Pummerer rearrangement of 1-deoxy-5-thioglucopyranose oxides; Novel synthesis of 5-thioglucopyranose derivatives

Article abstract of DOI:10.1016/S0040-4039(03)00842-6

The Pummerer rearrangement of 3,4-O-isopropylidene-1-deoxy-5-thiopyranose oxide derivatives took place at the C1 position regioselectively to give the corresponding 5-thiopyranoses. The reaction mechanism of this reaction is also discussed.

Full text of DOI:10.1016/S0040-4039(03)00842-6

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 44 (2003) 4089–4093
Pummerer rearrangement of 1-deoxy-5-thioglucopyranose oxides;
novel synthesis of 5-thioglucopyranose derivatives
Hiroko Matsuda,a Junji Fujita,a Yasuharu Morii,a Masaru Hashimoto,a,* Toshikatsu Okunoa
and Kimiko Hashimotob
aFaculty of Agriculture and Life Science, Hirosaki University, 3 Bunkyo-Cho, Hirosaki, Aomori 036-8561, Japan
bBiochemical Resources Laboratory, Plant Science Center, RIKEN, 2-1, Hirosawa, Wako-shi, Saitama, 351-0198, Japan
Received 26 February 2003; revised 27 March 2003; accepted 28 March 2003
Abstract—The Pummerer rearrangement of 3,4-O-isopropylidene-1-deoxy-5-thiopyranose oxide derivatives took place at the C1
position regioselectively to give the corresponding 5-thiopyranoses. The reaction mechanism of this reaction is also discussed.
© 2003 Elsevier Science Ltd. All rights reserved.
Merrer et al.,7,8 thiepane 1 was subjected to a ring
Since Pummerer rearrangement provides a synthetic
equivalent of carbonyl compounds from sulfoxides, an
equivalent for alcohols under non-oxidative conditions,
it has been utilized in the total syntheses of natural
products as an alternative protocol for oxidation of the
alcohols.1,2 This reaction might be desirable for intro-
duction of the C1 hemithioacetal group of thiosugars if
we can perform the reaction of 1-deoxy-5-thiopyranose
oxides (or thiofuranose oxides3) at the C1 position
regioselectively. In the course of our synthetic studies
on oligothiosaccharides,4 we revealed that the Pum-
merer rearrangement of 1-deoxy-5-thioglucopyranose
with an acetonide protective group at the C3-, C4-alco-
hols proceeds at the C1 position regioselectively,
providing 5-thioglucose derivatives in good yields. The
mechanism of this selectivity is also discussed in this
paper.
contraction under the Mitsunobu conditions to give
1-deoxy-5-thioglucopyranose 2 in 80% yield (Scheme 1).
This was first converted to tetrabenzoate 3a by acidic
removal of the acetonide group of 2 and subsequent
treatment with BzCl in pyridine. Since acidic treatment
of 2 in acetone resulted in partial migration of the
acetonide, the 2,3-O-acetonide thus obtained was also
converted into acetate 3b. The recovered 3,4-O-ace-
tonide was transformed to dibenzoate 3c and MOM
ether 3d under the usual conditions. Oxidation of 3a–d
In spite of extensive studies by Oae5 and Crucianelli6 on
Pummerer rearrangements, the regioselectivity for
asymmetric sulfoxides has not been discussed well.
Recently, Naka et al. investigated regio- and stereo-
selective formation of thionucleosides via Pummerer-
type glycosidation.3 However, their studies did not
provide enough information to predict the regioselectiv-
ity for our substrates. Thus, 1-deoxy-5-thioglucopyran-
ose oxides (a)-4 and (b)-4 carrying various protective
groups were prepared as the substrates in order to
study the reactivity and regioselectivity in the
Pummerer rearrangement. Based on the information by
Scheme 1. Reagents and conditions: (a) PPh3, DEAD,
PhCOOH, THF, rt (80%); (b) for 3a; (i) cat. HCl, MeOH, rt
(84%), (ii) BzCl, pyridine, rt (100%); for 3b–d; (i) cat. p-
TsOH, rt, acetone, then separations, (ii) “3b, Ac2O, pyridine,
rt (100%); “3c, BzCl, pyridine, rt (95%); “3d, MOMCl,
(iPr)2NEt, CH2Cl2, 0°C (60%); (c) mCPBA, CH2Cl2, −20°C
(4a, 96%, 4b, 92%, 4c, 91%, 4d, 88%, isomeric ratios (a:b) 4a,
50:50, 4b, 60:40, 4c, 60:40, 4d, 50:50).
