Highly α- and β-selective radical C-glycosylation reactions using a controlling anomeric effect based on the conformational restriction strategy. A study on the conformation - Anomeric effect - Stereoselectivity relationship in anomeric radical reactions

Article abstract of DOI:10.1021/ja011321t

We hypothesized that, because the stereoselectivity of anomeric radical reactions was significantly influenced by the anomeric effect, which can be controlled by restricting the conformation of the radical intermediate, the proper conformational restriction of the pyranose ring of the substrates would therefore make highly α- and β-stereoselective anomeric radical reactions possible. Thus, the conformationally restricted 1-phenylseleno-D-xylose derivatives 9 and 10, restricted in a ⁴C₁-conformation, and 11 and 12, restricted in a ¹C₄-conformation, were designed and synthesized by introducing the proper protecting groups on the hydroxyl groups on the pyranose ring as model substrates for the anomeric radical reactions. The radical deuterations with Bu₃SnD and the C-glycosylation with Bu₃SnCH₂CH=CH₂ or CH₂=CHCN, using the ⁴C₁-restricted substrates 9 and 10, afforded the corresponding α-products (α/β = 97:3-85:15) highly stereoselectively, whereas the ¹C₄-restricted substrates 11 and 12 selectively gave the β-products (α/β = 1:99-0:100). Thus, stereoselectivity was significantly increased by conformational restriction and was completely inverted by changing the substrate conformation from the ⁴C₁-form into the ¹C₄-form. Ab initio calculations suggested that the radical intermediates produced from these substrates possessed the typical ⁴C₁- or ¹C₄-conformation, which was similar to that of the substrates, and that the anomeric effect in these conformations would be the factor controlling the transition state of the reaction. Therefore, the highly α- and β-selective reactions would occur because of the anomeric effect, which could be manipulated by conformational restriction of the substrates, as expected. This would be the first radical C-glycosylation reaction to provide both α- and β-C-glycosides highly stereoselectively.

Full text of DOI:10.1021/ja011321t

11870
J. Am. Chem. Soc. 2001, 123, 11870-11882
Highly R- and â-Selective Radical C-Glycosylation Reactions Using a
Controlling Anomeric Effect Based on the Conformational Restriction
Strategy. A Study on the Conformation-Anomeric Effect-
Stereoselectivity Relationship in Anomeric Radical Reactions
Hiroshi Abe, Satoshi Shuto,* and Akira Matsuda*
Contribution from the Graduate School of Pharmaceutical Sciences, Hokkaido UniVersity, Kita-12, Nishi-6,
Kita-ku, Sapporo 060-0812, Japan
ReceiVed May 30, 2001
Abstract: We hypothesized that, because the stereoselectivity of anomeric radical reactions was significantly
influenced by the anomeric effect, which can be controlled by restricting the conformation of the radical
intermediate, the proper conformational restriction of the pyranose ring of the substrates would therefore make
highly R- and â-stereoselective anomeric radical reactions possible. Thus, the conformationally restricted
1-phenylseleno-D-xylose derivatives 9 and 10, restricted in a 4C1-conformation, and 11 and 12, restricted in a
1C4-conformation, were designed and synthesized by introducing the proper protecting groups on the hydroxyl
groups on the pyranose ring as model substrates for the anomeric radical reactions. The radical deuterations
4
with Bu3SnD and the C-glycosylation with Bu3SnCH2CHdCH2 or CH2dCHCN, using the C1-restricted
substrates 9 and 10, afforded the corresponding R-products (R/â ) 97:3-85:15) highly stereoselectively, whereas
the 1C4-restricted substrates 11 and 12 selectively gave the â-products (R/â ) 1:99-0:100). Thus,
stereoselectivity was significantly increased by conformational restriction and was completely inverted by
4
1
changing the substrate conformation from the C1-form into the C4-form. Ab initio calculations suggested
4
1
that the radical intermediates produced from these substrates possessed the typical C1- or C4-conformation,
which was similar to that of the substrates, and that the anomeric effect in these conformations would be the
factor controlling the transition state of the reaction. Therefore, the highly R- and â-selective reactions would
occur because of the anomeric effect, which could be manipulated by conformational restriction of the substrates,
as expected. This would be the first radical C-glycosylation reaction to provide both R- and â-C-glycosides
highly stereoselectively.
Introduction
Scheme 1
C-Glycosides have gained increasing interest in view of their
occurrence as fragments in the structures of a number of natural
products.1 Moreover, because of their resistance to hydrolysis,
C-glycosides are expected to be stable mimics for natural
O-glycosides with biological activity.2,3 Consequently, the
methods for the synthesis of C-glycosides have been extensively
studied, via anomeric anion, cation, and radical intermediates.4
Radical C-glycosylations have the advantage of occurring under
mild neutral conditions, and those using an intramolecular
cyclization have been effective in constructing C-glycosidic
bonds highly stereoselectively. In recent years, we have been
working to develop radical C-glycosylations by intramolecular
cyclization with a silyl tether.3,5 However, while anomeric
pyranosyl radicals I, such as glucosyl radicals, stereoselectively
afford the corresponding R-product II6,7 in intermolecular radical
C-glycosylation reactions, as shown in Scheme 1, the â-selective
radical C-glycosylation is more difficult to effect.8
(1) (a) Humber, D. C.; Mulholland, K. R.; Stoodley, R. J. J. Chem. Soc.,
Perkin Trans. 1 1990, 283-292. (b) Murata, M.; Kumagai, M.; Lee, J. S.;
Yamamoto, T. Tetrahedron Lett. 1987, 28, 5869-5872. (c) Clayton, J. P.;
O’Hanlon, P. J.; Rogers, N. H. Tetrahedron Lett. 1980, 21, 881-884.
(2) (a) Watson, K. A.; Mitchell, E. P.; Johnson, L. N.; Son, J. C.; Bichard,
C. J. F.; Fleet, G. W. J.; Watkin, D. J.; Oikonomakos, N. G. J. Chem. Soc.,
Chem. Commun. 1993, 654-656. (b) Myers, R. W.; Lee, Y. C. Carbohydr.
Res. 1986, 152, 143-158. (c) Martin, J. L.; Johnson, L. N.; Withers, S. G.
Biochemistry 1990, 29, 10745-1057.
(3) Our studies on C-glycosides as stable mimics of biologically active
O-glycosides: (a) Yahiro, Y.; Ichikawa, S.; Shuto, S.; Matsuda, A.
Tetrahedron Lett. 1999, 40, 5527-5531. (b) Shuto, S.; Yahiro, Y.; Ichikawa,
S.; Matsuda, A. J. Org. Chem. 2000, 65, 5547-5557. (c) Shuto, S.; Terauchi,
M.; Yahiro, Y.; Abe, H.; Ichikawa, S.; Matsuda, A. Tetrahedron Lett. 2000,
41, 4151-4155. (d) Abe, H.; Shuto, S.; Matsuda, A. Tetrahedron Lett. 2000,
41, 2391-2394. (e) Abe, H.; Shuto, S.; Matsuda, A. J. Org. Chem. 2000,
65, 4315-4325.
(5) (a) Shuto, S.; Kanazaki, M.; Ichikawa, S.; Matsuda, A. J. Org. Chem.
1997, 62, 5676-5677. (b) Shuto, S.; Kanazaki, M.; Ichikawa, S.; Minakawa,
N.; Matsuda, A. J. Org. Chem. 1998, 63, 746-754. (c) Ueno, Y.; Nagasawa,
Y.; Sugimoto, I.; Kojima, N.; Kanazaki, M.; Shuto, S.; Matsuda, A. J. Org.
Chem. 1998, 63, 1660-1667. (d) Sugimoto, I.; Shuto, S.; Matsuda, A. J.
Org. Chem. 1999, 64, 7153-7157. (e) Sugimoto, I.; Shuto, S.; Matsuda,
A. Synlett 1999, 1766-1768. (f) Shuto, S.; Sugimoto, I.; Abe, H.; Matsuda,
A. J. Am. Chem. Soc. 2000, 122, 1343-1351. (g) Kanazaki, M.; Ueno, Y.;
Shuto, S.; Matsuda, A. J. Am. Chem. Soc. 2000, 122, 2422-2432. (h)
Sukeda, M.; Shuto, S.; Sugimoto, I.; Ichikawa, S.; Matsuda, A. J. Org.
Chem. 2000, 65, 8988-8996.
(4) (a) Postema, M. H. D. Tetrahedron 1992, 48, 8545-8599. (b)
Jaramillo, C.; Kanapp, S. Synthesis 1994, 1-20. (c) Leavy, D. E.; Tang,
C. The Chemistry of C-Glycosides; Pergamon Press: Oxford, 1995. (d)
Postema, M. H. D. C-Glycoside Synthesis; CRC Press: Boca Raton, FL,
1995. (e) Du, Y.; Linhardt, R. J. Tetrahedron 1998, 54, 9913-9959.
10.1021/ja011321t CCC: $20.00 © 2001 American Chemical Society
Published on Web 11/09/2001

