Stereocontrolled synthesis of heterocyclic C-nucleosides. Protecting group effect and molecular modeling studies.

Article abstract of DOI:10.1021/jo016345x

We report herein a short stereocontrolled synthesis of heterocyclic C-nucleosides (indole, imidazole, benzimidazole, and 6-iodobenzimidazole). First, condensation of 2-lithiated heterocycles 2-5 with 5-(tert-butyldiphenylsilyl)-2,3-O-isopropylidene-D-gamma-ribonolactone (1) afforded the hemiacetals 6-9 in good yields. Then, borohydride reduction (NaBH(4)) of the protected hemiacetals proceeded stereoselectively to give predominantly the S diols 10-13, which upon Mitsunobu cyclization afforded the alpha-C-nucleosides 14-17. In contrast, the same PPh(3)/DEAD treatment of the 1:1 diastereomeric mixture of the free heterocyclic diols 10d and 11d gave exclusively the beta-anomers 14dbeta and 15dbeta, respectively, by a stereocontrolled process. The mechanisms of these stereocontrolled steps are discussed with the support of molecular modeling studies.

Full text of DOI:10.1021/jo016345x

3724
J . Org. Chem. 2002, 67, 3724-3732
Ster eocon tr olled Syn th esis of Heter ocyclic C-Nu cleosid es.
P r otectin g Gr ou p Effect a n d Molecu la r Mod elin g Stu d ies
Dominique Guianvarc’h, J ean-Louis Fourrey, Marie-Elise Tran Huu Dau, and
Vincent Gue´rineau
Institut de Chimie des Substances Naturelles, CNRS, Avenue de la Terrasse,
91198 Gif-Sur-Yvette, France
Rachid Benhida*
Laboratoire de Chimie Bioorganique, UMR 6001 CNRS, Universite´ de Nice Sophia Antipolis
Parc Valrose, 06108 Nice Cedex 2, France
benhida@unice.fr
Received December 5, 2001
We report herein a short stereocontrolled synthesis of heterocyclic C-nucleosides (indole, imidazole,
benzimidazole, and 6-iodobenzimidazole). First, condensation of 2-lithiated heterocycles 2-5 with
5-(tert-butyldiphenylsilyl)-2,3-O-isopropylidene-D-γ-ribonolactone (1) afforded the hemiacetals 6-9
in good yields. Then, borohydride reduction (NaBH4) of the protected hemiacetals proceeded
stereoselectively to give predominantly the S diols 10-13, which upon Mitsunobu cyclization
afforded the R-C-nucleosides 14-17. In contrast, the same PPh3/DEAD treatment of the 1:1
diastereomeric mixture of the free heterocyclic diols 10d and 11d gave exclusively the â-anomers
14d â and 15d â, respectively, by a stereocontrolled process. The mechanisms of these stereocontrolled
steps are discussed with the support of molecular modeling studies.
In tr od u ction
selective synthetic methodology providing one or the
other anomer would be of a great interest.
For many years, natural C-nucleosides and their
synthetic analogues have attracted wide interest in view
of the importance of their biological activities.1 Recently,
a renewed interest in these compounds arose as a
consequence of their potential applications in nucleic acid
chemistry.2 In view of our current research which pro-
poses to incorporate extended heterocyclic C-nucleosides
in triple-helix forming oligonucleotides, we needed to
design an efficient diastereoselective general procedure
toward the synthesis of highly functionalized imidazole,
benzimidazole and indole C-nucleosides. A large number
of synthetic approaches to a variety of C-nucleosides have
been reported,1 with several C-nucleosides being synthe-
sized by addition of an organometallic reagent to a
ribonolactone derivative followed by Lewis acid induced
hemiacetal deoxygenation.3 In many cases the deoxygen-
ation step was met with difficulties, leading to poor yield
and/or diastereoselectivity.4 Although most biologically
active C-nucleosides possess the â configuration, there
is growing interest in those of the R configuration, which
show promising biological activity.5 Therefore, a stereo-
We describe herein a short stereoselective synthesis
of either R or â heterocyclic C-nucleosides, by (i) conden-
sation of 2-lithiated heterocycles 2-5 (indole, benzimi-
dazole, imidazole, and 6-iodobenzimidazole) with pro-
tected ribonolactone 1, followed by (ii) borohydride
reduction and (iii) Mitsunobu cyclization of the resulting
diols. The diastereoselective access to the R-anomers was
controlled in the borohydride reduction step by just
keeping an appropriate protecting group on the hetero-
cycle. In contrast, the removal of the protecting group
resulted in total stereocontrol of the Mitsunobu ring-
closing step, even from a mixture of diastereomeric diols
affording exclusively the â anomers (Scheme 1). On the
basis of molecular modeling studies, we can propose a
rational explanation for the stereoselectivity of the reduc-
tion and for the stereocontrol observed in the Mitsunobu
ring-closing step.
Resu lts a n d Discu ssion
Con d en sa t ion of 2-Lit h ia t ed Het er ocycles w it h
5-(ter t-Bu tyld ip h en ylsilyl)-2,3-O-isop r op ylid en e-D-
γ-r ibon ola cton e. In our synthetic methodology we used
* To whom correspondence should be addressed. Fax:
(33) 0492076151.
(1) For reviews on the chemistry, biochemistry, and synthesis of
C-nucleoside analogues, see: (a) Buchanan, J . G. Prog. Chem. Org. Nat.
Prod. 1983, 44, 243-299. (b) Hacksell, U.; Daves, G. D., J r. Prog. Med.
Chem. 1985, 22, 1-65. (c) Postema, M. H. D. Tetrahedron 1992, 40,
8545-8599. (d) Watanabe, K. A. In Chemistry of Nucleosides and
Nucleotides; Townsend, L. B., Ed.; Plenum Press: New York, 1994;
Vol. 3, pp 421-535. (e) J aramillo, C.; Knapp, S. Synthesis 1994, 1-20.
(f) Yokoyama, M.; Toyoshima, H.; Shimizu, M.; Togo, H. J . Chem. Soc.,
Perkin Trans. 1 1997, 29-33. (g) Togo, H.; He, W.; Waki, Y.; Yokoyama,
M. Synlett 1998, 700-716.
(2) (a) Von Krosigk, U.; Benner, S. A. J . Am. Chem. Soc. 1995, 117,
5361-5362. (b) Matulic-Adamic, J .; Beigelman, L.; Portman, S.; Egli,
M.; Usman, N. J . Org. Chem. 1996, 61, 3909-3911. (c) Hildbrand, S.;
Leuman, C. Angew. Chem., Int. Ed. Engl. 1996, 35, 1968-1970. (d)
Chaudhuri, N. C.; Ren, R. X. F.; Kool, E. T. Synlett 1997, 341-347.
(3) (a) Piccirilli, J . A.; Kraush, T.; MacPherson, L. J .; Benner, S. A.
Helv. Chim. Acta 1991, 74, 397-406. (b) A similar approach was used
for the synthesis of imidazole-C-glycopyranosides: Granier, T.; Vasella,
A. Helv. Chim. Acta 1995, 78, 1738-1746. (c) Liu, W.; Walker, J . A.;
Chen, J . J .; Wise, D. S.; Townsend, L. B. Tetrahedron Lett. 1996, 37,
5325-5328. (d) Gudmundson, K. S.; Drach, J . C.; Townsend, L. B. J .
Org. Chem. 1997, 62, 3453-3459.
(4) (a) Pankiewicz, K. W.; Sochack, E.; Kabat, M. M.; Ciszewski, L.
A.; Watanabe, K. A. J . Org. Chem. 1988, 53, 3473-3478. (b) Gud-
mundson, K. S.; Drach, J . C.; Townsend, L. B. Tetrahedron Lett. 1996,
37, 2365-2368. (c) Matulic-Adamic, J .; Beigelman, L. Tetrahedron Lett.
1997, 38, 203-206.
(5) Bernardo, S. D.; Weigele, M. J . Org. Chem. 1976, 41, 287-290.