* Corresponding author. Tel.: +81-172-39-3782; fax: +81-172-39-3782;
e-mail: hmasaru@cc.hirosaki-u.ac.jp
0040-4039/03/$ - see front matter © 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0040-4039(03)00842-6

4090
H. Matsuda et al. / Tetrahedron Letters 44 (2003) 4089–4093
1
Table 1. The characteristic H and 13C NMR chemical
shifts of sulfoxides (a)- and (b)-4a–d, and their differences
Dl(=l(a)−l(b), italic)
Table 2. Pummerer rearrangement employing some sulfox-
ides
Signals
4aa
4bb
4cb
4d
a
b
a
b
a
b
a
b
C1Hax 3.22 2.72
+0.50
C1Heq 4.02 3.85
+0.17
2.57 1.63
+0.94
3.33 3.03
+0.30
3.10 4.61
−1.51
5.52 6.00
−0.48
2.66 1.58 3.01a 2.41a
+1.08 +0.60
3.24 3.33 3.79a 3.72a
−0.09 +0.07
5.32 6.13 4.03a 4.52a
−0.81 −0.49
3.25 4.39 3.50a 4.13a
−1.14 −0.63
2.82 2.65 3.23a 3.11a
+0.17 +0.12
C2H
C4H
C5H
C1
5.45 6.11
−0.70
5.79 6.23
−0.44
3.49 3.38
+0.11
2.73 2.53
+0.20
52.6 47.4 52.3 48.1 53.4 49.2 55.1b 50.6b
+5.2
65.3 67.3 67.1 68.9 67.8 69.3 68.9b 71.1b
−2.0 −1.8 −1.5 −2.2
64.8 67.9 69.2 71.5 71.4 72.9 70.7b 72.7b
−3.1 −2.3 −1.5 −2.0
65.7 59.0 65.1 60.6 64.8 59.2 63.9b 58.8b
+6.7 +4.5 +5.6 +5.1
+4.2
+4.2
+4.5
Run
Substrates
Products (%)
C2
5
6
7
8
9
C4
1
2
3
4
5
6
(a)-4a
(b)-4a
(a)-4b
(b)-4b
(a)-4c
(b)-4c
(a)-4c
(b)-4c
(a)-4d
(b)-4d
5.7
2.7
8.9
5.6
40
50
41
4.0
0.3
9.5
8.1
1.7
1.2
N.d.
N.d.
N.d.
N.d.
N.d.
N.d.
C5
N.d.b
N.d.
55
61
66
1.7
1.2
14
9.4
45
a Observed in CDCl3.
b Observed in C6D6.
Trace
Trace
N.d.
N.d.
N.d.
N.d.
Trace
Trace
N.d.
N.d.
N.d.
N.d.
Trace
Trace
N.d.
N.d.
Trace
N.d.
7a
8a
9
65
66c
84c
by mCPBA gave diastereomeric mixtures of a-sulfox-
ides (a)-4a–d and b-sulfoxides (b)-4a–d, respectively, in
high yields. These isomers were readily separated by
column chromatography. Their stereochemistry was
estimated by comparison of their 1H and 13C NMR
spectra.8–11 The characteristic signals are shown in
Table 1. Except for C1Heq, signs of the Dl value
[=l(a-isomer−l(b-isomer)] are consistent with the lit-
erature. There were no significant differences in the
chemical shifts in the C1Heq, probably due to both
gauche relationship of sulfoxide-C1Hax and sulfoxide-
C1Heq.