SelectiVe Radical C-Glycosylation Reactions
J. Am. Chem. Soc., Vol. 123, No. 48, 2001 11871
Scheme 2
Figure 1. Conformations of pyranosyl radical intermediates.
In SN1-like O- or C-glycosylation reactions of pyranoses via
an oxocarbenium intermediate, the nucleophilic attack at the
anomeric position often occurs stereoselectively from the R-side
to afford the corresponding R-glycosides as the major products.9
Experimental and theoretical studies showed that the stereo-
selectivity in these SN1-like glycosylation reactions is mainly
due to the kinetic anomeric effect.10,11 As pointed out by Giese,7c
in the radical C-glycosylation reaction shown in Scheme 1, the
R-selectivity can also be a result of the anomeric effect, similar
to that of the SN1-like glycosylations. The anomeric effect
should be influenced by the conformation of the sugar molecule,
because it is the stereoelectronic effect on the anomeric position
due to the nonbonding electrons on the ring oxygen.11b,c
Consequently, we speculated that the anomeric effect might be
employed to effectively control the stereoselectivity in anomeric
radical reactions by using conformationally restricted substrates.
On the basis of this idea, we performed anomeric radical
reactions as well as theoretical calculations with the conforma-
tionally restricted D-xylose derivatives as model substrates,
concentrating especially on the conformation-anomeric effect-
stereoselectivity relationship.12 Thus, we report here the ano-
meric effect-dependent highly R- and â-selective anomeric
radical reactions based on the conformational restriction of the
pyranose ring.
radicals possessed 4C1-chairlike conformation B. They explained
that these pyranosyl radicals were maximally stabilized in
conformations A or B because of the effective interaction
between the radical orbital (SOMO), the σ*-orbital of the
adjacent C2-O2 bond, and the p-orbital of a lone pair of the
ring oxygen in their periplanar arrangement. These results show
that the anomeric effect is likely to effectively stabilize the
anomeric radical intermediate. They also pointed out that only
4
in the case of axial attack on the C1-chairlike conformer B is
the overlap between the lone pair of the ring oxygen and the
unpaired electron of the radical orbital (or newly formed bond)
maintained en route to the transition state to form only
R-products, and that in the case of B2,5-boatlike conformer A,
where the stereoselectivity is somewhat smaller, the decrease
in the selectivity is probably due to the fact that boat conformers
are more flexible than chair conformers.7c
The important question in anomeric radical reactions is what
the transition state is and how the stereoselectivity is controlled.
Renaud reported pregnant results that radical deuteration of
R-phenylseleno cyclic sulfoxide 1 with Bu3SnD gave the
R-deuterated isomer 2 selectively14 (Scheme 2). He finely
explained that the energy of transition state 1′, in which the
radical center is pyramidal, is lowered by the interaction between
the antibonding orbital (σ*q ) of the newly forming carbon-
deuterium (C-D) bond and one of the nonbonded electron pairs
of the sulfur atom.15,16 This theory that the transition state energy
can be lowered by the orbital interaction due to σ*q of the newly
forming bond is originally presented by Cieplack for expounding
the stereoselectivity of nucleophilic additions.15 The explanation
by this kind of orbital interaction (the radical version of the
Cieplack theory) would afford the best account of the transition
state of anomeric radical reactions; the orbital interaction
between the σ*q of the newly forming bond at the anomeric
position and one of the nonbonded electron pairs of the ring
oxygen stabilizes the transition state, which is the kinetic
anomeric effect in the radical reactions.
Results and Discussion
Anomeric Effect in the Anomeric Radical Reactions. Giese
and co-workers have clarified conformations of the anomeric
radical intermediates produced from several O-acetylated pyra-
noses using ESR spectroscopy,13 and their results are sum-
marized in Figure 1. The glucosyl and xylosyl radicals possessed
B2,5-boatlike conformation A, and the mannosyl and lyxosyl
(6) (a) Curran, D. P.; Porter, N. A.; Giese, B. Stereochemistry of Radical
Reactions; VCH: Weinheim, 1996. (b) Descotes, G. J. Carbohydr. Chem.
1988, 7, 1-20.
(7) (a) Giese, B.; Dupes, J. Tetrahedron Lett. 1984, 25, 1349-1352. (b)
Giese, B.; Dupuis, J.; Leising, M.; Nix, M.; Lindner, H. J. Carbohydr. Res.
1987, 171, 329-341. (c) Giese, B. Angew. Chem., Int. Ed. Engl. 1989, 28,
969-980.
(8) â-selective radical C-glycosylation of xylose derivatives has been
reported: see ref 7b.
(9) (a) Lemieux, R.; Hendriks, K. B.; Stick, R. V.; James, K. J. Am.
Chem. Soc. 1975, 97, 4056-4062. (b) Lewis, M. D.; Cha, J. K.; Kishi, Y.
J. Am. Chem. Soc. 1982, 104, 4976-4978.
(10) Experimental studies: (a) Toshima, K.; Tatsuta, K. Chem. ReV 1993,
93, 1503-1531 andreferences therein. (b) Hanessian, S. PreparatiVe
Carbohydrate Chemistry; Marcel Dekker: New York, 1996 and references
therein.
(11) Theoretical studies: (a) Juaristi, E.; Cuevas, G. Tetrahedron 1992,
48, 5019-5087 and references therein. (b) The Anomeric Effect and
Associated Stereoelectronic Effects; ACS Symposium Series 539; Thatcher,
G. R. J., Ed.; American Chemical Society: Washington, DC, 1993. (c)
Thibaudeau, C.; Chattopadhyaya, J. Stereoelectronic Effects in Nucleosides
and Nucleotides and their Structural Implications; Uppsala University
Press: Uppsala, Sweden, 1999.
On the basis of these findings, the mechanism of anomeric
radical reactions is, using the deuteration of the tetra-O-
(13) (a) Dupuis, J.; Giese, B.; Ruegge, D.; Fischer, H.; Korth, H.-G.;
Sustman, R. Angew. Chem., Int. Ed. Engl. 1984, 23, 896-898. (b) Korth,
H.-G.; Sustmann, R.; Dupuis, J.; Giese, B. J. Chem. Soc., Perkin Trans. 2
1986, 1453-1459.
(14) Renaud, P. HelV. Chim. Acta 1991, 74, 1305-1313.
(15) Cieplack pointed out the importance of hyperconjugation in the
transition state of nucleophilic addition reactions to carbonyls, because the
energy of the transition state is lowered by delocalization of electrons from
an antiperiplanar vicinal σ-bond to the antibonding component (σ*q) of the
newly forming bond: (a) Cieplack, A. S. J. Am. Chem. Soc. 1981, 103,
4540-4552. (b) Johnson, C. R.; Tait, B. D.; Cieplack, A. S. J. Am. Chem.
Soc. 1987, 109, 5857-5876 and references therein.
(12) In hexopyranoses, such as glucose or mannose derivatives, the steric
effect due to the hydroxymethyl moiety attached at the 5-position would
affect the stereoselectivity of the anomeric radical reaction, which may
disturb the exact estimation of the anomeric effect on the stereoselectivity.
Therefore, we used the xylose derivatives as the model substrates in this
study, because they lack the carbon substituent at the 5-position.
(16) An explanation of the radical reaction transition state pertaining to
this kind of orbital interaction: Bodpudi, V. R.; le Noble, W. J. J. Org.
Chem. 1991, 56, 2001-2006.

11872 J. Am. Chem. Soc., Vol. 123, No. 48, 2001
Abe et al.
Figure 2. Possible reaction pathways of the anomeric glucosyl radical deuteration.
Scheme 3
Scheme 4
derivative 6 gave the usual R-product 7 highly stereoselectively
(Scheme 3b).7b Giese previously described that the â-isomer
might be produced via the 1,4B-boat or the 1C4-chair conformer
in the xylosyl radical intermediate because of the stereoelectronic
effect.7c On the other hand, we reasoned that the highly
â-selective reaction with 3 as the substrate would occur because
acetylated anomeric glucosyl radical by Bu3SnD7a as an
example, summarized in Figure 2. As reported by Giese,13
radical intermediate A in the B2,5-boat conformation has a planar
(sp2-like) radical center with a radical orbital having high
p-character. Bu3SnD could possibly attack it from either the R-
or the â-side. The important point in our mechanism is that in
the transition state the radical center should have more s-
character to be pyramidal (sp3-like),17 where the pyranose ring
1
of the kinetic anomeric effect via the C4-like transition state,
having the pyramidal radical center. The 1C4-like transition state
can be effectively stabilized by the σ*q-nO interaction, because
the 1C4-conformation might be more stable in xylose derivatives,
as compared with the usual hexopyranoses, because of the lack
of the hydroxymethyl moiety at the 5-position.
4
would adopt a C1 chairlike conformation. Thus, in the course
of forming the C-D bond, the radical center becomes pyramidal
to access to 4C1 chairlike transition state C or D. The transition
state for R-side attack C can be effectively stabilized by the
interaction between the antibonding σ*q of the newly forming
C-D bond and the p-orbital of a nonbonded electron pair (nO)
on the ring oxygen, while little stabilization would occur from
the σ*q-nO interaction in the transition state for â-side attack
D. This significant stabilization, due to the σ*q-nO interaction
in the C transition state, for example, would cause the selective
formation of the R-product rather than the â-product; this is
the kinetic anomeric effect in the anomeric radical reactions.
Interestingly, the anomeric radical addition reaction to CH2d
CHCN with D-xylose derivative 3 gave only â-products 4 and
5 (Scheme 3a),7b while a similar reaction with D-glucose
These experimental results and considerations led us to the
hypothesis that the stereoselectivity of anomeric radical reactions
was significantly affected by the kinetic anomeric effect, which
can be controlled by restricting the conformation of the radical
intermediate; the proper conformational restriction of the
pyranose ring of substrates would make highly R- and â-
stereoselective anomeric radical reactions possible, as shown
in Scheme 4. In the reaction of anomeric radical intermediate
4
E conformationally restricted in the C1-chair form, the con-
formation of which is analogous to transition state C shown in
Figure 2, the anomeric effect works effectively to give the
R-product highly selectively, because the intermediates smoothly
4
progress to the transition state in the sterically preferable C1-
(17) A calculation study showed that methyl radical had a productlike
pyramidal geometry in the transition state of its addition to a CdC double
bond: Dewar, M. J. S.; Olivella, S. J. Am. Chem. Soc. 1978, 100, 5290-
5295.
chairlike conformation. Similarly, when the conformation of
1
the radical intermediate is restricted to the unusual C4-chair
form F, the corresponding â-product should be selectively

SelectiVe Radical C-Glycosylation Reactions
J. Am. Chem. Soc., Vol. 123, No. 48, 2001 11873
Scheme 5
tricyclic sugar moiety of herbicidin B, a nucleoside antibiotic,
via a facially selective reduction of the enone system in a 1C4-
conformational substrate.21 We also developed an efficient
method for preparing 1-R-C-glucosides via a radical cyclization
reaction using 1C4-restricted glucose derivatives.3a-c On the basis
of these findings, we designed 2,3,4-tris-O-triisopropysilyl
(TIPS)-protected D-xylose derivatives 12 that would adopt a 1C4-
conformation because of the steric effect of the bulky silyl
groups.
Figure 3. Conformationally restricted and unrestricted xylose deriva-
tives as the anomeric radical reaction substrates.
The preparation of substrates 8-12 is summarized in Scheme
5. Tetra-O-acetylxylopyranoside 13 was treated with HBr/
AcOH, followed by PhSeH/Et3N to give 1-â-phenylselenide 8
as the sole product. Treatment of 8 with NaOMe in MeOH/
THF afforded triol 14, which was the common intermediate for
the desired substrates. When 14 was heated with 2,2,3,3-tetra-
methoxybutane (TMB) and CH(OMe)3 in MeOH in the presence
of catalytic 10-camphorsulfonic acid (CSA), the desired 2,3-
O-cyclic-diketal 9 (29%) and 3,4-O-cyclic-diketal 10 (65%)
were obtained.18 On the other hand, the 2,4-O-cyclic-phenyl-
boronate 11 was prepared by the procedure reported by Ferrier
and co-workers;19 namely, heating a solution of triol 14 and
phenylboronic acid in benzene under reflux with a Dean-Stark
apparatus quantitatively gave the desired 11.
obtained because of the kinetic anomeric effect. Therefore,
depending on the restricted conformation of the substrates, in
the 4C1- or the 1C4-form, the R- or â-products can be obtained
highly stereoselectively via anomeric radical reactions.
Design and Synthesis of D-Xylose Derivatives Restricted
4
1
to a C1- or a C4-Chair Conformation as Substrates. The
conformations of pyranoses can be restricted by introducing
proper protecting groups on the hydroxyl groups. We designed
the 1-â-phenylseleno-D-xylose derivatives 9-12 restricted in a
4C1- or a 1C4-conformation as the substrates for the experimental
and theoretical calculation studies. The conformationally un-
restricted tri-O-acetate 8 was also used in this study as the
reference substrate. The structures of these substrates are shown
in Figure 3.
3,4-Bis-O-silyl pyranoses have been synthesized via introduc-
tion of the silyl groups on the 3,4-trans-diol of the glycal.3a-c,20,21
We most recently developed an efficient method for directly
introducing the bulky silyl groups on the trans-vicinal-diol of
pyranoses with a TIPSOTf/NaH/THF system.22 Thus, treatment
of 14 with TIPSOTf/NaH in THF at room temperature suc-
cessfully gave 2,3,4-tris-O-TIPS substrate 12 in 78% yield.
The conformations of these substrates were investigated by
1H NMR (Figure 3). The large coupling constants (J > 9.3 Hz)
between the vicinal protons of 9 or 10 showed that these protons
were in an axial orientation, where the pyranose ring had the
4C1-conformation. On the other hand, the small coupling
constants (J < 3.8 Hz) between the vicinal equatorial protons
Substrates 9-11 have a bicyclic structure, because introducing
another ring into a pyranose would effectively restrict the
conformation of the pyranose ring. The conformation of the
pyranose ring of substrates 9 and 10 bearing a 2,3- or a 3,4-
O-cyclic-diketal group would be restricted in the 4C1-form
because of its trans-decalin-type ring system.18 Cyclic-phenyl-
boronate 11 was designed as a conformationally restricted
substrate in the 1C4-form, because pyranoses are known to adopt
1
a C4-conformation when a cyclic-phenylboronate system is
introduced at the 2,4-cis-diol.19
Substrate 12 was designed as a 1C4-restricted substrate without
introducing another ring. It has been recognized that introducing
a significantly bulky protecting group at the 3,4-trans-hydroxyl
groups of pyranoses causes a flip of their conformation leading
1
of 11 or 12 showed that these had the C4-conformation. The
conformationally unrestricted 8 had medium coupling constants
(4.4 < J < 6.7 Hz), compared with the other two types of
conformationally restricted substrates.
1
to a C4-form, in which the bulky substituents are in axial
positions because of mutual steric repulsion. Suzuki and co-
workers first reported this type of conformational feature of
pyranosides and efficiently synthesized aryl C-glycosides using
a 1C4-conformational donor.20 We successfully constructed the
Deuteration of the Conformationally Restricted Xylosyl
Radicals. We investigated the deuteration of the anomeric
radicals produced from unrestricted substrates 8, 4C1-restricted
substrates 9 and 10, and 1C4-restricted substrates 11 and 12 with
(18) (a) Montchamp, J. L.; Tian, F.; Hart, M. E.; Frost, J. W. J. Org.
Chem. 1996, 61, 3897-3899. (b) Hense, A.; Ley, S. V.; Osborn, H. M. I.;
Owen, D. R.; Poisson, J.-F.; Warriner, S. L.; Wesson, K. E. J. Chem. Soc.,
Perkin Trans. 1 1997, 2023-2031.
(19) Ferrier, R. J.; Prasad, D.; Rudowski, A.; Sangster, I. J. Chem. Soc.
1964, 3330-3334.
(20) Hosoya, T.; Ohashi, Y.; Matsumoto, T.; Suzuki, K. Tetrahedron
Lett. 1996, 37, 636-666.
(21) Ichikawa, S.; Shuto, S.; Matsuda, A. J. Am. Chem. Soc. 1999, 121,
10270-10280.
(22) Abe, H.; Shuto, S.; Tamura, S.; Matsuda, A. Tetrahedron Lett. 2001,
42, 6159-6161.

