2,3-Anhydro sugars in glycoside bond synthesis. Highly stereoselective syntheses of oligosaccharides containing α- and β-arabinofuranosyl linkages

Article abstract of DOI:10.1021/ja029302m

The ever-increasing discovery of biologically important events mediated by carbohydrates has generated great interest in the synthesis of oligosaccharides and the development of new methods for glycosidic bond formation. In this paper, we report that 2,3-

Full text of DOI:10.1021/ja029302m

Published on Web 03/14/2003
2,3-Anhydro Sugars in Glycoside Bond Synthesis. Highly
Stereoselective Syntheses of Oligosaccharides Containing r-
and â-Arabinofuranosyl Linkages
Rajendrakumar Reddy Gadikota,† Christopher S. Callam,† Timothy Wagner,‡
Brian Del Fraino,‡ and Todd L. Lowary*,†
Departments of Chemistry, The Ohio State UniVersity, 100 West 18th AVenue,
Columbus, Ohio 43210, and Youngstown State UniVersity, One UniVersity Plaza,
Youngstown, Ohio 44555
Received November 11, 2002; E-mail: lowary.2@osu.edu
Abstract: The ever-increasing discovery of biologically important events mediated by carbohydrates has
generated great interest in the synthesis of oligosaccharides and the development of new methods for
glycosidic bond formation. In this paper, we report that 2,3-anhydrofuranose thioglycosides (1, 5) and glycosyl
sulfoxides (2, 6), in which the hydroxyl groups C-2 and C-3 are “protected” as an epoxide, glycosylate
alcohols with an exceptionally high degree of stereocontrol. The predominant or exclusive product of
reactions with this fundamentally new class of glycosylating agent is that in which the newly formed glycosidic
bond is cis to the epoxide moiety. We further demonstrate that subsequent nucleophilic opening of the
epoxide moiety proceeds under basic conditions to give products in high yield and with good to excellent
regioselectivity. The major ring-opened products possess the arabino stereochemistry, and thus this
methodology constitutes a new approach for the synthesis of arabinofuranosides. In the epoxide opening
reactions of glycosides with the 2,3-anhydro-â-D-lyxo stereochemistry (e.g., 73), the addition of (-)-sparteine
(78) to the reaction mixture dramatically enhanced the regioselectivity in favor of the arabino product. This
represents the first example of the use of 78 to influence the regioselectivity of an epoxide ring opening
reaction with a non-carbon nucleophile. We have demonstrated the utility of this methodology through the
efficient synthesis of an arabinofuranosyl hexasaccharide, 7, which is a key structural motif in two
mycobacterial cell wall polysaccharides.
Introduction
methods for the synthesis of these important molecules, as
efficient and stereocontrolled routes to many glycosidic linkages
The diverse roles that oligosaccharides play in a number of
important biological events1 have underscored the importance
of glycosidic bond formation in organic chemistry.2 This field
has received increasing attention in recent years, and a number
of impressive achievements have been reported. Among these
are efficient routes for the synthesis of multimilligram quantities
of large oligosaccharides,3 the development of “one-pot”
protocols for the preparation of oligosaccharides,4 and the report
of the first automated solid-phase oligosaccharide synthesizer.5
However, despite these advances, there remains a need for new
are still not always available. In particular, the stereocontrolled
synthesis of oligosaccharides containing furanose residues is
largely unexplored and has only recently begun to be studied
in earnest.6,7 Although, as would be expected, the synthesis of
furanosides with the 1,2-trans stereochemistry can be achieved
in a straightforward manner through the use of donors with acyl
protecting groups on O-2, the stereoselective preparation of 1,2-
cis furanosides remains a challenging and unsolved problem.6,7
(4) (a) Zhang, Z. Y.; Ollmann, I. R.; Ye, X. S.; Wischnat, R.; Baasov, T.;
Wong, C. H. J. Am. Chem. Soc. 1999, 121, 734. (b) Douglas, N. L.; Ley,
S. V.; Lucking, U.; Warriner, S. L. J. Chem. Soc., Perkin Trans. 1 1998,
51.
† The Ohio State University.
‡ Youngstown State University.
(1) (a) Varki, A. Essent. Glycobiol. 1999, 57. (b) Dwek, R. A. Chem. ReV.
(5) Plante, O. J.; Palmacci, E. R.; Seeberger, P. H. Science 2001, 291, 1523.
(6) Some recent examples: (a) Mereyala, H. B.; Hotha, S.; Gurjar, M. K. Chem.
Commun. 1998, 685. (b) Sanchez, S.; Bamhaoud, T.; Prandi, J. Tetrahedron
Lett. 2000, 41, 7447. (c) Gurjar, M. K.; Reddy, L. K.; Hotha, S. J. Org.
Chem. 2001, 66, 4657. (d) Li, Y.; Singh, G. Tetrahedron Lett. 2001, 42,
6615. (e) Choudhury, A. K.; Roy, N. Carbohydr. Res. 1998, 308, 207. (f)
Pathak, A. K.; El-Kattan, Y. A.; Bansal, N.; Maddry, J. A.; Reynolds, R.
C. Tetrahedron Lett. 1998, 39, 1497. (g) Pathak, A. K.; Pathak, V.; Khare,
N. K.; Maddry, J. A.; Reynolds, R. C. Carbohydr. Lett. 2001, 4, 117. (h)
Sanchez, S.; Bamhaoud, T.; Prandi, J. Eur. J. Org. Chem. 2002, 3864.
(7) (a) Yin, H.; D’Souza, F. W.; Lowary, T. L. J. Org. Chem. 2002, 67, 892.
(b) D’Souza, F. W.; Ayers, J. D.; McCarren, P. R.; Lowary, T. L. J. Am.
Chem. Soc. 2000, 122, 1251. (c) D’Souza, F. W.; Cheshev, P. E.; Ayers,
J. D.; Lowary, T. L. J. Org. Chem. 1998, 63, 9037. (d) Yin, H.; Lowary,
T. L. Tetrahedron Lett. 2001, 42, 5829.
1996, 96, 683.
(2) (a) Davis, B. G. J. Chem. Soc., Perkin Trans. 1 2000, 2137. (b) Boons,
G.-J. Tetrahedron 1996, 52, 1095. (c) Toshima, K.; Tatsuta, K. Chem. ReV.
1993, 93, 1503. (d) Schmidt, R. R.; Kinzy, W., P. J. AdV. Carbohydr. Chem.
Biochem. 1994, 50, 21. (e) Danishefsky, S. J.; Bilodeau, M. T. Angew.
Chem., Int. Ed. Engl. 1996, 35, 1380.
(3) Some recent examples include: (a) Kim, H. M.; Kim, I. J.; Danishefsky,
S. J. J. Am. Chem. Soc. 2001, 123, 35. (b) Aly, M. R. E.; Ibrahim, E.-S. I.;
El-Ashry, E.-S. H. E.; Schmidt, R. R. Eur. J. Org. Chem. 2000, 2, 319. (c)
Mayer, T. G.; Schmidt, R. R. Eur. J. Org. Chem. 1999, 1153. (d) Nicolaou,
K. C.; Watanabe, N.; Li, J.; Pastor, J.; Winssinger, N. Angew. Chem., Int.
Ed. 1998, 37, 1559. (e) Pozsgay, V. J. Am. Chem. Soc. 1995, 117, 6673.
