6874
I. Suzuki et al. / Tetrahedron Letters 48 (2007) 6873–6876
OH
OH
OAc
OAc
these reactions, the enantioselectivity was enhanced with
OH
OH
(ii)
(i)
increase in 3a. This fact indicates that the minor enan-
tiomer of 2a suffered over-hydrolysis faster than did
the major enantiomer. When the reaction was conducted
at a higher concentration, 2a was obtained in 67% yield
(90% ee) without over-hydrolyzed product 3a along with
the recovery of 1a (entry 4). Since PLE was proved to be
the best catalyst amongst enzymes tested, we next exam-
ined desymmetrization reactions of other substrates and
the results are presented in Table 2.
Bn
N
Bn
N
HN
OAc
OAc
O
(iii)
N
Bn
1a
Scheme 1. Preparation of N-oxide 1a. Reagents and conditions: (i)
PhCH2Br, Et3N/CHCl3, rt, 12 h. (ii) AcCl, Et3N/benzene, 0 °C, 3 h,
87% (two steps). (iii) MCPBA/CHCl3, 0 °C, 0.5 h, 95%.
N-Oxide 1b, which has a larger substituent on nitrogen,
was hydrolyzed slowly to yield 2b only in 27% yield
along with 38% recovery of 1b, and the enantioselectiv-
ity of the reaction was moderate (63% ee). When N-
oxide 1c was treated with PLE in phosphate-buffer solu-
tion, enantiomeric excess of 2c reached to 99% ee, but
the yield was 34% due to instability of 2c, which was
gradually decomposed at room temperature even in buf-
fer solution. It is noteworthy that N-oxide 1d was not
hydrolyzed at all and was recovered almost completely.
It seems reasonable to assume that the lower lipophilic-
ity of N-oxide 1d compared to that of other N-oxides
tested prevents incorporation of 1d to the hydrophobic
reaction site of PLE. On the other hand, N-oxide 1e pos-
sessing a highly lipophilic cyclohexyl group on nitrogen
was consumed completely in 24 h and afforded 2e in 55%
along with 34% of 3e. These results indicate that the effi-
ciency of the PLE-catalyzed hydrolysis is considerably
dependent on a substituent on the nitrogen atom, and
we further investigated the PLE-catalyzed hydrolysis
of 1f–n, and the results are summarized in Table 3. N-
Oxides having an ortho-substituted benzyl group on
nitrogen exhibited 88–99% ee (entries 1, 4, 7 and 9),
whereas in the case of para-substituted N-oxides, to
our surprise, the enantioselectivity was only 6–13% ee
(entries 3, 6 and 8).8 For meta-substituted 1g and 1j,
while 1g gave 2g in 92% ee, 1j afforded 2j in only 36%
ee. With respect to reaction rates, considering the recov-
ery of diacetate 1, ortho-substituted N-oxides 1f, 1i, 1l
and 1n were smoothly hydrolyzed to give monoacetate
2f, 2i, 2l and 2n, respectively, while para-substituted
1h, 1k, 1m and meta-substituted 1g and 1j reacted more
slowly and larger amounts of starting diacetates
remained unreacted than the case of using ortho-substi-
tuted N-oxides.
NMR using (R)-BINOL (6–10 equiv) as a chiral solvat-
ing agent (CSA)7 and the results are summarized in
Table 1.
When PLE and lipase AS AMANO were used, hydroly-
sis of 1a proceeded to give 2a (entries 1–4 and 6),
whereas other enzymes did not work as catalysts for
the desymmetrization reactions (entries 5 and 7–11).
Since PLE gave a better result with respect to the enantio-
selectivity rather than lipase AS AMANO (entries 1
and 6), we further optimized reaction conditions by
using PLE. Since over-hydrolyzed product 3a was ob-
tained when the reaction was carried out for 24 h (entry
1), the reaction time was shortened to 12 h. In this case,
3a was not formed at all; however, recovery of 1a in-
creased to 44% and the yield and enantiomeric excess
of 2a were not greatly improved (entry 2). The enantio-
meric purity of 2a was enhanced up to 99% ee when the
reaction was conducted for 48 h, though the yield of 2a
was lowered to 28% with increase in 3a (entry 3). In
Table 1. Enzyme-catalyzed desymmetrizations of N-oxide 1a
OAc
OAc
OH
OH
OH
O
N
Bn
O
N
Bn
O
N
Bn
Enzyme
+
pH 7.2, 25 C
°
OAc
1a
2a
3a
Entry Enzyme
Time (h)
Yieldc (%)
Recovery 2a
3a
1
2
3
4
5
6
7
8
9
PLEa
24
12
48
24
66
24
66
66
66
66
24
44
9
25
>95
12
>95
>95
>95
>95
>95
48 (93)
51 (91)
28 (>99) 32
67 (90)
0
19
0
PLEa
PLEa
PLEa,d
0
0
0
0
0
0
0
0
PPLa
To determine the absolute configuration of N-oxides, we
first tried to obtain single crystals of the N-oxides 2a–n
in the presence or absence of chiral acids, such as cam-
phoric acid, camphorsulfonic acid, MTPA and (R)-BI-
NOL; however, none of them afforded crystals fine
enough for X-ray crystallographic analysis. Further-
more, we also attempted to convert 2a–n to diastereo-
meric esters with the chiral acids mentioned above, but
these efforts have been all failed due to low reactivity
of 2a–n.
Lipase ASb
Lipase AYSb
Lipase PSb
Lipase Mb
Lipase Gb
49 (60)
0
0
0
0
0
10
11
Lipase F-APb 66
960 Units of an enzyme and 0.43 mmol of 1a were mixed in 20 mL of
phosphate-buffer solution (pH 7.2) and the reaction was conducted for
indicated time at 25 °C.
a PLE and PPL were purchased from Sigma–Aldrich.
b Lipases were purchased from Amano Enzyme Inc.
c Values in parentheses indicate enantiomeric excesses determined by
1H NMR using (R)-BINOL as a CSA.
In conclusion, we found that PLE can be used as an effi-
cient catalyst for desymmetrization of prochiral N-oxi-
des and demonstrated that N-oxides 1 were hydrolyzed
by PLE efficiently to afford N-chirogenic tertiary amine
N-oxides up to 99% ee in moderate to good yields.
d 20 mg PLE (480 units) and 0.43 mmol of 1a were mixed in 10 mL of
phosphate-buffer solution (pH 7.2) and the reaction was performed
for indicated time at 25 °C.
Suzuki, Ichiro
