The stereoselective synthesis of 4'-β-thioribonucleosides via the pummerer reaction

Article abstract of DOI:10.1021/ja000541o

An efficient stereoselective synthesis of 4'-β-thioribonucleosides 14, 15, 27, and 30 using the Pummerer reaction as the key step is described. The Pummerer reaction of 1,4-anhydro-2-O-(2,4-dimethoxybenzoyl)-3,5-O-(1,1,3,3-tetraisopro pyldisiloxane-1,3-di

Full text of DOI:10.1021/ja000541o

J. Am. Chem. Soc. 2000, 122, 7233-7243
7233
The Stereoselective Synthesis of 4′-â-Thioribonucleosides via the
Pummerer Reaction†
Takashi Naka, Noriaki Minakawa, Hiroshi Abe, Daisuke Kaga, and Akira Matsuda*
Contribution from the Graduate School of Pharmaceutical Sciences, Hokkaido UniVersity, Kita-12,
Nishi-6, Kita-ku, Sapporo 060-0812, Japan
ReceiVed February 14, 2000
Abstract: An efficient stereoselective synthesis of 4′-â-thioribonucleosides 14, 15, 27, and 30 using the
Pummerer reaction as the key step is described. The Pummerer reaction of 1,4-anhydro-2-O-(2,4-
dimethoxybenzoyl)-3,5-O-(1,1,3,3-tetraisopropyldisiloxane-1,3-diyl)-4-sulfinyl-D-ribitol (R-10:S-10 ) 2.7:1)
in the presence of silylated uracil afforded the desired â-anomer of the 4′-thiouridine derivative 11 in 66%
yield without formation of its R-anomer. The reaction with R-10 gave 11 in 87% yield, while the one with
S-10 resulted in a 27% decrease of the desired product 11 along with a 22% yield of 3,6-O-(1,1,3,3-
tetraisopropyldisiloxane-1,3-diyl)-3-hydroxy-2-hydroxymethylthiophene (12). A likely explanation for the
observed difference in the reaction of R-10 and S-10 is that the reaction proceeds via an E2 type pathway,
which prefers anti elimination. Thus, R-10 would preferentially afford the R-thiocarbocation intermediate 21
via an E2 anti elimination under the reaction conditions. The resulting 21 would be expected to react with
silylated uracil stereoselectively to give 11 in good yield. However, formation of the more stable tertiary
R-thiocarbocation intermediate 23, which would prefer to give 12 and/or to decompose, would compete with
the formation of the desired 21 in the reaction with S-10. Consequently, this argument would explain the low
yields of the desired product 11 and the poor mass balance in the reaction with S-10. When the sulfoxide 10
(R-10:S-10 ) >16:1) prepared by oxidation of 9 with ozone was used for the Pummerer reaction, the desired
11 was obtained in 80% yield. Compound 11 was converted to 4′-â-thiouridine (14) by treatment of 11 with
ammonium fluoride, followed by methanolic ammonia. Similarly, 4′-â-thiocytidine (15) was prepared when
silylated N4-acetylcytosine was used in the Pummerer reaction. For the Pummerer reactions with purine bases,
6-chloropurine and 2-amino-6-chloropurine were found to be the most suitable. When the reactions were
conducted in a mixture of acetonitrile and 1,2-dichloroethane at room temperature, followed by reflux, the
desired products 25 and 28 were obtained in 65% and 56% yields, respectively. These compounds were then
converted to 4′-â-thioadenosine (27) and 4′-â-thioguanosine (30) under the usual conditions. This is therefore
the first time that the stereoselective synthesis of 4′-â-thioribonucleosides has been performed using the
neighboring group participation of the Pummerer reaction.
Introduction
herpes simplex virus type-1 and varicella zoster virus, BVDU
is rapidly metabolized to the inactive E-5-(2-bromovinyl)uracil
and 2-deoxyribose 1-phosphate by pyrimidine nucleoside phos-
Nucleoside antimetabolites occupy a pivotal position in the
search for effective anticancer and antiviral agents. Hence, much
attention has been focused on efforts to synthesize and evaluate
new nucleoside analogues. For example, 4′-thionucleosides, in
which the furanose ring oxygen is replaced by a sulfur atom,
have been studied extensively over the past 10 years because
of their potent biological activity.1 In 1964, 4′-thioadenosine
was synthesized as the first example of this class of compounds
by Reist et al.2 Although further examples were reported,3 stu-
dies in this area declined due to unfavorable results of biological
evaluation and difficulty in devising an efficient and large scale
preparation of the requisite 4′-thiosugars. The next significant
attempts to synthesize 4′-thionucleosides were initiated inde-
pendently by Dyson et al.1a with E-5-(2-bromovinyl)-4′-thio-
2′-deoxyuridine (4′-thioBVDU) and by Secrist et al.1b with their
report of 2′-deoxy-4′-thiopyrimidine nucleosides. Although the
parent compound, i.e., E-5-(2-bromovinyl)-2′-deoxyuridine
(BVDU),4 is known to be a potent and selective inhibitor of
(1) For examples, see: (a) Dyson, M. R.; Coe, P. L.; Walker, R. T. J.
Med. Chem. 1991, 34, 2782-2786. (b) Secrist, J. A., III.; Tiwari, K. N.;
Riordan, J. M.; Montgomery, J. A. J. Med. Chem. 1991, 34, 2361-2366.
(c) Van Draanen, N. A.; Freeman, G. A.; Short, S. A.; Harvey, R.; Jansen,
R.; Szczech, G.; Koszalka, G. W. J. Med. Chem. 1996, 39, 538-542. (d)
Rahim, S. G.; Trivedi, N.; Bogunovic-Batchelor M. V.; Hardy, G. W.; Mills,
G.; Selway, J. W. T.; Snowden, W.; Littler, E.; Coe, P. L.; Basnak, I.;
Whale, R. F.; Walker, R. T. J. Med. Chem. 1996, 39, 789-795. (e)
Yoshimura, Y.; Kitano, K.; Yamada, K.; Satoh, H.; Watanabe, M.; Miura,
S.; Sakata, S.; Sasaki, T.; Matsuda, A. J. Org. Chem. 1997, 62, 3140-
3152. (f) Yoshimura, Y.; Watanabe, M.; Satoh, H.; Ashida, N.; Ijichi, K.;
Sakata, S.; Machida, H.; Matsuda, A. J. Med. Chem. 1997, 40, 2177-2183.
(g) Secrist, J. A., III; Tiwari, K. N.; Shortnacy-Fowler, A. T.; Messini, L.;
Riordan, J. M.; Montgomery, J. A.; Meyers, S. C.; Ealick, S. E. J. Med.
Chem. 1998, 41, 3865-3871.
(2) Reist, E. J.; Gueffroy, D. E.; Goodman, L. J. Am. Chem. Soc. 1964,
86, 5658-5663.
(3) For examples, see; (a) Reist, E. J.; Fisher, L. V.; Goodman, L. J.
Org. Chem. 1968, 33, 189-192. (b) Bobek, M.; Whistler, R. L.; Bloch, A.
J. Med. Chem. 1970, 13, 411-413. (c) Ritchie, R. G. S.; Szarek, W. A. J.
Chem. Soc., Chem. Commun. 1973, 686-687. (d) Bobek, M.; Bloch, A.;
Parthasarathy, R.; Whistler, R. L. J. Med. Chem. 1975, 18, 784-787. (e)
Pickering, M. V.; Witkowski, J. T.; Robins, R. K. J. Med. Chem. 1976, 19,
841-842.
* To whom correspondence and reprint requests should be addressed.
Phone: +81-11-706-3228. Fax: +81-11-706-4980. E-mail: matuda@pharm.
hokudai.ac.jp.
† This paper constitutes Part 198 of Nucleosides and Nucleotides. Part
197: Abe, H.; Shuto, S.; Matsuda, A. J. Org. Chem. In press.
(4) Jones, A. S.; Verhelst, G.; Walker, R. T. Tetrahedron Lett. 1979,
4415-4418.
10.1021/ja000541o CCC: $19.00 © 2000 American Chemical Society
Published on Web 07/14/2000

7234 J. Am. Chem. Soc., Vol. 122, No. 30, 2000
Naka et al.
Scheme 1a
a Reagents: (a) 4 N HCl, dioxane, 100 °C; (b) NaBH4, MeOH; (c) TBDMSCl, imidazole, DMF; (d) p-nitrobenzoic acid, diisopropyl
azodicarboxylate, PPh3, THF; (e) NaOMe, MeOH; (f) TBAF, THF; (g) methanesulfonyl chloride, pyridine; (h) Na2S, DMF, 100 °C; (i) BCl3,
CH2Cl2, <-90 °C.
phorylase.5 In contrast, 4′-thioBVDU is resistant to pyrimidine
nucleoside phosphorylase, and showed a higher chemothera-
peutic index than BVDU.1a,6 Also the significant cytotoxicity
arising from replacement of the furanose ring oxygen in the
naturally occurring pyrimidine 2′-deoxyribonucleosides with a
sulfur atom reported by Secrist et al. attracted the interest of
medicinal chemists. These intriguing reports revived interest in
the synthesis and evaluation of 4′-thionucleosides, including the
replacement of the furanose ring oxygen in biologically active
nucleosides with a sulfur atom and investigation of their
L-isomers.1,7 Moreover, 4′-thionucleosides could be used as
antisense components, because oligonucleotides containing 4′-
thioribonucleosides showed high nuclease resistance and thermal
stability.8
The desired 4′-thionucleosides have generally been synthe-
sized by classical thioglycosidation of the corresponding thio-
sugars and nucleobases. As an alternative method, we1e,9 and
others10 developed the Pummerer reaction in order to condense
a nucleobase and a sulfoxide. However, despite numerous
attempts to synthesize 4′-thionucleoside analogues, little atten-
tion was paid to the stereoselectivity of the resulting 4′-
thionucleosides.11 Surprisingly, the stereocontrol in the thiogly-
cosidation is unsatisfactory even with the assistance of the
neighboring C-2 acetoxy group, unlike the normal glycosidation
between a ribofuranose derivative and a nucleobase.3d,7b Thus
far, there have been no systematic studies on the correlation
between the stereoselectivity and the neighboring group effect
on an adjacent acyloxy group. Based on these considerations,
we envisioned the stereoselective synthesis of 4′-â-thioribo-
nucleosides, which would be useful as ribonucleoside units for
not only sugar-modified 4′-thionucleosides synthesis but also
functionalized RNA molecule synthesis.
As part of our program, we recently reported the stereose-
lective coupling of thymine with meso-thiolane-3,4-diol-1-oxide
derivative via the Pummerer reaction.12 We indicated that the
R-thiocarbocation intermediates were less susceptible to neigh-
boring group effects than those of oxocarbocations. Conse-
quently, the stereoselective coupling was achieved by the intro-
duction of 2,4-dimethoxybenzoyl groups to the hydroxyl groups
of meso-thiolane-3,4-diol-1-oxide to give (2R*,3R*,4S*)-1-[3,4-
di-O-(2,4-dimethoxybenzoyl)thiolane-3,4-diol-2-yl]thymine.
Herein, we describe the efficient and stereoselective synthesis
of 4′-â-thioribonucleosides, based on our preliminary investiga-
tions using the Pummerer reaction. During our studies, we found
that a large difference exists between each diastereomer of the
starting sulfoxides in the Pummerer reaction. We also provide
some explanations for the observed differences from a mecha-
nistic point of view.
(5) Desgranges, C.; Razaka, G.; Rabaud, M.; Bricaud, H.; Balzarini, J.;
De Clercq, E. Biochem. Pharmacol. 1983, 32, 3583-3590.