(f) Depre, D.; Duffels, A.; Green, L. G.; Lenz, R.; Ley, S. V.; Wong,
C.-H. Chem. Eur. J. 1999, 5, 3326.
9
10.1021/ja029302m CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 4155-4165
4155

A R T I C L E S
Gadikota et al.
Chart 1
Figure 1. Synthesis of â-arabinofuranosides (4) from 2,3-anhydro sugar
glycosylating agents 1 and 2.
Over the past 5 years, we have described the synthesis of a
number of oligosaccharides that are fragments of two arabino-
furanose-containing polysaccharides found as key components
of the mycobacterial cell wall.7 These polysaccharides contain
both 1,2-trans (R-arabinofuranosyl) and 1,2-cis (â-arabinofura-
nosyl) linkages, and to date we have synthesized fragments
ranging in size from disaccharides to hexasaccharides. In the
synthesis of these targets, the installation of the â-arabinofura-
nosyl residues was generally problematic because these glyco-
sylation reactions, which employed traditional 2-O-benzylated
glycosyl donors, often showed poor stereoselectivity.7d
In our search for new methods to synthesize â-arabinofura-
nosides in a stereoselective manner, we discovered that epoxy
thioglycoside 1 and glycosyl sulfoxide 2, which possesses the
2,3-anhydro-D-lyxo stereochemistry, glycosylate alcohols with
a very high degree of stereoselectivity (Figure 1).8 The major,
or frequently only, product of this reaction is the one in which
the newly formed glycosidic bond is cis to the epoxide (e.g.,
3). We further demonstrated that the epoxide in 3 can be opened
by nucleophiles to afford â-arabinofuranosides (4). In our earlier
study,8 the regioselectivity of the ring-opening reaction was
rather modest, with C-3 attack being favored over C-2 attack
by an approximately 3:1 ratio. Nevertheless, if the second step
could be optimized, the route illustrated in Figure 1 represents
a very appealing approach for the stereocontrolled synthesis of
â-arabinofuranosides. From a broader perspective, anhydro
sugars 1 and 2 represent a fundamentally new class of gly-
cosylating agent. The epoxide moiety can be viewed as a non-
participating protecting group for OH-2. However, in con-
trast to more traditional nonparticipating hydroxyl protecting
groups (e.g., alkyl or silyl ethers), the epoxide is sterically
undemanding.
Scheme 1 a
a Legend: (a) DIAD, Ph3P, BzOH, THF, 0 °C f room temperature, 45
min, 82%; (b) polymer supported azodicarboxylate, BzOH, Ph3P, THF, room
temperature, 12 h, 72%; (c) m-CPBA, CH2Cl2, -78 °C f room temperature,
2.5 h, 78%.
In this paper we report a full account of our investigations
on the use of 1 and 2 in the synthesis of glycosidic bonds and
demonstrate the power of these reagents in the synthesis of
â-arabinofuranosides. A key feature of this novel route for the
preparation of these linkages is a highly regioselective (-)-
sparteine-mediated nucleophilic opening of the epoxide moiety
in the glycosides that are produced from 1 and 2. We also report
here that anhydro sugars 5 and 6 (Chart 1), possessing the 2,3-
anhydro-D-ribo stereochemistry, can be used as precursors to
R-arabinofuranosides. Finally, we demonstrate the utility of this
methodology through the synthesis of a hexasaccharide motif
(7, Chart 1) found in two mycobacterial cell wall polysaccha-
rides.9 In the synthesis of 7, three of the six residues were
installed by way of these anhydro sugar glycosylating agents.
Results and Discussion
Synthesis of Glycosylating Agents 1, 2, 5, and 6. The
successful application of synthetic methodology involving these
anhydro sugars requires that straightforward methods for their
preparation are available. We have developed such routes for
the synthesis of 1, 2, 5, and 6.
Illustrated in Scheme 1 is the synthesis of thioglycoside 1
and glycosyl sulfoxide 2 from 8, which was prepared in three
steps from D-arabinose.7b Reaction of 8 with diisopropyl
azodicarboxylate (DIAD), benzoic acid, and triphenylphosphine,
(9) Lowary, T. L. Mycobacterial Cell Wall Components. In Glycoscience:
Chemistry and Chemical Biology; Fraser-Reid, B., Tatsuta, K., Thiem, J.,
Eds.; Springer-Verlag: Berlin, 2001; pp 2005-2080.
(8) Gadikota, R. R.; Callam, C. S.; Lowary, T. L. Org. Lett. 2001, 3, 607.
9
4156 J. AM. CHEM. SOC. VOL. 125, NO. 14, 2003

2,3-Anhydro Sugars in Glycoside Bond Synthesis
A R T I C L E S
Chart 2
Scheme 2 a
a Legend: (a) 60% aqueous HOAc, H2SO4 (catalytic), 70 °C, 36 h; (b) Ac2O, pyridine, room temperature, 2 h, 88% (two steps); (c) p-TolSH, BF3‚OEt2,
CH2Cl2, 0 °C f room temperature, 2 h, 91%; (d) NaOCH3, CH3OH, room temperature, 8 h; (e) BzCl, pyridine, 2 h, 0 °C f room temperature, 89% (two
steps), 67% 5, 21% 14; (f) m-CPBA, CH2Cl2, -78 °C f room temperature, 2.5 h, 79%; (g) m-CPBA, CH2Cl2, -78 °C f room temperature, 2.5 h, 77%.
afforded epoxy thioglycoside 1 in 82% yield. Similar to the
Mitsunobu reaction leading to the corresponding methyl gly-
coside 2210 (Chart 2) none of the isomeric epoxide was isolated.
The structure of the product was determined by comparison of
the NMR data for 1 with those for 22.10 Additional proof of
the identity of 1 was obtained by synthesizing its corresponding
crystalline 5-O-p-nitrobenozate ester derivative 9 (Chart 1), the
structure of which was solved by X-ray crystallography (see
Supporting Information). The purification of 1 was complicated
by presence of the reduced DIAD that is produced during the
reaction. We therefore explored the use of a polymer-bound
azodicarboxylate11 and found that the reaction afforded the same
product, although the rate is slower and the yield somewhat
lower (72% yield from 8). Oxidation of 1 with m-CPBA
provided sulfoxide 2 in 78% yield as a >10:1 mixture of
diastereomers.
The synthesis of 5 and 6 (Scheme 2) was slightly more
involved than the preparation of 1 and 2 but nevertheless could
be carried out with minimal difficulty. Conversion of the 1,2-
O-isopropylidene xylose derivative 1112 into 12 was achieved
(10) Martin, M. G.; Ganem, B.; Rasmussen, J. R. Carbohydr. Res. 1983, 123,
332.
(11) Arnold, L. D.; Assil, H. I.; Vederas, J. C. J. Am. Chem. Soc. 1989, 111,
3973.
