPiv)2
ACHTUNGTRENNUNG
(264 mg, 0.65 mmol) was added in one portion. The reaction was slowly
warmed to RT and stirred for 4 h till the disappearance of 38 by TLC.
The MgO was filtered and washed thoroughly with CH2Cl2. The com-
bined filtrate was concentrated in vacuo with gentle heating and loaded
to a silica gel column. Eluting with 10:1 petroleum ether/ethyl acetate
yielded 44 as a white foam (206 mg, 86%). Rf =0.50 (2.5:1 petroleum
ether/ethyl acetate); m.p. 55–588C; [a]D =42.18 (CHCl3, c=1.08);
1H NMR (CDCl3, 500 MHz): d = 7.75
Scheme 30. Completion of the synthesis.
(dd, J=5.4, 3.1 Hz, 2H), 7.67 (dd, J=
5.5, 3.1 Hz, 2H), 7.15 (dd, J=4.5,
2.4 Hz, 1H), 4.90 (ddd, J=7.6, 3.5,
2.2 Hz, 1H), 4.14–4.07 (m, 2H), 3.47
(dd, J=6.6, 4.5 Hz, 1H), 3.40 (dt, J=
6.6, 1.7 Hz, 1H), 3.06–3.02 (m, 2H),
2.78 (dd, J=17.9, 3.3 Hz, 1H), 2.53
(ddd, J=17.7, 7.6, 2.6 Hz, 1H), 1.20 (t,
J=7.1 Hz, 3H), 1.04–1.00 (m, 2H),
0.005 ppm (s, 9H); 13C NMR (CDCl3,
125 MHz):
d = 167.5, 165.4, 134.3,
132.4, 131.4, 130.9, 123.3, 61.0, 49.2,
42.9, 40.3, 36.5, 26.1, 14.1, 9.5,
ꢀ2.1 ppm; IR (film): n˜ = 2955, 1775,
1715, 1468, 1388, 1327, 1263, 1144,
1109, 1005 cmꢀ1; HRMS: m/z: calcd
for C22H29N2O6SSi: 477.1516; found:
477.1494 [M+H]+.
Acknowledgements
Scheme 31. Final synthetic scheme.
We thank the National Institutes of
Health (GM 033049) and the National
Science Foundation for their generous
THF (70 mL) was injected into the flask and the mixture was stirred for
20 min at 408C to give a yellow solution. At 408C under N2, lactone 5
(2.48 g, 20 mmol) in 5 mL THF was cannulated into the solution. THF
(2ꢃ5 mL) was used to rinse the flask containing 8 and was also cannulat-
ed into the yellow solution. After 5 min, TMS-phthalimide (6.58 g,
30 mmol) in 15 mL THF was cannulated into the yellow solution under
N2. The reaction was then stirred at 408C under N2 for 8 h at which time
the TLC showed the disappearance of 8. THF was removed in vacuo. To
the residue was added TsOH·H2O (380 mg, 2.0 mmol) and 67 mL abso-
lute ethanol. The mixture was heated at reflux for 10 h monitored by
TLC. Ethanol was removed in vacuo, and the resulting solid mixture was
mixed evenly with 33 g silica gel by dissolving in CH2Cl2 and concentrat-
ing in vacuo. The silica gel containing crude product was loaded on top
of a silica gel column. Eluting with 15:1 petroleum ether/ethyl acetate
yielded 13 as a white solid (5.0 g, 84%). Rf =0.76 (1:1 petroleum ether/
ethyl acetate); HPLC OD 220 nm 90:10 heptane/2-propanol
0.8 mLminꢀ1, tmajor =11.44 min, tminor =15.54 min; [a]D =ꢀ93.88 (CHCl3,
c=1.05); m.p. 108–1128C; 1H NMR (CDCl3, 500 MHz): d = 7.75 (dd,
J=5.59, 3.07 Hz, 2H), 7.65 (dd, J=5.4, 3.1 Hz, 2H), 5.85 (m, 1H), 5.52
(d, 10.2, 1H), 4.92 (m, 1H), 4.07 (q, J=7.0 Hz, 2H), 2.72 (m, 1H), 2.34
(m, 2H), 2.27 (q, J=12.4 Hz, 1H), 2.13 (m, 1H), 1.18 ppm (t, J=7.6 Hz,
3H); 13C NMR (CDCl3, 125 MHz): d = 174.05, 167.8, 133.9, 131.9, 127.9,
126.6, 123.1, 60.5, 47.3, 39.4, 29.0, 27.0, 14.1 ppm; IR (neat): n˜ = 3033,
2984, 1770, 1731, 1702, 1613, 1464, 1387, 1179, 1105 cmꢀ1; elemental anal-
ysis calcd (%) for C17H17NO4: C, 68.21; H, 5.72; found: C, 67.88; H, 6.00;
HRMS: m/z: calcd for C17H17NO4: 299.1158; found: 299.1171 [M]+.