11874 J. Am. Chem. Soc., Vol. 123, No. 48, 2001
Table 1. Deuteration of the Xylosyl Anomeric Radicals
Abe et al.
Table 2. Allylation and Cyanoethylation of the Xylosyl Anomeric
Radicals
entry substrate conformation acceptorb product R/âc (yield, %)
1
2
3
4
5
9
10
11
12
12
4C1
4C1
1C4
1C4
1C4
A
A
A
A
B
17
17
17
17
18
85:15 (73)
91:9 (69)
1:99 (61)
1:99 (26)d
0:100 (66)
a Conditions were noted in Experimental Section. b A, Bu3SnCH2CHd
1
CH2; B, CH2dCHCN. c Determined by H NMR analysis. d Reduced
product 19 was obtained in 61% yield.
increases and inverts the stereoselectivity in the anomeric radical
reaction.
Stereoselective C-Glycosylation of the Xylosyl Radicals.
The conformational restriction strategy, which proved to be
effective in controlling the stereochemistry in the deuterium-
labeling reactions described above, was next applied to the
radical C-glycosylation reactions, that is, the allylation and the
cyanoethylation of the anomeric xylosyl radicals using the 4C1-
restricted substrates, 2,3-O-cyclic-diketal 9 and 3,4-O-cyclic-
diketal 10, and the 1C4-restricted substrates, 2,4-O-cyclic-
phenylboronate 11 and 2,3,4-tris-O-TIPS ether 12. The substrate
was heated with Bu3SnCH2CHdCH2/AIBN23 or CH2dCH2CN/
Bu3SnH/AIBN7b in benzene under reflux and deprotected under
appropriate conditions, and the resulting three hydroxyl groups
of the product were benzoylated with BzCl/pyridine. The
C-glycosylation product, 1-C-allylated 17 or 1-C-cyanoethylated
18, was isolated after purification by silica gel column chro-
matography, and the stereoselectivity was determined by the
1H NMR spectrum. The results are summarized in Table 2.
The radical C-allylation24 with substrate 9 restricted in the
4C1-conformation stereoselectively gave the R-product (R/â )
85:15) in 73% yield (entry 1). A similar result was observed in
the reaction with the other 4C1-restricted substrate 10 under the
same reaction conditions (entry 2, R/â ) 91:9, 69% yield).
When the 2,4-O-cyclic-phenylboronate 11 restricted in the 1C4-
conformation was treated under the same conditions, the
stereoselectivity was completely inverted, and the â-product was
obtained with extremely high stereoselectivity (entry 3, R/â )
a Conditions were noted in the Experimental Section. b For 1H NMR
assignment of the 1-deuterated product 15, the corresponding reduction
product (1,5-anhydro-2,3,4-tri-O-acetyl-D-xylitol, 16) was prepared by
treating 8 with Bu3SnH/AIBN.
2
c Determined by H NMR analysis.
Bu3SnD/AIBN. Such deuterium-labeling experiments would be
useful in estimating the stereoselectivity, leading to a clarifica-
tion of the anomeric effect in anomeric radical reactions.7a The
substrate (0.07 M) was heated with Bu3SnD (2.0 equiv)/AIBN
(0.5 equiv) in benzene under reflux and deprotected under
appropriate conditions, and the resulting three hydroxyl groups
of the product were acetylated with Ac2O/pyridine. Deuterium-
labeled product 15 was purified by silica gel column chroma-
2
tography, and the stereoselectivity was determined by the H
NMR spectrum. The results are summarized in Table 1.
The deuteration of unrestricted 8 as the substrate showed a
moderate R-selectivity (entry 1, R/â ) 65:35). The reaction with
substrate 9, restricted in the 4C1-conformation by a 2,3-O-cyclic-
diketal, showed high R-selectivity (entry 2, R/â ) 97:3). The
1
1:99, yield 61%). The reaction with the C4-restricted 2,3,4-
tris-O-TIPS derivative 12 as the substrate also gave the
â-product highly selectively, although the yield was low (entry
4, R/â ) 1:99, yield 26%), and the directly reduced product 19
was obtained as the major product in 61% yield.25 By using
CH2dCH2CN as the radical acceptor, the yield was significantly
improved without decreasing the stereoselectivity, and the
â-cyanoethylated product 18 was obtained as the sole product
in 66% yield from the substrate 12 (entry 5, R/â ) 0:100).
4
reaction with the other C1-restricted substrate 10 with a 3,4-
O-cyclic-diketal also showed the same high R-selectivity (entry
3, R/â ) 97:3). Thus, the conformational restriction of the
4
substrate in the C1-form resulted in significantly increased
R-selectivity compared with the result of the unrestricted
substrate 8. When 1C4-restricted substrates 11 and 12 were used,
the stereoselectivity was completely reversed. The radical
deuterations with both the 2,4-O-cyclic-phenylboronate 11 and
the 2,3,4-tris-O-TIPS derivative 12, irrespective of the hydroxyl
protecting groups, highly selectively gave the â-deuterated
products (entries 4 and 5, R/â ) 1:99).
(23) Keck, G. E.; Enholm, E. J.; Yates, J. B.; Wiley, M. R. Tetrahedron
1985, 41, 4079-4094.
(24) The radical allylation with unrestricted substrate 21 gave an unknown
product in low yield, which is probably derived from the acyloxy migration
from the 2-position to the 1-position.
(25) Reduced product 19 might be produced via the intramolecular
hydrogen abstraction from the 2-O-TIPS group.
These results clearly demonstrate that the conformational
4
1
restriction of the pyranose ring in the C1-form or C4-form

SelectiVe Radical C-Glycosylation Reactions
J. Am. Chem. Soc., Vol. 123, No. 48, 2001 11875
Table 3. Deuteration and Allylation of 2-Deoxyxylosyl Anomeric
Radicals
entry substrate conformation acceptorb product R/â (yield %)
1
2
3
4
5
20
21
21
22
22
unrestricted
A
A
B
A
B
25c
25c
26
84:16d (97)
99:1d (61)
98:2e (74)
72:28d (71)
46:54e (34)f
4C1
4C1
1C4
1C4
25c
26
a Conditions were noted in the Experimental Section. b A, Bu3SnD;
B, Bu3SnCH2CHdCH2. c For 1H NMR assignment of the 1-deuterated
product 25, the corresponding reduction product (1,5-anhydro-2-deoxy-
3,4-di-O-acetyl-D-xylitol, 27) was prepared by treating 20 with Bu3SnH/
AIBN.
Figure 4. Conformationally restricted and unrestricted 2-deoxyxylose
derivatives as the anomeric radical reaction substrates.
d Determined by 2H NMR analysis. e Determined by 1H NMR analysis.
f Reduced product 28 was obtained in 20% yield.
Scheme 6
The large and small coupling constants in the 1H NMR analyses
of 21 and 22, respectively (Figure 4), suggested that these
4
1
substrates preferred the typical C1- and C4-conformations.
The radical reactions of the 2-deoxy substrates with Bu3SnD/
AIBN or Bu3SnCH2CHdCH2/AIBN were carried out by the
same procedures described above, and the results are sum-
marized in Table 3. While the deuteration with unrestricted 20
showed moderate R-selectivity (entry 1, R/â ) 84:16), a highly
4
R-stereoselective deuteration occurred with C1-restricted sub-
As described, very effective stereocontrol by this conforma-
tional restriction method was observed in the radical C-
glycosylation reactions (Table 2), analogous to the deuteration
(Table 1). This is the first radical C-glycosylation reaction
that highly stereoselectively provides both the R- and â-C-
glycosides.
strate 21 (entry 2, R/â ) 99:1), similar to the deuteration with
4C1-restricted substrates 9 and 10 having a 2-hydroxyl moiety.
4
The radical allylation with C1-restricted 2-deoxy substrate 21
showed higher R-selectivity (entry 3, R/â ) 98:2), compared
to that of 9 and 10 (Table 2, entries 1, 2), to give C-allylated
product 26 in 74% yield, as expected. However, unexpectedly,
the radical reactions of 1C4-restricted 2-deoxy substrate 22 were
not stereoselective; the deuteration was moderately R-selective
(entry 4, R/â ) 72:28), and the allylation gave nonstereoselec-
tively C-allylated product 26 (R/â ) 46:54) in low yield (34%)
along with reduced product 28 in 20% yield.
Deuteration and C-Glycosylation of the 2-Deoxyxylosyl
Radicals. In the anomeric radical C-allylation, the stereoselec-
tivity with 4C1-restricted compounds 9 and 10 as the substrates
(Table 2, entries 1, 2) was slightly lower than in those with
1C4-restricted substrates 11 and 12 (Table 2, entries 3, 4). We
considered whether the stereoselectivity in the radical allylation
could be influenced by the 1,2-steric repulsion derived from
the 2-hydroxyl moiety on the R-face. To clarify the effect of
the 2-hydroxyl group on the stereoselectivity of the reaction,
we designed 2-deoxyxylose derivatives, that is, the unrestricted
Analysis of Conformation and Orbitals of the Anomeric
Radical Intermediates by Theoretical Calculations. Confor-
mational analysis of the anomeric radical intermediates derived
from the substrates would be helpful in understanding the
reaction mechanism. Orbital analysis of the radical intermediates
would also provide insight into the orbital interactions of the
transition states. Conformational analysis using 1H NMR spectra
4
substrate 20, the substrate 21 restricted in a C1-conformation
with a 3,4-O-cyclic-diketal, and the substrate 22 restricted in a
1C4-conformation with bulky TIPS groups (Figure 4), and
examined the radical deuteration and allylation with them.
The preparation of 2-deoxy-D-xylose derivatives 20-22 is
summarized in Scheme 6. Tri-O-acetyl-2-deoxy-D-xylopyranose
23, derived from tetra-O-acetyl-D-xylose,26 was treated with
PhSeH/BF3‚OEt2 to give 1-phenylseleno derivative 20 as an
anomeric mixture (R/â ) 1:1). After removal of the acetyl
groups of 20, the resulting diol 24 was further treated with TMB/
4
(Figure 3) showed that substrates 9-12 have the typical C1-
or 1C4-chair conformations. However, the conformations of the
radical intermediates produced from these substrates are unclear.
Thus, we performed conformational and orbital analyses by ab
initio calculations27 using the Gaussian 98 program.28 Final
optimization was performed by UHF/3-21G*. Single point
energies and NBO (natural bond orbital) analysis were calculated
by UB3LYP/6-31G*. The calculation results are summarized
in Figure 5 and Tables 4 and 5.
4
CSA/CH(OMe)3 or with TIPSOTf/2,6-lutidine to afford C1-
restricted substrate 21 or 1C4-restricted substrate 22, respectively.
(27) Theoretical calculations of anomeric radicals including the confor-
mational analysis of pyranosyl radicals: Rychnovsky, S. D.; Powers, J. P.;
LePage, T. J. J. Am. Chem. Soc. 1992, 114, 8375-8384.
(26) Giese, B.; Gro¨niger, K. S.; Witzel, T.; Korth, H.-G.; Sustmann, R.
Angew. Chem., Int. Ed. Engl. 1987, 26, 233-234.