10.1021/jo016345x CCC: $22.00 © 2002 American Chemical Society
Published on Web 05/07/2002

Stereocontrolled Synthesis of C-Nucleosides
J . Org. Chem., Vol. 67, No. 11, 2002 3725
Sch em e 1. Syn th etic Str a tegy
the ribonolactone 13a and the protected heterocycles 2-56
as starting materials. First, the conditions for the quan-
titative formation of 2-lithiated heterocycles 2-5 were
determined using hexachloroethane as a trapping agent,
indicating that, in all cases, the heterocycles (except 2d )
could be quantitatively metalated by 1.1 equiv of LDA
in THF at -78 °C in 45 min. Then, hemiacetals 6-9 were
obtained by slow addition of ribonolactone 1 to freshly
prepared lithiated heterocycles 2-5 in good yields (75-
91%, Table 1).
prepared by following the Katritzky method, which
utilized carbon dioxide for similtaneous protection of the
indole NH and subsequent 2-lithiation.7 This procedure,
when applied to the lithiated derivative of carboxy-
benzimidazole 3d (R ) CO2Li), was however unsuccessful
for the synthesis of the benzimidazole analogue 7d .
Alternatively, this free hemiacetal (7d ) could be quanti-
tatively obtained from the protected hemiacetal 7e by
selective removal of the (benzyloxy)methyl group using
catalytic hydrogenation (see Table 2, entry 6). Therefore,
we observed that the condensation step was very sensi-
tive to steric factors between the 2-lithiated heterocycles
2-5 and the bulky ribonolactone 1. This is further
supported by the fact that no significant reaction occurred
The 2-lithiated derivative of 1-carboxyindole 2d , used
in the synthesis of the free indole hemiacetal 6d , was
(6) Heterocycles 2-5 were protected using standard literature
protocols (Greene, T. W. Protective Groups in Organic Synthesis, 3rd
ed.; Wiley-Interscience: New York, 1999); see the Supporting Informa-
tion.
(7) Katritzky, A. R.; Akutagawa, K. Tetrahedron Lett. 1985, 26,
5935-5938.

3726 J . Org. Chem., Vol. 67, No. 11, 2002
Ta ble 1. Syn th esis of Hem ia ceta ls 6-9
Guianvarc’h et al.
opened form. The diastereomeric chemical ratio of all
hemiacetals 6-9 was further found to be concentration
and solvent dependent.8 At this step, we have first
examined the 1′-deoxygenation of the hemiacetals 6-9
under different reaction conditions. Unfortunately, treat-
ment of hemiacetals 6-9 with triethylsilane (Et3SiH) in
the presence of Lewis acid reagents (BF3.Et2O, ZnCl2,
TMSOTf) at -78 °C resulted in unchanged starting
materials. The increase of the concentration of the
reducing agent only induced product decomposition, as
shown by TLC and GC/MS analyses. Moreover, we failed
to acetylate or to mesylate the tertiary C1′-OH in order
to make the deoxygenation step more efficient. Such a
problem has been already encountered, and the stereo-
chemistry of the deoxygenation step remains unclear in
view of the heterogeneous results.3a,4 In our case, this
failure indicates that the C1′-OH group of the hemiacetals
is sterically constrained by both R and â faces.
entry
heterocycle
N1-R-Indole
product
yield (%)
1
2
3
4
2a , R ) SO2NMe2
2b, R ) SO2Ph
2c, R ) Boc
6a
6b
6c
91
89
a
2d , R ) CO2Li
6d (R ) H)
75
N1-R-Benzimidazole
5
6
7
3a , R ) SO2NMe2
3b, R ) SO2Ph
3e, R ) BOM
7a
7b
7e
90
81
75
N1-R-5-Iodobenzimidazole
Bor oh yd r id e Red u ction of Hem ia ceta ls 6-9. How-
ever, borohydride reduction of the ketone form of hemi-
acetals 6-9, using NaBH4 in methanol, proceeded in a
quantitative yield to afford a mixture of diastereomeric
diols 10-13 (S and R) in variable ratio, depending on
the nature of the heterocycle and its protecting group
(Table 2).
8
4a , R ) SO2NMe2
8a
85
N1-R-Imidazole
9
10
5a , R ) SO2NMe2
5f, R ) Tr
9a
9f
80
a
a
Only small amounts of these products were detected by GC/
MS analyses of the crude extracted mixtures.
Ta ble 2. Bor oh yd r id e Red u ction of Hem ia ceta ls 6-9
In most cases, except for 6d and 7d , the borohydride
reduction provided predominantly the S diol. For the
same protecting group, the diastereoselectivity is better
for imidazole and benzimidazoles than for the indole
series. For example, the reduction of indole hemiacetal
6a (Table 2, entry 1) led to a diastereoisomeric mixture
of diols 10a with a 75:25 S/R ratio, while reduction of
benzimidazole hemiacetal having the same sulfamoyl
protecting group 7a (Table 2, entry 4) gave a mixture of
diastereoisomeric diols 11a with a 90:10 S/R molecular
ratio. The diastereoselectivity is strongly dependent on
the protecting group of the heterocycles. Indeed, we
noticed an increased diastereoselectivity, especially with
more sterically hindered groups such as sulfamoyl (Table
2, entries 4, 8, and 9) or (benzyloxy)methyl, which gave
high diastereomeric excesses (Table 2, entry 7). In
contrast, the reduction of the hemiacetals 6d and 7d ,
having an unprotected heterocyclic moiety, gave a mix-
ture of diastereomeric diols 10d and 11d , respectively,
in nearly a 1:1 ratio (Table 2, entries 3 and 6). The
observed diastereoselective borohydride reduction for the
heterocycle-protected hemiacetals is in agreement with
the Felkin-Ahn model.9,10 All these results are further
supported by our molecular modeling study (see below),
which suggested that the presence of a hindered protect-
ing group on the heterocycle generates torsional con-
straints, resulting in a weakly twisted ketone-hetero-
cycle conformation (ketone out of the heterocycle plane),
thus inducing a diastereofacial hydride differentiation in
favor of the S diol. Our molecular modeling study also
indicated that the absence of diastereofacial differentia-
tion in the case of 6d and 7d may result from a ketone-
heterocycle planar conformation, allowing hydride ap-
proach by both faces.
entry
hemiacetal
product (%)
S/R ratioa,b
N1-R-Indole
1
2
3
6a , R ) SO2NMe2
6b, R ) SO2Ph
6d , R ) H
10a (97)
10b (94)
10d (96)
75/25
65/35
50/50
N1-R-Benzimidazole
4
5
6
7
7a , R ) SO2NMe2
7b, R ) SO2Ph
7d , R ) Hc
11a (98)
11b (97)
11d (95)
11e (95)
90/10
70/30
55/45
95/5
7e, R ) BOM
N1-R-5-Iodobenzimidazole
8a , R ) SO2NMe2 12a (92)
8
90/10
90/10
N1-R-Imidazole
9a , R ) SO2NMe2 13a (97)
9
a
The diastereomeric ratio was determined by reverse phase
b
HPLC analysis of the crude products. The stereochemistry of
each diol was determined on the basis of their cyclized products,
in which the anomeric configuration was assigned by 1H NMR and
2D COSY-NOESY experiments. c Compound 7d was obtained from
7e after removal of the (benzyloxy)methyl group: H2 (50 psi), Pd/
C, THF.
in the case of N-Boc- or N-trityl-protected indole or
imidazole (Table 1, entries 3 and 10), while their 2-lithi-
ated derivatives reacted with other classically used
electrophiles such as Bu3SnCl, C2Cl6, and cyclohexanone,
giving the expected 2-substituted derivatives. The NMR
spectra of each coupled products 6-9 revealed the
presence of three interconvertible species: i.e., the hemi-
acetal diastereoisomers (major products) and their open
(8) Similar observations, concerning equilibrium in solution of such
hemiacetals, were reported: Dondoni, A.; Schermann, M.-C. J . Org.
Chem. 1994, 59, 6404-6412.
(9) (a) Cherest, M.; Felkin, H. Tetrahedron Lett. 1968, 18, 2199-
2102. (b) Anh, N. T. Top. Curr. Chem. 1980, 88, 145.