9
J. AM. CHEM. SOC. VOL. 125, NO. 14, 2003 4157

A R T I C L E S
Gadikota et al.
Table 1. Glycosylations with 1 and 2
in 88% overall yield by hydrolysis of the acetal with aqueous
acetic acid followed by acetylation of the resulting diol.
Treatment of 12 with p-thiocresol and boron trifluoride ether-
ate afforded, in 91% yield, a 1:2.6 R:â mixture of thioglyco-
sides 13, which could not be separated. Reaction of 13 with
sodium methoxide followed by benzoylation of the intermediate
epoxy alcohol afforded thioglycosides 5 and 14, in a combined
yield of 89% over the two steps. These thioglycosides were
isolated pure following chromatography. The structures of 5
and 14 could not be assigned unambiguously using NMR
spectroscopy. Therefore, the minor product, which we expected
to be 14, was treated with sodium methoxide and the product
alcohol was converted to its 5-O-p-nitrobenzoate ester. X-ray
analysis of the resulting crystalline solid (see Supporting
Information) showed this compound to be 10 (Chart 1), thus
confirming that the minor epoxy thioglycoside produced by this
route is 14. Oxidation of 5 and 14 to sulfoxides 6 and 15
proceeded in 79% and 77% yields, respectively. As in the
preparation of 2, a mixture of both diastereomeric sulfoxides
was produced.
c
entry
R′
alcohol
activationa
product
yield (%)b
â:R ratio
1
2
3
4
5
6
7
8
STol
STol
STol
STol
STol
STol
STol
STol
STol
STol
STol
STol
STol
STol
STol
STol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
STol
18
19
20
21
22
24
25
26
27
28
29
30
31
32
33
34
22
22
24
24
25
26
26
29
29
31
31
33
33
34
34
18
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
B
C
B
C
B
B
C
B
C
B
C
B
C
B
C
D
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
38
38
40
40
41
42
42
45
45
47
47
49
49
50
50
35
79
81
83
80
77
81d
83
81
72
84
79
81
82e
51f
82
75g
82
84
80
82
78
82
83
83
79
74
80
71
77
78
81
77
â only
â only
6.5:1
8:1
â only
7:1
5:1
9:1
9
â only
â only
â only
â only
3:1
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
3:2
7:1
5:1
â only
â only
10:1
In an attempt to shorten the synthesis of 5 and 6, we
investigated the possibility of installing the thioglycoside directly
from 11, by reaction with p-thiocresol and boron trifluoride
etherate. Although under these conditions some of the desired
thioglycoside (16, Chart 1) was produced, the yields were not
reproducible and significant amounts of dithioacetal 17 (Chart
1) were sometimes also formed. We therefore view the route
that proceeds via 13 as superior.
â only
8:1
9:1
â only
5:1
â only
7:1
â only
â only
â only
8:1
Glycosylation Reactions. With donors 1, 2, 5, and 6 in hand,
we explored their use in glycosylation reactions. We first studied
the ability of 1 and 2 to glycosylate a panel of alcohols
(18-34, Chart 2).13 The results of these glycosylation reactions
are presented in Table 1. The thioglycosides were activated by
treatment with N-iodosuccinimide (NIS) and silver triflate
(AgOTf)14 at -40 °C in dichloromethane (activation method
A, Table 1). With the sulfoxide donors, we employed triflic
anhydride (Tf2O) activation in dichloromethane15 using the
protocol developed by Crich and co-workers.16 Under these
conditions, the sulfoxide was treated first with Tf2O in the
presence of 2,6-di-tert-butyl-4-methylpyridine (DTBMP) and,
after the mixture was stirred for a certain period of time, the
alcohol was added. Two variants of this method were explored.
Initially, we used a protocol in which after the sulfoxide was
treated with Tf2O at -78 °C, the reaction mixture was stirred
for 10 min at this temperature before the alcohol was added
(activation method B). Later, on the basis of the results of
â only
â only
a Legend: (A) stir thioglycoside (0.6 mmol), alcohol (0.5 mmol),
N-iodosuccinimide (0.6 mmol), silver triflate (0.15 mmol), and crushed 4
Å molecular sieves in 10 mL of CH2Cl2 at -40 °C; (B) stir sulfoxide (0.5
mmol), Tf2O (0.6 mmol), DTBMP (2.0 mmol), and crushed 4 Å molecular
sieves in 10 mL of CH2Cl2 at -78 °C for 10 min and then add the alcohol
(0.6 mmol); (C) stir sulfoxide (0.5 mmol), Tf2O (0.6 mmol), DTBMP (2.0
mmol), and crushed 4 Å molecular sieves in 10 mL of CH2Cl2 at -78 °C
for 10 min, raise the temperature to -40 °C and stir for 20 min, and then
add the alcohol (0.6 mmol); (D) stir thioglycoside (0.6 mmol), 1-benzene-
sulfinyl piperidine (BSP, 0.6 mmol), DTBMP (1.2 mmol), and crushed 4
Å molecular sieves in 10 mL of CH2Cl2 at -60 °C for 20 min and then
add Tf2O (0.7 mmol); after 15 min, warm the reaction mixture to -40 °C
and then add the acceptor (0.8 mmol) in CH2Cl2 (1 mL) and stir for 10
min. b Isolated yield following chromatography. c Ratio determined by
product yields following chromatography. d 5% of 51 obtained. e 11% of
52 obtained. f 20% of 53 obtained. g 20% of 54 obtained.
low-temperature NMR experiments,17 we modified the protocol
such that after the reaction mixture was stirred at -78 °C
for 10 min, the solution was warmed to -40 °C and then
stirred for 20 min prior to the addition of the alcohol (activation
method C).
From the results presented in Table 1, it is clear that that 1
and 2 do efficiently glycosylate a range of alcohols with a very
high degree of stereocontrol. Primary, secondary, and tertiary
simple alcohols are readily glycosylated (Table 1, entries 1-4),
as are primary and secondary carbohydrate alcohols (Table 1,
entries 5-32). In all examples, the major product is the
â-glycoside, the product in which the newly formed bond is
cis to the epoxide. Frequently, the stereoselectivity is total and
only the â-glycoside is formed.
(12) Chen, S.-H.; Lin, S.; King, I.; Spinka, T.; Dutschman, G. E.; Gullen, E.
A.; Cheng, Y.-C.; Doyle, T. W. Bioorg. Med. Chem. Lett. 1998, 8, 3245.
(13) Alcohols 18-21 are commercially available, while acceptors 22-29 and
31-33 were synthesized as previously reported. 22: Reference 10. 23:
Reference 19. 24: Reference 8. 25: Reference 6e. 26: Montgomery, J.
A.; Shortnacy, A. T.; Thomas, H. J. J. Med. Chem. 1974, 17, 1197. 27:
Sondheimer, S. J.; Eby, R.; Schuerch, C. Carbohydr. Res. 1978, 60, 187.
28, 29, 33: Ek, M.; Garegg, P. J.; Hultberg, H.; Oscarson, S. J. Carbohydr.