10
a Pyridine was employed as the solvent.
b N.d.=not detected.
c Yields of 10c obtained after acetylation employing Ac2O in pyridine.
were both tentatively assigned as 5R (carbohydrate
numbering) by considering the anomeric effect. How-
ever, conclusive evidence for their stereochemistry was
not provided by NOE studies. It can be supposed that
7a, 8a, and 9a are generated by degradation of TFA
ester 6a under the reaction conditions and/or purifica-
tion process.12 Accordingly, the rearrangement for 4a
proceeded at the C5 position regioselectively.
With these sulfoxides 4 in hand, their Pummerer rear-
rangement reactions were examined. Our preliminary
experiments disclosed that the rearrangement did not
proceed at room temperature when Ac2O was
employed. The heating conditions gave complex mix-
tures in the cases that 4a and 4c were employed as the
substrates. The conditions using trifluoroacetic anhy-
dride (TFAA) in place of Ac2O,2 the modified protocol
developed by one of the authors, was found to proceed
at room temperature. Thus, treatment of (a)-4a with
TFAA (5.0 equiv.) in the presence of pyridine (12
equiv.) in CH2Cl2 at room temperature for 3 h gave
5-thioglucopyranose derivative 5a in low yield (5.7%).
5-Trifluoroacetoxy derivative 6a (8.9%), the corre-
sponding alcohol 7a (50%), exo-olefin 8a (4.0%), and
endo-olefin 9a (9.5%) were also afforded by the reaction
as shown in Table 2. The product 5a was isolated as an
alcohol but not the corresponding TFA ester, which
will be discussed later. The products 6a, 7a, and 8a
were observed as single isomers, respectively, on their
1H NMR spectra. The stereochemistries of 6a and 7a
When 2,3-O-acetonide (a)-4b was employed, the re-
arrangement proceeded with higher regioselectivity at
the C5 position (carbohydrate numbering) to give a
mixture of 6b, 7b, 8b, and 9b. Thioglucopyranose 5b,
obtainable through the rearrangement at the C1 posi-
tion, was not observed under those conditions.
In contrast, similar treatment of 3,4-O-acetonide (a)-4c
effected the rearrangement at the C1 position in a
highly regioselective manner to give 5-thiogluco-
pyranose derivative 5c13 along with trace amounts of
6c–9c, caused by the rearrangement at the C5 position
(runs 5 and 6). The reactions gave 5c in slightly higher
yields without production of 6c–9c when pyridine was
employed as the solvents (runs 7 and 8). The existence
of the OH group in 5c was confirmed by observing
strong absorption at 3440 cm−1 (broad) in the IR
spectrum. Probably, the corresponding TFA ester of 5c

H. Matsuda et al. / Tetrahedron Letters 44 (2003) 4089–4093
4091
is not stable enough, so that the ester moiety might be
hydrolyzed during the work-up. The 1H NMR spec-
trum indicated that 5c consists of the two anomers
(a:b=90:10).
Since 5d appeared as a broad spot on the TLC, the
crude mixture was acetylated prior to purification, giv-
ing an anomeric mixture of 10d in 66% yield in two
steps. Preparative silica gel TLC could separate the
isomers.13
When a series of b-sulfoxides (b)-4a–d were subjected to
the Pummerer conditions, the same products as those
obtained from the corresponding a-sulfoxides were
afforded in slightly different yields. Noteworthy, (b)-4d
gave 10d in 84% yield after acetylation. It seems the
stereochemistry about the sulfoxide moiety is not
important for the regioselectivity in the rearrangements
according to these observations. This is inconsistent
with Naka’s report, disclosing the stereochemistry of
sulfoxides contributes significantly to the regioselectiv-
ity, although they used TMSOTf as the activator.3 By
performing at 0°C for 20 min, the reaction transformed
(a)-4c into (b)-4c in 60% yield along with the rear-
ranged product 5c (27%). This observation suggests
that the rearrangements may proceed through the com-
mon intermediates in our cases. The isomerization of
sulfonium ion A may occur through intermolecular
path a or intramolecular path b as shown in Scheme 2,
although we can not figure out at this time which
pathway contributes predominantly. The sulfonium B
seemed to be hydrolyzed to sulfoxides during work-up,
giving sulfoxide (b)-4c. Inversion by a hydroxy anion
can be ignored because of excess TFAA which must
consume H2O quickly. Since Oae et al. have disclosed
that the Pummerer rearrangement of cyclic sulfoxides
proceeds through E2 1,2-elimination,5 carbenium ion C
might be formed from sulfonium B. A trifluoroacetate
ion attacks the C1 to terminate the reaction. Subse-
quent hydrolysis of the ester moiety during work-up
afforded 5c. Isomerizations to a-sulfoxides (a)-4a–d
were not observed by those TLC analyses in the cases
of the reaction for the corresponding b-sulfoxides.