(10) Guianvarc’h, D.; Benhida, R.; Fourrey, J .-L. Tetrahedron Lett.
2001, 42, 647-650.
1
carbonyl form. For example, the H NMR spectrum (300
MHz) of compound 9a shows the signals at 4.80 and 5.55
ppm (singlets) for the 1′-OH of the hemiacetals and at
5.95 ppm (doublet) for the proton H2′ of the carbonyl

Stereocontrolled Synthesis of C-Nucleosides
J . Org. Chem., Vol. 67, No. 11, 2002 3727
F igu r e 1. Proposed intermediate for the stereocontrolled cyclization.
Yokoyama et al.,12 in which the cyclization of a mixture
(S/R ) 1/8) of an NH free indole diol similar to 10d (5′-
trityl instead of 5′-TBDPS) provided the R- and â-C-
nucleoside anomers in an unchanged ratio (R/â ) 1/8),
through an intramolecular SN2 process. In our case, we
have verified that the PPh3/DEAD treatment of separated
pure diols 10d S and 10d R gave only the same cyclized
â-anomer 14d â, in high yields (Table 3, entries 4 and 5).
In addition, we did not observe any R vs â epimerization
under the present reaction conditions. In our preliminary
communication,10 we suggested that the stereocontrol
may result from benzylic elimination (Ph3PdO) from 10d
(or 11d in the benzimidazole series) and spontaneous
cyclization via the more favorable rotamer intermediate
10d J (or 11d J ) to give the â-anomer (Figure 1).13 In
accordance with our experimental results, molecular
modeling studies for this ring-closing step (10d to 14d â
and 11d to 15d â) show clearly that the cyclization
leading to the â-anomers 14d â and 15d â is a highly
favorable process, while the formation of the R-anomer
is strongly constrained in view of the calculated energies
and geometrical parameters of their respective transition
states TSâ and TSR.
Ta ble 3. Mitsu n obu Rin g-Closin g Step Lea d in g to
C-Nu cleosid es 14-17
entry
diol
S/R ratioa
product (yield, %)
N1-R-Indole
1
2
3
4
5
R ) SO2NMe2 10a
100:0
100:0
50:50
100:0
0:100
14a r (85)
14br (82)
14d â (92)
14d â (92)
14d â (90)
R ) SO2Ph
R ) H
10b
10d
10d
10d
R ) H
R ) H
N1-R-Benzimidazole
6
7
8
9
R ) SO2NMe2 11a
90:10
90:10
55:45
95:5
15a r (75)
15br (70)
15d â (90)
15er (75)
R ) SO2Ph
R ) H
11b
11d
11e
R ) BOM
N1-R-5-Iodobenzimidazole
10
R ) SO2NMe2 12a
90:10
16a r (80)
N1-R-Imidazole
11
R ) SO2NMe2 13a
90:10
17a r (75)
r/â Con figu r a tion Assign m en t. After the cyclization
of diols 10-13 under Mitsunobu conditions, the anomeric
configuration for each series was established by NOESY
experiments realized for compounds 14br and 18d r and
for 14d â and 18d â. 18d r was obtained from 15er by
successive 1-(benzyloxy)methyl and 5′-TBDPS protecting
group cleavage (Figure 2). 18d â was obtained from 15d â
after 5′-deprotection. The most significant correlations
are represented in Figure 2. Indeed, the â configuration
of compounds 14d â and 18d â was clearly evidenced by
the observed NOE correlation between H-1′ and H-4′. In
the same way, we observed for the other anomers (14br
and 18d r) an NOE correlation between H-1′ and H-2′
(cis relationship), in accordance with an R configuration.
The final protecting group cleavage of C-nucleosides
14-17 was achieved in one step, under acidic conditions.
Thus, treatment of the cyclized products 14a r, 14d â,
15a r, 15d â, 16a r, and 17a r with 2 N HCl in dioxane or
by using TFA/EtOH solution (see Experimental Section)
afforded the free C-nucleosides 19r, 19â, 20r, 20â, 21r,
and 22r, respectively, in high yields (Figure 3).
a
The S/R ratio was determined by 1H NMR and HPLC analyses
of the pure diols.
The diastereoselectivity of the borohydride reduction
seems also to depend on the chemical nature of the
heterocycle. Among the heterocycle sulfamoyl protected
derivatives, the diastereoselectivity is higher for the
(benz)imidazole series (Table 2, entries 4, 8, and 9) than
for the indole analogues 6a (Table 2, entry 1).
Mitsu n obu Rin g-Closin g Step . The treatment of
diols 10-13 under standard Mitsunobu conditions (DEAD/
PPh3/THF reflux) led to the cyclized products 14-17 in
good yields (70-92%). The anomeric configuration of the
cyclized products was found to depend on the nature of
the starting heterocyclic diol (Table 3).
Thus, the diols whose heterocyclic moiety is protected
gave, upon cyclization, the corresponding R-protected
C-nucleosides (Table 3, entries 1, 2, 6, 7, and 9-11)
through an intramolecular SN2 process and the R epimer
(minor product) remained unchanged.11 In contrast,
treatment of a 1:1 mixture of diastereomeric free diols
10d (Table 3, entry 3) and 11d (Table 3, entry 8) under
the same conditions afforded exclusively the â-C-nucleo-
sides 14d â and 15d â in good yields, respectively. This
observation is in disagreement with the findings of
(12) (a) Yokoyama, M.; Tanabe, T.; Toyoshima, A.; Togo, H. Synthesis
1993, 517-520. (b) Yokoyama, M.; Toyoshima, A.; Akiba, T.; Togo, H.
Chem. Lett. 1994, 265-268.
(13) Similar observations were reported in the imidazole series but,
with the formation of a small amount of the R-anomer (R/â ) 1/26).
See: (a) Harusawa, S.; Murai, Y.; Moriyama, H.; Ohishi, H.; Yoneda,
R.; Kurihara, T. Tetrahedron Lett. 1995, 36, 3165-3168. (b) Harusawa,
S.; Murai, Y.; Moriyama, H.; Imazu, T.; Ohishi, H.; Yoneda, R.;
Kurihara, T. J . Org. Chem. 1996, 61, 4405-4411.
(11) When the reaction time was prolonged, we observed the
formation of a complex mixture from which we could not isolate the
â-protected C-nucleosides.

3728 J . Org. Chem., Vol. 67, No. 11, 2002
Guianvarc’h et al.
F igu r e 4. Structure and atom numbering of the hemiacetals
(ketone form) studied by molecular modeling.
is localized in one face of the indole ring and, in the
second case, it is localized in the other face. As nucleo-
philic attack on the positive face is favored at first
because of Coulombic attraction, the face occupied by the
5-CH2OTBDPS side chain is disfavored.
Figure 5 displays the two lowest energy conformers of
6d (ES ) -139.89 kJ /mol and ER ) -139.05 kJ /mol). In
these conformations, the OdC(1)-C(2′)-N(1′) dihedral
angles are -160.4 and +176.0°, respectively, and the
C(2′)-C(1)-C(2)-C(3) dihedral angles are +165.5 and
-66.4°. These two geometries suggest that borohydride
reduction of the most stable conformer leads to diol 10d S
(Figure 5a) and that borohydride reduction of the second
lowest energy conformer leads to diol 10d R (Figure 5b).
As their calculated steric energy difference is only 0.84
kJ /mol (∆(ES - ER)), the observed mixture of diastereo-
meric diols 10d (S/R 1:1) is explained.
For the ketone form of compound 6a , the geometry
analysis of all the possible conformations within 3 kcal/
mol from the global minimum reveals that, in contrast
to 6d , because of the steric effect due to the protecting
group, there are only a few conformers having the double
bond of the carbonyl group approximately in the plane
of the bicycle and, for these conformers, the C(2′)-C(1)-
C(2)-C(3) dihedral angles vary from 150 to 180°. This
means that only one face of the bicycle is hindered by
the TBDPS group. Figure 6 displays the most stable
energy conformer having the double bond of the carbonyl
group nearly in the plane of the bicycle of 6a . In this
conformation, the OdC(1)-C(2′)-N(1′) dihedral angle is
-153.7° and the C(2′)-C(1)-C(2)-C(3) dihedral angle is
F igu r e 2. R/â configuration assignment by NOESY experi-
ments.