support of our programs. T.Z. thanks Novartis and Roche for graduate
fellowships. Mass spectra were provided, in part, by the Mass Spectrome-
try Regional Center of the University of California, San Francisco. Palla-
dium salts were generously supplied by Johnson-Matthey. We thank
Gilead Sciences for a reference sample of Tamiflu.
[1] For information on H5N1 avian flu and H5N1 virus, visit: http://
www.nature.com/avianflu/index.html. For information on H1N1 flu
and H1N1 virus, visit: http://www.cdc.gov/h1n1flu/.
[2] For reviews on Tamiflu synthesis, see: a) V. Farina, J. D. Brown,
Angew. Chem. Int. Ed. 2006, 45, 7330; b) M. Shibasaki, M. Kanai,
Eur. J. Org. Chem. 2008, 11, 1839; c) J. Magano, Chem. Rev. 2009,
109, 4398.
[3] Contributions from Gilead Sciences: a) C. U. Kim, W. Lew, M. A.
Williams, H. Liu, L. Zhang, S. Swaminathan, N. Bischofberger, M. S.
Chen, D. B. Mendel, C. Y. Tai, G. Laver, R. C. Stevens, J. Am.
Chem. Soc. 1997, 119, 681; b) J. C. Rohloff, K. M. Kent, M. J. Pos-
tich, 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; c) N. W. Bischofberger, C. U.
Kim, W. Lew, H. Liu, M. A. Williams, US Patent 5763483, 1998. Ex-
tensive studies at Roche: d) S. Abrecht, P. Harrington, H. Iding, M.
Karpf, R. Trussardi, B. Wirz, U. Zutter, Chimia 2004, 58, 621; e) M.
Karpf, R Trussardi Angew. Chem. 2009, 121, 5871; Angew. Chem.
Int. Ed. 2009, 48, 5760; Angew. Chem. Int. Ed. 2009, 48, 5760.
3642
www.chemeurj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3630 – 3643

Synthesis of (ꢀ)-Oseltamivir (Tamiflu)
FULL PAPER
[4] a) Y.-Y. Yeung, S. Hong, E. J. Corey, J. Am. Chem. Soc. 2006, 128,
6310; b) Y. Fukuta, T. Mita, N. Fukuda, M. Kanai, M. Shibasaki, J.
Am. Chem. Soc. 2006, 128, 6312; c) X. Cong, Z.-J. Yao, J. Org.
Chem. 2006, 71, 5365; d) T. Mita, N. Fukuda, F. X. Roca, M. Kanai,
M. Shibasaki, Org. Lett. 2007, 9, 259; e) K. Yamatsugu, S. Kamijo,
Y. Suto, M. Kanai, M. Shibasaki, Tetrahedron Lett. 2007, 48, 1403;
f) K. M. Bromfield, H. Graden, D. P. Hagberg, T. Olsson, N. Kann,
Chem. Commun. 2007, 3183; g) N. Satoh, T. Akiba, S. Yokoshima, T.