Scheme 2.
Table 3. Bond angles ÚSꢀC1ꢀC2 and ÚSꢀC5ꢀC4 of sulf-
oxides suggested by molecular modeling calculations (6-
31G*)
ÚSꢀC1ꢀC2
Sulfide (°)
a-Sulfoxide (°)
b-Sulfoxide (°)
112.2
112.1
112.6
107.3
106.5
107.7
115.3
116.3
116.2
ÚSꢀC5ꢀC4
Sulfide (°)
a-Sulfoxide (°)
b-Sulfoxide (°)
110.9
110.6
111.1
113.9
114.6
114.3
106.3
105.4
106.5
occurred mainly in the experiments (ÚSꢀC5ꢀC4 for Y,
ÚSꢀC1ꢀC2 for Z), were suggested to be around 115°,
which approximates that for sp2 carbons. Thus, these
carbons can be easily transformed to the planar carbo-
cation by losing the proton attached. In contrast, the
angles for other sites (ÚSꢀC1ꢀC2 for Y, ÚSꢀC5ꢀC4
for Z) were estimated to be around 105°, which is
rather narrower than that of standard sp3. So, these
carbons might stay during the reactions. While there
was no remarkable difference between the angles
ÚSꢀC1ꢀC2 and ÚSꢀC5ꢀC4 for monocyclic X accord-
ing to similar calculations, the Pummerer rearrange-
ment of 4a proceeded selectively at C5. It can still be
explained by taking the Saytzeff rule into account.15
The regioselectivities are discussed next. Acetonides 4c
and 4d carry an electron-withdrawing benzoate ester
and an electron-donating MOM ether, respectively, at
their C2 positions (carbohydrate numbering). However,
these provided similar results. Accordingly, an electro-
static factor might not contribute to the regioselectivity.
On the other hand, the position of the acetonide group
dramatically influenced the selectivity as mentioned
above. Thus, a relationship between the selectivity and
conformations of the ring moiety of 4 was investigated
using molecular modeling calculations. Model com-
pounds X, Y, and Z were chosen in order to save the
time required for the calculations. The results are sum-
marized in Table 3.14
As described, the Pummerer rearrangement of 1-deoxy-
5-thioglucopyranoses with an acetonide group at the
C3-, C4-hydroxy groups were found to provide 5-
thioglucose derivatives efficiently. Since the products 5c
and 5d carry nonidentical hydroxyl protective groups
which can be distinguished in the deprotecting steps,
those can be utilized for our synthetic studies on
oligothiosaccharides.
In the cases of bicyclic Y and Z, the annular bond
angles of the carbons, where the rearrangement

4092
H. Matsuda et al. / Tetrahedron Letters 44 (2003) 4089–4093
1.27, 1.30 (each 3H%a, s, C(CH3)2 (a-anomer)), 3.66 (1H%a,
Acknowledgements
dt, J=3.9, 11.2 Hz, C5H (a-anomer)), 3.82 (1H%a, dd,
J=11.2, 11.7 Hz, C4H (a-anomer)), 4.25 (1H%a, t, J=
11.7 Hz, C3H (a-anomer)), 4.39 (1H%a, dd, J=3.9, 11.2
Hz, C6H (a-anomer)), 4.48 (1H%b, t, J=10.3 Hz, C3H
(b-anomer)), 4.71 (1H%b, dd, J=6.8, 11.2 Hz, C6H (b-
anomer)), 4.78 (1H%a, dd, J=3.9, 11.2 Hz, C6H (a-
anomer)), 5.08 (1H, brs, C1H), 5.54 (1H%a, dd, J=3.4,
11.7 Hz, C2H (a-anomer)), 5.65 (1H%b, dd, J=7.8, 10.3
Hz, C2H (b-anomer)), 6.94–7.20 (6H, aromatic protons),
8.08–8.20 (4H, aromatic protons). FDMS m/z=445 ([M+
H]+), EIMS (rel. int.) m/z=429 (1.8, [M−CH3]+), 322
(4.5, [M−PhCOOH]+), 105 (100, [PhCO]+), EIHRMS
calcd for C22H21O7S ([M−CH3]+): 429.1008, found m/z=
429.1012.