Molecu la r Mod elin g Stu d ies. (a ) Molecu la r Mod -
elin g Stu d ies of th e Red u ction Step (6-9 f 10-13).
The molecular modeling studies were performed on
ketones 6a ,d (indole series) and on 7a ,d (benzimidazole
series) (Figure 4) and gave similar results for both series
(the calculated energies and geometrical parameters are
summarized in Tables 4 and 7 in the Supporting Infor-
mation). For clarity, only the discussions concerning the
indole series are given in this section.
The geometry analysis of all the possible conformations
of the reactive ketone form of 6d (Figure 4) within 3 kcal/
mol from the global minimum reveals that, in all cases,
the double bond of the carbonyl group is practically in
the plane of the indole ring: the OdC(1)-C(2′)-N(1′)
dihedral angles are about -30, +30° or -150, +170°. The
approach of the hydride ion is not hampered by the indole
ring. Moreover, it is noteworthy that the C(2′)-C(1)-
C(2)-C(3) dihedral angles vary, on one hand, from about
-60 to -140° and, on the other hand, from +80 to +180°.
In the first case, the side chain bearing the TBDPS group
F igu r e 3. Structures of the obtained free C-nucleosides.

Stereocontrolled Synthesis of C-Nucleosides
J . Org. Chem., Vol. 67, No. 11, 2002 3729
F igu r e 6. Ball and stick drawing of the most stable conformer
of 6a (ConfS precursor of diol 10a S).
the â-anomer 14d â is more stable than the R-anomer
14d r; i.e., the energy difference is 7.15 kJ /mol with the
molecular mechanics method (Table 5) and 6.31 kJ /mol
with the molecular orbital method at the RHF/AM1 level
(Table 6).
Moreover, since the activation energies for conforma-
tional changes in large molecules may be considerable,
a process which can lower the rotations may be expected
to lead to a substantial acceleration of the overall rate of
the reaction. Therefore, only the possible conformations
of the intermediate 10d J that have similar torsional
angles related to the R- and â-anomers 14d were con-
sidered. The geometry analysis of these selected confor-
mations of the intermediate 10d J and the two lowest
energy conformations of the R- and â-anomers 14d
suggests that the reaction leading to R-anomer 14d r is
strongly hampered. In addition, the geometry analysis
reveals that the second lowest energy conformation of the
intermediate 10d J is comparable to the most stable
conformation of the â-anomer 14d â. The energy differ-
ence between the second lowest energy conformation and
the most stable conformation of the intermediate 10d J
is only 0.84 kJ /mol. The geometry analysis also indicates
that no conformation of the intermediate 10d J within 3
kcal/mol from the global minimum is comparable to the
most stable conformation of the R-anomer 14d r (E )
-82.59 kJ /mol) and that the most stable conformation
of the intermediate 10d J (E ) -134.19 kJ /mol) is
comparable to the conformation of the R-anomer 14d r
with steric energy equal to -80.83 kJ /mol (Table 5). The
energy difference between the R- and â-anomers 14d that
can be obtained is 8.91 kJ /mol. Moreover, the energy
difference between the constrained transition structures
TSR and TSâ (Table 5) corresponding to the cyclization
leading respectively to the R- and â-anomers 14d is
significantly important (∆E ) 11.73 kJ /mol). AM1 cal-
culations lead to similar results. The energy differences
between the R- and â-anomers 14d that can be obtained
and between the constrained transition structures TSR
and TSâ are 13.04 and 24.79 kJ /mol, respectively (Table
6). Similar results were obtained in the benzimidazole
series for the studied compounds 11d J , 15d r, and 15d â
(Figure 1) (Tables 8 and 9, see the Supporting Informa-
tion).
F igu r e 5. Ball and stick drawings of (a, top) the most stable
conformer of 6d (ConfS precursor of diol 10d S) and (b, bottom)
the second lowest energy conformer of 6d (ConfR precursor of
diol 10d R).
+156.2°. Borohydride reduction of this conformer led to
the diol 10a S. The difficulty in forming the diol 10a R is
due to the nonexistence of the conformer having the
double bond of the carbonyl group in the plane of the
indole ring with the C(2′)-C(1)-C(2)-C(3) dihedral angle
varying from about -60 to -140°.
(b) Molecu la r Mod elin g Stu d ies of th e Cycliza tion
Step . Similar results were obtained for both indole and
benzimidazole series. Therefore, only the discussions for
the indole series are given in this section.
The calculated steric energies (MM2) and the geo-
metrical parameters of the most relevant conformations
of the intermediate precursor 10d J (Figure 1) and of the
R- and â-anomers 14d are summarized in Table 5 (see
the Supporting Information). The formation enthalpies
(AM1) and the most relevant geometrical parameters of
the intermediate 10d J and of the corresponding con-
strained transition structures (TSR and TSâ) of the R-
and â-anomers 14d in the ground state are summarized
in Table 6 (see the Supporting Information). Both meth-
ods (molecular mechanics and orbital molecular methods)
lead to similar results. The calculation results reveal that

3730 J . Org. Chem., Vol. 67, No. 11, 2002
Guianvarc’h et al.
P r ep a r a tion of Hem ia ceta ls 6-9. These compounds were
obtained in pure form as an unseparable diastereomeric
mixture (in equilibrium; see text). However, the signals of the
All together, these considerations correlate nicely with
the experimental findings.
1
two diastereoisomeric hemiacetals were clearly shown by H,
13C, and 2D COSY NMR and are given in this section as M
and m , for the major and the minor diastereoisomers, respec-
tively.
Con clu sion
A short stereocontrolled approach for the synthesis of
heterocyclic C-nucleosides has been developed for both
R- and â-anomers. Our synthesis started by condensation
of a range of 2-lithiated heterocycles with ribonolactone
1 to give the hemiacetals 6-9 in good yields. We observed
that the choice of an appropriate protecting group on the
heterocycle allowed a high stereoselective borohydride
reduction of the hydroxy ketone to the corresponding diol
in quantitative yield which, upon Mitsunobu cyclization,
gave the R-anomer. In contrast, removal of the protecting
group allowed the total stereocontrol of the ring closure
of diol, leading to the â-anomer exclusively. In view of
the given examples, the strategy should hold great
promise for stereoselective synthesis of a wide range of
C-nucleosides and in particularly the purine C-nucleo-
sides.
2-[5′-O-(ter t-Bu tyld ip h en ylsilyl)-1′-h yd r oxy-2′,3′-O-iso-
p r op ylid en e-D-r ib ofu r a n osyl-1′]-1-(N,N-d im et h ylsu lfa -
m oyl)in d ole (6a ). To a stirred solution of N-(N′,N′-dimeth-
ylsulfamoyl)indole (2a ; 1.5 g, 4.4 mmol) in dry tetrahydrofuran
(50 mL) cooled to -78 °C was added LDA (2 M solution in
hexane, 2.42 mL, 1.1 equiv) dropwise. After 45 min a solution
of 5-(tert-butyldiphenylsilyl)-2,3-O-isopropylidene-D-ribono-1,4-
lactone (1; 1.5 g, 0.8 equiv) in dry THF (25 mL) was slowly
added. The mixture, which was allowed to react at room
temperature over 3 h, was quenched with a solution of
ammonium chloride (10 mL) and extracted with methylene
chloride (2 × 100 mL). The combined organic layers were dried
(MgSO4) and evaporated under reduced pressure to give an
oil. Silica gel column chromatography purification using 10%
of ethyl acetate in heptane afforded 6a as a white foam (2 g,
1
91%). Rf ) 0.50 (7:3 heptane/ethyl acetate). H NMR (CDCl3):