(6) Rahim, S. G.; Littler, E.; Powell, K. L.; Collins, P.; Coe, P. L.; Dyson,
M. R.; Walker, R. T. Proc. 31st ICAAC (Chicago) 1991, 1232.
(7) For examples: (a) Secrist, J. A., III; Riggs, R. M.; Tiwari, K. N.;
Montgomery, J. A. J. Med. Chem. 1992, 35, 533-538. (b) Tiwari, K. N.;
Secrist, J. A., III; Montgomery, J. A. Nucleosides Nucleotides 1994, 13,
1819-1828. (c) Branalt, J.; Kvarnstrom, I.; Svensson, S. C. T.; Classon,
B.; Samuelsson, B. J. Org. Chem. 1994, 59, 4430-4432. (d) Uenishi, J.;
Takahashi, K.; Motoyama, M.; Akashi, H.; Sasaki, T. Nucleosides Nucle-
otides 1994, 13, 1347-1361. (e) Young, R. J.; Shaw-Ponter, S.; Thomson,
J. B.; Miller, J. A.; Cumming, J. G.; Pugh, A. W.; Rider, P. Bioorg. Med.
Chem. Lett. 1995, 5, 2599-2604. (f) Marquez, V. E.; Jeong, L. S.; Nicklaus,
M. C.; George, C. Nucleosides Nucleotides 1995, 14, 555-558. (g)
Yoshimura, Y.; Kitano, K.; Watanabe, M.; Satoh, H.; Sakata, S.; Miura,
S.; Ashida, N.; Machida, H.; Matsuda, A. Nucleosides Nucleotides 1997,
16, 1103-1106. (h) Jeong, L. K.; Moon, H. R.; Choi, Y. J.; Chun, M. W.;
Kim, H. O. J. Org. Chem. 1998, 63, 4821-4825.
(8) For examples: (a) Leydier, C.; Bellon, L.; Barascut, J. L.; Imbach,
J. L. Nucleosides Nucleotides 1995, 14, 1027-1030. (b) Boggon, T. J.;
Hancox, E. L.; McAuley-Hecht, K. E.; Connolly, B. A.; Hunter, W. N.;
Brown, T.; Walker, R. T.; Leonard, G. A. Nucleic Acids Res. 1996, 24,
951-961. (c) Jones, G. D.; Altmann, K.-H.; Husken, D.; Walker, R. T.
Bioorg. Med. Chem. Lett. 1997, 7, 1275-1278.
(9) Nishizono, N.; Koike, N.; Yamagata, Y.; Matsuda, A. Tetrahedron
Lett. 1996, 37, 7569-7572.
(10) O’Neil, I. A.; Hamilton, K. M. Synlett 1992, 791-792.
Results and Discussion
Preparation of the 4′-Thiosugar Portion. To conduct the
Pummerer reaction, 1,4-anhydro-4-thio-D-ribitol (7) is needed.
Although Althenbach et al. have reported the synthesis of 7
from achiral thiophene-2-carboxylic acid,13 the method does not
provide 7 efficiently because the strategy consists of enzymatic
resolution and long reaction times for oxidation of the double
bond. Thus, an alternative method was necessary for large-scale
preparation. Among the efforts to synthesize the 4′-thioribo-
furanose derivatives, the synthetic tactics involving two con-
secutive SN2 reactions of D-ribose reported by Dyson et al.14
and Leydier et al.15 seemed most promising. Accordingly,
compound 7 was prepared as shown in Scheme 1. Methyl 2,3,5-
(12) Naka, T.; Nishizono, N.; Minakawa, N.; Matsuda, A. Tetrahedron
Lett. 1999, 40, 6297-6300.
(13) Altenbach, H.-J.; Brauer, D. J.; Merhof, G. F. Tetrahedron 1997,
53, 6019-6026.
(14) Dyson, M. R.; Coe, P. L.; Walker, R. T. Carbohydr. Res. 1991,
216, 237-248.
(11) (a) Haraguchi, K.; Nishikawa, A.; Sasakura, E.; Tanaka, H.;
Nakamura, K. T.; Miyasaka, T. Tetrahedron Lett. 1998, 39, 3713-3716.
(b) Inguaggiato, G.; Jasamai, M.; Smith, J. E.; Slater, M.; Simons, C.
Nucleosides Nucleotides 1999, 18, 457-467.
(15) Leydier, C.; Bellon, L.; Barascut, J.-L.; Deydier, J.; Maury, G.;
Pelicano, H.; Alaoui, M. A. E.; Imbach, J.-L. Nucleosides Nucleotides 1994,
13, 2035-2050.

StereoselectiVe Synthesis of 4′-â-Thioribonucleosides
J. Am. Chem. Soc., Vol. 122, No. 30, 2000 7235
tri-O-benzyl-D-ribofuranoside (1) was synthesized from com-
mercially available D-ribose using the Barker and Fletcher
procedure.16 Acidic methanolysis of 1, followed by reduction
of the resulting lactol derivative with sodium borohydride, gave
diol 2 in 87% yield. After the selective protection of the primary
hydroxyl group of 2 with tert-butyldimethylsilyl chloride, 3 was
converted to the L-lyxose derivative 4 by the Mitsunobu reaction
using p-nitrobenzoic acid. Deprotection of the p-nitrobenzoyl
and tert-butyldimethylsilyl groups was performed by treatment
of 4 with NaOMe, followed by tetrabutylammonium fluoride
(TBAF) to give the diol 5 in 81% yield. Reaction of 5 with
methanesulfonyl chloride in pyridine gave the dimesylate, which
was treated with sodium sulfide in DMF to give 6 in 84% yield.
When the tribenzylated derivative 6 was treated with boron
trichloride in dichloromethane at -78 °C, the usual conditions
for debenzylation of 4′-thiosugar derivatives,1e,7b the reaction
gave a complex mixture, with the desired 7 being isolated in
poor yield. The reaction was thus carried out at -98 °C and
quenched by addition of MeOH below -90 °C. As a result, 7
was obtained in 79% yield. Control of the temperature is critical
and quenching the reaction above -90 °C resulted in a reduced
isolated yield of 7.17 Consequently, 7 was obtained in 29% yield
by an 11-step synthesis from inexpensive D-ribose.
Scheme 2a
a Reagents: (a) 1,3-dichloro-1,1,3,3-tetraisopropyldisiloxane, pyri-
dine; (b) 2,4-dimethoxybenzoyl chloride, pyridine; (c) m-CPBA,
CH2Cl2, -40 °C; (d) O3, CH2Cl2, -78 °C.
To achieve the stereoselective synthesis of 4′-â-thioribo-
nucleosides, introduction of the 2,4-dimethoxybenzoyl group
on the hydroxyl group at the 2-position was necessary for the
Pummerer reaction.12 Compared to reaction with the meso-
thiolane-3,4-diol-1-oxide derivative described in the Introduc-
tion, control of not only the stereoselectivity (at the anomeric
position), but also the regioselectivity (at the anomeric position
or the 4-position), is required. The regioselectivity of the
Pummerer reaction is likely to depend on the acidity of the
R-proton.18 In addition, steric hindrance may also affect ste-
reoselectivity.19 Hence, electron-donating and sterically hinder-
ing protecting groups would be preferable as protecting groups
of the hydroxyl groups at the 3- and 5-positions to avoid
formation of the regioisomer. Accordingly, 7 was treated with
1,3-dichloro-1,1,3,3-tetraisopropyldisiloxane in pyridine to give
8, which was subsequently converted to 9 (79% and 99%,
respectively). Oxidation of 9 by m-chloroperoxybenzoic acid
(m-CPBA) gave the desired sulfoxide 10 in 82% yield as a
diastereomeric mixture (R/S ) 2.7:1) (Scheme 2). The configu-
rations of each diastereomer, R-10 and S-10 (Figure 1), were
assigned through a study of solvent- and Eu(dpm)3-induced
shifts in their 1H NMR spectra as reported by Folli et al.20 After
separation of the diastereomers, assignment of proton signals,
especially the Ha and Hb protons of each compound, was based
on two-dimensional NMR and nOe experiments. In the major
isomer, the Ha signal was observed at 3.57 ppm in CDCl3 while
the Hb signal was observed upfield at 2.89 ppm. Unlike the
major isomer, in the minor isomer, the Ha signal was observed
at 3.05 ppm, which appeared farther upfield than the Hb signal
(3.70 ppm). Since a greater deshielding is expected for the
Figure 1. Structures of R-10 and S-10.
Figure 2. Chemical shifts of Ha and Hb of S-10 in CDCl3, as a function
of concentration of Eu(dpm)3.
proton at the R-position closer to the sulfinyl oxygen atom, the
configuration of the major isomer was assigned as R, while S
was assigned for the minor isomer. Further evidence was
obtained by a comparison of the downfield shifts of the Ha and
Hb signals in the presence of increasing amounts of Eu(dmp)3.
As shown in Figure 2, a larger downfield shift of the Ha signal
of S-10, which is situated in a cis-orientation relative to the
sulfinyl oxygen, was observed on increasing the concentration
of the shift reagent while the opposite effect was observed in
the major isomer, and consequently, its configuration was
assigned as R (data not shown).
Synthesis of 4′-â-Thioribopyrimidine Nucleosides. In a
previous report,12 we achieved the â-selective coupling of
thymine with meso-3,4-di-O-(2,4-dimethoxybenzoyl)thiolane-
3,4-diol-1-oxide using the Pummerer reaction. The optimized
conditions were used for the reaction of 10 with the silylated
uracil. Accordingly, a solution of the silylated uracil in a mixture
of toluene-CH2Cl2 containing an excess amount of triethyl-
amine and trimethylsilyl trifluoromethanesulfonate (TMSOTf)
was added to a solution of 10 (R/S ) 2.7:1) in CH2Cl2 at 0 °C.
The reaction proceeded immediately, and the desired 4′-â-
(16) Barker, R.; Fletcher, H. G., Jr. J. Org. Chem. 1961, 26, 4605-4609.
(17) Hancox et al. also reported the importance of maintaining low
temperature throughout the debenzylation reaction and during the quenching
process. In our reaction with boron trichloride, the quenching of the reaction
by adding MeOH is an extremely exothermic process. This may contribute
to forming a bicyclic episulfonium intermediate, which gives a complex
mixture, but not the desired product; see: Hancox, E. L.; Walker, R. T.
Nucleosides Nucleotides 1996, 15, 135-148.
(18) For examples, see: (a) Johnson, C. R.; Sharp, J. C.; Phillips, W. G.
Tetrahedron Lett. 1967, 5299-5302. (b) Davenport, D. A.; Moss, D. B.;
Rhodes, J. E.; Walsh, J. A. J. Org. Chem. 1969, 34, 3353-3359.
(19) Jones, D. N.; Helmy, E.; Whitehouse, R. D. J. Chem. Soc., Perkin
Trans. 1 1972, 1329-1335.
(20) Folli, U.; Iarossi, D.; Taddei, F. J. Chem. Soc., Perkin Trans. 2
1974, 933-937.

7236 J. Am. Chem. Soc., Vol. 122, No. 30, 2000
Naka et al.
Table 1. The Pummerer Reaction of 10 with Pyrimidine Bases
entry
R/S ratio of 10
nucleobase
uracil
uracil
uracil
uracil
conditions
11 or 13 (%)
12 (%)
1
2
3
4
5
6
7
2.7:1
2.7:1
2.7:1
R only
S only
>16:1
>16:1
TMSOTf (8 equiv), Et3N (8 equiv), 0 °C
TMSOTf (8 equiv), Et3N (8 equiv), rt
TMSOTf (8 equiv), Et3N (4 equiv), followed by Et3N (4 equiv), rt
TMSOTf (8 equiv), Et3N (4 equiv), followed by Et3N (4 equiv), rt
TMSOTf (8 equiv), Et3N (4 equiv), followed by Et3N (4 equiv), rt
TMSOTf (8 equiv), Et3N (4 equiv), followed by Et3N (4 equiv), rt
TMSOTf (8 equiv), Et3N (2 equiv), followed by Et3N (6 equiv), rt
52
57
66
87
27
80
75
7
8
9
0
uracil
22
trace
trace
uracil
N4-acetylcytosine
Scheme 3a
Scheme 4a
a Reagents: (a) 1,3-dichloro-1,1,3,3-tetraisopropyldisiloxane, pyri-
dine; (b) 2,4-dimethoxybenzoyl chloride, Et3N, DMAP, CH3CN, reflux.
a Reagents: (a) NH4F, MeOH, reflux; (b) NH3/MeOH.
reaction. To confirm the absence of the R-isomer, compound
18 was prepared using an alternative method (Scheme 4). The
4′-R-thiouridine (16) was prepared according to the method
reported by Bellon et al.21 After protection of the hydroxyl
groups at the 3′- and 5′-position by a TIPDS group to give 17,
introduction of a 2,4-dimethoxybenzoyl group on the remaining
hydroxyl group was examined. When 17 was treated with 2,4-
dimethoxybenzoyl chloride in pyridine at room temperature,
none of the desired 18 was obtained due to steric hindrance.
