M. Ohtawa et al.
Bull. Chem. Soc. Jpn. Vol. 87, No. 1 (2014)
115
yield, accompanied by a small amount of undesired secondary
alcohol 9a and diol 10a (Entry 1). The reactions of uridine
derivative 7b,12 which is a ribonucleoside, and methyl β-D-
glucofuranoside 7c12 with NH4F in MeOH at 0 °C also
furnished the primary alcohols 8b and 8c in yields of 83%
and 86%, respectively, with trace amounts of diols 10b and
10c13 (Entries 2 and 3). The number of equivalents of NH4F in
Entries 1-3 was optimized to afford the corresponding primary
alcohols in high yields.14 In contrast, the reactions of methyl α-
D-glucopyranoside 7d12 and methyl β-D-galactopyranoside 7e15
with NH4F (10 equiv) in MeOH at room temperature afforded
the desired primary alcohols 8d and 8e in moderate yields, with
about 40% of recovered starting materials16 (Entries 4 and 6).
The starting materials were not consumed completely because
the mono-deprotected 8d and 8e could be recyclized under
the reaction conditions to provide the corresponding starting
materials 7d and 7e with high thermodynamic stability.17 To try
to prevent recyclization, Dess-Martin periodinane (DMP) was
added to the reaction mixtures to oxidize the resulting primary
alcohols to the corresponding aldehydes (Entries 5 and 7).
DMF was used as the solvent instead of MeOH. The corre-
sponding aldehyde products were not obtained, but, surpris-
ingly, the yields of the desired 8d and 8e increased, and only
small amounts of the starting materials 7d and 7e, respectively,
were recovered. It seems likely that temporary protection of
the resulting hydroxy group with an active species originating
from hypervalent iodine prevented recyclization; mechanistic
studies of the effect of DMP in the NH4F-mediated regiose-
lective mono-deprotection reaction are currently in progress
in our laboratory. In contrast, the reaction of 3-deoxyglucose
derivative 7f12 with NH4F (10 equiv) in MeOH at room tem-
perature for 0.5 h gave the desired 8f in 39% yield, accom-
panied by diol 10f in 14% yield and recovered starting material
in 42% yield (Entry 8). Prolonging the reaction time increased
the yield of the corresponding diol 10f and decreased the yields
of the desired 8f and recovered starting material 7f (Entry 9).
Moreover, the reactions of di-tert-butylsilylene acetals 7g4b and
7h,12 derived from simpler 1,3-diols, with NH4F (10 equiv)
in MeOH or DMF at room temperature afforded only trace
amounts of the desired 8g and 8h, regardless of the reaction
time (Entries 10-13), and the yields of diols 10g and 10h
and recovered starting materials 7g and 7h followed the same
trends as those shown by 10f and 7f. These results indicate that
moderate steric bulkiness around the secondary hydroxy group
is important for avoiding over-deprotection of the desired
primary alcohol with the fluorosilyl ether.
O
H
O
N
HO
H
O
O
O
O
t-Bu
t-Bu
Si
CN
O
F–
+
NH4
Figure 2. Proposed transition-state in regioselective mono-
deprotection of di-tert-butylsilylene acetal.
not been reported so far.6 We expected that hydrogen-bond
formation by the fluoride anion in the reaction would reduce its
nucleophilicity to silicon and result in regioselective mono-
deprotection. Initially, tetra-n-butylammonium fluoride, the
most typical desilylating reagent, in the presence of acetic acid
was examined (Entries 1 and 2). Consequently, triol 6 was
obtained as the major product, without the desired primary
alcohol 4. Furthermore, treatment with Et3N¢3HF7 also gave
the same results (Entries 3 and 4). The use of inorganic fluoride
reagents such as CsF8 and KF9 in aqueous solution led to no
reaction (Entries 5 and 6). Subsequent treatment with NH4F
(1 equiv) in MeOH afforded the desired 4 (Entry 7), but in poor
yield (18%), with recovered 3 (49%). Therefore, optimization
of the reaction conditions using NH4F was conducted. An
increase in the number of equivalents of NH4F improved
the yield of the desired 4, and the best yields were obtained
by treatment with 10 equiv of NH4F (Entries 8 and 9). Next,
various solvents were investigated in the regioselective mono-
deprotection of 3. The use of dimethylformamide (DMF) and a
1:1 mixture of MeOH and DMF instead of MeOH gave almost
the same results (Entries 10 and 11). In contrast, the reaction
in dimethyl sulfoxide (DMSO) gave only a low yield of 4,
with a large amount of the undesired 1,3-diol 6 (Entry 12).
Moreover, the use of ethers such as tetrahydrofuran (THF) and
Et2O led to no reaction and a very low yield (10%) of 4 with
recovered 3 (75%) (Entries 13 and 14). Although the reactions
in MeCN and acetone afforded the desired 4 in good yield
regioselectively, the reactions proceeded a little more slowly
than those in MeOH and DMF. As a result, MeOH and DMF
proved to be the most suitable solvents for the regioselective
mono-deprotection of di-tert-butylsilylene acetal.
The regioselectivity observed in these reactions probably
comes from kinetically controlled ring cleavage, in which the
ammonium ion is located beside the more sterically accessible
oxygen through Coulombic forces, prior to delivery of fluoride,
as shown in Figure 2. In addition, reduction of the nucleophi-
licity of the fluoride anion toward silicon, caused by formation
of strong hydrogen-bonds with ammonium ions in the reac-
tion,10 will contribute to the regioselective mono-deprotection.
We next investigated the scope of this reaction (Table 2).
First, the regioselective mono-deprotection reactions of di-tert-
butylsilylene acetal-protected D-glucofuranosides with NH4F
were examined. The thymidine derivative 7a,11 which is a
deoxyribonucleoside, was treated with NH4F in MeOH at
0 °C, and provided the desired primary alcohol 8a in 81%
Next, we focused on the regioselective mono-deprotection of
other types of di-tert-butylsilylene acetals (Table 3). Although
treatment of di-tert-butylsilylene acetal 7i,12 consisting of
primary and tertiary alcohols, with NH4F (10 equiv) in MeOH
at room temperature provided the desired primary alcohol 8i,
the reaction proceeded very slowly and gave 8i in only low
yield (35%), accompanied by unreacted 7i (31%) and the diol
10i (30%) (Entry 1). Formation of the primary alcohol 8i, i.e.,
the di-tert-butylfluorosilyl ether of the tertiary alcohol, would
probably be difficult because of its steric bulkiness. In contrast,
the reaction of di-tert-butylsilylene acetal 7j, derived from a
1,4-diol with NH4F (10 equiv) in MeOH at room temperature,
led to no reaction (Entry 2), even though the reaction of di-tert-