δ 1.04 (s, 9Hm), 1.11 (s, 9HM), 1.26 (s, 6HM), 1.41 (s, 3Hm), 1.65
(s, 3Hm), 2.80 (s, 6Hm), 2.88 (s, 6HM), 3.84 (m, 4Hm+M), 4.42
(m, 2Hm+M), 4.72 (dd, 1Hm, J ) 6.9 and 4.1 Hz), 4.82 (dd, 1HM,
J ) 5.7 and 1.1 Hz), 5.07 (d, 1HM, J ) 5.7 Hz), 5.07 (s, 1HM),
5.14 (d, 1Hm, J ) 6.9 Hz), 5.42 (s, 1Hm), 7.02 (br s, 2Hm+M),
7.19-7.70 (m, 26Hm+M), 7.89 (m, 1HM), 8.00 (m, 1Hm). 13C NMR
(CDCl3): δ 19.32, 25.54, 26.58, 26.89, 27.01, 38.49, 38.78, 63.56,
65.52, 81.41, 82.81, 82.97, 84.44, 85.50, 88.00, 100.76, 106.24,
112.01, 112.56, 114.74, 115.26, 121.37, 121.52, 122.94, 123.25,
124.43, 124.81, 127.32, 127.84, 127.97, 128.12, 129.87, 130.06,
130.15, 132.68, 133.29, 135.70, 135.74, 135.82, 137.66, 140.12,
151.00. MS (ES): m/z 673 [M + Na]+, 633 [M - H2O + H]+.
Anal. Calcd for C34H42N2O7SiS: C, 62.74; H, 6.50; N, 4.30.
Found: C, 62.37; H, 6.83; N, 4.74.
Exp er im en ta l Section
Reactions were monitored by TLC on silica gel 60 F-254,
and compounds were visualized with UV light, iodine, or 10%
sulfuric acid in ethanol followed by heating. 1H (300 MHz) and
13C (75 MHz) NMR spectra were recorded with CDCl3 or CD3-
OD as solvent and are so indicated: chemical shifts are
reported in ppm (δ) from tetramethylsilane; coupling constants
are reported in Hz. Flash chromatography was carried out on
silica gel 60 (230-400 mesh) using heptane/ethyl acetate and
methylene chloride/methanol mixtures as eluents. The dia-
stereomeric ratios were determined by reverse phase analytical
HPLC analysis using Waters Symmetry 4.6 × 250 mm column
and acetonitrile/water (75/25 v/v) as eluent. All moisture-
sensitive reactions were carried out under a nitrogen atmo-
sphere. All solvents were dried over standard drying agents
and freshly distilled prior to use.
Com p u t a t ion a l P r oced u r e. Molecu la r Mech a n ics
Meth od . Five thousand conformations of each studied com-
pound, i.e. the reactive ketone forms of the hemiacetals 6a ,d ,
the intermediate 10d J , and the R- and â-anomers of 14d , were
generated by a random search Monte Carlo method14 and
optimized by the TNCG truncated Newton molecular mechan-
ics minimization method15 using the Macromodel (version 5.5)
program16 with an MM2 force field.17 The search was carried
out on blocks of 500 Monte Carlo steps until no additional
conformation was found to be of lower energy than the current
minimun. Duplicated conformations as well as those that had
chirality changes were discarded. From these conformational
searches, all the possible conformations within 3 kcal/mol from
the global minimum were analyzed. Therefore, for each
cyclization leading to R- and â-anomers 14d , the constrained
transition structures with the forming O- - -C bond equal to
about 2.1 Å were performed.
2-[5′-O-(ter t-Bu tyld ip h en ylsilyl)-1′-h yd r oxy-2′,3′-O-iso-
p r op ylid en e-D-r ibofu r a n osyl-1′]in d ole (6d ). The lithium
salt of indole-1-carboxylic acid (2d ) was prepared from indole
(0.5 g, 4.27 mmol), following the procedure described by
Katritzky.7 To a solution of 2d in dry THF (20 mL) cooled to
-78 °C was added dropwise t-BuLi (1.5 M solution in n-
pentane, 1.2 equiv). The reaction mixture was stirred at the
same temperature for 1.5 h, and a solution of ribonolactone 1
(1 g, 0.8 equiv) in dry THF (10 mL) was slowly added. The
mixture was stirred and warmed slowly to room temperature
over 3 h and then quenched with ammonium chloride solution
(10 mL) and extracted with methylene chloride (2 × 50 mL).
The combined organic layers were dried (MgSO4) and evapo-
rated under reduced pressure to give a crude oil. Silica gel
column chromatography purification using 10% of ethyl acetate
in heptane afforded 6d as a white foam (955 mg, 75%). Rf )
1
0.44 (7:3 heptane/ethyl acetate). H NMR (CDCl3): δ 0.99 (s,
9HM), 1.10 (s, 9Hm), 1.39 and 1.48 (2s, 6Hm), 1.43 and 1.64
(2s, 6HM), 3.68 (m, 2Hm+M), 3.97 (m, 3Hm+M), 4.43 (m, 1Hm),
4.56 (dd, 1HM, J ) 6.4 and 8.7 Hz), 4.73 (d, 1Hm, J ) 5.6 Hz),
4.92 (dd, 1Hm, J ) 5.6 and 1.6 Hz), 5.48 (d, 1HM, J ) 6.4 Hz),
6.65 (m, 1Hm), 6.66 (m, 1HM), 7.04-7.74 (m, 28Hm+M), 8.58 (br
s, 1Hm), 9.35 (br s, 1HM). 13C NMR (CDCl3): δ 19.35, 25.00,
25.64, 26.62, 26.77, 26.91, 27.01, 27.35, 63.73, 64.97, 65.53,
69.88, 78.72, 80.93, 81.98, 82.65, 85.38, 86.44, 88.06, 100.09,
100.46, 101.52, 110.37, 110.73, 111.22, 112.34, 113.01, 115.77,
119.68, 119.88, 121.14, 122.14, 123.45, 126.77, 127.66, 127.84,
128.10, 129.90, 130.12, 130.24, 130.39, 133.01, 134.21, 134.99,
135.59, 135.58, 135.68, 135.89, 137.46, 139.18. MS (ES): m/z
566 [M + Na]+. Anal. Calcd for C32H37NO5Si: C, 70.69; H, 6.86;
N, 2.58. Found: C, 70.82; H, 6.81; N, 2.37.
Or bita l Molecu la r Sem iem p ir ica l Meth od . The geom-
etries for all the selected structures of the intermediate 10d J
and of the R- and â-anomers 14d summarized in Tables 4 and
5 were studied using the orbital molecular method and were
optimized by means of gradient techniques at the RHF/AM1
level18 using the MOPAC program (version 5.0).19 Therefore,
for each cyclization, the constrained transition structures were
performed by freezing the forming O- - -C bond at 2.1 Å.
2-[5′-O-(ter t-Bu tyld ip h en ylsilyl)-1′-h yd r oxy-2′,3′-O-iso-
p r op ylid en e-D-r ibofu r a n osyl-1′]ben zim id a zole (7d ). A so-
(14) Chang, G.; Guida, W. C.; Still, W. C. J . Am. Chem. Soc. 1989,
111, 4379-4386.
(15) Ponder, J . W.; Richards, F. M. J . Comput. Chem. 1987, 8, 1016.
(16) Mohamadi, F.; Richards, N. J . G.; Guida, W. C.; Liskamp, R.;
Lipton, M. C.; Caufield, M.; Chang, G.; Hendrickson, T.; Still, W. C. J .
Comput. Chem. 1990, 11, 440.
(18) Dewar, M. J . S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J . J . P.
J . Am. Chem. Soc. 1985, 107, 3902-3909.
(19) Stewart, J . J . P. MOPAC 5.0, QCPE Program No. 455 and J .
Comput. Chem. 1989, 101, 221.
(17) Allinger, N. L. J . Am. Chem. Soc. 1977, 99, 8127-8134.