However, when the reaction was carried out in the presence of
triethylamine and DMAP in acetonitrile under reflux conditions,
compound 18 was obtained in 39% yield. Although both 11
and 18 showed almost the same Rf values on TLC analysis,
characteristic differences were observed in the 1H NMR spectra.
For example, the anomeric proton signal of 11 was observed at
6.00 ppm as a singlet arising from the predominance of the C-3′-
endo conformation, whereas that of 18 was observed at 6.52
ppm as a doublet. No proton signals corresponding to 18 were
observed in compound 11 obtained from the Pummerer reaction.
Furthermore, the other possible regioisomers, having a thiogly-
cosidic linkage at the 4′-position, were not obtained either. These
results therefore imply that the stereoselective and regioselective
coupling of the sulfoxide 10 with pyrimidines was achieved
via the Pummerer reaction.
We were next interested in determining if there was a
difference between the Pummerer reaction and the classical
thioglycosidation. Both reactions are thought to proceed via
R-thiocarbocation intermediates, although the mechanism of
their formation differs in each. Thus, similar stereoselectivity
would be expected with the assistance of the 2,4-dimethoxy-
benzoyl group even in the classical thioglycosidation. As shown
in Scheme 5, we examined the thioglycosidation of the silylated
uracil with the 1-acetoxy derivative 19. Compound 10 was
converted into 19 by treatment with acetic anhydride. Interest-
ingly, the neighboring group assistance of the 2,4-dimethoxy-
benzoyl group was not affected and the reaction gave 19 as a
diastereomeric mixture (R:â ) 1:1), which was then subjected
thiouridine derivative 11 was obtained stereoselectively in 52%
yield, along with a 7% yield of the thiophene derivative 12
(Table 1, entry 1). When the reaction was run at room
temperature, 11 was obtained in 57% yield (entry 2). As
described in the previous communication, formation of the
thiophene derivative 12 was caused by the presence of triethyl-
amine. Accordingly, the reaction was carried out by adding
triethylamine in two portions as shown in entry 3 (see the
Experimental Section). Although formation of the unfavorable
12 was not suppressed, the yield of 11 was increased to 66%.
Further changing conditions such as solvent, base, and reaction
temperature did not improve the overall reaction. During the
course of our investigations, we found that a large difference
in reactivity existed between each diastereomer, i.e., R-10 and
S-10. Surprisingly, when the separated R-10 was subjected to
the Pummerer reaction, 11 was obtained in 87% yield (entry
4). Furthermore, none of the undesired product 12 was observed.
In contrast, the reaction with S-10 gave 11 in only 27% yield,
along with a 22% yield of 12 (entry 5). As can be readily seen
from these results, R-10 is much more suitable for the Pummerer
reaction to give 11 than S-10. Hence, we examined the oxidation
of 9 with the idea of improving the R-selectivity. When 9 was
treated with ozone in CH2Cl2 at -78 °C, the sulfoxide 10 was
obtained in 82% yield with >16:1 R/S ratio (see the Experi-
mental Section). The Pummerer reaction using 10 (R/S ) >16:
1) with the silylated uracil gave 11 in 80% yield with only a
trace amount of 12 (entry 6). For the synthesis of the 4′-â-
thiocytidine derivative 13, the optimal results were obtained
when the reaction was conducted with the silylated N4-
acetylcytosine, as shown in entry 7. The 4′-â-thiouridine (14)
and the 4′-â-thiocytidine (15) were obtained by treatment of
11 and 13 with ammonium fluoride in MeOH under reflux,
followed by methanolic ammonia, respectively (Scheme 3). The
structures of 14 and 15 were confirmed by comparison of the
analytical data with those reported by Imbach et al.15,21
In all of our attempts, none of the 4′-R-thiouridine derivative
18 or its cytidine derivative was detected in the Pummerer
(21) Bellon, L.; Barascut, J.-L.; Imbach, J.-L. Nucleosides Nucleotides
1992, 11, 1467-1479.

StereoselectiVe Synthesis of 4′-â-Thioribonucleosides
J. Am. Chem. Soc., Vol. 122, No. 30, 2000 7237
Scheme 5a
should exist in the subsequent E2 elimination step to give the
R-thiocarbocation intermediates. In the E2 elimination, it is well
known that anti elimination is greatly favored over syn
elimination. This tendency is also maintained in the Pummerer
reaction and has been demonstrated by Oae et al.25 and Kita et
al.26 in the reaction of conformationally rigid cyclic sulfoxides,
that is, deuterated 1-thiadecalin 1-oxides. In the silylated
sulfoxide 20, there is a single proton, i.e., H-1â, which has an
anti orientation with the leaving group to give the R-thiocar-
bocation intermediate 21 predominantly via an E2 elimination
(path a). The resulting intermediate 21 is expected to react with
a silylated nucleobase with neighboring group participation to
give 4′-â-thioribonucleoside 11 in good yield. In contrast, two
properly positioned protons for E2 anti elimination, i.e., H-1R
and H-4, are present in the silylated sulfoxide 22. Consequently,
formation of the more stable tertiary cation intermediate 23 via
path b would compete with the formation of the desired 21.
Since 23 is thought to be less reactive than 21 toward attack by
a silylated nucleobase due to steric hindrance, 23 would prefer
to give the thiophene 12 and/or to decompose, but not give
possible regioisomers, which have a thioglycosidic linkage at
the 4′-position. This would explain the low yields of the desired
11 and the poor mass balance of the Pummerer reaction with
S-10. R-10 and S-10 differ from the six-membered cyclic
sulfoxides such as 1-thiadecalin 1-oxides, in that they are rather
perturbational molecules, and it is difficult to estimate the exact
dihedral angles between the hydrogen atoms and the leaving
group for E2 elimination. In addition, the cyclic protecting group
of the hydroxyl groups at the 3- and 5-position of 10, which
forces the sugar pucker mode to the C-3-endo conformation,27
may also be a contributing factor in explaining the differences;
however, the considerations as illustrated in Scheme 6 would
be one of the explanations of the observed chemical reactivity
differences.
a Reagents: (a) Ac2O, 110 °C; (b) uracil, N,O-bis(trimethylsilyl)-
acetamide, TMSOTf, CH3CN.
to thioglycosidation conditions. According to the method
reported by Bellon et al.,21 19 was treated with the silylated
uracil in acetonitrile in the presence of TMSOTf. However, the
starting material 19 was not consumed even after 39 h, and the
desired product 11 was obtained in only 35% yield along with
a 13% yield of 19. Although the stereochemistry of the resulting
11 was â-selective as expected, the isolated yield was not
satisfactory. When subjected to the same conditions, the reaction
between 2,3,5-tri-O-benzyl-1-O-acetyl-4-thio-D-ribofuranose and
the silylated uracil gave the 4′-thiouridine derivative in 74%
yield.21 However, the product was an R/â mixture. These results
may be explained in terms of the so-called armed-disarmed
principle reported by Fraser-Reid et al.22 It is well known that
the glycosyl acceptors possessing an ether type of protecting
group on the C-2 hydroxyl group are much more reactive than
those possessing an ester type of protecting group. The same
differences would be expected in the case of 4-thiosugar
derivatives. No improvement was observed even using the
Vorbru¨ggen method.23 Although further attempts with other
Lewis acids, solvents, and leaving groups at the C-1 position
have not been conducted, it can be concluded that the Pummerer
reaction is an efficient alternative method of synthesizing 4′-
â-thioribopyrimidine nucleosides stereoselectively.
Considerations of the Differences in Chemical Reactivity
between the Two Diastereomeric Sulfoxides. As described
in the previous communication,12 the Pummerer reactions
between each diastereomer of the meso-3,4-di-O-(2,4-dimethoxy-
benzoyl)thiolane-3,4-diol-1-oxides, that is, between the cis- and
trans-sulfoxides and the silylated thymine, resulted in similar
products distribution. In contrast, striking differences were
observed between R-10 and S-10 in a similar Pummerer reaction.
Thus, the Pummerer reaction with R-10 gave the desired 11 in
good yield while the reaction with S-10 gave 11 in only 27%
yield along with a 22% yield of the undesired thiophene 12.
Moreover, the mass balance of the reaction with S-10 was poor.
It was presumed that these observed chemical reactivity dif-
ferences would arise from differences in the ease of formation
of the R-thiocarbocation intermediates, which would afford the
desired compound 11 from each diastereoisomer.
We further examined the molecular orbital theoretical argu-
ment for the above hypothesis. Fukui and Fujimoto have
demonstrated good parallelism between the reactivity of hy-
drogen atoms and the frontier electron density of the LUMO in
E2 eliminations.28 Thus, a hydrogen atom which possesses a
larger frontier electron density of the LUMO is expected to
participate in E2 elimination. Based on these considerations,
the electron density of the LUMO at each hydrogen atom
(H-1R,â and H-4) of the intermediates 20 and 22 was calcu-
lated.29 As can be seen in Figure 3, the resulting electron
densities agree well with the experimental results in the
Pummerer reaction. In the intermediate 20, the H-1â possesses
a much larger frontier electron density of the LUMO than H-1R
and H-4, implying that elimination of H-1â is preferable to
elimination of H-1R and H-4, as we had speculated. In contrast,
To date, numerous investigations have been carried out to
clarify the mechanism of the Pummerer reaction.24 Depending
on the structures of the starting sulfoxides, both E1cb and E2
type pathways are known to give the R-thiocarbocation inter-
mediates from sulfoxides. The differences in our reaction can
be explained if the formation of the R-thiocarbocation inter-
mediates occurs via the E2 type pathway. As illustrated in
Scheme 6, trimethylsilylation of the sulfoxide by TMSOTf is
the initial step of the Pummerer reaction, and this process is
expected to occur immediately in both R-10 and S-10 to give
20 and 22, respectively. It appears that the definitive differences
(25) Oae, S.; Itoh, O.; Numata, T.; Yoshimura, T. Bull. Chem. Soc. Jpn.
1983, 56, 270-279.
(26) Kita, Y.; Shibata, N.; Yoshida, N.; Kawano, N.; Fujimori, C.;
Yoshikawa, N.; Fujita, S. J. Chem. Soc., Perkin Trans. 1 1995, 2829-
2834.
(27) The sugar pucker modes of R-10 and S-10 were predicted from the
coupling constants of J1R,2 and J3,4 which were the same as for the ribose.
The coupling constants of J1R,2 and J3,4 of R-10 were 0 and 12.0 Hz, and
those of S-10 were 0.9 and 10.3 Hz, respectively. Consequently, the sugar
pucker modes of R-10 and S-10 were both estimated to be preferentially
C-3-endo conformations.
(28) (a) Fukui, K.; Fujimoto, H. Tetrahedron Lett. 1965, 4303-4307.
(b) Fukui, K.; Fujimoto, H. Frontier Orbitals and Reaction Path; World
Scientific: Singapore, 1997; pp 171-202.
(22) Mootoo, D. R.; Konradsson, P.; Udodong, U.; Fraser-Reid, B. J.
Am. Chem. Soc. 1988, 110, 5583-5584.
(23) Vorbru¨ggen, H.; Krolikiewicz, K.; Bennua, B. Chem. Ber. 1981,
114, 1234-1255.
(29) The preoptimal geometries of the intermediates 20 and 22 were
calculated by MM2 methods. The final optimal conformations and a
coefficient of the LUMO at each hydrogen atom were obtained by PM3
methods. The frontier electron density was calculated from 2 times the
square of the coefficient of the LUMO.
(24) Grierson, D. S.; Husson, H.-P. In ComprehensiVe Organic Synthesis;
Trost, B. M., Ed.; Pergamon Press: Oxford, 1991; Vol. 6, pp 909-947
and references therein.

7238 J. Am. Chem. Soc., Vol. 122, No. 30, 2000
Naka et al.