11876 J. Am. Chem. Soc., Vol. 123, No. 48, 2001
Abe et al.
Table 4. Hybridized Orbital Analysis of the Anomeric Carbon (C1) and the Ring Oxygen (O5) in the Xylosyl Radical Conformersa
8′′a
8′′b
8′′c
atom
O5
R spin
â spin
R spin
â spin
R spin
â spin
LP(1) sp1.6
LP(2) sp20.8
LP(1) sp8.3
LP(1) sp1.7
LP(2) sp31.3
LP*(1) sp65.1
LP(1) sp1.6
LP(2) sp36.5
LP(1) sp57.5
LP(1) sp1.5
LP(2) sp82.9
LP*(1) p
LP(1) sp1.82
LP(2) sp17.3
LP(1) sp8.7
LP(1) sp1.7
LP(2) sp13.3
LP*(1) p
C1
10′a
10′b
11′a
atom
O5
R spin
â spin
R spin
â spin
LP(1) sp1.7
BD(C1-O5) p
BD(C1-O5) p
R spin
â spin
LP(1) sp1.7
LP(2) sp22.2
LP(1) sp8.6
LP(1) sp1.7
LP(1) sp2.5
LP(2) sp7.9
LP(1) sp14.6
LP(1) sp1.56
LP(2) sp47.3
LP(1) sp8.3
LP(1) sp1.7
LP(2) sp10.9
LP*(1) p
BD(C1-O5) sp31.3
BD(C1-O5) sp65.1
C1
a Hybridized orbitals based on the NBO theory were calculated by UB3LYP/6-31G*.
Table 5. Stabilization Energies by the Hyperconjugation
Interactions between LP(2) O5 and LP*(1) C1 and between LP (1)
C1 and BD* (C2-O2) in the Xylosyl Radical Conformers
energya (kcal/mol)
donor
LP(2) O5 LP*(1) C1
LP(2) C1 BD*(C2-O2) 1.87 14.98 9.71 1.52
acceptor
8′′a 8′′b 8′′c
32.11 39.01 39.60 π-bond π-bond 38.69
9.58 9.15
10′a
10′b 11′a
a Second-order perturbation energies based on the NBO theory were
calculated by UB3LYP/6-31G*.
the stable conformer 11′a in a typical 1C4-conformation. Instead
of the radical intermediate produced from the unrestricted tri-
O-acetyl substrate 8 used in the above experiments, the
corresponding 3,4-di-O-methylxylosyl radical 8′′ was used in
this calculation as an unrestricted model intermediate to simplify
the calculations.29 The conformationally unrestricted xylosyl
radical would exist in a great number of conformations because
of the extremely high conformational flexibility. Therefore, we
selected the three typical conformers that are likely to be stable,
4
that is, 8′′a in a C1-form, 8′′b in a B2,5-form, and 8′′c in a
1C4-form, for the energy calculations, and compared their
energies.30 The structures and calculated energies of 8′′a-c are
also shown in Figure 5. The relative energies of the unrestricted
three conformers 8′′a, 8′′b, and 8′′c were 2.70, 0.94, and 0.00
kcal/mol, respectively. The energy difference between 8′′b and
8′′c might not be so important, especially under the reaction
conditions at the high temperature (80 °C).
Figure 5. Conformers in the xylosyl radical intermediate and their
calculated energies.
We next analyzed the hybridized orbitals and the orbital
interactions leading to stabilization of the above conformers
8′′a-c, 10′a,b, and 11′a on the basis of NBO theory.31 The
hybridized orbitals of these conformers were calculated for the
two lone pairs of the O5-atom and the radical orbital of the C1-
atom, and the results are shown in Table 4. In all the conformers,
one lone pair (LP) of the O5-atom was in an equatorial sp2-like
LP(1) orbital and the other in an axial p-like LP(2) orbital.
Furthermore, in all the conformers, a radical electron at the C1-
atom existed, not in the â-spin-orbital, but in the R-spin-orbital;
the R-spin had a high occupation (>0.91), whereas the â-spin-
orbital was empty.32 In B2,5-boat conformer 8′′b and 4,5H-half
boat conformer 10′b, the radical orbital of the R-spin had a high
We first calculated the energies of the conformers of the
radical intermediates produced from 4C1-restricted substrate 10
and 1C4-restricted substrate 11. The calculated stable structures
and their energies are shown in Figure 5. For the radical
intermediate derived from 10, which was protected with a 3,4-
O-cyclic-diketal, the two stable conformers 10′a in a 4C1-
conformation with the equatorial C2-O2 bond and 10′b in a
4,5H-conformation with the pseudoaxial-like C2-O2 bond were
4
obtained, and the C1-conformer 10′a was 3.55 kcal/mol more
stable than the 4,5H-conformation 10′b. Calculation of the radical
intermediate of the 2,4-O-cyclic-phenylboronate 11 gave only
(28) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.;
Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A.
D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi,
M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.;
Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick,
D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.;
Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi,
I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.;
Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M.
W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon,
M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.6; Gaussian,
Inc.: Pittsburgh, PA, 1998.
(29) Although we first tried to calculate the conformer energies using
the radical intermediate produced from tri-O-acetyl substrate 8, it gave a
number of local minima because of the bond rotational barrier of the three
O-acetyl moieties, which were energetically similar and had a similar
pyranose conformation. A search of all of these minimum conformers was
likely to be troublesome, and therefore, we used simplified model
intermediate 8′′ in the calculations.
4
1
(30) The C1- and the C4-conformers are the typical conformations in
pyranoses. The B2,5-conformer was found to be a stable conformer in the
tri-O-acetylxylosyl radical intermediate by ESR analysis: see ref 13.
(31) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. ReV. 1988, 88,
899-926.

SelectiVe Radical C-Glycosylation Reactions
J. Am. Chem. Soc., Vol. 123, No. 48, 2001 11877
π-character (p>94%), and the radical center at the C1-atom was
planar. On the other hand, in the chair conformers, 8′′a, 8′′c,
10′a, and 11′a, the radical orbital of the R-spin had relatively
high s-character, and the radical center at the C1-atom was more
pyramidal (sp3-like), compared to boatlike conformers 8′′b and
10′b.
Stabilization energies of the orbital interactions between the
LP(2) O5 and the LP*(1) C1 or between the LP(1) C1 and the
BD*(C2-O2) of conformers 8′′a-c, 10′a,b, and 11′a were
calculated on the basis of NBO theory,31,33 and the results are
summarized in Table 5. All of the conformers were significantly
stabilized by the orbital interaction between the LP(2) O5 and
the LP*(1) C1; the stabilized energies in the conformers 8′′a-c
and 11′a were 32.11-39.6 kcal/mol, and the interaction in the
conformers 10′a and 10′b had π-bonding character.34 The orbital
interaction between the LP(1) C1 and the BD*(C2-O2) signifi-
cantly stabilized the conformers 8′′b, 8′′c, 10′b, and 11′a with
an axial-like C2-O2 bond (9.15-14.98 kcal/mol), while only
minimal stabilization was observed in conformers 8′′a and 10′a
with an equatorial-like C2-O2 bond (1.87 and 1.52 kcal/mol,
respectively).
Figure 6. Transition state model of the reaction of the 4C1-conforma-
tionally restricted xylosyl radicals for an understanding of the stereo-
selectivity.
The calculations on unrestricted intermediate 8′′ suggested
that the intermediate would exist predominantly in B2,5-boat
conformer 8′′b and 1C4-chair conformer 8′′c, in which the orbital
between the LP(1) C1 and the BD*(C2-O2) interacts effectively.
stereoselectively, while the 1C4-conformationally restricted
substrates give the â-products. To understand the reaction
mechanism, we should examine the reaction transition states
from the standpoint of steric hindrance and stereoelectronic
effect, on the basis of the results of the experimental and
theoretical calculation studies.
4
The C1-chair conformer 8′′a would be relatively unimportant
in the equilibrium. Giese and co-workers investigated the
conformation of the tri-O-acetylxylosyl radical A (shown in
Figure 1) derived from unrestricted substrate 8 by an ESR
study.13 They reported that the B2,5-boat conformer was the most
preferred in the xylosyl radical, although it could change into
In the reactions of anomeric radicals, a radical acceptor can
attack from both axial and equatorial directions.35 As an
example, in the deuteration with 4C1-restricted substrate 10 and
1C4-restricted substrate 11, the highly stereoselective attack of
Bu3SnD proceeded from the R- or â-axial direction, respectively.
We will first consider the R-selective reaction with radical
intermediate 10′a in the 4C1-conformation, as shown in Figure
6a. Taking into account only steric hindrance in the attack on
1
other conformers, such as the 1,4B-boat or the C4-chair con-
former, because of the high conformational flexibility. Therefore,
our energy calculation results on the unrestricted radical
intermediate were in accord with those of the ESR study.13
The calculation results indicated that the radical intermediate
derived from 10 with a 3,4-O-cyclic-diketal structure would
4
the radical intermediate 10′a in the C1-conformation (Figure
6b, steric hindrance), the 1,2-steric repulsion in the R-axial attack
would be more significant than that in the â-equatorial attack,
because the axial attack of Bu3SnD would proceed through the
pseudoaxial direction,35 where the newly forming C-D bond
and the C2-O2 bond are nearly eclipsed. Moreover, the axial
attack on 10′a is accompanied by rather significant 1,3- and
1,5-diaxial repulsion. Hence, the equatorial attack should be
preferred over the axial attack from the viewpoint of steric
hindrance. On the other hand, considering the stereoelectronic
effect in the radical reaction of 10′a, the R-axial attack would
be preferred to the equatorial attack. In the reaction of the radical
intermediate in the 4C1-conformation, the transition state is likely
4
preferentially exist in C1-chair conformer 10′a and that the
intermediate derived from 11 protected with a cyclic-phenyl-
boronate would also exist in 1C4-chair conformer 11′a as
virtually the sole conformer. Both conformers, 10′a and 11′a,
restricted in the chair conformations, possessed a lone pair with
high p-character in the axial direction and also the pyramidal
(sp3-like) radical center at the C1-position. Thus, the conforma-
tion of the anomeric radical intermediates can be manipulated
on the basis of the conformational restriction of the substrates,
as we hypothesized.
Discussion. We have shown, using the D-xylose and the 2-
deoxy-D-xylose derivatives as substrates, that the stereoselec-
tivity of deuteration and C-glycosylation reactions of pyranosyl
radicals can be effectively controlled by restricting the confor-
mation of the pyranose ring of the substrates. As expected, the
stereoselectivity is completely inverted by flipping the substrate
4
to assume the C1-like conformation because of the confor-
mational restriction, and the axial attack from the R-side could
be significantly stabilized by the interaction between the anti-
bonding σ*q of the newly forming anomeric C-D bond and
the axial-directed nonbonding electrons of the ring oxygen which
have high p-character (Figure 6c, anomeric effect). The stereo-
electronic stabilization would effectively work in the transition
4
1
4
conformations from the C1-form into the C4-form; the C1-
conformationally restricted substrates yield the R-products highly
4
state, because of the restricted C1-like conformation, to yield
(32) The R-spin occupation of radical electron at the C1-position: 8′′a
(0.96246), 8′′b (0.91382), 8′′c (0.94286), 12′a (0.96417), 12′b (0.94635),
13′a (0.93575).
(33) A similar analysis using the NBO method: Senyurt, N.; Aviyente,
V. J. Chem. Soc., Perkin Trans. 2 1998, 1463-1470.
(34) π-Bonding character means double bond. Using the default setting
of the NBO program, when the single electron occupation between two
atoms is higher than 0.9 e (occupancy threshold), it is judged on bonding.
In this case, the π-bond in 10′b and 10′b would be energetically higher
[p-BD(C1-O5) occupation: 10′a (0.98381), 10′b (0.98337)] than the
nonbonded interaction in 8′′a-c, and 11′a
the R-product highly selectively; this is the kinetic anomeric
effect in anomeric radical reactions.
(35) Theoretical calculations by Giese and Houk showed that attack of
radical acceptors on the cyclohexyl radical occurred through the two
pathways of axial and equatorial directions. They concluded that the axial
attack occurred from the direction slightly diverted from the vertical, in
other words, a pseudoaxial direction: Damm, W.; Giese, B.; Hartung, J.;
Hasskerl, T.; Houk, K. N.; Hu¨ter, O.; Zipse, H. J. Am. Chem. Soc. 1992,
114, 4067-4079.