Stereocontrolled Synthesis of C-Nucleosides
J . Org. Chem., Vol. 67, No. 11, 2002 3731
lution of 7e (400 mg, 0.6 mmol) in EtOH/THF (1/1 v/v, 8 mL)
was hydrogenated over 10% palladium on carbon (40 mg) for
15 h at 50 psi. The catalyst was removed by filtration through
Celite and washed several times with ethanol. The filtrate was
evaporated to give a crude residue, which was purified by silica
gel column chromatography, using 20% of ethyl acetate in
heptane, to afford 7d as a white foam (293 mg, 90%). Rf )
Calcd for C32H39NO5Si: C, 70.43; H, 7.20; N, 2.57. Found: C,
70.71; H, 6.95; N, 2.41.
Diol 11d . By the same procedure as described above for the
preparation of 10a , 7d (400 mg, 0.73 mmol) was reduced with
sodium borohydride to give an inseparable mixture of diaste-
reoisomers 11d (S/R 55/45; 380 mg, 95%) as a white foam. Rf
) 0.50 (1:1 heptane/ethyl acetate). 1H NMR (CDCl3): δ 1.08
(s, 9H), 1.09 (s, 9H), 1.32 (s, 3H), 1.36 (s, 3H), 1.37 (s, 3H),
1.46 (s, 3H), 3.87 (dd, 1H, J ) 10.6 and 5.7 Hz), 3.95 (m, 2H),
3.98 (dd, 1H, J ) 10.6 and 2.5 Hz), 4.07 (m, 2H), 4.45 (m, 2H),
4.50 (dd, 1H, J ) 8.5 and 5.3 Hz), 4.56 (dd, 1H, J ) 5.3 and
5.5 Hz), 5.17 (d, 1H, J ) 8.5 Hz), 5.30 (d, 1H, J ) 5.3 Hz),
7.26 (m, 4H), 7.40 (m, 10H), 7.60 (m, 4H), 7.72 (m, 10H). 13C
NMR (CDCl3): δ 19.46, 25.33, 25.62, 26.99, 27.55, 28.12, 65.51,
65.66, 66.09, 66.67, 69.94, 69.70, 76.88, 77.37, 78.73, 79.25,
109.18, 122.98, 127.83, 127.89, 129.86, 133.38, 135.78, 154.21.
MS (ES): m/z 547 [M + H]+. Anal. Calcd for C31H38N2O5Si: C,
68.10; H, 7.01; N, 5.12. Found: C, 67.93; H, 6.86; N, 4.82.
P r ep a r a tion of Cyclized P r od u cts 14-22. 2-[5′-O-(ter t-
Bu tyld ip h en ylsilyl)-2′,3′-O-isop r op ylid en e-r-D-r ibofu r a -
n osyl-1′]-1-(N,N-d im eth ylsu lfa m oyl)in d ole (14a r). By the
typical procedure of the Mitsunobu reaction, 10a S (155 mg,
0.23 mmol) and PPh3 (1.5 equiv, 93 mg) were dissolved in THF
and the mixture was refluxed. Diethylazodicarboxylate (1.5
equiv, 50 mL) was then added dropwise to the refluxing
mixture, and heating was continued for 1 h. The solvent was
evaporated to give a residual oil, which was chromatographed
on a silica gel column using 5% of ethyl acetate in heptane to
give 14a r (123 mg, 85%) as a white foam. Rf ) 0.46 (7:3
1
0.25 (4:1 heptane/ethyl acetate). H NMR (CDCl3): δ 1.03 (s,
9Hm), 1.12 (s, 9HM), 1.25 (s, 3HM), 1.36 (s, 3HM), 1.37 (s, 3Hm),
1.66 (s, 3Hm), 3.74 (dd, 1Hm, J ) 5.1 and 11.6 Hz), 3.82 (dd,
1Hm, J ) 3.8 and 11.6 Hz), 3.92 (dd, 1HM, J ) 5.3 and 10.6
Hz), 4.07 (dd, 1HM, J ) 7.8 and 10.6 Hz), 4.44 (m, 2Hm+M),
4.85 (d, 1Hm, J ) 6.8 Hz), 4.87 (d, 1HM, J ) 5.6 Hz), 4.92 (d,
1HM, J ) 5.6 Hz), 5.24 (d, 1Hm, J ) 6.8 Hz), 7.12-7.93 (m,
28Hm+M), 9.30 (s, 1HM), 9.63 (s, 1Hm). 13C NMR (CDCl3): δ
19.43, 24.73, 25.10, 26.65, 26.98, 27.13, 63.63, 65.62, 81.60,
83.08, 83.80, 86.95, 87.58, 100.03, 104.11, 112.77, 128.01,
130.05, 133.25, 135.77, 151.89. MS (ES): m/z 567 [M + Na]+,
545 [M + H]+, 527 [M - H2O + H]+. Anal. Calcd for
C31H36N2O5Si: C, 68.35; H, 6.66; N, 5.14. Found: C, 68.69; H,
6.45; N, 4.86.
P r ep a r a tion of Diols 10-13. Diol 10a . Compound 6a (700
mg, 1.07 mmol) was dissolved in dry methanol (10 mL), and
sodium borohydride (100 mg, 2.5 equiv) was added slowly. The
mixture was stirred at room temperature for 1 h, quenched
with ice-water, and neutralized with HCl solution (0.1 N).
The aqueous phase was extracted with ethyl acetate (2 × 75
mL). The combined organic layers were dried (MgSO4) and
evaporated under reduced pressure to give a crude solid. Silica
gel column chromatography purification using 5% of ethyl
acetate in heptane afforded 10a S (513 mg) and 10a R (169 mg)
as a white foam (S/R: 75/25; 97%). 10a S: Rf ) 0.50 (7:3
1
heptane/ethyl acetate). H NMR (CDCl3): δ 1.11 (s, 9H), 1.24
(s, 3H), 1.26 (s, 3H), 2.73 (s, 6H), 3.75 (dd, 1H, J ) 3.2 and
11.2 Hz), 3.92 (dd, 1H, J ) 3.4 and 11.2 Hz), 4.35 (m, 1H),
4.94 (d, 1H, J ) 6.0 Hz), 5.25 (t, 1H, J ) 3.2 Hz), 5.84 (d, 1H,
J ) 4.2 Hz), 6.87 (s, 1H), 7.30 (m, 2H), 7.40 (m, 7H), 7.53 (m,
1H), 7.67 (m, 3H), 7.90 (m, 1H). 13C NMR (CDCl3): δ 19.30,
25.13, 26.12, 27.01, 38.38, 66.05, 80.57, 82.90, 83.48, 83.85,
108.93, 112.47, 114.41, 120.78, 122.95, 123.63, 127.95, 129.22,
129.93, 130.05, 133.02, 135.60, 135.73, 137.07, 138.83. MS
(ES): m/z 657 [M + Na]+.