Scheme 6
product was assigned as the N-7 isomer 24, while the minor
was the desired N-9 isomer 25.34 Since the kinetic N-7 isomer
is expected to rearrange into the N-9 isomer, the reaction
conditions were further investigated. Interestingly, rearrangement
of 24 into the desired 25 was observed on TLC analysis when
the Pummerer reaction was conducted first at room temperature
then under reflux conditions. Accordingly, 25 was obtained in
65% yield under the conditions shown in entry 4. Conversion
of 25 into 4′-â-thioadenosine (27) by treatment with TBAF,
followed by ethanolic ammmonia,35 gave 27 in good yield
(Scheme 7). The structure of 27 was confirmed by comparison
of the analytical data with those of reported by Leyder et al.15
The synthesis of 4′-â-thioguanosine (30) was achieved as
shown in Scheme 8. The Pummerer reaction of 10 with 2-amino-
6-chloropurine gave the desired N-9 isomer 28 in 56% yield
when the reaction was conducted under reflux conditions.36 The
structure of 28 was deduced from its UV spectrum, which was
similar to the spectra of the N-9 glycosyl isomers of 2-amino-
6-chloropurine37 but not those of the N-7 isomers.36 The
resulting 28 was then deprotected by TBAF to give 29
quantitatively. Conversion of 29 into 4′-â-thioguanosine (30)
was done by treatment with 2-mercaptoethanol at room tem-
perature, followed by sodium methoxide under reflux conditions
to give 30 in 55% yield. The UV spectrum of 30 in H2O showed
two absorption maxima at 284 and 253 nm, which are identical
with those of guanosine. In addition, the â configuration of 30
was confirmed by NOE experiment. Thus, the expected NOEs
were observed at H-1′ (3.5%) and H-2′ (6.0%) upon irradiation
of H-8.
Figure 3. Frontier electron densities of LUMO in 20 and 22.
a larger frontier electron density of the LUMO was observed at
H-1R as well as at H-4, implying that E2 elimination of these
protons takes place in preference to that of H-1â in the
intermediate 22. It appears that the observed chemical reactivity
differences between the two diastereomeric sulfoxides would
also be explained by these calculations.
Synthesis of 4′-â-Thioribopurine Nucleosides. In contrast
to reactions with pyrimidine bases, those with purine bases are
somewhat more complex due to the possible formation of
regioisomers, i.e., N-3, N-7, and the desired N-9 isomers. In
the usual glycosidation conditions using Lewis acids, the N-3
and N-7 isomers, which are kinetically controlled products, are
known to rearrange subsequently to the thermodynamically most
stable N-9 isomer.30 Accordingly, the Pummerer reaction was
conducted in the presence of various purine bases.31 Among
the nucleobases examined, the reaction with 6-chloropurine gave
the simplest result; i.e., two separable compounds were obtained
in 43% and 3% yields, respectively (Table 2, entry 1). Since
the anomeric proton signals in both compounds were observed
at 6.41 ppm (major) and 6.07 ppm (minor) as singlets, the
stereochemistries at the anomeric positions were considered to
be the â configuration.32 In the glycosidation with 6-chloropu-
rine, the N-7 isomer is generally formed along with the N-9
In conclusion, we have developed a multigram synthesis of
the thiosugar 7 via the Mitsunobu reaction from inexpensive
D-ribose. Using the sulfoxide 10 prepared from 7, the first
stereoselective synthesis of 4′-â-thiouridine, -cytidine, -adeno-
sine, and -guanosine was accomplished with the assistance of
neighboring group participation using the Pummerer reaction.
Since thioglycosidation of 19, which has a 2,4-dimethoxyben-
zoyl group on the C-2 hydroxyl group, was unsuccessful, the
Pummerer reaction was shown as to be an efficient alternative
isomer,33 and these isomers are typically differentiated by H
NMR by the characteristic downfield shifts for the H-1′ and
nucleobase proton signals of the N-7 isomer relative to those
of the N-9 isomer.33b Based on this information, the major
1
(30) (a) Vorbru¨ggen, H.; Hofle, G. Chem. Ber. 1981, 114, 1256-1268.
(b) Boryski, J. Nucleosides Nucleotides 1996, 15, 771-791.
(31) Adenine, hypoxanthine, and N6-benzoyladenine were also used in
place of 6-chloropurine. The Pummerer reactions in the presence of both
adenine and hypoxanthine gave complex mixtures, while the reaction with
N6-benzoyladenine gave the coupling product in 66% yield as a mixture of
regioisomers, which probably correspond to N-3, N-7, and the desired N-9
isomers. However, the resulting mixture was inseparable, and the N-9 isomer
was not obtained preferentially under any conditions.
(34) The proton signals of the nucleobase and H-1′ were observed at
8.76 and 8.59 ppm (H-2 and H-8) and 6.07 ppm (H-1′) for N-9 isomer 25,
while those of 24 were observed at 9.11 and 8.88 ppm (H-2 and H-8) and
6.41 pp (H-1′), respectively.
(35) When 26 was treated with methanolic ammonia, a 6-methoxypurine
derivative was observed along with the 4′-thioadenosine (27).
(36) When the Pummerer reaction was carried out in a mixture of
toluene-CH2Cl2 at room temperature, the N-7 isomer was obtained in 58%
yield preferentially. The UV spectrum of the N-7 isomer in MeOH showed
absorption maxima at 325, 294, and 256 nm. Among them, the maximum
at 325 nm is characteristic of the N-7 glycosyl isomers of 2-amino-6-
chloropurine; see: Hanna, N. B.; Ramasamy, K.; Robins, R. K.; Revankar,
G. R. J. Heterocycl. Chem. 1988, 25, 1899-1903.
(32) The assignment of the configuration at the anomeric carbon as R
or â has been often based on the splitting pattern of H-1′. In the
ribonucleosides possessing 1,1,3,3-tetraisopropyldisiloxane-1,3-diyl group
on their 3′- and 5′-hydroxyl groups, a somewhat broad singlet is generally
observed for the â isomer, while a doublet is observed for the R isomer. In
comparison with those of 11 and 18, the same tendency appeared to be
maintained in the 4′-thioribonucleoside analogues.
(33) (a) Kazimierczuk, Z.; Cottam, H. B.; Revankar, G. R.; Robins, R.
K. J. Am. Chem. Soc. 1984, 106, 6379-6382. (b) Hildebrand, C.; Wright,
G. E. J. Org. Chem. 1992, 57, 1808-1813.
(37) The UV spectrum of 28 in MeOH showed absorption maxima at
315 (shoulder), 298, and 256 nm; see ref 36.

StereoselectiVe Synthesis of 4′-â-Thioribonucleosides
Table 2. The Pummerer Reaction of 10 with 6-Chloropurine
J. Am. Chem. Soc., Vol. 122, No. 30, 2000 7239
entry
1
6-chloropurine (equiv)
2
conditions
yield of 24 (%)
yield of 25 (%)
TMSOTf (8 equiv), Et3N (4 equiv), then Et3N (4 equiv),
toluene-CH2Cl2, rt
43
3
2
3
4
4
4
4
TMSOTf (8 equiv), Et3N (4 equiv), then Et3N (4 equiv),
toluene-CH2Cl2, rt
46
9
42
65
TMSOTf (8 equiv), Et3N (4 equiv), then Et3N (4 equiv),
toluene-ClCH2CH2Cl, rt then 83 °C
trace
trace
TMSOTf (8 equiv), Et3N (4 equiv), then Et3N (4 equiv),
CH3CN-ClCH2CH2Cl, rt then 83 °C
Scheme 7a
expressed as s (singlet), d (doublet), t (triplet), q (quartet), m (multiplet),
or br (broad). All exchangeable protons were detected by addition of
D2O. Assignment of H signals was based on two-dimensional NMR
1
and NOE experiments. TLC was done on Merck Kieselgel F254
precoated plates. Silica gel used for column chromatography was YMC
gel 60A (70-230 mesh).
2,3,5-Tri-O-benzyl-D-ribitol (2). To a solution of 116 (101.3 g, 0.23
mol) in 1,4-dioxane (800 mL) was added 4 N aqueous HCl (800 mL),
and the whole was heated under reflux. The reaction mixture was
partitioned between ether and saturated aqueous NaHCO3. The separated
organic layer was washed with H2O, followed by brine. The organic
layer was dried (Na2SO4) and concentrated in vacuo. The residue was
dissolved in MeOH (1.0 L), and sodium borohydride (35.3 g, 0.93 mol)
was added at 0 °C. After being stirred for 3 h at room temperature, the
reaction mixture was concentrated in vacuo, and the residue was
coevaporated with MeOH. The residue was purified by a silica gel
column, eluted with hexane/AcOEt (1:1), to give 2 (85.4 g, 87% as a
colorless oil): FAB-LRMS m/z 423 (MH+, 12.4%); 1H NMR (270
MHz, CDCl3) δ 7.35-7.20 (m, 15 H), 4.73-4.48 (m, 6 H), 3.98-
3.56 (m, 7 H), 2.98 (br s, 1 H), 2.61 (br s, 1 H); 13C NMR (100 MHz,
CDCl3) δ 137.81, 137.78, 137.64, 128.34, 128.31, 128.28, 127.93,
127.78, 127.73, 127.69, 79.24, 79.17, 73.89, 73.34, 71.89, 70.94, 60.88.
Anal. Calcd for C26H30O5: C, 73.91; H, 7.16. Found: C, 73.76; H,
7.20.
a Reagents: (a) TBAF, AcOH, THF; (b) NH3/EtOH.
Scheme 8a
2,3,5-Tri-O-benzyl-1-O-tert-butyldimethylsilyl-D-ribitol (3). A mix-
ture of 2 (20.0 g, 47 mmol), imidazole (14.2 g, 208 mmol), and
TBDMSCl (7.9 g, 52 mmol) in dry DMF (200 mL) was stirred at 0 °C
for 30 min. The reaction was quenched by addition of ice, and the
reaction mixture was partitioned between ether and H2O. The separated
organic layer was washed with H2O, followed by brine. The organic
layer was dried (Na2SO4) and concentrated in vacuo. The residue was
purified by a silica gel column, eluted with hexane/AcOEt (20:1-5:
1), to give 3 (24.7 g, 97% as a colorless oil): FAB-LRMS m/z 537
a
Reagents: (a) 2-amino-6-chloropurine, TMSOTf, Et3N, CH3CN-
ClCH2CH2Cl, room temperature then 83 °C; (b) TBAF, AcOH, THF;
(c) 2-mercaptoethanol, NaOMe, MeOH, reflux.
1
(MH+, 1.4%); H NMR (500 MHz, CDCl3) δ 7.34-6.34 (m, 15 H),
4.74-4.50 (m, 6 H), 4.02-3.60 (m, 7 H), 2.93 (br s, 1 H), 0.83 (s, 9
H), -0.01 (s, 6 H); 13C NMR (125 MHz, CDCl3) δ 138.47, 138.16,
128.36, 128.31, 128.29, 127.95, 127.82, 127.73, 127.64, 127.58, 127.54,
80.69, 78.93, 73.65, 73.35, 72.66, 71.25, 71.08, 62.57, 25.91, 18.27,
-5.41. Anal. Calcd for C32H44O5Si: C, 71.60; H, 8.26. Found: C,
71.55; H, 8.27.
2,3,5-Tri-O-benzyl-1-O-tert-butyldimethylsilyl-4-O-p-nitrobenzoyl-
L-lyxitol (4). A solution of 3 (21.2 g, 39.5 mmol) in THF (150 mL)
containing p-nitrobenzoic acid (13.2 g, 79 mmol) and triphenylphos-
phine (20.7 g, 79 mmol) was cooled to 0 °C, and a THF solution of
diisopropyl azodicarboxylate (15.5 mL, 79 mmol in 100 mL) was added
to the mixture over 3 h. After being stirred for 12 h at room temperature,
the reaction mixture was partitioned between ether and H2O. The
separated organic layer was washed with saturated aqueous NaHCO3
(three times), followed by brine. The organic layer was dried
(Na2SO4) and concentrated in vacuo. The residue was purified by a
silica gel column, eluted with hexane/AcOEt (50:1-10:1), to give 4
method for synthesizing 4′- â-thioribonucleosides stereoselec-
tively. In addition, we demonstrated the striking differences
between R-10 and S-10 in the Pummerer reaction. The differ-
ences in E2 elimination behavior to give R-thiocarbocation
intermediates from R-10 and S-10 was postulated as one of the
possible interpretations for the observed chemical reactivity
differences. This should be one of the best methods to date for
producing 4′-â-thioribonucleosides.