11878 J. Am. Chem. Soc., Vol. 123, No. 48, 2001
Abe et al.
Figure 8. R-Attack to the 2-deoxyxylosyl radical in a 4C1-conformation
controlled by the anomeric effect.
the 2-hydroxyl moiety in the axial R-attack on the anomeric
radical intermediate produced from 10 might decrease the
stereoselectivity, as shown in Figure 6b, and the highly
stereoselective attack from the R-side by Bu3SnCH2CHdCH2
on the 2-deoxyxylosyl radical intermediate 21′ occurred, as
shown in Figure 8, probably because of a decrease in 1,2-steric
repulsion.
In the radical C-glycosylation with 2,3,4-O-silylated substrate
12, the yield was low (26%, R/â ) 1:99) when Bu3SnCH2-
CHdCH2 was used as the acceptor (Table 2, entry 4), while a
similar reaction with CH2dCHCN as the acceptor gave the
corresponding â-C-glycoside in 66% yield as the sole product
(Table 2, entry 5). This may be due to the difference in reactivity
between the radical acceptors CH2dCHCN and Bu3SnCH2CHd
CH2. According to frontier molecular orbital (FMO) theory, the
π* of a radical acceptor at the lower energy level should
effectively interact with a singly occupied molecular orbital
(SOMO) at a higher level.37 Our calculations by RHF/3-21G*
showed that the π* of CH2dCHCN was located at an energy
level lower than that of Bu3SnCH2CHdCH2.38 The SOMO of
anomeric radicals is known to be located at a raised energy
level,6 probably because of the anomeric effect. Therefore, the
relatively high π* energy level of Bu3SnCH2CHdCH2 would
not interact as effectively with the SOMO of the anomeric
radical produced from 12, resulting in the low yield of the
C-glycoside 17. However, a greater stabilization of the transition
state would occur in the reaction with CH2dCHCN by the
interaction between the higher level SOMO of the anomeric
radical and the π* of CH2dCHCN, so that the radical addition
reaction might proceed smoothly. Because the anomeric effect
raises the SOMO energy of the anomeric radical, the transition
state may be significantly stabilized by this kind of interaction
between the SOMO and the π* of the acceptor. Accordingly,
the kinetic anomeric effect on these addition reactions of
anomeric radicals may depend on the electronic property of the
radical acceptor used.
Figure 7. Transition state model of the reaction of the 1C4-conforma-
tionally restricted xylosyl radicals for an understanding of the stereo-
selectivity.
We will next consider the â-selective reaction with the radical
intermediate 11′a in a 1C4-conformation (Figure 7). With respect
to steric hindrance in the attack on the radical 11′a, the â-axial
attack of Bu3SnD proceeding through the pseudoaxial direction,
analogous to that on 10′a described above, encounters 1,3- and
1,5-diaxial repulsion (Figure 7b, steric hindrance). However,
1
in the reaction of 11′a in the C4-conformation, the 1,2-steric
repulsion derived from the 2-hydroxyl moiety would prevent
the R-equatorial attack, at least to some extent. Therefore, from
the standpoint of steric hindrance, we cannot judge whether the
equatorial or the axial attack would be favored. On the other
hand, the axial attack by Bu3SnD from the â-side of 11′a would
be preferred because of the kinetic anomeric effect contributing
to stabilize the transition state having the 1C4-like conformation
(Figure 7c, anomeric effect).
We, therefore, concluded that the dominant factor controlling
the stereoselectivity of these radical reactions is not steric
hindrance but the kinetic anomeric effect. Because of the
conformational restriction, the transition state conformation
would be analogous to that of the intermediate radical, that is,
the 4C1- or 1C4-forms, where the anomeric effect is most
effective in causing the highly stereoselective axial attack on
the radical intermediate. The transition state model of attack
on the pyranosyl radical in the 4C1- or 1C4-conformations,
illustrated by Newman projection formula in Figures 6 and 7,
nicely explains the stereochemical results.
Although radical deuterations with the 4C1-restricted xylosyl
and 2-deoxyxylosyl substrates (10 and 21, respectively), having
a 3,4-O-cyclic-diketal structure, likewise gave the corresponding
R-deuterated products highly selectively (R/â ) 97:3 or 99:1),
different stereoselectivity was observed in the radical allylation
with the two 4C1-restricted substrates; these anomeric C-allyla-
tions give 17 (R/â ) 91:9) from 10 and 26 (R/â ) 98:2) from
21, respectively. Steric hindrance would influence the reaction
course more in the allylation with Bu3SnCH2CHdCH2 than in
the deuteration with Bu3SnD.23,36 Therefore, the 1,2-steric
repulsion between the radical acceptor Bu3SnCH2CHdCH2 and
Unexpectedly, the radical deuteration and the allylation with
2-deoxyxylosyl substrate 22 showed no â-selectivity (Table 3,
1
entries 4, 5). Although H NMR data (Figure 4) indicated that
substrate 22 was likely to prefer the 1C4-conformation, the
reaction results suggested that the 1C4-conformation in the
radical intermediate of 22 was not sufficiently stable, especially
at the higher reaction temperature (80 °C). We examined the
stability of the 1C4-conformations in the intermediates by MM3
calculation39 with the corresponding 1-deoxy derivatives, that
is, 3,4-bis-O-TIPS-1,2-dideoxyxylose (30) and 2,3,4-tris-O-
TIPS-1-deoxyxylose (29), as model compounds of the radical
(37) Fleming, I. Frontier Orbitals and Organic Chemical Reactions;
Wiley: London, 1976.
(38) Calculated energies of the π* of CH2dCHCN and Bu3SnCH2CHd
CH2 by RHF/3-21G* were +0.18142 and +0.23263 eV, respectively.
(39) Mohanmadi, F.; Richard, N. G. J.; Guida, W. C.; Liskamp, R.;
Caufield, C.; Chang, G.; Hendrikson, T.; Still, W. C. J. Comput. Chem.
1990, 11, 440-467.
(36) In allylation of pyranosyl radicals using Bu3SnCH2CHdCH2 as the
acceptor, the stereoselectivity was significantly influenced by the steric
hindrance of the 2-substituent of the substrates: Roe, B. A.; Boojamara,
C. G.; Griggs, J. L.; Bertozzi, C. R. J. Org. Chem. 1996, 61, 6442-6445.