Fukuyama. Angew. Chem. 2007, 119, 5836; Angew. Chem. Int. Ed.
2007, 46, 5734; Angew. Chem. Int. Ed. 2007, 46, 5734; Angew. Chem.
Int. Ed. 2007, 46, 5734; Angew. Chem. Int. Ed. 2007, 46, 5734; 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; i) B. M. Trost, T. Zhang Angew. Chem. 2008, 120, 3819;
Angew. Chem. Int. Ed. 2008, 47, 3759; Angew. Chem. Int. Ed. 2008,
47, 3759; j) H. Ishikawa, T. Suzuki, Y. Hayashi, Angew. Chem. 2009,
121, 1330; Angew. Chem. Int. Ed. 2009, 48, 1304; k) T. Mandai, T.
Oshitari, Synlett 2009, 787; l) L.-D. Nie, X.-X. Shi, Tetrahedron:
Asymmetry 2009, 20, 124; m) M. Matveenko, A. C. Willis, M. G.
Banwell, Tetrahedron Lett. 2008, 49, 7018; n) K. Yamatsugu, L. Yin,
S. Kamijo, Y. Kimura, M. Kanai, M. Shibasaki Angew. Chem. 2009,
121, 1090; Angew. Chem. Int. Ed. 2009, 48, 1070; Angew. Chem. Int.
Ed. 2009, 48, 1070; o) N. Satoh, T. Akiba, S. Yokoshima, T. Fukuya-
ma, Tetrahedron 2009, 65, 3239; p) B. Sullivan, I. Carrera, M.
Drouin, T. Hudlicky Angew. Chem. 2009, 121, 4293; Angew. Chem.
Int. Ed. 2009, 48, 4229; Angew. Chem. Int. Ed. 2009, 48, 4229; q) K.
Yamatsugu, M. Kanai, M. Shibasaki, Tetrahedron 2009, 65, 6017;
r) H. Sun, Y.-J. Lin, Y.-L. Wu, Y. Wu, Synlett 2009, 2473; s) H.
Osato, I. L. Jones, A. Chen, C. L. L. Chai, Org. Lett. 2010, 12, 60;
t) J. Ma, Y. Zhao, S. Ng, J. Zhang, J. Zeng, A. Than, P. Chen, X.-W.
Liu, Chem. Eur. J. 2010, 16, 4533; u) A. Kamimura, T. Nakano, J.
Org. Chem. 2010, 75, 3133; v) S. Zhu, S. Yu, Y. Wang, D. Ma Angew.
Chem. 2010, 122, 4760; Angew. Chem. Int. Ed. 2010, 49, 4656;
Angew. Chem. Int. Ed. 2010, 49, 4656.
d) S. A. Glover, A. Goosen, C. W. McCleland, J. L. Schoonraad, Tet-
rahedron 1987, 43, 2577; e) S. A. Glover, C. A. Rowbottom, A. P.
Scott, Tetrahedron 1990, 46, 7247.
[19] For aziridination reactions, see a recent review and references there-
in: I. D. G. Watson, L. Yu, A. K. Yudin, Acc. Chem. Res. 2006, 39.
194.
[20] B. M. Trost, G. Dong, J. Am. Chem. Soc. 2006, 128, 6054.
[21] A. Mordini, F. Russo, M. Valacchi, L. Zani, A. Degl’Innocenti, G.
Reginato, Tetrahedron 2002, 58, 7153.
[22] P. O’Brien, C. M. Rosser, D. Caine, Tetrahedron 2003, 59, 9779.
[23] A. Mordini, L. Sbaragli, M. Valacchi, F. Russo, G. Reginato, Chem.
Commun. 2002, 778.
[24] K. C. Nicolaou, T. Montagnon, P. S. Baran, Y.-L. Zhong, J. Am.
Chem. Soc. 2002, 124, 2245.
[25] a) J.-i. Kato, N. Iwasawa, T. Mukaiyama, Chem. Lett. 1985, 743; b) S.