Experimental Section
General Methods. Physical data were measured as follows. Melting
points are uncorrected. 1H and 13C NMR spectra were recorded on 270,
400, or 500 MHz and 67.5, 100, or 125 MHz instruments in CDCl3 or
DMSO-d6 as the solvent with tetramethylsilane as an internal standard.
Chemical shifts are reported in parts per million (δ), and signals are

7240 J. Am. Chem. Soc., Vol. 122, No. 30, 2000
Naka et al.
(22.6 g, 83% as a yellow oil): FAB-LRMS m/z 686 (MH+, 20.9%);
1H NMR (270 MHz, CDCl3) δ 8.15-8.03 (m, 4 H), 7.27-7.11 (m, 15
H), 5.64 (dd, 1 H, J ) 5.1, 9.5 Hz), 4.72-4.39 (m, 6 H), 4.04-3.57
(m, 6 H), 0.84 (s, 9 H), 0.00 (s, 6 H); 13C NMR (125 MHz, CDCl3) δ
164.08, 150.40, 138.22, 138.10, 137.78, 135.76, 130.82, 128.34, 128.28,
128.22, 127.87, 127.80, 127.69, 127.66, 127.52, 79.17, 76.59, 74.18,
73.41, 73.14, 72.43, 68.42, 61.64, 25.88, 18.21, -5.40, -5.46. Anal.
Calcd for C39H47NO8Si: C, 68.30; H, 6.91; N, 2.04. Found: C, 68.12;
H, 6.95; N, 2.12.
2,3,5-Tri-O-benzyl-L-lyxitol (5). To a solution of 4 (103 g, 150
mmol) in MeOH (300 mL) was added NaOMe (28% solution, 2 mL),
and the whole was stirred at room temperature for 2 h. The reaction
mixture was concentrated in vacuo, and the residue was partitioned
between ether and H2O. The separated organic layer was washed with
brine, dried (Na2SO4), and concentrated in vacuo. The residue was
suspended in ether, and insoluble materials were filtered off. The filtrate
was concentrated in vacuo, and the residue was dissolved in THF (300
mL). To the solution was added TBAF (1 M in THF, 150 mL, 150
mmol), and the mixture was stirred at room temperature for 1 h. The
reaction mixture was partitioned between ether and H2O, and the
separated organic layer was washed with brine. The organic layer was
dried (Na2SO4) and concentrated in vacuo. The residue was purified
by a silica gel column, eluted with hexane/AcOEt (5:1-4:1), to give 5
(51.5 g, 81% as a colorless oil): FAB-LRMS m/z 423 (MH+, 35.4%);
1H NMR (270 MHz, CDCl3) δ 7.37-7.25 (m, 15 H), 4.75-4.44 (m,
6 H), 4.00-3.43 (m, 7 H), 2.59 (br s, 1 H), 2.27 (br s, 1 H); 13C NMR
(125 MHz, CDCl3) δ 137.88, 137.79, 128.39, 128.31, 128.11, 127.81,
127.76, 127.66, 79.59, 77.09, 74.23, 73.29, 72.29, 71.20, 69.65, 60.44.
Anal. Calcd for C26H30O5: C, 73.91; H, 7.16. Found: C, 73.81; H,
7.19.
1,4-Anhydro-2,3,5-tri-O-benzyl-4-thio-D-ribitol (6). To a solution
of 5 (51.5 g, 121 mmol) in dry pyridine (300 mL) was added
methanesulfonyl chloride (28.0 mL, 363 mmol) at 0 °C. After the
mixture was stirred for 2 h at the same temperature, the reaction was
quenched by addition of ice. The reaction mixture was partitioned
between AcOEt and H2O. The separated organic layer was washed with
saturated aqueous NaHCO3 (three times), followed by brine. The organic
layer was dried (Na2SO4) and concentrated in vacuo, and the residue
was coevaporated several times with toluene. The residue was dissolved
in dry DMF (500 mL), and sodium sulfide nonahydrate (32.0 g, 133
mmol) was added to the solution. The mixture was heated at 100 °C
for 2 h. After being cooled to room temperature, the mixture was diluted
with ether and washed with H2O (three times), followed by brine. The
organic layer was dried (Na2SO4) and concentrated in vacuo. The
residue was purified by a silica gel column, eluted with hexane/AcOEt
(10:1), to give 6 (43.2 g, 84% as a colorless oil): FAB-LRMS m/z
421 (MH+, 19.3%); 1H NMR (270 MHz, CDCl3) δ 7.32-7.22 (m, 15
H), 4.65-4.44 (m, 6 H), 4.02 (ddd, 1 H, J ) 6.9, 5.6, 3.6 Hz), 3.95
(dd, 1 H, J ) 3.6, 4.3 Hz), 3.67 (m, 1 H), 3.48 (m, 2 H), 3.03 (dd, 1
H, J ) 6.9, 10.6 Hz), 2.88 (dd, 1 H, J ) 5.6, 10.6 Hz); 13C NMR (125
MHz, CDCl3) δ 137.88, 137.84, 137.82, 128.20, 128.15, 127.77, 127.54,
127.47, 80.82, 79.48, 72.93, 71.79, 71.71, 71.68, 47.07, 30.61. Anal.
Calcd for C26H28O3S: C, 74.25; H, 6.71. Found: C, 74.15; H, 6.75.
1,4-Anhydro-4-thio-D-ribitol (7). A solution of 6 (7.0 g, 17 mmol)
in dry CH2Cl2 (100 mL) was cooled to -98 °C, and 1 M BCl3 in
CH2Cl2 (100 mL, 100 mmol) was added over 2 h. After the solution
was stirred at the same temperature for 1 h, MeOH (100 mL) was added
to the reaction mixture under -90 °C. The solvent was removed in
vacuo, and the residue was purified by a silica gel column, eluted
with hexane/AcOEt (5:1-4:1), to give 7 (2.0 g, 79% as a yellow oil):
FAB-LRMS m/z 151 (MH+, 12.5%); 1H NMR (270 MHz, CD3OD) δ
4.12 (ddd, 1 H, J ) 5.0, 5.3, 3.6 Hz), 3.85 (dd, 1 H, J ) 3.6, 5.6 Hz),
3.63 (dd, 1 H, J ) 5.9, 11.2 Hz), 3.46 (dd, 1 H, J ) 6.3, 11.2 Hz),
3.25 (ddd, 1 H, J ) 5.6, 5.9, 6.3 Hz), 2.82 (dd, 1 H, J ) 5.0, 10.9 Hz),
2.65 (dd, 1 H, J ) 5.3, 10.9 Hz); 13C NMR (68 MHz, CD3OD) δ 77.73,
75.85, 65.46, 52.68, 33.46. Anal. Calcd for C5H10O3S: C, 39.99; H,
6.71. Found: C, 39.85; H, 6.59.
at room temperature. The reaction was quenched by addition of ice,
and the mixture was partitioned between AcOEt and H2O. The separated
organic layer was washed with saturated aqueous NaHCO3 (three times),
followed by brine. The organic layer was dried (Na2SO4) and
concentrated in vacuo, and the residue was coevaporated with toluene.
The residue was purified by a silica gel column, eluted with hexane/
AcOEt (50:1), to give 8 (3.2 g, 79% as a colorless oil): FAB-LRMS
1
m/z 393 (MH+, 14.6%); H NMR (270 MHz, CDCl3) δ 4.33 (ddd, 1
H, J ) 1.6, 0.8, 4.0 Hz), 4.24 (dd, 1 H, J ) 4.0, 8.3 Hz), 4.05 (dd, 1
H, J ) 3.2, 12.3 Hz), 3.92 (dd, 1 H, J ) 4.4, 12.3 Hz), 3.50 (ddd, 1 H,
J ) 8.3, 3.2, 4.4 Hz), 3.04 (dd, 1 H, J ) 1.6, 12.3 Hz), 2.86 (dd, 1 H,
0.8, 12.3 Hz), 2.67 (s, 1 H), 1.12-1.05 (m, 28 H); 13C NMR (68 MHz,
CDCl3) δ 77.58, 74.39, 61.08, 49.24, 32.36, 17.34, 17.25, 17.06, 13.37,
13.25, 12.69, 12.65. Anal. Calcd for C17H36O4SSi2: C, 52.00; H, 9.24.
Found: C, 52.00; H, 9.06.
1,4-Anhydro-2-O-(2,4-dimethoxybenzoyl)-3,5-O-(1,1,3,3-tetraiso-
propyldisiloxane-1,3-diyl)-4-thio-D-ribitol (9). To a solution of 8 (2.8
g, 7.2 mmol) in dry pyridine (35 mL) was added 2,4-dimethoxybenzoyl
chloride (2.9 g, 14.3 mmol) at 0 °C, and the mixture was stirred at
room temperature for 12 h. The reaction was quenched by addition of
ice, and the mixture was partitioned between AcOEt and H2O. The
separated organic layer was washed with saturated aqueous NaHCO3
(three times), followed by brine. The organic layer was dried
(Na2SO4) and concentrated in vacuo, and the residue was coevaporated
with toluene. The residue was purified by a silica gel column, eluted
with hexane/AcOEt (10:1-5:1), to give 9 (3.9 g, 99% as a colorless
oil): FAB-LRMS m/z 557 (MH+, 6.7%); 1H NMR (270 MHz, CDCl3)
δ 7.88 (d, 1 H J ) 9.5 Hz), 6.49 (m, 2 H), 5.71 (dd, 1 H, J ) 4.8, 4.0
Hz), 4.33 (dd, 1 H, J ) 4.0, 9.5 Hz), 4.11 (dd, 1 H, J ) 2.8, 12.3 Hz),
3.96 (dd, 1 H, J ) 3.2, 12.3 Hz), 3.88, 3.85 (each s, each 3 H,), 3.66
(ddd, 1 H, J ) 9.5, 2.8, 3.2 Hz), 3.21 (dd, 1 H, J ) 4.8, 12.7 Hz), 2.89
(d, 1 H, J ) 12.7 Hz), 1.13-0.96 (m, 28 H); 13C NMR (100 MHz,
CDCl3) δ 164.42, 164.12, 161.38, 133.74, 112.38, 104.36, 98.82, 75.58,
74.92, 59.67, 55.91, 55.44, 49.61, 31.08, 17.44, 17.37, 17.34, 17.31,
17.27, 17.07, 13.45, 13.35, 12.78, 12.75. Anal. Calcd for C26H44O7-
SSi2: C, 56.08; H, 7.96. Found: C, 55.99; H, 7.99.
1,4-Anhydro-2-O-(2,4-dimethoxybenzoyl)-3,5-O-(1,1,3,3-tetraiso-
propyldisiloxane-1,3-diyl)-4-sulfinyl-D-ribitol (10). Method A. To a
solution of 9 (978 mg, 1.76 mmol) in dry CH2Cl2 (10 mL) was added
m-CPBA (328 mg, 1.90 mmol) at -40 °C, and the mixture was stirred
at the same temperature for 30 min. The reaction was quenched by
addition of saturated aqueous sodium thiosulfate, and the reaction
mixture was partitioned between AcOEt and H2O. The separated organic
layer was washed with saturated aqueous NaHCO3 (three times),
followed by brine. The organic layer was dried (Na2SO4) and
concentrated in vacuo. The residue was purified by a silica gel column,
eluted with hexane/AcOEt (1:1), to give 10 as a mixture of diastere-
omers (743 mg, 82%; R/S ) 1:2.7). Separation of R-10 and S-10 was
performed by preparative TLC, developed with benzene/AcOEt (3:1).