SelectiVe Radical C-Glycosylation Reactions
J. Am. Chem. Soc., Vol. 123, No. 48, 2001 11879
was done on Merck silica gel 7734 or 9385. Reactions were carried
out under an argon atmosphere.
Phenyl 1-Seleno-2,3,4-tri-O-acetyl-â-D-xylopyranoside (8). A
mixture of 13 (15.7 g, 49 mmol) and HBr (33% in AcOH, 10 mL) in
CH2Cl2 (30 mL) was stirred at room temperature for 2 h. The resulting
mixture was partitioned between AcOEt and H2O, and the organic layer
was washed with H2O, aqueous saturated NaHCO3, and brine, dried
(Na2SO4), evaporated, and dried in vacuo at room temperature for 1 h.
A solution of the residue, Et3N (9.6 mL, 69 mmol), and PhSeH (6.2
mL, 58 mmol) in CH3CN (150 mL) was stirred at room temperature
for 1 h. The resulting mixture was partitioned between AcOEt and
aqueous saturated NaHCO3, and the organic layer was washed with
H2O and brine, dried (Na2SO4), and evaporated. The resulting residue
was purified by column chromatography (SiO2, hexane/AcOEt, 2:1)
to give 8 (14.7 g, 72% as an oil): 1H NMR (CDCl3, 270 MHz) δ
7.64-7.20 (m, 5 H), 5.18 (d, 1 H, J ) 6.7), 5.12 (dd, 1 H, J ) 6.7,
6.7), 5.03 (dd, 1 H, J ) 6.7, 6.7), 4.87 (ddd, 1 H, J ) 4.4, 6.7, 6.7),
4.36 (dd, 1 H, J ) 4.4, 12.3), 3.54 (dd, 1 H, J ) 6.7, 12.3), 2.08 (m,
9 H); FAB-HRMS calcd for C17H20O7SeNa 439.0272 (MNa+), found
439.0296. Anal. Calcd for C17H20O7Se: C, 49.17; H, 4.85. Found: C,
49.32; H, 4.88.
Phenyl 1-Seleno-â-D-xylopyranoside (14). A mixture of 8 (11.0 g,
26.5 mmol) and NaOMe (28% in MeOH, 500 µL) in THF/MeOH (20
mL/30 mL) was stirred at room temperature for 2 h and then neutralized
with Diaion PK 212 resin (H+ form). The resin was filtered off, and
the filtrate was evaporated to give 14 (7.66 g, 99% as a yellow solid):
1H NMR (CD3OD, 400 MHz) δ 7.55-7.17 (m, 5 H), 4.83 (d, 1 H, J
) 7.9), 3.97 (dd, 1 H, J ) 4.7, 11.4), 3.43-3.24 (m, 3 H), 3.20 (dd,
1 H, J ) 8.5, 11.4); FAB-HRMS calcd for C11H14O4SeNa 312.9955
(MNa+), found 313.0005. Anal. Calcd for C11H14O4Se: C, 45.69; H,
4.88. Found: C, 45.65; H, 4.96.
1
4
Figure 9. Relative energies of the C4-conformers based on the C1-
conformer calculated by the MM3 force field.
intermediates derived from 22 and 12, to estimate the steric
effect of the O-silyl groups on the pyranose conformation. As
shown in Figure 9, the calculated energies suggested that in
3,4-bis-O-silylated 30, the 1C4-conformer would not be so
4
stable (0.06 kcal/mol) compared to the flipped C1-conformer,
whereas the 1C4-conformer was 2.21 kcal/mol more stable than
4
the C1-conformer in 2,3,4-tris-O-silylated 29. Therefore, the
nonstereoselective experimental results with 2-deoxyxylose
derivative 22 as the substrate may be attributed to insufficient
conformational restriction to the 1C4-form under the high-
temperature reaction conditions.
As described, we have proved that the conformational
restriction strategy made possible the control of the stereo-
selectivity of the anomeric radical reactions. In the O- or
C-glycosylations via SN1 displacement reactions on the oxo-
carbenium intermediate, highly stereoselective reactions may
also be realized, on the basis of the conformation restriction
strategy controlling the anomeric effect.40
Conclusion
Phenyl 2,3-O-[(2S,3S)-2,3-Dimethoxybutan-2,3-diyl]-1-seleno-
â-D-xylopyranoside (9) and Phenyl 3,4-O-[(2S,3S)-2,3-Dimethoxy-
butan-2,3-diyl]-1-seleno-â-D-xylopyranoside (10). A mixture of 14
(150 mg, 519 µmol), TMB (82 µL, 1.04 mmol), CH(OMe)3 (454 µL,
4.15 mmol), and CSA (6 mg, 25 µmol) in MeOH (2 mL) was heated
under reflux for 1.5 h. The mixture was partitioned between AcOEt
and aqueous saturated NaHCO3, and the organic layer was washed
with H2O and brine, dried (Na2SO4), and evaporated. The resulting
residue was purified by column chromatography (SiO2, hexane/AcOEt,
3.5:1-2:1) to give 9 (61 mg, 29% as a white solid) and 10 (136 mg,
Introduction of protecting groups, such as 2,3- or 3,4-O-
cyclic-diketal, 2,4-O-cyclic-phenylboronate, and 2,3,4-tris-O-
TIPS, on the hydroxyl groups effectively restricted the pyranose
conformation in a 4C1- or a 1C4-form. Radical deuterations with
1
4C1- or C4-restricted substrates highly stereoselectively gave
the corresponding R- or â-products, respectively. Selectivity was
increased by the conformational restriction and completely
1
inverted by flipping the substrate conformation from the C4-
form into the 4C1-form, because of the kinetic anomeric effect.
Analogously, the radical C-glycosylations, that is, the allylation
and the cyanoethylation, showed high stereoselectivity by
the conformational restriction method. Ab initio calculations
suggested that the radical intermediates produced from these
1
65% as a white solid). 9: H NMR (CDCl3, 500 MHz) δ 7.62-7.27
(m, 5 H, Ar), 4.95 (d, 1 H, H-1, J ) 9.3), 4.13 (dd, 1 H, H-5a, J ) 5.9,
11.4), 3.89 (m, 1 H, H-4), 3.65 (m, 2 H, H-2, H-3), 3.29 (m, 4 H,
H-5b, OMe), 3.21 (s, 3 H, OMe), 2.27 (d, 1 H, 4-OH, J ) 2.9), 1.34
(m, 6 H, Me × 2); FAB-HRMS calcd for C17H24O6SeNa 427.0636
(MNa+), found 427.0631. Anal. Calcd for C17H24O6Se: C, 50.62; H,
4
1
substrates possessed typical C1- or C4-conformations which
were similar to those of the substrates and that the anomeric
effect was operative in these conformations. These results
suggest that the highly R- and â-selective reactions occurred
because of the anomeric effect, which was manipulated by the
conformational restriction of the substrates, as expected. This
method is the first radical C-glycosylation reaction that highly
stereoselectively provides both the R- and â-C-glycosides and
would be applicable to stereoselective construction of various
biologically important C-glycosides.
1
6.00. Found: C, 50.98; H, 6.12. 10: H NMR (CDCl3, 500 MHz) δ
7.64-7.30 (m, 5 H, Ar), 4.69 (d, 1 H, H-1, J ) 9.4), 3.98 (dd, 1 H,
H-5a, J ) 4.9, 10.7), 3.71 (ddd, 1 H, H-4, J ) 4.9, 10.0, 10.0), 3.65
(dd, 1 H, H-3, J ) 9.4, 9.4), 3.49 (ddd, 1 H, H-2, J ) 1.9, 9.4, 9.4),
3.44 (dd, 1 H, H-5b, J ) 10.0, 10.0), 3.30 (s, 3 H, OMe), 3.24 (s, 3 H,
OMe), 2.41 (d, 1 H, 2-OH, J ) 1.9), 1.32 (s, 3 H, Me), 1.28 (s, 3 H,
Me); FAB-HRMS calcd for C17H24O6SeNa 427.0636 (MNa+), found
427.0609. Anal. Calcd for C17H24O6Se: C, 50.62; H, 6.00. Found: C,
50.77; H, 6.04.
Phenyl 1-Seleno-â-D-xylopyranoside 2,4-O-cyclic-phenylboronate
(11). A suspension of 14 (375 mg, 1.30 mmol) and phenylboronic acid
(174 mg, 1.42 mmol) in benzene (30 mL) was heated under reflux
using a Dean-Stark apparatus for 1 h. The resulting clear solution
was evaporated, and the residual white powder was treated with hexane
and filtrated to give 11 (485 mg, 100% as a white powder): 1H NMR
(DMSO-d6, 400 MHz) δ 7.94-7.24 (m, 10 H, Ar), 6.36 (d, 1 H, 3-OH,
J ) 3.8), 5.56 (s, 1 H, H-1), 4.44 (d, 1 H, H-5a, J ) 12.3), 4.42 (dd,
1 H, H-3, J ) 1.9, 3.8), 4.08 (br s, 1 H, H-2), 4.05 (d, 1 H, H-4, J )
3.5), 3.68 (d, 1 H, H-5b, J ) 12.3); FAB-HRMS calcd for C17H17BO4-
Se 376.0385 (M+), found 376.0361. Anal. Calcd for C17H17BO4Se: C,
54.44; H, 4.57. Found: C, 54.43; H, 4.41.
Experimental Section
General Methods. 1H and 2H spectra were recorded at 270 and 500
MHz (1H) and at 400 MHz (2H). Chemical shifts are reported in ppm
downfield from TMS (1H) and CDCl3 (2H), and J values are given in
hertz. The 1H NMR assignments were in agreement with COSY spectra.
Mass spectra were obtained by fast atom bombardment (FAB) and
electron spray ionization (ESI) methods. Thin-layer chromatography
was done on Merck coated plate 60F254. Silica gel chromatography
(40) Examples of highly R-stereoselective O-mannosylation using con-
formationally restricted substrates with a 4,6-O-benzylidene group have been
reported: (a) Crich, D.; Sun, S. J. Am. Chem. Soc. 1997, 119, 11217-
11223. (b) Crich, D.; Sun, S. J. Am. Chem. Soc. 1998, 120, 435-436. (c)
Crich, D.; Cai, W.; Dai, Z. J. Org. Chem. 2000, 65, 1291-1297.
Phenyl 1-Seleno-2,3,4-tris-O-triisopropylsilyl-â-D-xylopyranoside
(12). To a suspension of 14 (300 mg, 1.04 mmol) and NaH (60% in
mineral oil, 326 mg, 8.15 mmol) in THF (5 mL) was slowly added

11880 J. Am. Chem. Soc., Vol. 123, No. 48, 2001
Abe et al.
TIPSOTf (1.12 mL, 4.17 mmol) over 20 min, and the resulting mix-
ture was stirred at room temperature for 2 h. The reaction mixture
was neutralized with AcOH and partitioned between AcOEt and
H2O, and the organic layer was washed with aqueous saturated Na-
HCO3 and brine, dried (Na2SO4), and evaporated. The resulting residue
was purified by column chromatography (SiO2, hexane/benzene, 20:1)
to give 12 (615 mg, 78% as an oil): 1H NMR (CDCl3, 400 MHz) δ
7.58-7.24 (m, 5 H, Ar), 5.66 (s, 1 H, H-1), 4.58 (d, 1 H, H-5a, J )
12.3), 4.20 (s, 1 H, H-3), 3.99 (s, 1 H, H-2), 3.75 (s, 1 H, H-4),
3.19 (d, 1 H, H-5b, J ) 12.3), 1.08 (m, 63 H, TIPS × 3); ESI-
HRMS calcd for C38H74O4SeSi3Na 781.3957 (MNa+), found 781.3984.
Anal. Calcd for C38H74O4SeSi3: C, 60.20; H, 9.84. Found: C, 59.92;
H, 9.59.
(â-anomer), 2.79 (R-anomer); FAB-HRMS calcd for C11H16DO7
262.1037 (MH+), found 262.1041.
1,5-Anhydro-2,3,4-tri-O-acetyl-D-xylitol (16). Compound 16 (66
mg, 91% as an oil) was prepared from 8 (116 mg, 279 µmol) by the
procedure for 15 (Table 1, entry 1) with Bu3SnH instead of Bu3SnD:
1H NMR (CDCl3, 400 MHz) δ 5.16 (dd, 1 H, H-3, J ) 8.5, 8.5), 4.91
(ddd, 2 H, H-2, H-4, J ) 5.0, 8.5, 8.5), 4.03 (dd, 2 H, H-1â, H-5a, J
) 5.0, 11.4), 3.34 (dd, 2 H, H-1R, H-5b, J ) 8.8, 11.4), 2.06 (m, 9 H,
Ac × 3); FAB-HRMS calcd for C11H17O7 261.0974 (MH+), found
261.0929.
General Procedure for the Radical Allylation with Allyltributyl-
tin. To a solution of a substrate (301 µmol, 0.10 M) and allyltributyltin
(467 µL, 1.51 mmol) in benzene (3 mL) was added AIBN (5 mg, 30
µmol) at 80 °C, and the mixture was heated at the same temperature.
After disappearance of the starting material on TLC, the mixture was
evaporated, and the residue was treated by the procedure as described
General Procedure for the Radical Deuteration with Bu3SnD.
To a solution of a substrate (140 µmol, 0.07 M) and Bu3SnD (113 µL,
418 µmol) in benzene (2 mL) was added AIBN (5 mg, 30 µmol) at 80
°C, and the mixture was heated at the same temperature. After the dis-
appearance of the starting material on TLC, the mixture was evaporated,
and the residue was treated by the procedure as described to give 15
1
to give 17 or 26. The R/â ratio of the product was determined by H
NMR.
3-(2,3,4-Tri-O-benzoyl-D-xylopyranosyl)propene (17) from 9 (Table
2, Entry 1). After the treatment of 9 (121 mg, 301 µmol) according to
the general procedure for the allylation described above, the residue
was shortly filtrated through a column (SiO2, hexane/AcOEt, 2:1) to
give the crude product. A solution of the crude product in aqueous
TFA (80%, 4 mL) was stirred at room temperature for 15 min,
evaporated, and azeotroped with toluene (3 times). A mixture of the
resulting residue, BzCl (210 µL, 1.80 mmol), and DMAP (37 mg, 301
µmol) in pyridine (3 mL) was stirred at room temperature for 2 h.
After the addition of ice, the mixture was partitioned between AcOEt
and aqueous 1 M HCl, and the organic layer was washed with aqueous
saturated NaHCO3, H2O, and brine, dried (Na2SO4), and evaporated.
The resulting residue was purified by column chromatography (SiO2,
hexane/AcOEt, 10:1) to give 17 (107 mg, 73% as a white solid, R/â
ratio ) 85:15); FAB-HRMS calcd for C29H27O7 487.1757 (MH+), found
481.1791. The R/â mixture of 17 (105 mg) was further purified by
column chromatography (SiO2, hexane/AcOEt, 12:1) to give the
R-anomer (85 mg, as a white solid) and the â-anomer (15 mg, as a
white solid). R-Anomer: 1H NMR (CDCl3, 400 MHz) δ 8.17-7.15
(m, 15 H, Ar), 5.88 (m, 1 H, CHdCH2), 5.62 (br s, 1 H, H-3), 5.21 (br
s, 1 H, H-3), 5.11 (m, 3 H, H-4, CHdCH2), 4.31 (d, 1 H, H-5a, J )
13.2), 4.13 (dd, 1 H, H-5b, J ) 2.0, 13.2), 4.07 (ddd, 1 H, H-1, J )
1.5, 7.0, 7.0), 2.59-2.41 (m, 2 H, CH2); FAB-HRMS calcd for C29H27O7
487.1757 (MH+), found 487.1730. Anal. Calcd for C29H26O7: C, 71.59;
H, 5.39. Found: C, 71.30; H, 5.36. â-Anomer: 1H NMR (CDCl3, 400
MHz) δ 7.94-7.25 (m, 15 H, Ar), 5.90 (m, 1 H, CHdCH2), 5.83 (dd,
1 H, H-3, J ) 9.6, 9.6), 5.40 (dd, 1 H, H-2, J ) 9.6, 9.6), 5.38 (m, 1
H, H-4), 5.11 (m, 2 H, CHdCH2), 4.40 (dd, 1 H, H-5a, J ) 5.6, 11.1),
3.72 (ddd, 1 H, H-1, J ) 3.8, 7.6, 9.6), 3.52 (dd, 1 H, H-5b, J ) 11.1,
11.1), 2.41 (m, 2 H, CH2); FAB-HRMS calcd for C29H27O7 487.1757
(MH+), found 487.1733. Anal. Calcd for C29H26O7‚0.7H2O: C, 69.78;
H, 5.53. Found: C, 71.01; H, 5.93.
Compound 17 from 10 (Table 2, Entry 2). Compound 17 (83
mg, 69% as a white solid, R/â ratio ) 91:9) was prepared from 10
(121 mg, 301 µmol) by the procedure described above for 9 (Table 2,
entry 1); FAB-HRMS calcd for C29H27O7 487.1757 (MH+), found
487.1769.
Compound 17 from 11 (Table 2, Entry 3). After the treatment of
11 (87 mg, 301 µmol) according to the general procedure described
above, a mixture of the residue and 1,3-propanediol (55 µL, 761 µmol)
in acetone (1.5 mL) was stirred at room temperature for 15 min and
evaporated to give the crude deprotected product, which was benzo-
ylated by the procedure described for 9 (Table 2, entry 1) to give 17
(90 mg, 61% as a white solid, R/â ratio ) 1:99); FAB-HRMS calcd
for C29H27O7 487.1757 (MH+), found 487.1787.
2
or 25. The R/â ratio of the product was determined by H NMR.
1-[2H]-1,5-Anhydro-2,3,4-tri-O-acetyl-D-xylitol (15) from 8 (Table
1, Entry 1). After the treatment of 8 (56 mg, 140 µmol) according to
the general procedure for the deuteration described above, the resulting
residue was purified by column chromatography (SiO2, hexane/AcOEt,
3.5:1) to give 15 (24 mg, 66% as a white solid, R/â ratio ) 65:35): 1H
NMR (CDCl3, 400 MHz) δ 5.16 (dd, 1 H, H-3, J ) 8.5, 8.5), 4.91 (m,
2 H, H-3, H-4), 4.03 (m, 1.6 H, H-1â, H-5a), 3.34 (m, 1.4 H, H-1R,
2
H-5b), 2.06 (m, 9 H, Ac × 3); H NMR (CHCl3, 400 MHz) δ 3.48
(â-anomer), 2.79 (R-anomer); FAB-HRMS calcd for C11H16DO7
262.1037 (MH+), found 262.1019. Anal. Calcd for C11H15DO7: C,
50.57; H, 6.56. Found: C, 50.32; H, 6.22.
Compound 15 from 9 (Table 1, Entry 2). After the treatment of 9
(56 mg, 140 µmol) according to the general procedure for the
deuteration described above, the residue was shortly filtrated through
a column (SiO2, hexane/AcOEt, 1:1) to give a crude product. A solution
of the obtained product in aqueous TFA (80%, 3 mL) was stirred at
room temperature for 15 min, and the solvent was evaporated and
azeotroped with toluene (3 times). A mixture of the resulting residue,
AcCl (50 µL, 703 µmol), and DMAP (4-(dimethylamino)pyridine) (17
mg, 140 µmol) in pyridine (2 mL) was stirred at room temperature for
2 h. After addition of ice, the mixture was partitioned between AcOEt
and aqueous 1 M HCl, and the organic layer was washed with aqueous
saturated NaHCO3, H2O, and brine, dried (Na2SO4), and evaporated.
The resulting residue was purified by column chromatography (SiO2,
hexane/AcOEt, 3.5:1) to give 15 (30 mg, 83% as a white solid, R/â
ratio ) 97:3): 2H NMR (CHCl3, 400 MHz) δ 3.48 (â-anomer), 2.79
(R-anomer); FAB-HRMS calcd for C11H16DO7 262.1037 (MH+), found
262.1044.
Compound 15 from 10 (Table 1, Entry 3). Compound 15 (30 mg,
83% as a white solid, R/â ratio ) 97:3) was obtained from 10 (56 mg,
140 µmol) by the procedure described for 9 (Table 1, entry 2): 2H
NMR (CHCl3, 400 MHz) δ 3.48 (â-anomer), 2.79 (R-anomer); FAB-
HRMS calcd for C11H16DO7 262.1037 (MH+), found 262.1053.
Compound 15 from 11 (Table 1, Entry 4). After the treatment of
11 (40 mg, 140 µmol) according to the general procedure for the
deuteration described above, the mixture of the residue and 1,3-
propanediol (50 µL, 690 µmol) in acetone (2 mL) was stirred at 0 °C
for 15 min, and the solvent was evaporated to give the crude deprotected
product, which was acetylated by the procedure described for 9 (Table
1, entry 2) to give 15 (26 mg, 72% as a white solid, R/â ratio ) 1:99):
2H NMR (CHCl3, 400 MHz) δ 3.48 (â-anomer), 2.79 (R-anomer); FAB-
HRMS calcd for C11H16 DO7 262.1037 (MH+), found 262.1032.
Compound 15 from 12 (Table 1, Entry 5). After the treatment of
12 (56 mg, 140 µmol) according to the general procedure for the
deuteration described above, the residue was shortly filtrated through
a column (SiO2, hexane/benzene, 8:1) to give a crude product. A
solution of the crude product and TBAF (1 M in THF, 556 µL, 556
µmol) was stirred at room temperature for 1 h and evaporated to give
the crude deprotected product, which was acetylated by the procedure
described for 9 (Table 1, entry 2) to give 15 (29 mg, 80% as a
white solid, R/â ratio ) 1:99): 2H NMR (CHCl3, 400 MHz) δ 3.48
Compound 17 and 1,5-Anhydro-2,3,4-tri-O-benzoyl-D-xylitol (19)
from 12 (Table 2, Entry 4). After treatment of 12 (228 mg, 301 µmol)
according to the general procedure for the allylation described above,
the residue was shortly filtrated through a column (SiO2, hexane/
benzene, 50:1-10:1) to give a crude product. A solution of the crude
product and TBAF (1 M in THF, 1.35 mL, 1.35 mmol) in THF (2 mL)
was stirred at room temperature for 2 h and evaporated to give the
crude deprotected product, which was benzoylated by the procedure