Chem. 1983, 2, 305. 31: Mulard, L. A.; Kovac, P.; Glaudemans, C. P. J.
Carbohydr. Res. 1994, 259, 21. 32: Eby, R.; Schuerch, C. Carbohydr.
Res. 1982, 100, C41. The synthesis of 30 and 34 is described in the
Experimental Section.
(14) Garegg P. J. AdV. Carbohydr. Chem. Biochem. 1997, 52, 179.
(15) (a) Gildersleeve, J.; Smith, A.; Sakurai, K.; Raghavan, S.; Kahne, D. J.
Am. Chem. Soc. 1999, 121, 6176. (b) Gildersleeve, J.; Pascal, R. A., Jr.;
Kahne, D. J. Am. Chem. Soc. 1998, 120, 5961. (c) Yan, L.; Kahne, D. J.
Am. Chem. Soc. 1996, 118, 9239. (d) Kahne, D.; Walker, S.; Cheng, Y.;
Van Engen, D. J. Am. Chem. Soc. 1989, 111, 6881.
(16) (a) Crich, D.; Sun, S. J. Org. Chem. 1997, 62, 1198. (b) Crich, D.; Sun, S.
J. Org. Chem. 1996, 61, 4506.
(17) Callam, C. S.; Gadikota, R. R.; Krein, D. M.; Lowary, T. L. J. Am. Chem.
Soc., submitted.
9
4158 J. AM. CHEM. SOC. VOL. 125, NO. 14, 2003

2,3-Anhydro Sugars in Glycoside Bond Synthesis
A R T I C L E S
Figure 2. Proposed formation of 2-p-thiotoluyl-â-D-xylofuranosides (69) from p-thiotoluyl glycosides with the 2,3-anhydro-R-D-lyxo stereochemistry (1).
Table 2. Glycosylations with 5 and 6
It was not possible to determine the C-1 stereochemistry of
the product glycosides from the NMR parameters usually used
for the assignment of anomeric stereochemistry in furanosides,
.
the chemical shift of C-1, or the magnitude of 3JH-1,H-2 18 We
have, however, previously shown that an unambiguous param-
eter that can be used for assigning the stereochemistry in 2,3-
anhydro-O-glycosides is the magnitude of the one-bond C-1,H-1
coupling constant.19 In glycosides in which H-1 is trans to the
epoxide moiety, 1JC-1,H-1 ) 163-168 Hz; when this hydrogen
is cis to the oxirane ring, 1JC-1,H-1 ) 171-174 Hz. This method
was used for assigning the anomeric stereochemistry of the
glycosylation products.
c
entry
R’
alcohol
activationa
product
yield (%)b
â:R ratio
1
2
3
4
5
6
7
8
STol
STol
STol
STol
STol
STol
STol
STol
STol
STol
STol
STol
STol
STol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
S(O)Tol
STol
18
19
20
21
22
23
24
25
26
27
29
32
33
34
22
25
27
27
29
29
32
32
33
33
34
34
18
A
A
A
A
A
A
A
A
A
A
A
A
A
A
B
B
B
C
B
C
B
C
B
C
B
C
D
55
56
57
58
59
60
61
62
63
64
65
66
67
68
59
62
64
64
65
65
66
66
67
67
68
68
55
80
82
81
80
78
82
78
80
88
84
81
88
82
84
84
82
77
84
86
87
63
75
82
85
85
85
77
R only
R only
R only
R only
R only
R only
9:1
9:1
14:1
9
In analyzing the yields and â:R ratios presented in Table 1,
some trends can be noted. First, in general, the â-selectivity
decreases with increasing steric hindrance on the alcohol.
Similarly, electron-withdrawing protecting groups on the ac-
ceptor decrease the â-selectivity (compare entries 6 and 8, Table
1). However, even with relatively unreactive secondary carbo-
hydrate alcohols the â:R ratios are better than 5:1 in most cases.
The worst selectivity (3:2 â:R) is observed in the reaction of
32 with thioglycoside 1. Second, a comparison of the glyco-
sylations employing sulfoxide 2 with those that use thioglycoside
1 reveals that the former provides not only better yields of the
products but also higher â:R selectivities. Third, in the activation
of sulfoxide 2, the length of time the reaction mixture is stirred
prior to the addition of the alcohol significantly influences the
outcome of the reaction. Better yields and â:R selectivity result
from stirring the reaction mixture for a longer time and by
warming it to -40 °C before adding the alcohol (compare
reactions involving activation conditions B vs C for the same
acceptor: e.g., entries 26 and 27). We also observed that in
some of the glycosylations with thioglycoside 1 that small
amounts of 2-deoxy-2-thiocresyl-â-D-xylofuranosides (e.g., 69,
Figure 2) were produced in addition to the desired 2,3-anhydro
sugar glycosides. Presumably, these byproducts are formed by
the pathway outlined in Figure 2, which is promoted by the
triflic acid that is produced as the reaction proceeds.20 We
attempted to circumvent this rearrangement process by adding
DTBMP to the reaction mixture to neutralize any triflic acid
that was generated. However, this resulted in very low conver-
sion of the thioglycosides. Therefore, ensuring that the reaction
mixture is alkaline is not a viable method for preventing the
formation of the 2-thiocresyl glycoside byproducts. Taken
together, these observations point to sulfoxide 2 being a superior
glycosylating agent when compared to thioglycoside 1.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
R only
R only
8:1
8:1
R only
R only
R only
R only
R only
R only
R only
6:1
R only
R only
R only
R only
R only
R only
a Legend: (A) stir thioglycoside (0.6 mmol), alcohol (0.5 mmol),
N-iodosuccinimide (0.6 mmol), and silver triflate (0.15 mmol) in 10 mL of
CH2Cl2 at -40 °C; (B) stir sulfoxide (0.5 mmol), Tf2O (0.6 mmol), and
DTBMP (2.0 mmol) in 10 mL of CH2Cl2 at -78 °C for 10 min and then
add the alcohol (0.6 mmol); (C) stir sulfoxide (0.5 mmol), Tf2O (0.6 mmol),
and DTBMP (2.0 mmol) in 10 mL of CH2Cl2at -78 °C for 10 min, raise
the temperature to -40 °C and stir for 20 min, and then add the alcohol
(0.6 mmol; (D) stir thioglycoside (0.6 mmol), 1-benzenesulfinyl piperidine
(BSP, 0.6 mmol), DTBMP (1.2 mmol), and crushed 4 Å molecular sieves
in 10 mL of CH2Cl2 at -60 °C for 20 min and then add Tf2O (0.7 mmol);
after 15 min, warm the reaction mixture to -40 °C and then add the acceptor
(0.8 mmol) in CH2Cl2 (1 mL) and stir for 10 min. b Isolated yield following
chromatography. c Ratio determined by product yields.
used for the activation of these susbstrates were identical with
those used with 1 and 2, and the magnitude of the JC,H was
1
again used to determine the stereochemistry at the anomeric
center in the product glycosides.19 As with 1 and 2, the
glycosylations with 5 and 6 provided, as the major product, the
glycoside in which the aglycone is oriented cis to the epoxide
moiety. Given the stereochemistry of the epoxide ring in 5 and
6, these reactions produce predominantly the R-glycoside.