Physical data for R-10 (a colorless oil): FAB-LRMS m/z 573 (MH+,
10.4%); 1H NMR (500 MHz, CDCl3) δ 7.92 (d, 1 H, Ar, J ) 8.3 Hz),
6.49 (m, 2 H, Ar), 5.79 (dd, 1 H, H-2, J2,1â ) 5.4, J2,3 ) 3.6 Hz), 4.59
(d, 1 H, H-5a, J5a,5b ) 12.8 Hz), 4.22 (dd, 1 H, H-5b, J5b 4 ) 2.8, J5b,5a
) 12.8 Hz), 4.12 (dd, 1 H, H-3, J3,2 ) 3.6, J3,4 ) 12.0 Hz), 3.88, 3.86
(each s, each 3 H, MeO), 3.57 (dd, 1 H, H-1â, J1â,2 ) 5.4, J1â,1R
)
15.5 Hz), 3.49 (dd, 1 H, H-4, J4,3 ) 12.0, J4,5b ) 2.8 Hz), 2.89 (d, 1 H,
H-1R, J1R,1â ) 15.5 Hz), 1.11-0.94 (m, 28 H, TIPDS); 13C NMR (125
MHz, CDCl3) δ 164.47, 164.14, 161.62, 133.99, 111.70, 104.52, 98.80,
77.20, 72.54, 72.40, 67.89, 55.80, 55.36, 54.28, 17.22, 17.14, 17.08,
17.03, 17.00, 16.84, 16.82, 13.28, 13.04, 12.55, 12.51 Anal. Calcd for
C26H44O8SSi2‚0.5 H2O: C, 53.67; H, 7.80. Found: C, 53.35; H, 7.53.
Physical data for S-10 (a colorless oil): FAB-LRMS m/z 573 (MH+,
29.2%); FAB-HRMS Calcd for C26H44O8SSi2 (MH+) 573.2373. Found
1
573.2392. H NMR (500 MHz, CDCl3) δ 7.76 (d, 1 H, Ar, J ) 9.3
Hz), 6.44 (m, 2 H, Ar), 5.96 (ddd, 1 H, H-2, J2,1â ) 5.4, J2,1R ) 0.9,
J2,3 ) 3.9 Hz), 5.36 (dd, 1 H, H-3, J3,2 ) 3.9, J3,4 ) 10.3 Hz), 4.49
(dd, 1 H, H-5a, J5a,4 ) 3.6, J5a,5b ) 12.8 Hz), 4.43 (dd, 1 H, H-5b, J5b,4
) 4.6, J5b,5a ) 12.8 Hz), 3.83, 3.82 (each s, each 3 H, MeO), 3.70 (dd,
1,4-Anhydro-3,5-O-(1,1,3,3-tetraisopropyldisiloxane-1,3-diyl)-4-
thio-D-ribitol (8). To a solution of 7 (1.54 g, 10.3 mmol) in dry pyridine
(20 mL) was added 1,3-dichloro-1,1,3,3-tetraisopropyldisiloxane (3.6
mL, 11.0 mmol) at 0 °C, and the reaction mixture was stirred for 12 h
1 H, H-1R, J1R,2 ) 0.9, J1R,1â ) 15.0 Hz), 3.05 (dd, 1 H, H-1â, J1â,2
)
5.4, J1â,1R ) 15.0 Hz), 3.01 (ddd, 1 H, H-4, J4,3 ) 10.3, J4,5b ) 3.6,
J4,5b ) 4.6 Hz), 1.08-0.87 (m, 28 H, TIPDS); 13C NMR (125 MHz,

StereoselectiVe Synthesis of 4′-â-Thioribonucleosides
J. Am. Chem. Soc., Vol. 122, No. 30, 2000 7241
CDCl3) δ 164.55, 164.12, 161.48, 133.81, 111.70, 104.56, 98.87, 74.95,
72.58, 62.28, 58.51, 55.85, 54.45, 17.30, 17.26, 17.23, 17.20, 17.15,
16.95, 16.88, 13.09, 12.97, 12.68. Anal. Calcd for C26H44O8SSi2‚0.5
H2O: C, 53.67; H, 7.80. Found: C, 53.35; H, 7.53.
Method B. Ozone was bubbled through a solution of 9 (2.1 g, 3.8
mmol) in CH2Cl2 (40 mL) at -78 °C. After 30 min, N2 gas was bubbled
through the solution to remove excess ozone. The reaction mixture was
allowed to warm to room temperature and concentrated in vacuo. The
residue was purified by a silica gel column, eluted with hexane/AcOEt
(1:1), to give 10 as a mixture of diastereomers (1.8 g, 82%; R/S )
>16:1).
ammonia (saturated at 0 °C, 20 mL). The reaction mixture was kept
for 24 h at room temperature, and the solvent was removed in vacuo.
The residue was purified by a silica gel column, eluted with 25% MeOH
in CHCl3, to give 14 (152 mg, 85% as a pale brown foam, crystallized
from EtOH): mp 195-196 °C (lit.21 mp 195-196 °C); FAB-LRMS
m/z 261 (MH+, 8.3%); 1H NMR (500 MHz, DMSO-d6) δ 11.09 (br s,
1 H, NH), 7.98 (d, 1 H, H-6, J6,5 ) 8.2 Hz), 5.88 (d, 1 H, H-1′, J1′,2′
)
7.4 Hz), 5.68 (d, 1 H, H-5, J5,6 ) 8.2 Hz), 5.46 (d, 1 H, 2′-OH, J2′OH,2′
) 6.1 Hz), 5.24 (d, 1 H, 3′-OH, J3′-OH,3′ ) 4.2 Hz), 5.16 (t, 1 H, 5′-
OH, J5′-OH,5′ ) 5.3 Hz), 4.13 (ddd, 1 H, H-2′, J2′,2′-OH ) 6.1, J2′,1′
)
7.4, J2′,3′ ) 3.5 Hz), 4.00 (ddd, 1 H, H-3′, J3′,3′-OH ) 4.2, J3′,2′ ) 3.5,
J3′,4′ ) 3.0 Hz), 3.61 (ddd, 1 H, H-5′a, J5′a,5′-OH ) 5.3, J5′a,4′ ) 6.6,
J5′a,5′b ) 11.4 Hz), 3.53 (ddd, 1 H, H-5′b, J5′b,5′-OH ) 5.3, J5′b,4′ ) 5.4,
J5′b,5′a ) 11.4 Hz), 3.18 (ddd, 1 H, H-4′, J4′,3′ ) 3.0, J4′,5′a ) 6.6, J4′,5′b
) 5.4 Hz); 13C NMR (100 MHz, DMSO-d6) δ 163.03, 151.16, 141.60,
102.25, 76.43, 73.11, 63.24, 62.39, 53.12.
1-[2-O-(2,4-Dimethoxybenzoyl)-3,5-O-(1,1,3,3-tetraisopropyldisi-
loxane-1,3-diyl)-4-thio-â-D-ribofuranosyl]uracil (11) (Entry 6). To
a suspension of uracil (35 mg, 0.31 mmol) in dry toluene (2 mL) was
added triethylamine (87 µL, 0.62 mmol) and TMSOTf (241 µL, 1.25
mmol), and the mixture was stirred at room temperature until two-
phase clear solution was obtained. Dry CH2Cl2 (1 mL) was added to
the above solution, which gave a one-phase clear solution, and the whole
was added to a solution of 10 (100 mg, 0.18 mmol, R/S ) >16:1) in
dry CH2Cl2 (1 mL) dropwise over 15 min via a cannula. An additional
amount of triethylamine (87 µL, 0.16 mmol) in dry toluene (0.5 mL)
was added dropwise to the reaction mixture to initiate the Pummerer
reaction. After the mixture was stirred for 5 min at room temperature,
the reaction was quenched by addition of ice, and the reaction mixture
was partitioned between AcOEt and H2O. The separated organic layer
was washed with saturated aqueous NaHCO3 (three times), followed
by brine. The organic layer was dried (Na2SO4) and concentrated in
vacuo. The residue was purified by a silica gel column, eluted with
hexane/AcOEt (5:1-1:1), to give 11 (84 mg, 80% as a white foam):
1-(4-Thio-â-D-ribofuranosyl)cytosine (15). In a manner similar to
that described for 14, 13 (324 mg, 0.46 mmol) was treated with
ammonium fluoride, followed by methanolic ammonia to give 15 (75
mg, 63% as a pale brown foam): FAB-LRMS m/z 260 (MH+, 63.5%);
1H NMR (400 MHz, DMSO-d6) δ 7.94 (d, 1 H, H-6, J6,5 ) 7.5 Hz),
7.12 (d, 2 H, NH2), 5.92 (d, 1 H, H-1′, J1′,2′ ) 6.6 Hz), 5.78 (d, 1 H,
H-5, J5,6 ) 7.5 Hz), 5.30 (d, 1 H, 2′-OH, J2′OH,2′ ) 6.1 Hz), 5.14 (d, 1
H, 3′-OH, J3′-OH,3′ ) 4.2 Hz), 5.10 (t, 1 H, 5′-OH, J5′-OH,5′ ) 5.1 Hz),
4.04 (ddd, 1 H, H-2′, J2′,2′-OH ) 6.1, J2′,1′ ) 6.6, J2′,3′ ) 3.6 Hz), 3.96
(ddd, 1 H, H-3′, J3′,3′-OH ) 4.2, J3′,2′ ) 3.6, J3′,4′ ) 3.7 Hz), 3.61 (ddd,
1 H, H-5′a, J5′a,5′-OH ) 5.1, J5′a,4′ ) 6.3, J5′a,5′b ) 11.3 Hz), 3.53 (ddd,
1 H, H-5′b, J5′b,5′-OH ) 5.1, J5′b,4′ ) 5.6, J5′b,5′a ) 11.3 Hz), 3.19 (ddd,
1 H, H-4′, J4′,3′ ) 3.7, J4′,5′a ) 6.3, J4′,5′b ) 5.6 Hz). Anal. Calcd for
C9H13N3O4S‚0.8MeOH: C, 41.31; H, 5.73; N, 14.75. Found: C, 41.05;
H, 5.35 N, 14.67.
1
FAB-LRMS m/z 667 (MH+, 6.5%); H NMR (500 MHz, CDCl3) δ
8.82 (br s, 1 H, NH), 8.15 (d, 1H, H-6, J6,5 ) 8.1 Hz), 7.85 (d, 1 H,
Ar, J ) 8.6 Hz), 6.50 (m, 2 H, Ar), 6.00 (s, 1 H, H-1′), 5.74 (d, 1 H,
H-5, J5,6 ) 8.1 Hz), 5.61 (d, 1 H, H-2′, J2′,3′ ) 3.7 Hz), 4.45 (dd, 1 H,
H-3′, J3′,2′ ) 3.7, J3′,4′ ) 9.5 Hz), 4.16 (dd, 1 H, H-5′a, J5′a,4′ ) 2.7,
J5′a,4′ ) 12.7 Hz), 4.07 (d, 1 H, H-5′b, J5′b,5′a ) 12.7 Hz), 3.87, 3.86
(each s, each 3 H, MeO), 3.73 (dd, 1 H, H-4′, J4′,3′ ) 9.5, J4′,5′a ) 2.7
Hz), 1.15-0.90 (m, 28 H, TIPDS); 13C NMR (68 MHz, CDCl3) δ
164.51, 163.45, 162.73, 161.61, 150.01, 140.86, 133.93, 111.82, 104.64,
102.25, 99.01, 77.68, 71.45, 62.70, 58.03, 55.94, 55.49, 50.93, 17.45,
17.33, 17.25, 17.02, 16.84, 13.32, 13.14, 12.55. Anal. Calcd for
C30H46N2O9SSi2: C, 54.03; H, 6.95; N, 4.20. Found: C, 54.03; H, 6.90;
N, 4.11.
Physical data for 3,6-O-(1,1,3,3-tetraisopropyldisiloxane-1,3-diyl)-
3-hydroxy-2-hydroxymethylthiophene (12): FAB-LRMS m/z 372
(MH+, 31.3%); FAB-HRMS calcd for C17H33O3SSi2 (MH+) 372.1611,
found 373.1672; 1H NMR (270 MHz, CDCl3) δ 7.10 (d, 1 H, J ) 5.3
Hz), 6.66 (d, 1 H, J ) 5.3 Hz), 4.74 (s, 2 H), 1.25-0.75 (m, 28 H);
13C NMR (125 MHz, CDCl3) δ 149.24, 122.86, 122.25, 122.03, 55.03,
16.89, 16.87, 16.67, 16.62, 13.00, 12.78.
