PHARMACEUTICAL INDUSTRY IN SWITZERLAND
627
CHIMIA 2004, 58, No. 9
to produce the Boc-substituted triazole 47
(Scheme 11). The diphenylphosphoryl de-
rivative obtained was transesterified in the
same pot to the diethylphosphoryl ester 53,
a conversion which is essential to prevent
extensive side product formation in the en-
suing P,N-bond cleavage to 56.
Surprisingly and interestingly, the aziri-
dine intermediate 53 undoubtedly repre-
sented the unexpected endo-aziridine iso-
mer as confirmed by selective hydrolysis to
the corresponding crystalline carboxylic
acid (LiOH in THF/MeOH/H2O at 60 °C)
and X-ray crystallography [17].Attempts to
gain insight into the mechanism of this un-
expected selectivity, the configuration of
the two regio-isomeric precursor triazoles
57 and 58 formed in an about 2:1 ratio by
treatment of (–)-46 with DPPA at 30 °C for
72 h was examined. However, only the ma-
jor isomer 57 was isolated by chromatogra-
phy, while 58 decomposed on the column.
Crystallization and X-ray analysis of 57 re-
vealed the exo-position of triazole moiety,
as already suggested for both regioisomers
by 1H-NMR analysis of the mixture. Heat-
ing the isolated triazole 57 to 70 °C provid-
ed the endo-aziridine 59 (Fig. 3).
Scheme 12. Furan-ethyl acrylate Diels-Alder approach – final route
Although a full explanation for the ob-
served formal ‘inversion’ is still pending
[18], this unexpected result opened the door
to a very short and effective synthesis of os-
eltamivir phosphate (1) since the endo-
aziridine 53 smoothly underwent elimina-
tive ring opening to form 54 followed by di-
rect O-mesylation and regio- and
stereoselective introduction of the 3-pentyl-
ether side chain applying the conditions
used in the discovery chemistry synthesis
(Scheme 1) efficiently leading to 55. P,N-
bond cleavage in 55 required somewhat
harsh conditions and led to the crystalline
hydrochloride 56. With the specific config-
uration of this intermediate, the transforma-
tion to the drug substance 1 applying an
analogous azide-free ‘allylamine’ protocol
as described above became feasible, lead-
ing to the optically pure drug substance 1 in
good overall yield.
Fig. 3. Configurations of intermediates 57, 58, and 59
steps had obviously to be avoided. The re-
sults of these efforts are summarized in
Scheme 12.
razyme L-2 was improved by intensive pa-
rameter optimization. Low temperatures
and non-polar co-solvents favored high
enantioselectivity, whereas high pH and
high substrate concentration adversely af-
fected selectivity. Best results under techni-
cally relevant conditions (5% conc.) were
obtained with methylcyclohexane as a
biphasic co-solvent in aqueous buffer pH
8.0 at 1°C, providing (–)-46 in 97% ee at
75% conversion. Almost complete removal
of the remaining endo-isomer, both enan-
tiomers of which proved to be inert against
enzymatic hydrolysis, was then achieved by
distillation.
In order to explore and improve the iso-
mer ratio of the Diels-Alder reaction in fa-
vor of the desired exo-isomer 46, zinc- and
magnesium-based catalysts were evaluated
which proved active enough to promote the
cycloaddition without causing significant
decomposition or polymerization of the
starting materials. These investigations led
to the use of inexpensive zinc chloride as
the catalyst of choice. Further optimiza-
tions revealed a clear dependence of the
exo/endo-ratio with increasing reaction
times confirming the kinetically preferred
formation of the endo-isomer of 46 fol-
lowed by a steady increase of the share of
the thermodynamically favored exo-isomer
asymptotically reaching the equilibrium ra-
tio of 9:1. This exo/endo-ratio along with
the obtained yield and the simplicity of the
process were superior to the best results
known in literature for 46 or its methyl es-
ter analogue [15].
3.2.2. Aromatic Ring Transformations:
The Desymmetrization Concept [19]
Taking advantage of the desymmetriza-
tion protocol over racemate cleavage re-
garding effectiveness, the ‘desymmetriza-
tion concept’ based on a potential enzym-
atic monohydrolysis of an all-cis
meso-diester of type 62 to the optically ac-
tive mono-acid 63 outlined in Scheme 13
was proposed. Besides the desymmetriza-
tion step 62 → 63, the concept is based on
the intended transformation of a pyrogallol
derivative of type 60 to a diester 62 includ-
ing a selective cis-hydrogenation of the
symmetrically substituted aromatic iso-
phthalic diester 61, as well as on the effi-
cient introduction of the two amino func-
The safety limitations associated with
the use of Boc-azide were overcome by the
use of diphenylphosphoryl azide (DPPA)
which, to our knowledge, represented the
first example of its use for the generation of
an aziridine moiety from an olefin [16]. The
use of DPPA, a commercially available
azide with a high decomposition tempera-
ture of 190 °C, not only added in [3+2]-
fashion to the double bond of (–)-46 already
at 70 °C to form a mixture of the corre-
sponding exo-triazoles, but also allowed for
the thermal extrusion of nitrogen thus
avoiding the technically problematic photo-
chemical nitrogen extrusion step required
For the optical resolution of rac-46 via
enantioselective ester hydrolysis 83 mi-
croorganisms and 50 enzymes were
screened. The moderate enantioselectivity
of the best enzyme found, namely Chi-