We next turned our attention to glycosylations involving
thioglycoside 5 and glycosyl sulfoxide 6 (Table 2). The methods
(20) Analogous migrations of thioglycosides have been previously reported: (a)
Yu, B.; Yang, Z. Org. Lett. 2001, 3, 377. (b) Johnston, B. D.; Pinto, B. M.
J. Org. Chem. 2000, 65, 4607. (c) Zuurmond, H. M.; van der Klein, P. A.
M.; van der Marel, G. A.; van Boom, J. H. Tetrahedron 1993, 49, 6501.
(d) Auzanneau, F. I.; Bundle, D. R. Carbohydr. Res. 1991, 212, 13.
(18) Mizutani, K.; Kasai, R.; Nakamura, M.; Tanaka, O.; Matsuura, H.
Carbohydr. Res. 1989, 185, 27.
(19) Callam, C. S.; Gadikota, R. R.; Lowary, T. L. J. Org. Chem. 2001, 66,
4549.
9
J. AM. CHEM. SOC. VOL. 125, NO. 14, 2003 4159

A R T I C L E S
Gadikota et al.
Figure 4. Possible products resulting from the nucleophilic opening of
the epoxide moiety of 2,3-anhydro-R-D-ribofuranosides (72) and 2,3-
anhydro-â-D-lyxofuranosides (73).
in essentially identical yield. Analogous results were obtained
upon glycosylation with the 2,3-anhydro-D-ribo donors; coupling
of n-octanol with 5, 6, 14, or 15 gave the R-glycoside in nearly
the same yield. Thus, the stereochemistry at the anomeric center
in the donor does not influence the â:R ratio of the products.17
In other experiments, we determined that the stereochemistry
at sulfur in the sulfoxides had no impact on the R:â ratio in the
products (see Table S1 in the Supporting Information).
Epoxide Ring Opening Reactions. If anhydro sugar deriva-
tives 1, 2, 5, and 6 are to be useful reagents for the synthesis of
glycosidic bonds, then it is critical that methods be available
for the regioselective opening of the oxirane ring in the glycoside
products. 2,3-Anhydro sugars have proven to be useful inter-
mediates for the preparation of a number of modified sugar
derivatives, in particular in the area of nucleoside synthesis.24
On the basis of previous studies24,25 addressing the nucleophilic
opening of the epoxide rings in these anhydro sugars, we
expected that the regioselectivity of the reaction with the 2,3-
anhydro-D-ribo glycosides (e.g., 72, Figure 4) would be high.
Nucleophilic attack at C-2 would be preferred because the
protected hydroxymethyl substituent at C-4 hinders C-3 attack.
In contrast, we expected that the regioselectivity of the epoxide
openings in the 2,3-anhydro-D-lyxo glycoside series (e.g., 73,
Figure 4) would be lower. Solely on the basis of steric hindrance
considerations, nucleophilic attack at either C-2 or C-3 would
be expected to be equally likely, as the bottom face of the
furanose ring has no substituents to bias the approach of the
nucleophile to the epoxide. However, previous results24,25 did
suggest that, although the regioselectivity of the reaction with
73 would likely be reduced relative to 72, attack at C-3 would
be preferred with most nucleophiles. The factors that govern
the regioselectivity preference in the nucleophilic opening of
2,3-anhydro-â-D-lyxo furanosides are not completely understood.
Figure 3. Reactions carried out to determine if the stereochemistry at the
anometric center in the donor influences the stereochemical outcome of
the glycosylation reactions. Note that the yields are high and essentially
identical regardless of the anomeric stereochemistry in the donor. Legend:
(a) n-octanol, NIS, AgOTf, CH2Cl2, -40 °C, see conditions A in Table 1
for protocol; (b) Tf2O, DTBMP, CH2Cl2, -78 °C, then n-octanol, see
conditions B in Table 1 for protocol.
When these reactions are compared with those employing 1 and
2, some differences are observed. First, the stereoselectivity of
the reactions with 5 and 6 is generally higher than those with
1 and 2.21 Second, whereas glycosyl sulfoxide 2 is a superior
donor to thioglycoside 1, the reverse situation is true with 5
and 6. In general, sulfoxide 6 provides the products in lower
yield than thioglycoside 5. Finally, with thioglycoside 5, only
trace amounts of the 1-thiocresyl rearrangement products are
produced. The formation of these rearranged byproducts was
more significant with 1.
We also demonstrated that both thioglycosides 1 and 5 can
be converted to O-glycosides with high stereoselectivity upon
activation with 1-benzenesulfinyl piperidine (BSP)22 (see Table
1, entry 32, and Table 2, entry 27).
In further studies, we explored if the stereochemistry at the
anomeric center in the donor influences the stereochemical
outcome of the reaction (Figure 3). To this end, we glycosylated
n-octanol with 1 and 2, as well as their â-anomer counterparts
70 and 71.23 With all four donors, the â-glycoside was obtained
(22) Crich, D.; Smith, M. J. Am. Chem. Soc. 2001, 123, 9015.
(23) These compounds were synthesized using the transformations outlined in
Scheme 1 from the â-anomer of 8, which was obtained as a minor byproduct
in its synthesis (see Supporting Information).
(24) For some examples see: (a) Huryn, D. M.; Okabe, M. Chem. ReV. 1992,
92, 1745. (b) Habich, D.; Barth, W. Synthesis 1988, 943 (c) Mengel, R.;
Griesser, H. Tetrahedron Lett. 1977, 18, 1177. (d) Robins, M. J.; Janda,
K. D.; Hansske, F.; Chen, J. J. Org. Chem. 1991, 56, 3410. (e) Roussev,
C. D.; Simeonov, M. F.; Petkov, D. D. J. Org. Chem. 1997, 62, 5239. (f)
Reichman, U.; Hollenberg, D. H.; Chu, C. K.; Fox, J. J. J. Org. Chem.
1976, 41, 2043. (g) Robins, M. J.; Fouron, Y.; Mengel, R. J. Org. Chem.
1974, 39, 1564. (h) Miah, A.; Reese, C. B.; Song, Q.; Sturdy, Z.; Neidle,
S.; Simpson, I. J.; Read, M.; Rayner, E. J. Chem. Soc., Perkin Trans. 1
1998, 3277.
(21) The relatively increased stereoselectivity may by due to neighboring group
participation from the O-5 benzoate group. Such an event would lead to
the formation of an acyloxonium ion over the â-face of the furanose ring,
thus forcing the alcohol to attack C-1 from the R-face. However, we see
no evidence of the formation of this intermediate. In particular, no ortho
ester byproduct was detected in the glycosylations with sulfoxide 6. Under
the basic reaction conditions of these reactions, the formation of an
acyloxoium ion of this type would be expected to lead to the formation of
significant amounts of this ortho ester byproduct.
(25) Williams, N. R. AdV. Carbohydr. Chem. Biochem. 1970, 25, 109.