We would like to thank Professor Jun Kawabata and
Dr. Eri Fukushi of Hokkaido University for measure-
ments of mass spectra and Emeritus Professor Haruhisa
Shirahama of Hokkaido University for fruitful
discussions.
References
1. Matsumoto, T.; Shirahama, H.; Ichihara, A.; Shin, H.;
Kagawa, S.; Sakan, F.; Matsumoto, S.; Nishida, S. J.
Am. Chem. Soc. 1968, 90, 3280.
2. Konno, K.; Hashimoto, K.; Ohfune, Y.; Shirahama, H.;
Matsumoto, T. J. Am. Chem. Soc. 1988, 110, 4807.
3. Naka, T.; Minakawa, N.; Abe, H.; Kaga, D.; Matsuda,
A. J. Am. Chem. Soc. 2000, 122, 7233.
4. Matsuda, H.; Ohara, K.; Morii, Y.; Hashimoto, M.;
Miyairi, K.; Okuno, T. Bioorg. Med. Chem. Lett. 2003,
13, 1063.
5. Oae, S.; Itoh, O.; Numata, T.; Yoshimura, T. Bull. Chem.
Soc. Jpn. 1983, 56, 270.
As described in the text, 5d appeared as a broad spot on
the TLC. So, the crude residue including 5d was acety-
lated under the usual conditions, providing acetate 10d in
66% in two steps. Product 10d consisted of a 34:66
anomeric mixture. These were separated by preparative
silica gel TLC (AcOEt:benzene=3:97).
Spectral data for 10d are as follows: (a-anomer) [h]D23
1
+92° (c 9.5, CHCl3). H NMR (400 MHz, C6D6) l 1.28,
6. Crucianelli, M.; Bravo, P.; Arnone, A.; Corradi, E.;
Meille, S. V.; Zanda, M. J. Org. Chem. 2000, 65, 2965.
7. Fuzier, M.; Merrer, Y. L.; Depezay, J.-C. Tetrahedron
Lett. 1995, 36, 6443.
1.29 (each 3H, s, C(CH3)2), 1.64 (3H, s, CH3CO-), 3.20
(3H, s, CH3OCH2O-), 3.60–3.68 (2H, C4H, C5H), 4.01
(1H, dd, J=8.3, 10.8 Hz, C3H), 4.10 (1H, dd, J=3.9,
10.8 Hz, C2H), 4.32 (1H, dd, J=7.9, 11.8 Hz, C6H),
4.57, 4.72 (each 1H, d, J=6.8 Hz, CH3OCH2O-), 4.81
(1H, dd, J=3.4, 11.8 Hz, C6H), 6.41 (1H, d, J=3.9 Hz,
C1H), 6.97–8.14 (5H, aromatic protons). FDMS m/z=
426 (M+), EIMS (rel. int.) m/z=411 (4.1, [M−CH3]+), 304
(8.0, [M−PhCOOH]+), 105 (100, [PhCO]+), EIHRMS
calcd for C19H23O8S ([M−CH3]+): 411.1114, found m/z=
411.1121.
8. Merrer, Y. L.; Fuzier, M.; Dosbaa, I.; Foglietti, M.-J.;
Depezay, J.-C. Tetrahedron 1997, 53, 16731.
9. Buck, K. W.; Foster, A. B.; Pardoe, W. D.; Qadir, M. H.;
Webber, J. M. J. Chem. Soc., Chem. Commun. 1966, 759.