Hatakeyama, H. Mori, K. Kitano, H. Yamada, M. Nishizawa, Tetra-
hedron Lett. 1994, 35, 4367; c) S. H. Lee, Y. J. Park, C. M. Yoon, Tet-
rahedron Lett. 1999, 40, 6049; d) K. Iwanami, H. Seo, Y. Tobita, T.
Oriyama, Synthesis 2005, 183; e) K. Iwanami, K. Yano, T. Oriyama,
Chem. Lett. 2007, 36, 38.
[26] D. C. Hill, L. A. Flugge, P. A. Petillo, J. Org. Chem. 1997, 62, 4864.
[27] V. Kumar, N. G. Ramesh, Chem. Commun. 2006, 4952.
[28] B. M. Trost, G. S. Massiot, J. Am. Chem. Soc. 1977, 99, 4405.
[29] A similar diene intermediate bearing an NH-Boc group rather than
an N-phthaloyl group was reported in Coreyꢂs synthesis (Y.-Y.
Yeung, S. Hong, E. J. Corey, J. Am. Chem. Soc. 2006, 128, 6310). In
their approach, the synthesis of the diene required 8 steps, whereas
in our approach, compound 38 took only 3 steps. From the conju-
gate diene, they applied a two-step haloamination-SN2 displacement
sequence to form the desired aziridine while we developed a direct
aziridination process.
[30] Examples for 1,4-conjuate addition to form aziridines: a) S. Fiora-
vanti, D. Colantoni, L. Pellacani, P. A. Tardella, J. Org. Chem. 2005,
70, 3296; b) D. Colantoni, S. Fioravanti, L. Pellacani, P. A. Tardella,
Org. Lett. 2004, 6, 197; c) S. Fioravanti, A. Morreale, L. Pellacani,
P. A. Tardella, Synlett 2004, 6, 1083; d) R. Vyas, G.-Y. Gao, J. D.
Harden, X. P. Zhang, Org. Lett. 2004, 6, 1907; e) S. Fioravanti, A.
Morreale, L. Pellacani, P. A. Tardella, Eur. J. Org. Chem. 2003,
4549; f) S. L. Jain, B. Sain, Tetrahedron Lett. 2003, 44, 575; g) T. Gas-
peri, M. A. Loreto, P. A. Tardella, A. Gambacorta, Tetrahedron Lett.
2002, 43, 3017; h) J. Aires-de-Sousa, S. Prabhakar, A. M. Lobo,
A. M. Rosa, M. J. S. Gomes, M. C. Corvo, D. J. Williams, A. J. P.
White, Tetrahedron Asymmetry 2001, 12, 3349; i) S. Fioravanti, A.
Morreale, L. Pellacani, P. A. Tardella, J. Org. Chem. 2002, 67, 4972;
j) T. Ishikawa, T. Kudoh, J. Yoshida, A. Yasuhara, S. Manabe, S.
Saito, Org. Lett. 2002, 4, 1907.
[5] The lactone can also be synthesized in two steps: P. A. Bartlett,
L. A. McQuaid, J. Am. Chem. Soc. 1984, 106, 7854.
[6] B. M. Trost, J.-P. Surivet, Angew. Chem. 2000, 112, 3252; Angew.
Chem. Int. Ed. 2000, 39, 3122;.
[7] B. M. Trost, R. C. Bunt, J. Am. Chem. Soc. 1996, 118, 235.
[8] L. Grehn, U. Ragnarsson, Synthesis 1987, 275.
[9] Pd-catalyzed AAA with HNACTHNUTRGNE(NUG Boc)2: D. A. Evans, K. R. Campos, J. S.
Tedrow, F. E. Michael, M. R. Gagne, J. Am. Chem. Soc. 2000, 122,
7905.
[10] Pd-catalyzed stereospecific allylic alkylation with HNACTHNUTRGNEUNG(Cbz)2: R. A.