SelectiVe Radical C-Glycosylation Reactions
J. Am. Chem. Soc., Vol. 123, No. 48, 2001 11881
described above for 9 (Table 2, entry 1) to give 17 (38 mg, 26% as a
white solid, R/â ratio ) 1:99) and 19 (83 mg, 61% as a white solid).
17: FAB-HRMS calcd for C29H26O7Na 509.1576 (MNa+), found
509.1604. 19: 1H NMR (CDCl3, 400 MHz) δ 8.13-7.36 (m, 15 H,
Ar), 5.82 (dd, 1 H, H-3, J ) 7.9, 7.9), 6.68 (ddd, 2 H, H-2, H-4, J )
4.7, 8.1, 8.1), 5.39 (dd, 2 H, H-1a, H-5a, J ) 4.7, 11.6), 4.58 (dd, 2 H,
H-1b, H-5b, J ) 8.1, 11.6); FAB-HRMS calcd for C26H22O7Na
469.1263 (MNa+), found 469.1263.
3-(2,3,4-Tri-O-benzoyl-D-xylopyranosyl)propionitrile (18) from
12 (Table 2, Entry 5). To a solution of 12 (228 mg, 301 µmol),
acrylonitrile (199 µL, 3.0 mmol), and Bu3SnH (162 µL, 600 µmol) in
benzene (3 mL) was added AIBN (10 mg, 61 µmol) at 80 °C, and the
mixture was heated at the same temperature. After disappearance of
the starting material on TLC, the mixture was evaporated, and the
residue was treated by the procedure described above for the allylation
of 12 (Table 2, entry 4). After purification by column chromatography
(SiO2, hexane/Et2O, 20:1), 18 was obtained (28 mg, 66% as an oil,
R/â ) 0:100): 1H NMR (CDCl3, 500 MHz) δ 8.10-7.20 (m, 15 H,
Ar), 5.88 (dd, 1 H, H-3, J ) 5.6, 10.8), 5.39 (ddd, 1 H, H-4, J ) 5.6,
10.8, 10.8), 5.34 (dd, 1 H, H-2, J ) 9.6, 9.6), 4.42 (dd, 1 H, H-5a,
J ) 5.6, 10.8), 3.76 (ddd, 1 H, H-1, J ) 2.9, 9.4, 9.4), 3.56 (dd,
1 H, H-5b, J ) 10.8, 10.8), 2.58 (m, 2 H, CH2CH2CN), 1.96 (m,
2 H, CH2CH2CN); FAB-HRMS calcd for C29H26NO7 500.1709 (MH+),
found 500.1689. Anal. Calcd for C29H25NO7‚0.3H2O: C, 68.98; H, 5.11;
N, 2.77. Found: C, 69.05; H, 5.11; N, 2.54.
Phenyl 2-Deoxy-3,4-di-O-acetyl-1-seleno-â-D-xylopyranoside (20).
To a solution of 2326 (2.82 g, 10.8 mmol) and PhSeH (2.39 mL, 21.6
mmol) in CH2Cl2 (72 mL) was added BF3‚OEt2 (2.33 mL, 18.4 mmol)
at 0 °C, and the mixture was stirred at room temperature for 30 min.
The mixture was partitioned between Et2O and aqueous saturated
NaHCO3, and the organic layer was washed with H2O and brine, dried
(Na2SO4), and evaporated. The resulting residue was purified by column
chromatography (SiO2, hexane/AcOEt, 5:1) to give 20 (3.36 g, 87%
as a white solid, R/â ratio ) ∼1:1 based on 1H NMR): 1H NMR
(CDCl3, 400 MHz) δ 7.58-7.27 (m, 5 H), 5.68 (dd, 0.55 H, J ) 4.4,
4.4), 5.55 (dd, 0.45 H, J ) 4.7, 4.7), 5.14-3.50 (m, 4 H), 2.65-2.00
(m, 8 H); FAB-HRMS calcd for C15H18O5SeNa 381.0217 (MNa+),
found 381.0220. Anal. Calcd for C15H18O5Se: C, 50.43; H, 5.08.
Found: C, 50.46; H, 5.16.
(MNa+), found 411.0702. Anal. Calcd for C17H24O5Se: C, 52.57; H,
6.21. Found: C, 52.72; H, 6.25.
Phenyl 3,4-Bis-O-triisopropylsilyl-2-deoxy-1-seleno-â-D-xylopyrano-
side (22). A mixture of 24 (369 mg, 1.35 mmol), TIPSOTf (1.09 mL,
4.06 mmol), and 2,6-lutidine (1.57 mL, 13.5 mmol) in CH2Cl2 (10 mL)
was stirred at room temperature for 30 min. After the addition of MeOH
(550 µL), the mixture was partitioned between AcOEt and aqueous 1
M HCl, and the organic layer was washed with H2O, aqueous saturated
NaHCO3, and brine, dried (Na2SO4), and evaporated. The resulting
residue was purified by column chromatography (SiO2, hexane/benzene,
15:1-10:1) to give 22 (736 mg, 93% as an oil, R/â ratio ) 1:1 based
1
on H NMR). The R/â mixture of 22 (736 mg) was further purified
by column chromatography (SiO2, hexane/benzene, 15:1) to give the
R-anomer (209 mg), the â-anomer (382 mg), and the R/â mixture
(145 mg). R-Anomer: 1H NMR (CDCl3, 400 MHz) δ 7.59-7.19
(m, 5 H, Ar), 5.81 (d, 1 H, H-1, J ) 5.5), 4.55 (d, 1 H, H-5a, J )
12.0), 4.01 (d, 1 H, H-3, J ) 2.8), 3.69 (br s, 1 H, H-4), 3.58 (d,
1 H, H-5b, J ) 12.0), 2.77 (ddd, 1 H, H-2a, J ) 2.8, 5.5, 14.0), 2.06
(d, 1 H, H-2b, J ) 14.0), 1.10 (m, 42 H, TIPS × 2); FAB-HRMS
calcd for C29H54O3SeSi2Na 609.2675 (MNa+), found 609.2682. Anal.
Calcd for C29H54O3SeSi2: C, 59.45; H, 9.29. Found: C, 59.17; H,
9.05. â-Anomer: 1H NMR (CDCl3, 400 MHz) δ 7.62-7.25 (m,
5 H, Ar), 5.46 (dd, 1 H, H-1, J ) 2.2, 10.5), 3.99 (br s, 1 H, H-3), 3.98
(dd, 1 H, H-5a, J ) 1.0, 11.8), 3.83 (dd, 1 H, H-5b, J ) 2.4, 11.8),
3.61 (br s, 1 H, H-4), 2.38 (ddd, 1 H, H-2a, J ) 2.5, 10.5, 13.3), 1.90
(m, 1 H, H-2b), 1.05 (m, 42 H, TIPS × 2); FAB-HRMS calcd for
C29H54O3SeSi2Na 609.2675 (MNa+), found 609.2645. Anal. Calcd for
C29H54O3SeSi2‚0.3H2O: C, 58.91; H, 9.31. Found: C, 58.98; H, 8.95.
1-[2H]-1,5-Anhydro-2-deoxy-3,4-di-O-acetyl-D-xylitol (25) from
20 (Table 3, Entry 1). After treatment of 20 (50 mg, 140 µmol)
according to the procedure described for the deuteration of 8 (Table 1,
entry 1), 25 (28 mg, 97% as an oil, R/â ratio ) 84:16) was obtained:
1H NMR (CDCl3, 500 MHz) δ 5.00 (ddd, 1 H, H-3, J ) 4.6, 8.1, 8.1),
4.83 (ddd, 1 H, H-4, J ) 4.2, 7.5, 7.5), 3.98 (dd, 1 H, H-5a, J ) 4.2,
11.7), 3.87 (m, 0.84 H, H-1â), 3.54 (m, 0.16 H, H-1R), 3.39 (dd, 1 H,
H-5b, J ) 7.5, 11.7), 2.12 (m, 1 H, H-2a), 2.07 (m, 6 H, Ac × 2), 1.73
(m, 1 H, H-2b): 2H NMR (CHCl3, 400 MHz) δ 3.33 (â-anomer), 3.00
(R-anomer); ESI-HRMS calcd for C9H13DO5Na 226.0802 (MNa+),
found 226.0806.
Compound 25 from 21 (Table 3, Entry 2). After treatment of 21
(54 mg, 104 µmol) according to the procedure described for the
deuteration of 9 (Table 1, entry 2), 25 (18 mg, 61% as an oil, R/â ratio
) 99:1) was obtained: 2H NMR (CHCl3, 400 MHz) δ 3.33 (â-anomer),
3.00 (R-anomer); ESI-HRMS calcd for C9H13DO5Na 226.0802 (MNa+),
found 226.0806.
Compound 25 from 22 (Table 3, Entry 4). After treatment of 22
(82 mg, 140 µmol) according to the procedure described for the
deuteration of 12 (Table 1, entry 5), 25 (20 mg, 71% as an oil,
R/â ratio ) 72:28) was obtained: 2H NMR (CHCl3, 400 MHz) δ 3.33
(â-anomer), 3.00 (R-anomer); ESI-HRMS calcd for C9H13DO5Na
226.0802 (MNa+), found 226.0808.
Phenyl 2-Deoxy-1-seleno-â-D-xylopyranoside (24). A mixture
of 20 (1.98 g, 5.54 mmol) and NaOMe (28% in MeOH, 100 µL) in
THF/MeOH (5 mL/5 mL) was stirred at room temperature for 2 h
and neutralized with Diaion PK 212 resin (H+ form). The resin was
filtered off, and the filtrate was evaporated and dried in vacuo at
room temperature for 1 h to give 24 (1.51 g, 99% as a white solid, R/â
ratio ) 1:1): 1H NMR (CDCl3, 400 MHz) δ 7.55 (m, 2 H), 7.26 (m,
3 H), 5.84 (dd, 0.54 H, J ) 2.0, 5.1), 5.19 (dd, 0.46 H, J ) 2.8, 8.7),
4.23-3.25 (m, 4 H), 2.51-1.80 (m, 4 H); FAB-HRMS calcd for
C11H14O3SeNa 297.0006 (MNa+), found 297.0028. Anal. Calcd for
C11H14O3Se: C, 48.36; H, 5.17. Found: C, 48.26; H, 5.05.
Phenyl 2-Deoxy-3,4-O-[(2S,3S)-2,3-dimethoxybutan-2,3-diyl]-1-
seleno-â-D-xylopyranoside (21). Compound 21 (1.68 g, 89% as a white
solid, R/â ratio ) 1:1 based on 1H NMR) was obtained from 24 (1.34
g, 4.9 mmol) as described for the synthesis of 10, after purification by
column chromatography (SiO2, hexane/AcOEt, 20:1). The R/â mixture
of 24 (600 mg) was further purified by silica gel chromatography (SiO2,
hexane/AcOEt, 25:1) to give the R-anomer (223 mg), the â-anomer
(111 mg), and the R/â mixture (252 mg). R-Anomer: 1H NMR (CDCl3,
400 MHz) δ 7.57-7.25 (m, 5 H, Ar), 5.93 (d, 1 H, H-1, J ) 4.7),
4.10-3.65 (m, 4 H, H-3, H-4, H-5a, H-5b), 3.31 (s, 3 H, OMe), 3.28
(s, 3 H, OMe), 2.30 (ddd, 1 H, H-2b, J ) 1.7, 5.0, 11.7), 2.23 (ddd, 1
H, H-2a, J ) 5.3, 11.7, 11.7), 1.32 (s, 3 H, Me), 1.31 (s, 3 H, Me);
FAB-HRMS calcd for C17H24O5SeNa 411.0687 (MNa+), found 411.0681.
Anal. Calcd for C17H24O5Se‚0.1H2O: C, 52.47; H, 6.27. Found: C,
52.24; H, 6.16. â-Anomer: 1H NMR (CDCl3, 400 MHz) δ 7.62-7.30
(m, 5 H, Ar), 5.00 (dd, 1 H, H-1, J ) 2.1, 11.7), 4.00 (dd, 1 H, H-5a,
J ) 5.0, 10.6), 3.74 (ddd, 1 H, H-3, J ) 4.4, 9.7, 9.7), 3.66 (ddd, 1 H,
H-4, J ) 5.0, 9.8, 9.8), 3.36 (dd, 1 H, H-5b, J ) 10.6, 10.6), 3.25 (s,
3 H, OMe), 3.07 (s, 3 H, OMe), 2.23 (ddd, 1 H, H-2a, J ) 2.1, 4.4,
11.7), 1.89 (ddd, 1 H, H-2b, J ) 11.7, 11.7, 11.7), 1.29 (s, 3 H, Me),
1.28 (s, 3 H, Me); FAB-HRMS calcd for C17H24O5SeNa 411.0687
3-(2-Deoxy-2,3,4-tri-O-benzoyl-D-xylopyranosyl)propene (26) from
21 (Table 3, Entry 3). After treatment of 21 (117 mg, 301 µmol)
according to the procedure described for the allylation of 9 (Table 2,
entry 1), 26 (61 mg, 74% as an oil, R/â ratio ) 98:2) was obtained:
1H NMR (CDCl3, 500 MHz) δ for the R-anomer 8.15-7.29 (m, 10 H,
Ar), 5.86 (m, 1 H, CHdCH2), 5.42 (d, 1 H, H-3, J ) 1.5), 5.11 (m, 3
H, H-4, CHdCH2), 4.19 (d, 1 H, H-5a, J ) 13.2), 4.09 (dd, 1 H, H-5b,
J ) 1.3, 13.2), 3.86 (m, 1 H, H-1), 2.37 (m, 2 H, CH2CHdCH2), 2.04
(m, 2 H, H-2a, H-2b); 1H NMR (CDCl3, 500 MHz) δ for the â-anomer
8.15-7.29 (m, 10 H, Ar), 5.86 (m, 1 H, CHdCH2), 5.41 (ddd, 1 H,
H-3, J ) 5.3, 10.7, 10.7), 5.33 (ddd, 1 H, H-4, J ) 5.4, 10.7, 10.7),
5.11 (m, 2 H, CHdCH2), 4.35 (dd, 1 H, H-5a, J ) 5.3, 10.7), 3.63 (m,
1 H, H-1), 3.44 (dd, 1 H, H-5b, J ) 10.7, 10.7), 2.37 (m, 2 H, CH2-
CHdCH2), 1.71 (m, 2 H, H-2); FAB-HRMS calcd for C22H22O5Na
389.1365 (MNa+), found 389.1369. Anal. Calcd for C22H22O5: C, 72.12;
H, 6.05. Found: C, 72.03; H, 6.22.
Compound 26 and 1,5-Anhydro-2-deoxy-3,4-di-O-benzoyl-D-
xylitol (28) from 22 (Table 3, Entry 5). After treatment of 22 (176
mg, 301 µmol) according to the procedure described for the allylation
of 12 (Table 2, entry 4), 26 (36 mg, 34% as an oil, R/â ratio ) 46:54)