N4-Acetyl-1-[2-O-(2,4-dimethoxybenzoyl)-3,5-O-(1,1,3,3-tetraiso-
propyldisiloxane-1,3-diyl)-4-thio-â-D-ribofuranosyl]cytosine (13). In
the similar manner as described for 11, the Pummerer reaction of 10
(88 mg, 0.15 mmol, R/S ) >16:1) with N4-acetylcytosine (47 mg, 0.31
mmol) using trietylamine (43 µL, 0.31 mmol, and an additional 129
µL, 0.92 mmol) and TMSOTf (238 µL, 1.23 mmol) gave 13 (82 mg,
75% as a yellow solid): FAB-LRMS m/z 708 (MH+, 2.1%); 1H NMR
(500 MHz, CDCl3) δ 9.44 (br s, 1 H), 8.57 (d, 1 H, J ) 7.6 Hz), 7.82
(d, 1 H, J ) 8.7 Hz), 7.41 (d, 1 H, J ) 7.6 Hz), 6.46 (m, 2 H), 6.07
(s, 1 H), 5.66 (d, 1 H, J ) 3.7 Hz), 4.40 (dd, 1 H, J ) 3.7, 9.5 Hz),
4.14 (dd, 1 H, J ) 3.0, 12.8 Hz), 4.06 (d, 1 H, J ) 12.8 Hz), 3.83 (s,
6 H), 3.74 (dd, 1 H, J ) 9.5, 3.0 Hz), 2.24 (s, 3 H), 1.14-0.87 (m, 28
H); 13C NMR (125 MHz, CDCl3) δ 171.48, 164.30, 163.22, 163.02,
161.41, 155.21, 145.74, 133.82, 112.08, 104.54, 98.94, 96.95, 77.44,
71.20, 64.31, 58.08, 55.88, 55.40, 50.68, 24.82, 17.40, 17.31, 17.29,
17.26, 16.96, 16.95, 16.81, 16.79, 13.27, 13.07, 12.48. Anal. Calcd for
C32H49N3O9SSi2‚0.25H2O: C, 53.94; H, 7.00; N, 5.90. Found: C, 54.20;
H, 7.08; N, 5.43.
1-[3,5-O-(1,1,3,3-Tetraisopropyldisiloxane-1,3-diyl)-4-thio-r-D-ri-
bofuranosyl]uracil (17). In the same manner as described for 8, 16 (7
mg, 0.03 mmol) was treated with 1,3-dichloro-1,1,3,3-tetraisopropyl-
disiloxane (40 µL, 0.1 mmol) to give 17 (10 mg, 67% as a white
solid): FAB-LRMS m/z 503 (MH+, 42.7%); FAB-HRMS calcd for
C21H39N2O6SSi2 (MH+) 503.2067, found 503.2049; 1H NMR (270 MHz,
CDCl3) δ 8.90 (br s, 1 H, NH), 7.98 (d, 1H, H-6, J6,5 ) 8.3 Hz), 6.32
(d, 1 H, H-1′, J1′,2′ ) 4.9 Hz), 5.70 (d, 1 H, H-5, J5,6 ) 8.3 Hz), 4.49
(dd, 1 H, H-2′, J2′,1′ ) 4.9, J2′,3′ ) 3.7 Hz), 4.32 (dd, 1 H, H-3′, J3′,2′
)
3.4, J3′,4′ ) 9.5 Hz), 4.10 (dd, 1 H, H-5′a, J5′a,4′ ) 3.0, J5′a,5′b ) 12.9
Hz), 3.95 (dd, 1 H, H-5′b, J5′b,4′ ) 2.0, J5′b,5′a ) 12.9 Hz), 3.81 (ddd, 1
H, H-4′, J4′,3′ ) 9.5, J4′,5′a ) 3.0, J4′,5′b ) 2.0 Hz), 3.04 (br s, 1 H,
2′-OH), 1.11-1.05 (m, 28 H, TIPDS).
1-[2-O-(2,4-Dimethoxybenzoyl)-3,5-O-(1,1,3,3-tetraisopropyldisi-
loxane-1,3-diyl)-4-thio-r-D-ribofuranosyl]uracil (18). To a solution
of 17 (10 mg, 0.02 mmol) in dry CH3CN (2 mL) were added 2,4-
dimethoxybenzoyl chloride (11 mg, 0.06 mmol), triethylamine (10 µL,
0.06 mmol), and DMAP (23 mg, 0.19 mmol), and the mixture was
heated under reflux for 12 h. The reaction was quenched by addition
of ice, and the reaction mixture was partitioned between AcOEt and
H2O. The separated organic layer was washed with saturated aqueous
NaHCO3 (three times), followed by brine. The organic layer was dried
(Na2SO4) and concentrated in vacuo. The residue was purified by a
silica gel column, eluted with hexane/AcOEt (1:1), to give 18 (5 mg,
39% as a white solid): FAB-LRMS m/z 667 (MH+, 14.8%); FAB-
HRMS calcd for C30H47N2O9SSi2 (MH+) 667.2540, found 667.2522;
1H NMR (500 MHz, CDCl3) δ 8.00 (br s, 1 H, NH), 7.87 (d, 1H, H-6,
J6,5 ) 8.3 Hz), 7.79 (d, 1 H, Ar, J ) 8.6 Hz), 6.52 (d, 1 H, H-1′, J1′,2′
) 4.9 Hz), 6.48 (m, 2 H, Ar), 5.85 (dd, 1 H, H-2′, J2′,1′ ) 4.9, J2′,3′
)
3.4 Hz), 5.46 (dd, 1 H, H-5, J5,6 ) 8.3, J5,1′ ) 2.3 Hz), 4.44 (dd, 1 H,
H-3′, J3′,2′ ) 3.4, J3′,4′ ) 9.6 Hz), 4.15 (dd, 1 H, H-5′a, J5′a,4′ ) 2.7,
J5′a,5′b ) 12.7 Hz), 3.98 (dd, 1 H, H-5′b, J5′b,4′ ) 1.7, J5′b,5′a ) 12.7 Hz),
3.95 (ddd, 1 H, H-4′, J4′,3′ ) 9.6, J4′,5′a ) 2.7, J4′,5′b ) 1.7 Hz), 1.14-
0.95 (m, 28 H, TIPDS).
2-O-(2,4-Dimethoxybenzoyl)-3,5-O-(1,1,3,3-tetraisopropyldisilox-
ane-1,3-diyl)-1-O-acetyl-4-thio-D-ribofuranose (19). A solution of 10
(210 mg, 0.36 mmol) in Ac2O (10 mL) was heated at 110 °C for 6 h.
The reaction was quenched by addition of H2O, and the mixture was
partitioned between AcOEt and H2O. The separated organic layer was
washed with saturated aqueous NaHCO3 (three times), followed by
1-(4-Thio-â-D-ribofuranosyl)uracil (14). A solution of 11 (460 mg,
0.69 mmol) in MeOH (10 mL) containing ammonium fluoride (510
mg, 13.8 mmol) was heated under reflux for 12 h. The solvent was
removed in vacuo, and the residue was dissolved in methanolic

7242 J. Am. Chem. Soc., Vol. 122, No. 30, 2000
Naka et al.
brine. The organic layer was dried (Na2SO4) and concentrated in vacuo.
The residue was purified by a silica gel column, eluted with hexane/
AcOEt (7:1), to give 19 (93 mg, 41% as a colorless oil): FAB-LRMS
55.59, 54.34. Anal. Calcd for C19H19N4O6SCl: C, 48.87; H, 4.10; N,
12.00. Found: C, 48.54; H, 4.10; N, 11.76.
9-(4-Thio-â-D-ribofuranosyl)adenine (27). A solution of 26 (37 mg,
0.08 mmol) in ethanolic ammonia (saturated at 0 °C, 5 mL) was heated
for 24 h at 100 °C in a steel container. The solvent was removed in
vacuo, and the residue was purified by a silica gel column, eluted with
25% MeOH in CHCl3, to give 27 (18 mg, 83%, crystallized from
MeOH-H2O): mp 254-256 °C (lit.15 mp 246-248 °C); FAB-LRMS
1
m/z 555 (M+ - OAc, 6.0%); H NMR (270 MHz, CDCl3) δ 7.86 (d,
0.5 H, J ) 8.6 Hz), 7.82 (d, 0.5 H, J ) 9.2 Hz), 6.48 (m, 2 H), 5.77
(s, 0.5 H), 5.62 (d, 0.5 H, J ) 3.3 Hz), 5.57 (m, 0.5 H), 4.63 (dd, 0.5
H, J ) 3.3, 9.8 Hz), 4.58 (m, 1H), 4.15-4.06 (m, 1.5 H), 3.88, 3.87,
3.86, 3.85 (each s, each 1.5 H), 3.66 (m, 0.5 H), 3.26 (dd, 0.5 H, J )
9.2, 10.6 Hz), 3.09 (dd, 0.5 H, 6.6, 9.2 Hz), 2.07, 2.05 (each s, each
1.5 H), 1.13-0.98 (m, 28 H); 13C NMR (125 MHz, CDCl3) δ 169.33,
169.22, 164.59, 164.53, 164.15, 161.75, 161.72, 133.97, 133.74, 111.87,
111.74, 104.66, 104.36, 99.01, 98.82, 96.97, 78.92, 77.58, 74.90, 72.41,
61.19, 59.26, 55.97, 55.86, 55.49, 49.77, 30.44, 21.90, 20.86, 17.45,
17.32, 17.29, 17.25, 17.22, 17.19, 17.16, 17.03, 13.48, 13.36, 13.07,
13.00, 12.77, 12.72, 12.46.
1
m/z 285 (MH+, 33.3%); H NMR (400 MHz, DMSO-d6) δ 8.42, 8.12
(each s, each 1 H, H-2 and 8), 7.27 (br s, 2 H, NH2), 6.59 (d, 1 H,
H-1′, J1′,2′ ) 6.6 Hz), 5.57 (d, 1 H, 2′-OH, J2′OH,2′ ) 6.1 Hz), 5.35 (d,
1 H, 3′-OH, J3′-OH,3′ ) 4.6 Hz), 5.23 (t, 1 H, 5′-OH, J5′-OH,5′ ) 5.6
Hz), 4.63 (ddd, 1 H, H-2′, J2′,2′-OH ) 6.1, J2′,1′ ) 6.6, J2′,3′ ) 3.4 Hz),
4.17 (ddd, 1 H, H-3′, J3′,3′-OH ) 4.6, J3′,2′ ) 3.4, J3′,4′ ) 3.4 Hz), 3.76
(ddd, 1 H, H-5′a, J5′a,5′-OH ) 5.6, J5′a,4′ ) 6.6, J5′a,5′b ) 11.2 Hz), 3.59
(ddd, 1 H, H-5′b, J5′b,5′-OH ) 5.6, J5′b,4′ ) 5.9, J5′b,5′a ) 11.2 Hz), 3.29
(ddd, 1 H, H-4′, J4′,3′ ) 3.4, J4′,5′a ) 6.6, J4′,5′b ) 5.9 Hz). Anal. Calcd
for C10H13N5O3S‚0.6H2O: C, 40.84; H, 4.87; N, 23.81. Found: C, 41.01;
H, 4.86; N, 23.45.
6-Chloro-9-[2-O-(2,4-dimethoxybenzoyl)-3,5-O-(1,1,3,3-tetraiso-
propyldisiloxane-1,3-diyl)-4-thio-â-D-ribofuranosyl]-9H-purine (25).