9
4160 J. AM. CHEM. SOC. VOL. 125, NO. 14, 2003

2,3-Anhydro Sugars in Glycoside Bond Synthesis
A R T I C L E S
Table 3. Opening of R-ribo-Epoxides
Monosaccharide 35 and disaccharides 42 and 45 were used
as model substrates when exploring the regioselectivity of the
nucleophilic opening of the epoxide in species containing a 2,3-
anhydro-â-D-lyxofuranoside moiety (Table 4). Our initial in-
vestigations showed that, as previously reported,8 relatively
disappointing levels of regioselectivity resulted from the reac-
tions in which NaN3, LiAlH4, NaOBn, or NaOAll was used as
the nucleophile. Although C-3 attack was favored, the ratio of
â-arabinofuranoside to â-xylofuranoside was at best 3.7:1. While
this regioselectivity was disappointing, we were nevertheless
encouraged enough to explore a number of modifications of
the reaction conditions. Given our interest in synthesizing
naturally occurring oligosaccharides via this route, our attention
was focused on the use of the alkoxides as nucleophiles.
Through variation of the reaction conditions we have discovered
that the regioselectivity can be dramatically improved by
changing the counterion and through the use of additives. We
initially carried out the reaction with NaOBn in the presence
of 18-crown-6, but no change in regioselectivity was observed
relative to the reaction in the absence of the additive (compare
entries 3 and 5, Table 4). Changing the counterion to potassium,
in either the presence or the absence of 18-crown-6, reduced
the regioselectivity (Compare entries 3, 6, and 7, Table 4). In
contrast, replacement of sodium with lithium provided a
significant increase in regioselectivity, in favor of C-3 attack.
This substitution increased the arabino:xylo ratio from 3.5:1 to
6.5:1 (compare entries 3 and 8, Table 4). When the reaction
with the lithium alkoxide was carried out in the presence of
15-crown-5 (1.2 equiv relative to the amount of LiOBn) the
regioselectivity was increased further to 10:1 in favor of the
â-arabinofuranoside (Table 4, entry 9). In addition to enhancing
the regioselectivity, another advantage of using the crown ether
is that the reaction can be carried out at a significantly lower
temperature; 80 °C vs 120 °C. The use of TMEDA as the
additive gave results inferior to those obtained with the crown
ether (Table 4, entry 10), as both the regioselectivity and the
rate of reaction were reduced. The most impressive results were
obtained when (-)-sparteine (78, Chart 3) was added to the
reaction mixture. Heating 35, 42, or 45 with LiOBn in benzyl
alcohol in the presence of (-)-sparteine (1.2 equiV relatiVe to
the amount of the lithium alkoxide) proVided the â-arabino-
furanoside as the sole product in excellent yield (Table 4, entries
11-15). Additionally, in the presence of (-)-sparteine the
reaction was substantially faster than the reactions both in the
absence of the additive and in the presence of 15-crown-5.
Whereas the reaction with no additive required 3 h at 120 °C
to complete, when 78 was added, the reaction was done in 20
min at 70 °C.
temp
time
(h)
yield
(%)a
arabino:xylo
ratio
c
entry
nucleophile
solvent
(°C)
b
1
2
3
4
NaN3
EtOH
THF
BnOH
AllOH
reflux
reflux
120
6
3
2
2
91
84
87
85
9:1 (74a:75a)
10:1 (74b:75b)
8:1 (74c:75c)
8:1 (74d:75d)
LiAlH4
NaOBn
NaOAll
120
a Isolated yield. b NH4Cl added to the reaction mixture. c Ratio deter-
mined by by 1H NMR spectroscopy, through integration of the resonances
for the anomeric hydrogen of the nonreducing sugar residue.
One possible explanation for this experimental observation is
that the partial positive charge that develops in the SN2 transition
state is favored at C-3 relative to C-2, due to the latter’s closer
proximity to the electron-withdrawing anomeric center.25
Mindful of these issues, we explored the nucleophilic opening
of the epoxides in the some of the product glycosides. At the
outset, we note that one characteristic shared by all of the epoxy
glycosides is the stability of the oxirane ring. In order for the
ring-opening reactions to proceed at a reasonable rate, it was
necessary that they be carried out at elevated temperatures.
Despite having to heat these reaction mixtures to moderately
high temperatures, no significant amount of product or substrate
degradation was observed and the ring-opened products were
always isolated in good to excellent yield. We also point out
that in the reactions with some nucleophiles (hydride, alkoxides)
that the O-5-benzoyl group is cleaved during the course of the
reaction.
We are currently unsure as to the origin of the remarkable
regioselectivity that is observed when 78 is added to these
reaction mixtures. The use of (-)-sparteine in conjunction with
alkyllithiums to carry out chiral deprotonation reactions is well-
known,27 and this methodology has recently been extended to
R-deprotonations of meso-epoxides (e.g., 79, Chart 3) for the
enantioselective synthesis of bicyclic alcohols (e.g. 80)28
Another recent report describes the use of 78 as a ligand for
We first explored the opening of the glycosides with the 2,3-
anhydro-D-ribo stereochemistry, using 65 as a model substrate
(Table 3). As predicted, the regioselectivity of the openings is
high with all the nucleophiles used (NaN3, LiAlH4, NaOBn,
NaOAll). Attack at C-2 is preferred, and the R-arabinofuranoside
is produced as the major product over the R-xylofuranoside. In
all cases, the regioselectivity is better than 8:1 and the combined
yields of 74 and 75 are excellent. The structures of the major
products could be readily determined through the use of 1H and
13C NMR spectroscopy.26
(27) (a) Hoppe, D.; Hense, T. Angew. Chem., Int. Ed. 1997, 36, 2282. (b) Beak,
P.; Basu, A.; Gallagher, D. J.; Park, Y. S.; Thayumanavan, S. Acc. Chem.
Res. 1996, 29, 552.
(28) (a) Hodgson, D. M.; Cameron, I. D.; Christlien, M.; Green, R.; Lee, P. G.;
Robinson, L. A. J. Chem. Soc., Perkin Trans. 1 2001, 2161. (b) Alexakis,
A.; Vrancken, E.; Mangeney, P. J. Chem. Soc., Perkin Trans. 1 2000, 3354.
(26) Mizutani, K.; Kasai, R.; Nakamura, M.; Tanaka, O.; Matsuura, H.
Carbohydr. Res. 1989, 185, 27.
9
J. AM. CHEM. SOC. VOL. 125, NO. 14, 2003 4161

A R T I C L E S
Gadikota et al.