11882 J. Am. Chem. Soc., Vol. 123, No. 48, 2001
Abe et al.
and 28 (21 mg, 20% as an oil) were obtained. 26: FAB-HRMS calcd
for C22H22O5Na 389.1365 (MNa+), found 389.1345. 28: 1H NMR
(CDCl3, 400 MHz) δ 8.05 (m, 10 H, Ar), 5.39 (ddd, 1 H, H-3, J ) 4.4,
7.6, 7.6), 5.26 (ddd, 1 H, H-4, J ) 4.1, 7.6, 7.6), 4.21 (dd, 1 H, H-5a,
J ) 4.4, 11.7), 4.01 (ddd, 1 H, H-1a, J ) 4.4, 4.4, 11.7), 3.67 (ddd, 1
H, H-1b, J ) 2.9, 9.4, 10.6), 3.60 (dd, 1 H, H-1a, J ) 7.9, 11.7), 2.35
(m, 1 H, H-2a), 1.96 (m, 1 H, H-2b); FAB-HRMS calcd for C9H19O5
327.1232 (MH+), found 327.1208.
1,5-Anhydro-2-deoxy-3,4-di-O-acetyl-D-xylitol (27). Compound 27
(26 mg, 96% as an oil) was prepared from 20 (50 mg, 140 µmol) by
the procedure described for the deuteration of 20 with Bu3SnH instead
of Bu3SnD: 1H NMR (CDCl3, 500 MHz) δ 5.00 (ddd, 1 H, H-3, J )
4.6, 8.1, 8.1), 4.83 (ddd, 1 H, H-4, J ) 4.2, 7.5, 7.5), 3.98 (dd, 1 H,
H-5a, J ) 4.2, 11.7), 3.87 (ddd, 1 H, H-1â, J ) 4.6, 4.6, 11.8), 3.54
(ddd, 1 H, H-1R, J ) 3.0, 9.2, 11.8), 3.39 (dd, 1 H, H-5b, J ) 7.5,
11.7), 2.12 (m, 1 H, H-2a), 2.07 (m, 6 H, Ac × 2), 1.73 (m, 1 H,
H-2b); ESI-HRMS calcd for C9H14O5Na 225.0739 (MNa+), found
225.0741.
Calculations. All ab initio and semiempirical PM3 calculations were
performed using the Gaussian 98 program28 on an SGI O2 workstation.
MM3 calculations were performed using the Macromodel 5.0 program.37
The preoptimized geometries by UHF/STO-3G* or PM3 were taken
to be the input geometries for final optimization by UHF/3-21G*. The
stationary points were characterized by frequency analysis (minimum
with 0). Single point energies and properties based on the natural bond
orbital (NBO) theory30 were calculated by UB3LYP/6-31G*. Hybrid
orbitals and stabilization energy by orbital interactions were analyzed
by the NBO method.32 Orbital interactions were examined in terms of
second-order perturbation from filled orbitals to empty neighboring
orbitals.
Acknowledgment. This investigation was supported by a
Grant-in-Aid for Creative Scientific Research (13NP0401) from
the Japan Society for Promotion of Science.We thank the Japan
Society for Promotion of Science (H.A.) for support of this
research. We are also grateful to Ms. H. Matsumoto, A. Maeda,
S. Oka, and N. Hazama (Center for Instrumental Analysis,
Hokkaido University) for technical assistance with NMR, MS,
and elemental analysis.
Supporting Information Available: 1H NMR spectral
charts of 15 (from 8), 16, 17 (R-anomer), 17 (â-anomer), 17
(from 9, 10, 11, and 12), 18, 19, 21 (R-anomer), 21 (â-anomer),
22 (R-anomer), 22 (â-anomer), 25 (from 20), 26 (from 21 and
2
22), 27, and 28; H NMR spectral charts of 15 (from 8, 9, 10,
11, and 12) and 25 (from 20, 21, and 22) (PDF). This material
is available free of charge via the Internet at http://pubs.acs.org.
JA011321T