To a suspension of 6-chloropurine (70 mg, 0.46 mmol) in a mixture of
dry CH3CN (2 mL) and 1,2-dichloroethane (1 mL) were added
triethylamine (65 µL, 0.46 mmol) and TMSOTf (180 µL, 0.93 mmol),
and the mixture was stirred at room temperature until the solution was
clear. The resulting solution was added to a solution of 10 (67 mg,
0.12 mmol, R/S ) >16:1) in dry 1,2-dichloroethane (1 mL) dropwise
over 15 min via a cannula. An additional amount of triethylamine (65
µL, 0.46 mmol) in dry 1,2-dichloroethane (0.5 mL) was added dropwise
to the reaction mixture to initiate the Pummerer reaction. After being
stirred at room temperature for 5 min, the reaction mixture was heated
at 83 °C for 24 h. The reaction was quenched by addition of ice, and
the reaction mixture was partitioned between AcOEt and H2O. The
separated organic layer was washed with saturated aqueous NaHCO3
(three times), followed by brine. The organic layer was dried (Na2-
SO4) and concentrated in vacuo. The residue was purified by a silica
gel column, eluted with hexane/AcOEt (5:1-1:1), to give 25 (53 mg,
65% as a yellow solid): FAB-LRMS m/z 709 (MH+, 1.0%); 1H NMR
(400 MHz, CDCl3) δ 8.76, 8.59 (each s, each 1 H, H-2 and 8), 7.93 (d,
1 H, Ar, J ) 8.8 Hz), 5.96 (m, 2 H, Ar), 6.07 (s, 1 H, H-1′), 5.84 (d,
1 H, H-2′, J2′,3′ ) 3.9 Hz), 4.96 (dd, 1 H, H-3′, J3′,2′ ) 3.9, J3′,4′ ) 9.8
Hz), 4.56 (dd, 1 H, H-5′a, J5′a,4′ ) 2.9, J5′a,5′b ) 12.9 Hz), 4.12 (d, 1 H,
H-5′b, J5′b,5′a ) 12.9 Hz), 3.90, 3.88 (each s, each 3 H, MeO × 2),
3.86 (dd, 1 H, H-4′, J4′,3′ ) 9.8, J4′,5′a ) 2.9 Hz), 1.16-0.89 (m, 28 H,
TIPDS); 13C NMR (125 MHz, CDCl3) δ 164.64, 163.59, 161.69,
152.07, 151.18, 151.10, 144.24, 134.02, 132.17, 111.24, 104.63, 98.88,
77.87, 72.22, 60.65, 58.10, 55.98, 55.56, 51.36, 17.52, 17.43, 17.12,
17.08, 16.98, 13.36, 13.24, 13.08, 12.65. Anal. Calcd for C31H46N4O7-
SSi2Cl: C, 52.49; H, 6.39; N, 7.90. Found: C, 52.38; H, 6.40; N, 7.56.
1H NMR data for 6-chloro-7-[2-O-(2,4-dimethoxybenzoyl)-3,5-O-
(1,1,3,3-tetraisopropyldisiloxane-1,3-diyl)-4-thio-â-D-ribofuranosyl]-
2-Amino-6-chloro-9-[2-O-(2,4-dimethoxybenzoyl)-3,5-O-(1,1,3,3-
tetraisopropyldisiloxane-1,3-diyl)-4-thio-â-D-ribofuranosyl]-9H-pu-
rine (28). In the same manner as described for 25, the Pummerer
reaction of 10 (179 mg, 0.31 mmol, R/S ) >16:1) with 2-amino-6-
chloropurine (106 mg, 0.62 mmol) gave 28 (127 mg, 56% as a white
solid): FAB-LRMS m/z 724 (MH+, 0.3%); 1H NMR (500 MHz, CDCl3)
δ 8.25 (s, 1 H, H-8), 7.87 (d, 1 H, Ar, J ) 8.4 Hz), 6.49 (m, 2 H, Ar),
5.84 (s, 1 H, H-1′), 5.77 (d, 1 H, H-2′, J2′, 3′ ) 3.6 Hz), 5.29 (br s, 2 H,
NH2), 4.71 (dd, 1 H, H-3′, J3′,2′ ) 3.6, J3′,4′ ) 9.5 Hz), 4.17 (dd, 1 H,
H-5′a, J5′a,4′ ) 2.9, J5′a,5′b ) 12.6 Hz), 4.08 (d, 1 H, H-5′b, J5′b,5′a
)
12.6 Hz), 3.87, 3.84 (each s, each 3 H, MeO × 2), 3.79 (m, 1 H, H-4′),
1.13-0.86 (m, 28 H, TIPDS); 13C NMR (125 MHz, CDCl3) δ 164.70,
163.55, 161.72, 159.19, 153.27, 151.45, 140.79, 134.01, 125.57, 111.52,
104.73, 99.00, 77.83, 72.39, 59.59, 58.26, 55.99, 55.52, 50.93, 17.45,
17.34, 17.04, 17.02, 16.87, 13.35, 13.18, 13.05, 12.59. Anal. Calcd for
C31H46N5O7SSi2Cl: C, 51.33; H, 6.53; N, 9.65. Found: C, 51.45; H,
6.52; N, 9.23.
2-Amino-6-chloro-9-[2-O-(2,4-dimethoxybenzoyl)-4-thio-â-D-ri-
bofuranosyl]-9H-purine (29). In the same manner as described for
26, treatment of 28 (127 mg, 0.18 mmol) with TBAF (1 M in THF,
350 µL, 0.35 mmol) gave 29 (84 mg, 99% as a white solid): FAB-
LRMS m/z 482 (MH+, 2.5%); 1H NMR (400 MHz, DMSO-d6) δ 8.55
(s, 1 H, H-8), 7.66 (d, 1 H, Ar, J ) 9.3 Hz), 7.04 (br s, 2 H, NH2),
6.54 (m, 2 H, Ar), 6.14 (d, 1 H, H-1′, J1′,2′ ) 7.3 Hz), 5.75 (m, 2 H,
H-2′ and 3′-OH), 5.32 (br s, 1 H, 5′-OH), 4.57 (m, 1 H, H-3′), 3.86
(m, 1 H, H-5′a), 3.79 (s, 3 H, MeO), 3.72 (m, 1 H, H-5′b), 3.64 (s, 3
H, MeO), 3.43 (m, 1 H, H-4′); 13C NMR (100 MHz, DMSO-d6) δ
164.17, 163.36, 160.90, 159.70, 153.99, 149.50, 141.63, 133.40, 123.24,
110.73, 105.33, 98.74, 78.37, 71.22, 62.95, 58.67, 55.69, 55.64, 54.32.
Anal. Calcd for C19H20N5O6SCl‚0.5H2O: C, 46.49; H, 4.31; N, 14.27.
Found: C, 46.64; H, 4.19; N, 14.11.
1
9H-purine (24): H NMR (270 MHz, CDCl3) δ 9.11, 8.88 (each s,
each 1 H, H-2 and 8), 7.82 (d, 1 H, Ar, J ) 8.4 Hz), 6.45 (m, 2 H, Ar),
6.41 (s, 1 H, H-1′), 5.83 (d, 1 H, H-2′, J2′,3′ ) 3.2 Hz), 4.60 (dd, 1 H,
H-3′, J3′,2′ ) 3.2, J3′,4′ ) 9.2 Hz), 4.19 (dd, 1 H, H-5′a, J5′a,4′ ) 3.2,
J5′a,5′b ) 13.0 Hz), 4.10 (d, 1 H, H-5′b, J5′b,5′a ) 13.0 Hz), 3.86, 3.84
(each s, each 3 H, MeO × 2), 3.82 (m, 1 H, H-4′), 1.17-0.87 (m, 28
H, TIPDS).
9-(4-Thio-â-D-ribofuranosyl)guanine (30). A solution of 29 (80
mg, 0.17 mmol) in MeOH (10 mL) containing 2-mercaptoethanol (47
µL, 0.66 mmol) and NaOMe (28% solution, 130 µL) was heated under
reflux for 24 h. The reaction mixture was neutralized with 1 N HCl,
and the solution was concentrated in vacuo. The residue was dissolved
in H2O, and the aqueous layer was washed with AcOEt. The aqueous
layer was concentrated in vacuo, and the residue was crystallized from
H2O to give 30 (28 mg, 55% as pale brown crystals): mp 240 °C dec;
FAB-LRMS m/z 300 (MH+, 8.3%); 1H NMR (400 MHz, DMSO-d6) δ
10.64 (br s, 1 H, NH), 8.02 (s, 1 H, H-8), 6.50 (br s, 2 H, NH2), 5.65
(d, 1 H, H-1′, J1′,2′ ) 6.8 Hz), 5.48 (d, 1 H, 2′-OH, J2′OH,2′ ) 5.6 Hz),
5.28 (m, 1 H, 3′-OH), 5.15 (t, 1 H, 5′-OH, J5′-OH,5′ ) 5.4 Hz), 4.46
(ddd, 1 H, H-2′, J2′,2′-OH ) 5.6, J2′,1′ ) 6.8, J2′,3′ ) 3.4 Hz), 4.15 (m, 1
6-Chloro-9-[2-O-(2,4-dimethoxybenzoyl)-4-thio-â-D-ribofurano-
syl]-9H-purine (26). To a solution of 25 (52 mg, 0.07 mmol) in THF
(2 mL) containing AcOH (9 µL, 0.15 mmol) was added TBAF (1 M
in THF, 150 µL, 0.15 mmol). The mixture was stirred at room
temperature for 10 min. The solvent was removed in vacuo, and the
residue was purified by a silica gel column, eluted with acetone, to
give 26 (34 mg, 99% as a white solid): FAB-LRMS m/z 467 (MH+,
1
22.7%); H NMR (400 MHz, CDCl3) δ 8.81, 8.34 (each s, each 1 H,
H-2 and 8), 7.73 (d, 1 H, Ar, J ) 8.6 Hz), 6.50 (m, 2 H, Ar), 6.27 (d,
1 H, H-1′, J1′,2′ ) 6.3 Hz), 6.23 (dd, 1 H, H-2′, J2′,1′ ) 6.3, J2′,3′ ) 9.8
Hz), 4.81 (dd, 1 H, H-3′, J3′,2′ ) 9.8, J3′,4′ ) 3.7 Hz), 4.27 (m, 1 H,
5′-OH), 4.15 (m, 1 H, H-5′a, J5′a,4′ ) 3.2, J5′a,5′b ) 12.0 Hz), 4.01 (m,
1 H, H-5′b, J5′b,4′ ) 2.9, J5′b,5′a ) 12.0 Hz), 3.88, 3.85 (each s, each 3
H, H-3′), 3.72 (ddd, 1 H, H-5′a, J5′a,5′-OH ) 5.4, J5′a,4′ ) 7.1, J5′a,5′b
11.4 Hz), 3.53 (ddd, 1 H, H-5′b, J5′b,5′-OH ) 5.4, J5′b,4′ ) 5.9, J5′b,5′a
)
)
11.4 Hz), 3.24 (m, 1 H, H-4′); 13C NMR (100 MHz, DMSO-d6) δ
156.61, 153.39, 151.53, 135.84, 116.40, 76.99, 73.03, 63.18, 60.16,
52.96. Anal. Calcd for C10H13N5O4S‚0.5H2O: C, 38.90; H, 4.58; N,
22.71. Found: C, 39.28; H, 4.47; N, 22.82.
H, MeO × 2), 3.79 (ddd, 1 H, H-4′, J4′,3′ ) 3.7, J4′,5′a ) 3.2, J4′,5′b
)
2.9 Hz), 3.20 (d, 1 H, 3′-OH, J3′-OH,3′ ) 3.2 Hz); 13C NMR (100 MHz,
CDCl3) δ 165.00, 164.40, 160.80, 151.69, 151.44, 150.88, 144.95,
134.37, 132.86, 110.34, 105.07, 99.01, 79.67, 77.12, 75.30, 63.53, 56.06,

StereoselectiVe Synthesis of 4′-â-Thioribonucleosides
J. Am. Chem. Soc., Vol. 122, No. 30, 2000 7243
Acknowledgment. This investigation was supported in part
by a Grant-in-Aid for Scientific Research on Priority Areas from
the Ministry of Education, Science, Sports, and Culture of Japan.
We thank Ms. H. Matsumoto and Ms. A. Maeda (Center for
Instrumental Analysis, Hokkaido University) for elemental
analysis. We also thank Ms. S. Oka and Ms. N. Hazama (Center
for Instrumental Analysis, Hokkaido University) for measure-
ment of mass spectra.
JA000541O