Table 4. Opening of â-lyxo-Epoxides
entry
substrate
nucleophile
additive
NH4Cl
solvent
temp (°C)
time
3 h
3 h
3 h
3 h
3 h
3 h
3 h
3 h
3 h
5 h
20 min
20 min
20 min
3 h
yield (%)a
arabino:xylo ratiob
1
2
3
4
5
6
7
8
45
45
45
45
45
45
45
45
45
45
45
42
35
45
NaN3
EtOH
THF
reflux
reflux
120
120
120
120
120
120
80
70
70
70
70
90
86
81
83
67
73
70
79
71
70
85
83
86
62
3:1 (76a:77a)
3:1 (76b:77b)
LiAlH4
NaOBnc
NaOAllc
NaOBnc
KOBnc
KOBnc
LiOBnc
LiOBnc
LiOBnc
LiOBnc
LiOBnc
LiOBnc
NaOBnc
BnOH
AllOH
BnOH
BnOH
BnOH
BnOH
BnOH
BnOH
BnOH
BnOH
BnOH
BnOH
3.5:1 (76c:77c)
2.9:1 (76d:77d)
3.7:1 (76c:77c)
2.6:1 (76c:77c)
2.7:1 (76c:77c)
6.5:1 (76c:77c)
10:1 (76c:77c)
7:1 (76c:77c)
only arabino (76c)
only arabino (76e)
only arabino (76f)
3.5:1 (76c:77c)
18-crown-6d
18-crown-6d
9
15-crown-5d
TMEDA
10
11
12
13
14
(-)-sparteined
(-)-sparteined
(-)-sparteined
(-)-sparteined
70
1
a Isolated yield. b Ratio determined by product yields or by H NMR spectroscopy, through integration of the resonances for the anomeric hydrogen of
the nonreducing sugar residue. c 3.0 equiv of alkoxide relative to epoxide. d 1.2 equiv of additive relative to alkoxide.
Chart 3
However, to the best of our knowledge, the work described in
this report is the first example of the use of (-)-sparteine as a
method for controlling the regioselectivity of an epoxide opening
using a non-carbon nucleophile.
Our initial supposition was that the chirality of 78 was an
important factor in determining the outcome of the reaction
through the formation of a matched pair complex with the
epoxide and the lithium ion. Similar complexes have been pro-
posed in other (-)-sparteine-mediated reactions of epoxides.32
This is supported by the observation that when these additions
are carried out with either 15-crown-5 or TMEDA, the regio-
selectivities of the reactions are lower than when 78 is used.
Unfortunately, (+)-sparteine is not readily available,33 and so
we were unable to test this hypothesis by carrying out the
reaction in the presence of the enantiomer of 78. However, both
D- and L-arabinose are commercially available at low cost, and
the easy access to these sugars allowed us to determine if the
absolute stereochemistry of 78 was an important factor in the
regioselectivity of these reactions. Three disaccharides, 84-86,
were synthesized (see the Supporting Information for details),
the synthesis of R,â-epoxy silanes29 (81, Chart 3). However,
and the epoxide ring in each was opened with LiOBn in the
presence of (-)-sparteine. The results, together with those for
disaccharide 42, are presented in Figure 5. From these investiga-
tions, it is clear that glycosides containing the 2,3-anhydro-â-
L-lyxofuranoside moiety (e.g., 84 and 86) are also opened at
C-3 by LiOBn with similarly high regioselectivity. Furthermore,
when the stereochemistry of the reducing-end residue is changed
reports of the use of (-)-sparteine as a means for influencing
addition reactions to epoxides are much less frequent. Alexakis
and co-workers have described that the reaction of organolithium
reagents to meso-epoxides in the presence of 78 and boron
trifluoride affords ring-opened products with moderate enan-
tiomeric excess.30 More recent work by Beak and co-workers
has described the asymmetric synthesis of N-Boc-protected
pyrrolidines and piperidines (e.g., 82) via treatment of N-Boc
protected amines (e.g., 83) with n-butyllithium and 78.31
(30) Alexakis, A.; Vrancken, E.; Mangeney, P. Synlett 1998, 1165.
(31) Serino, C.; Stehle, N.; Park, Y. S.; Florio, S.; Beak, P. J. Org. Chem. 1999,
64, 1160.
(32) Hodgson, D. W.; Gibbs, A. R. Lee, G. P. Tetrahedron 1996, 52, 14361.
(33) The first asymmetric synthesis of (+)-sparteine has been reported re-
cently: Smith, B. T.; Wendt, J. A.; Aube, J. Org. Lett. 2002, 4, 2577.
(29) Hodgson, D. M.; Norsikian, S. L. M. Org. Lett. 2001, 3, 461.
9
4162 J. AM. CHEM. SOC. VOL. 125, NO. 14, 2003

2,3-Anhydro Sugars in Glycoside Bond Synthesis
A R T I C L E S
Figure 5. Demonstration that the chirality of the (-)-sparteine is not important in determining the regiochemical outcome of the nucleophilic opening of
2,3-anhydro-â-lyxofuranosides.
Figure 6. Retrosynthetic analysis of 7. A previous synthesis7a employed monosaccharides 90-93 (left-hand side). The synthesis described here uses 2, 5,
26, and 94 (right-hand side).
Scheme 3 a
from D to L (85, 86), the regioselectivity is unchanged. It appears,
therefore, that the chirality of 78 is unimportant in determining
the regiochemistry of the epoxide ring opening reaction.
Consideration of the results in Figure 5, together with the fact
that the selectivity of the ring opening is reduced when TMEDA
(Table 4, entry 10) is used suggests that the rigid structure of
the (-)-sparteine may be a key issue leading to the high
a Legend: (a) BzCl, pyridine, 30 min, 0 °C, 94%; (b) p-TolSH, BF3‚OEt2,
CH2Cl2, 0 °C f room temperature, 1 h, 93%.
regioselectivity. Also possible is that the bite angle between
the nitrogens is important in the binding of the lithium and that
the decrease in selectivity with TMEDA is a consequence of
the two-carbon tether between the nitrogens that cannot achieve
the required geometry.
(benzoyl) protecting group that was likely immediately cleaved
upon the addition of the alkoxide. It is plausible that a complex
involving the lithium ion, (-)-sparteine, the epoxide oxygen,
and the alkoxide derived from OH-5 is formed and that this is
critical in determining the regioselectivity of the reaction.
However, at this point we have no evidence to support the
formation of such a complex in these reactions.
We are currently exploring the scope of this epoxide ring-
opening reaction with regard to solvent, additive, and the
protecting group on OH-5. In the cases we have examined to
date, OH-5 was either unprotected or protected by a base-labile
9
J. AM. CHEM. SOC. VOL. 125, NO. 14, 2003 4163

A R T I C L E S
Gadikota et al.
Scheme 4 a
a Legend: (a) 5, NIS, AgOTf, CH2Cl2, -40 °C, 10 min, 82%; (b) NaOBn, BnOH, 120 °C, 3 h, 84%; (c) 94, NIS, AgOTf, CH2Cl2, 0 °C, 20 min, 81%;
(d) NaOCH3, CH2Cl2:CH3OH (3:1), 8 h, room temperature, 98%; (e) 2, Tf2O, DTBMP, CH2Cl2, -78 °C f -40 °C, then 98, -40 °C f room temperature,
30 min, 74%; (f) LiOBn, (-)-sparteine, BnOH, 20 min, 70 °C, 71%. (g) 10% Pd/C, H2, AcOH:H2O (4:1), 4 h, 91%.