1
heptane/ethyl acetate). H NMR (CDCl3): δ 1.08 (s, 9H), 1.17
(s, 3H), 1.32 (s, 3H), 2.85 (s, 6H), 3.61 (br s, 1H), 3.83 (dd, 1H,
J ) 10.3 and 6.8 Hz), 3.98 (dd, 1H, J ) 2.9 and 10.3 Hz), 4.03
(m, 1H), 4.29 (dd, 1H, J ) 9.4 and 5.1 Hz), 4.50 (br s, 1H),
4.65 (dd, 1H, J ) 9.6 and 5.1 Hz), 5.64 (d, 1H, J ) 9.6 Hz),
6.80 (s, 1H), 7.25 (m, 3H), 7.39 (m, 5H), 7.54 (m, 1H), 7.65 (m,
4H), 8.01 (d, 1H, J ) 7.2 Hz). 13C NMR (CDCl3): δ 19.43, 25.64,
27.01, 27.92, 38.50, 64.01, 65.49, 69.88, 77.65, 79.40, 108.67,
109.18, 114.96, 121.31, 123.13, 124.53, 127.93, 128.71, 130.00,
133.03, 135.69, 141.18. MS (ES): m/z 698 [M + 2Na]+, 675
[M + Na]+, 635 [M - H2O + H]+. 10a R: Rf ) 0.45 (7:3 heptane/
2-[5′-O-(ter t-Bu tyld iph en ylsilyl)-2′,3′-O-isopr op ylid en e-
â-D-r ibofu r a n osyl-1]in d ole (14d â). 10d S (150 mg, 0.27
mmol) and 10d R (150 mg, 0.27 mmol) were independently
cyclized, by the same procedure as described for the prepara-
tion of 14a r, to give the same compound 14d â (92% from 10d S
and, 90% from 10d R) as a white foam. Rf ) 0.50 (7:3 heptane/
1
ethyl acetate). H NMR (CDCl3): δ 1.09 (s, 9H), 1.34 (s, 3H),
1.46 (s, 3H), 2.81 (s, 6H), 3.28 (br s, 1H), 3.89 (m, 2H), 4.27
(m, 2H), 4.70 (d, 1H, J ) 6.5 Hz), 5.89 (br s 1H), 6.85 (s, 1H),
7.25 (m, 3H), 7.40 (m, 5H), 7.53 (m, 1H), 7.68 (m, 4H), 8.01 (d,
1H, J ) 7.0 Hz). 13C NMR (CDCl3): δ 19.45, 24.71, 26.98, 38.54,
64.82, 65.45, 70.01, 76.54, 78.56, 108.72, 108.82, 115.02,
121.07, 123.26, 124.24, 127.85, 127.90, 128.94, 129.94, 133.04,
135.66, 135.72, 142.82. MS (ES): m/z 698 [M + 2Na]+, 675
[M + Na]+. Anal. Calcd for 10a S, C34H44N2O7SSi: C, 62.55; H,
6.79; N, 4.29. Found: C, 62.45; H, 6.74; N, 4.25.
1
ethyl acetate). H NMR (CDCl3): δ 1.11 (s, 9H), 1.37 (s, 3H),
1.62 (s, 3H), 3.73 (dd, 1H, J ) 11.4 and 4.8 Hz), 3.85 (dd, 1H,
J ) 3.5 and 11.4 Hz), 4.30 (m, 1H), 4.77 (dd, 1H, J ) 2.8 and
6.3 Hz), 4.83 (dd, 1H, J ) 3.6 and 6.3 Hz), 5.24 (d, 1H, J ) 3.6
Hz), 6.44 (m, 1H), 6.97 (m, 1H), 7.08 (m, 2H), 7.30-7.47 (m,
6H), 7.55 (m, 1H), 7.75 (m, 4H), 8.50 (br s, 1H). 13C NMR
(CDCl3): δ 19.48, 25.54, 27.07, 27.49, 64.30, 81.38, 81.93, 85.21,
86.94, 111.10, 114.03, 119.83, 120.53, 121.93, 128.04, 130.09,
132.91, 133.14, 135.57, 135.69, 135.98, 137.36. MS (FAB): m/z
528 [M + H]+. Anal. Calcd for C32H37NO4Si: C, 72.83; H, 7.07;
N, 2.65. Found: C, 72.74; H, 7.05; N, 2.61.
Diol 10d . By the same procedure as described for the
preparation of 10a , 6d (410 mg, 0.75 mmol) was reduced with
sodium borohydride to give 10d (195 mg, F1, R or S) and 10d
(201 mg, F2, S or R) as a white foam (S/R: 1/1; 96%). 10d (F1):
2-[5′-O-(ter t-Bu tyld iph en ylsilyl)-2′,3′-O-isopr op ylid en e-
â-D-r ibofu r a n osyl-1′]ben zim id a zole (15d â). 11d (S/R mix-
ture 55/45; 460 mg, 0.84 mmol) was cyclized by the same
procedure as described for the preparation of 14a r, to give
15d â (401 mg, 90%) as a white foam. Rf ) 0.40 (1:1 heptane/
1
Rf 0.52 (7:3 heptane/ethyl acetate). H NMR (CDCl3): δ 1.08
(s, 9H), 1.28 (s, 3H), 1.36 (s, 3H), 3.02 (br s, 1H), 3.17 (br s,
1H), 3.81 (dd, 1H, J ) 10.5 and 5.3 Hz), 3.92 (dd, 1H, J ) 3.0
and 10.5 Hz), 4.10-4.23 (m, 2H), 4.52 (dd, 1H, J ) 2.6 and
6.0 Hz), 5.36 (m, 1H), 6.47 (s, 1H), 7.12 (m, 2H), 7.35 (m, 5H),
7.65 (m, 7H), 8.84 (br s, 1H). 13C NMR (CDCl3): δ 19.42, 24.92,
26.98, 27.07, 65.37, 65.76, 69.74, 76.45, 79.78, 99.97, 108.93,
111.12, 119.75, 120.72, 121.93, 127.91, 127.96, 128.11, 130.04,
132.93, 135.65, 136.055, 139.25. MS (ES): m/z 568 [M + Na]+,
546 [M + H]+. 10d (F2): Rf ) 0.50 (7:3 heptane/ethyl acetate).
1H NMR (CDCl3): δ 1.08 (s, 9H), 1.27 (s, 6H), 3.78 (dd, 1H, J
) 10.5 and 7.4 Hz), 3.98 (m, 2H), 4.17 (dd, 1H, J ) 5.2 and
9.7 Hz), 4.34 (dd, 1H, J ) 9.2 and 5.2 Hz), 5.13 (d, 1H, J ) 9.2
Hz), 6.56 (s, 1H), 7.14 (m, 2H), 7.33 (m, 5H), 7.65 (m, 7H),
8.61 (br s, 1H). 13C NMR (CDCl3): δ 19.41, 25.71, 27.03, 28.13,
65.39, 66.43, 69.67, 77.58, 80.61, 100.17, 109.41, 111.03,
119.66, 120.68, 121.66, 128.04, 128.50, 130.23, 135.70, 135.93,
138.84. MS (ES): m/z 568 [M + Na]+, 546 [M + H]+. Anal.
1
ethyl acetate). H NMR (CDCl3): δ 1.08 (s, 9H), 1.35 (s, 3H),
1.60 (s, 3H), 3.62 (dd, 1H, J ) 5.9 and 11.6 Hz), 3.84 (dd, 1H,
J ) 3.6 and 11.6 Hz), 4.38 (m, 1H), 4.67 (dd, 1H, J ) 6.1 and
2.8 Hz), 5.17 (dd, 1H, J ) 2.8 and 2.8 Hz), 5.40 (d, 1H, J ) 2.8
Hz), 7.00 (m, 1H), 7.13-7.48 (m, 8H), 7.74 (m, 1H), 7.62 (m,
4H), 9.85 (s, 1H). 13C NMR (CDCl3): δ 19.22, 25.26, 26.91,
27.14, 64.34, 80.08, 82.25, 85.02, 86.49, 114.34, 114.86, 125.00,
128.00, 130.11, 132.71, 133.58, 139.70, 151.60. MS (ES): m/z
529 [M + H]+. Anal. Calcd for C31H36N2O4Si: C, 70.42; H, 6.86;
N, 5.30. Found: C, 70.51; H, 6.78; N, 5.11.
2-(5′-Hydr oxy-2′,3′-O-isopr opyliden e-r-D-r ibofu r an osyl-
1′)ben zim id a zole (18d r). A solution of 15er (220 mg, 0.34
mmol) in 1:1 EtOH/THF (v/v; 5 mL) was hydrogenated over
10% palladium on carbon (22 mg) for 15 h at 50 psi. The

3732 J . Org. Chem., Vol. 67, No. 11, 2002
Guianvarc’h et al.
catalyst was removed by filtration through Celite and washed
several times with ethanol. The filtrate was evaporated to give
a crude residue which was dissolved in THF (5 mL). After
addition of 1 M TBAF in THF (0.5 mL, 1.5 equiv), the reaction
mixture was stirred at room temperature for 1 h and then
evaporated. The residue was purified by column chromatog-
raphy, using 50% of ethyl acetate in heptane, to afford 18d r
as a white foam (84 mg, 85%). Rf ) 0.20 (1:4 heptane/ethyl
acetate). 1H NMR (CDCl3): δ 1.35 (s, 3H), 1.55 (s, 3H), 3.89
(m, 1H), 3.97 (dd, 1H, J ) 3.7 and 12.0 Hz), 4.06 (dd, 1H, J )
6.3 and 12.0 Hz), 4.83 (dd, 1H, J ) 3.8 and 6.0 Hz), 5.41 (s,
7.43 (d, 1H, J ) 7.7 Hz). 13C NMR (CD3OD): δ 63.18, 72.49,
77.41, 80.71, 86.02, 100.79, 111.92, 120.08, 120.99, 123.32,
129.42, 137.88, 138.39. MS (IC, isobut): m/z 250 [M + H]+.