Fairhurst, R. J. Taylor, WO Patent 2008006563, 2008.
[11] Pd-catalyzed AAA with HNACHTNUGTRENNUNG
Stanford University, 2006.
[12] Pd-catalyzed AAA with NaNACHTNUGTRENNUNG
[31] Example of 1,6-addition onto diene esters: L. Bernardi, J. Lopez-
Cantarero, B. Niess, K. A. Jorgensen, J. Am. Chem. Soc. 2007, 129,
5772.
Wang, K, Ding, Tetrahedron Lett. 2001, 42, 7659.
[13] Pd-catalyzed AAA with sodium phthalimide: B. M. Trost, R. Radi-
nov, J. Am. Chem. Soc. 1997, 119, 5962.
[32] a) T. Schirmeister, Liebigs Ann. 1997, 1895; b) D. A. Clark, F. De
Riccardis, K. C. Nicolaou, Tetrahedron 1994, 50, 11391.
[33] For a recent review on metal-catalyzed aziridination reactions, see:
J. A. Halfen, Curr. Org. Chem. 2005, 9, 657.
[34] a) For a recent report on Ag catalyzed aziridination reactions, see:
Y. Cui, C. He, J. Am. Chem. Soc. 2003, 125, 16202; b) for a recent
report on Au-catalyzed aziridination reactions, see: Z. Li, X. Ding,
C. He, J. Org. Chem. 2006, 71, 5876.
[35] a) K. Guthikonda, J. Du Bois, J. Am. Chem. Soc. 2002, 124, 13672;
b) K. Guthikonda, P. M. Wehn, B. J. Caliando, J. Du Bois, Tetrahe-
dron 2006, 62, 11331.
[36] 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] Pd-catalyzed AAA with phthalimide: a) B. M. Trost, R. C. Bunt, J.
Am. Chem. Soc. 1994, 116, 4089; b) B. M. Trost, R. C. Bunt, R. C.
Lemoine, T. L. Calkins, J. Am. Chem. Soc. 2000, 122, 5968; c) B. M.
Trost, A. C. Krueger, R. C. Bunt, J. Zambrano, J. Am. Chem. Soc.
1996, 118, 6520; d) B. M. Trost, C. Jiang, K. Hammer, Synthesis
2005, 3335; e) B. M. Trost, D. B. Horne, M. J. Woltering, Angew.
Chem. 2003, 115, 6169; Angew. Chem. Int. Ed. 2003, 42, 5987;
f) B. M. Trost, D. B. Horne, M. J. Woltering, Chem. Eur. J. 2006, 12,
6607; g) B. M. Trost, A. Aponick, J. Am. Chem. Soc. 2006, 128, 3931;
h) B. M. Trost, A. Aponick, B. N. Stanzl, Chem. Eur. J. 2007, 13,
9547; i) B. M. Trost, B. M. OꢂBoyle, Org. Lett. 2008, 10, 1369.
[15] R. C. Bunt, Ph. D. thesis, Stanford University, 1996.
[16] W. Adam, N. Bottke, O. Krebs, C. R. Sahao-Moller, Eur. J. Org.
Chem. 1999, 1963.
[17] G. E. Keck, R. R. Webb, J. B. Yates, Tetrahedron 1981, 37, 4007.
[18] a) J. Lessard, M. Mondon, D. Touchard, Can. J. Chem. 1981, 59, 431;
b) J. Lessard, J. Tuaillon, J. Org. Chem. 1982, 47, 5013; c) G. F.
Koser, P. B. Kokil, M. Shah, Tetrahedron Lett. 1987, 28, 5431;
[37] a) J. Chan, K. D. Baucom, J. A. Murry, J. Am. Chem. Soc. 2007, 129,
14106; b) D. N. Zalatan, J. Du Bois, J. Am. Chem. Soc. 2009, 131,
7558.
Received: November 29, 2010
Published online: March 1, 2011
Chem. Eur. J. 2011, 17, 3630 – 3643
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3643