We have also explored the nucleophilic opening of these
epoxy glycosides under acidic conditions, but with very little
success. No reaction was observed when 35 or 45 was treated
with a range of Lewis acids (e.g., Sm(OTf)3, Ti(O-i-Pr)4, LiClO4,
Sc(OTf)3, Yb(OTf)3) and alcohols (benzyl alcohol, allyl alcohol).
Similar results were observed when the chromium salen catalysts
developed by Jacobsen and co-workers34 were used. It appears
that these epoxides are sufficiently unreactive to undergo ring-
opening reactions promoted by mild Lewis acids. We subse-
quently used stronger Lewis acids (e.g., TiCl4) for these
reactions; however, under these conditions, only decomposition
of the epoxy glycoside was observed. Accordingly, on the basis
of our results to date, we do not view the nucleophilic opening
of these epoxides under acidic conditions as viable.
Application to Oligosaccharide Synthesis. The investiga-
tions described above clearly demonstrate the potential of 2,3-
anhydrofuranose derivatives in the synthesis of oligosaccharides
containing furanose residues. The stereoselectivities of the
glycosylation reactions involving 1, 2, 5, and 6 are high, and
we have identified conditions under which the epoxide rings in
the product glycosides can be opened in a highly regioselective
manner. The major products produced from the epoxide ring-
opening reactions possess the arabino stereochemistry, and
hence, these reagents are extremely useful for the synthesis of
arabinofuranosides. In particular, the conversion of 1 and 2 into
â-arabinofuranosides (Figure 1) is, we believe, the method of
choice for the synthesis of these glycosidic linkages, which have
previously been notoriously difficult synthetic targets.6a,b,7a To
further demonstrate the potential of these anhydrosugars in
glycoside bond synthesis, we have applied the methodology
described here to the synthesis of hexasaccharide 7 (Chart 1),
a key structural motif in two mycobacterial cell wall polysac-
charides: arabinogalactan and lipoarabinomannan.9
One previous synthesis of 7 has been reported, via a route
that employs the four monosaccharide building blocks 90-93
(Figure 6).7a,35 The target was obtained from 90-93 in seven
steps in 21% overall yield. In that synthesis, the key step was
a low-temperature glycosylation reaction of a tetrasaccharide
diol with 2 equiv of thioglycoside 93. We envisioned that 7
could be obtained from four other monosaccharide building
blocks (2, 5, 26, and 94, Figure 6), also in seven steps.
We first turned our attention to the synthesis of the appropriate
monosaccharide precursors. Methyl glycoside 26 is known,13
and the synthesis of 2 and 5 is described above. Thioglycoside
94 was synthesized as outlined in Scheme 3 via a route that
started with methyl glycoside 34.36
With the building blocks in hand, the synthesis of hexasac-
charide 7 was readily achieved as outlined in Scheme 4. First,
glycosylation of alcohol 26 with epoxy thioglycoside 5 provided
disaccharide 63 in 82% yield, together with a 6% yield of the
corresponding â-glycoside. The oxirane ring in the product was
subsequently opened by heating 63 in a solution of sodium
benzylate in benzyl alcohol to afford, in 84% yield, disaccharide
96. Reaction of this diol with an excess of thioglycoside 94
(35) Syntheses of two pentasaccharides related to 7 have also been reported.6a,b
(36) Synthesized from the known methyl 5-O-benzyl-2,3-anhydro-R-D-lyxo-
furanoside (Wright, J. A.; Taylor, N. F. Carbohydr. Res. 1967, 3, 333) as
described in the Supporting Information.
(34) Jacobsen, E. N. Acc. Chem. Res. 2000, 33, 421.
9
4164 J. AM. CHEM. SOC. VOL. 125, NO. 14, 2003

2,3-Anhydro Sugars in Glycoside Bond Synthesis
A R T I C L E S
provided tetrasaccharide 97, which was deprotected to give the
corresponding diol 98 (79% from 96). This tetrasaccharide was
then glycosylated with glycosyl sulfoxide 2, providing a 74%
yield of hexasaccharide 99. This reaction also gave small
amounts of other stereoisomers of 99 that were not characterized.
These byproducts could be easily separated from the desired
hexasaccharide by chromatography. Opening of the two ep-
oxides in 99 was achieved by the (-)-sparteine-mediated
reaction described above, which provided hexasaccharide 100
in 71% yield. The final product 7 was obtained in 91% yield
by hydrogenation of 100. Using this route, it was possible to
synthesize 7 on a >50 mg scale, with the overall yield over the
seven steps being 26%. In comparing the route shown in Scheme
4 with the previous synthesis of 7, the number of steps is
equivalent, but the overall yield is slightly higher (26% vs 21%).
This synthesis clearly demonstrates that these epoxide-containing
glycosylating agents are viable reagents for the synthesis of
complex oligosaccharides.
lithium alkoxides when the reaction is carried out in the presence
of (-)-sparteine (78). This represents the first example of the
use of 78 to influence the regioselectivity of an epoxide ring
opening reaction with a non-carbon nucleophile. We have further
demonstrated the potential of these reagents through the
synthesis of hexasaccharide 7, via a route that is superior to
one that we reported previously.7a Investigations currently
underway include further exploration of the utility of other 2,3-
anhydro sugars in glycoside bond synthesis, the determination
of the scope of the (-)-sparteine-mediated epoxide ring opening
reaction, the elucidation of the mechanisms by which these
processes occur,17 and the application of this methodology to
the synthesis of other complex oligosaccharides.
Experimental Section
See the Supporting Information.
Acknowledgment. This work was supported by the National
Institutes of Health (Grant No. AI44045-01) and the National
Science Foundation (Grant No. CHE-9875163). The contribu-
tions of C.S.C. to this work were supported first by a GAANN
fellowship from the U.S. Department of Education, later by an
American Chemical Society Division of Organic Chemistry
Fellowship sponsored by Aventis Pharmaceuticals, and finally
by a Presidential Fellowship provided by the Ohio State Uni-
versity Graduate School.
Conclusions
In conclusion, we report here that 2,3-anhydrofuranose
thioglycosides and glycosyl sulfoxides 1, 2, 5, and 6 can be
used in highly stereoselective glycosylation reactions. The major
products are those in which the newly formed glycosidic bond
is cis to the epoxide moiety. These species represent a
fundamentally new class of glycosylating agent in which the
hydroxyl groups at C-2 and C-3 are “protected” as an epoxide.
We have also demonstrated that post-glycosylation opening of
the oxirane moiety can be carried out in a highly regioselective
manner providing products with the arabino stereochemistry.
A particularly notable discovery is that the epoxide moiety in
glycosides possessing the 2,3-anhydro-â-lyxo stereochemistry
(e.g., 42, Table 4) can be regioselectively opened at C-3 by
Supporting Information Available: NMR spectra and char-
acterization data for all new compounds, details on the prepara-
tion of 84-86, and X-ray crystallographic data for 9 and 10.
This material is available free of charge via the Internet at
http://pubs.acs.org.
JA029302M
9
J. AM. CHEM. SOC. VOL. 125, NO. 14, 2003 4165