10. Foster, A. B.; Inch, T. D.; Qadir, M. H.; Webber, J. M.
J. Chem. Soc., Chem. Commun. 1968, 1086.
11. Lambert, J. B.; Netzel, D. A.; Sun, H.-n.; Lilianstrom, K.
K. J. Am. Chem. Soc. 1976, 98, 3778.
(b-isomer): [h]D23 −43° (c 14.5, CHCl3). 1H NMR (400
MHz, C6D6) l 1.15, 1.17 (each 3H, s, C(CH3)2), 1.41 (3H,
s, CH3CO-), 3.00 (1H, ddd, J=5.3, 7.3, 9.3 Hz, C5H),
3.11 (3H, s, CH3OCH2O-), 3.48 (1H, t, J=9.3 Hz, C3H),
3.78 (1H, t, J=9.3 Hz, C4H), 4.13 (1H, dd, J=4.9, 9.3
Hz, C2H), 4.20 (1H, dd, J=7.3, 11.7 Hz, C6H), 4.57,
4.73 (each 1H, d, J=6.8 Hz, CH3OCH2O-), 4.60 (1H, dd,
J=5.3, 11.7 Hz, C6H), 6.10 (1H, d, J=4.9 Hz, C1H),
6.85–8.08 (5H, aromatic protons). FDMS m/z=426 (M+),
EIMS (rel. int.) m/z=411 (7.3, [M−CH3]+), 304 (39,
[M−PhCOOH]+), 105 (100, [PhCO]+), EIHRMS calcd for
C19H23O8S ([M−CH3]+): 411.1114, found m/z=411.1137.
14. PC Spartan Pro version 1.08 by Wavefunction Inc. was
used for the calculations. Since some model compounds
employed for the calculations involved a sulfoxide func-
tion, ab initio method based on the 6-31G* basis set was
employed by taking the accuracy into account. When
6-31G* was employed, optimization for these model com-
pounds required 20–60 h by our systems (Athron XP
1800+, 256 Mb RAM). Since conformation search by
AM1 for sulfide X gave many stable conformations (15
conformations), only the conformation found as the most
stable by the above calculations was re-optimized by
6-31G* for sulfide and sulfoxides X. Conformation
searches for model sulfides Y and Z could be performed
directly with 6-31G*, providing three and five stable
conformations, respectively. The conformations with
12. Actually, treatment of 6b with triethylamine in MeOH
gave 8b (8.0%), 9b (17%), recovered 6b (25%), and methyl
ether 11 (12%).
13. Typical reaction conditions are as follows:
A mixture of (a)-4c (40.5 mg, 91.2 mmol) and TFAA (30
mL, 212 mmol) in pyridine (1.5 mL, 594 mmol) was
stirred at room temperature for 2 h. After MeOH (1.0
mL) was added, the mixture was concentrated in vacuo
to give the crude 6d as an oil, which was treated with
Ac2O (300 mL, excess) in pyridine (600 mL) for 2 h. After
concentration in vacuo, silica gel column chromatogra-
phy of the residue (AcOEt:hexane=10:90) gave
a
diastereomeric mixture of 6c (26.5 mg, 66%) as an oil. IR
(film) 3440, 2985, 2930, 1720, 1270, 1110, 1070, 710 cm−1
.
The 1H NMR spectrum of this sample showed that it
consists of the two anomers (a:b=90:10). Assignments of
signals for the main a-anomer and some for the minor
1
b-isomer are described. H NMR (400 MHz, C6D6, a:b=
90:10) 1.24, 1.26 (each 3H%b, s, C(CH3)2 (b-anomer)),

H. Matsuda et al. / Tetrahedron Letters 44 (2003) 4089–4093
4093
minimum energies found by the above calculations were
employed as the initial geometries for optimization of
the corresponding sulfoxides.
anion (TFA−) also might be important for this kind of
consideration.
15. Our preliminary calculations for carbenium ions 12 and
13 based on 6-31G* suggested that 12 is 9 kcal/mol
more stable than that of 13. This result may explain
the selectivity of the rearrangement. However, that
energy difference obtained is too large to explain the
ratio in the experiment. The existence of the counter