Anal. Calcd for C13H15NO4: C, 62.64; H, 6.07; N, 5.62. Found:
C, 62.75; H, 6.29; N, 5.38.
2-r-D-Ribofu r a n osylben zim id a zole (20r). Treatment of
15a r (200 mg, 0.31 mmol) under the same conditions as
described for 19â gave 20r (71 mg, 90%). Rf ) 0.25 (9:1
1
methylene chloride/methanol). H NMR (CD3OD): δ 3.90 (td,
2H, J ) 11.5, 6.4, and 4.8 Hz), 4.31 (m, 1H), 4.35 (m, 1H),
4.40 (m, 1H), 5.21 (d, 1H, J ) 8.0 Hz), 7.45 (m, 2H), 7.68 (m,
2H). 13C NMR (CD3OD): δ 62.24, 73.52, 77.83, 79.39, 84.68,
115.49, 126.95. MS (IC, isobut): m/z 251 [M + H]+.
1H), 5.69 (d, 1H, J ) 3.8 Hz), 7.24 (m, 4H), 7.58 (br s, 1H). 13
C
NMR (CDCl3): δ 24.52, 26.09, 61.30, 79.98, 81.35, 81.65, 84.50,
113.10, 123.01, 151.19. MS (IC, isobut): m/z )291 [M + H]+.
2-(5′-Hydr oxy-2′,3′-O-isopr opyliden e-â-D-r ibofu r an osyl-
1′)ben zim id a zole (18d â). 15d â (210 mg; 0.39 mmol) was
dissolved in THF (5 mL), and a 1 M TBAF solution in THF
(0.6 mL, 1.5 equiv) was added. The reaction mixture was
stirred at room temperature for 1 h and then evaporated. The
residue was purified by column chromatography, using 50%
of ethyl acetate in heptane, to afford 18d â as a white foam
(112 mg, 96%). Rf ) 0.25 (1:4 heptane/ethyl acetate). 1H NMR
(CDCl3): δ 1.34 (s, 3H), 1.58 (s, 3H), 3.77 (dd, 1H, J ) 12.3
and 3.3 Hz), 4.03 (dd, 1H, J ) 2.1 and 12.3 Hz), 4.48 (m, 1H),
4.95 (dd, 1H, J ) 2.3 and 6.1 Hz), 5.02 (dd, 1H, J ) 3.1 and
6.0 Hz), 5.30 (d, 1H, J ) 3.1 Hz), 7.27 (m, 4H), 7.58 (br s, 1H).
13C NMR (CDCl3): δ 25.38, 27.38, 63.25, 82.09, 82.44, 86.25,
87.20, 113.90, 123.13, 143.58, 153.90. MS (ES): m/z 313 [M +
Na]+, 291 [M + H]+.
2-â-D-Ribofu r a n osylben zim id a zole (20â). By the same
procedure as described above for the preparation of 19â, 15d â
(250 mg, 0.47 mmol) was converted to 20â (108 mg, 92%). Rf
1
) 0.31 (9:1 methylene chloride/methanol). H NMR (CD3OD):
δ 3.77 (dd, 1H, J ) 3.6 and 12.2 Hz), 3.95 (dd, 1H, J ) 2.9 and
12.2 Hz), 4.08 (m, 1H), 4.19 (m, 1H), 4.24 (m, 1H), 5.06 (d,
1H, J ) 5.0 Hz), 7.26 (m, 2H), 7.56 (m, 2H). 13C NMR (CD3-
OD): δ 62.61, 72.09, 77.66, 80.78, 86.27, 115.85, 123.76. MS
(IC, isobut): m/z 251 [M + H]+.
2-r-D-Ribofu r a n osyl-5-iod oben zim id a zole (21r). By the
same procedure as described above for the preparation of 19â,
16a r (300 mg, 0.4 mmol) was converted to 21r (127 mg, 86%).
1
Rf ) 0.30 (9:1 methylene chloride/methanol). H NMR (CD3-
OD): δ 3.86 (2dd, 2H, J ) 11.6, 6.1, and 4.8 Hz), 4.31 (m, 1H),
4.38 (m, 1H), 4.44 (dd, 1H, J ) 4.4 and 7.2 Hz), 5.02 (d, 1H, J
) 7.2 Hz), 7.35 (d, 1H, J ) 8.5 Hz), 7.52 (dd, 2H, J ) 1.8 and
8.5 Hz), 7.91 (br s, 1H). 13C NMR (CD3OD): δ 62.03, 73.27,
78.64, 79.19, 83.20, 117.02, 132.39, 156.68. MS (ES): m/z 377
[M + H]+. Anal. Calcd for C12H13IN2O4: C, 38.32; H, 3.48; N,
7.45. Found: C, 38.59; H, 3.70; N, 7.72.
2-r-D-Ribofu r a n osylin d ole (19r). 14a r (150 mg, 0.23
mmol) was dissolved in EtOH (3 mL), and a solution of TFA/
H2O (4/1, 3 mL) was added. The reaction mixture was stirred
for 2 h at 60 °C, cooled to 0 °C, and neutralized with NaHCO3
solution. The mixture was evaporated, and the crude residue
was purified by silica gel column chromatography using 5%
of methanol in methylene chloride, to afford 19r (48 mg, 82%).
2-r-D-Ribofu r a n osylim id a zole (22r). By the same pro-
cedure as described above, 17a r (585 mg, 1 mmol) was
converted to 22r (165 mg, 83%). Rf ) 0.15 (9:1 methylene
1
1
Rf ) 0.40 (9:1 methylene chloride/methanol). H NMR (CD3-
chloride/methanol). H NMR (CD3OD): δ 3.85 (2dd, 2H, J )
OD): δ 3.64 (dd, 1H, J ) 12.0 and 4.5 Hz), 3.68 (dd, 1H, J )
2.3 and 12.0 Hz), 4.01 (m, 1H), 4.29 (dd, 1H, J ) 8.6 and 4.5
Hz), 4.56 (dd, 1H, J ) 3.4 and 4.5 Hz), 5.70 (dd, 1H, J ) 1.3
and 3.4 Hz), 6.83 (s, 1H), 7.38 (t, 1H, J ) 7.2 Hz), 7.50 (t, 1H,
J ) 7.3 Hz), 7.74 (d, 1H, J ) 7.2 Hz), 7.96 (d, 1H, J ) 7.9 Hz).
13C NMR (CD3OD): δ 62.99, 73.57, 74.63, 79.89, 83.13, 112.88,
115.56, 121.80, 124.83, 125.15, 131.45, 138.55, 139.59. MS (IC,
isobut): m/z 250 [M + H]+. Anal. Calcd for C13H15NO4: C,
62.64; H, 6.07; N, 5.62. Found: C, 62.89; H, 6.32; N, 5.86.
2-â-D-Ribofu r an osylin dole (19â). 14dâ (64 mg, 0.26 mmol)
was dissolved in dioxane (3 mL), and 2 N HCl (0.5 mL) was
added. The reaction mixture was stirred at room temperature
for 2 h, cooled to 0 °C, and neutralized by a saturated solution
of NaHCO3. The mixture was evaporated, and the crude
residue was purified by silica gel column chromatography
using 5% of methanol in methylene chloride, to afford 19â as
an oil (59 mg, 91%). Rf ) 0.50 (9:1 methylene chloride/
methanol). 1H NMR (CD3OD): δ 3.68 (dd, 1H, J ) 12.0 and
4.1 Hz), 3.85 (dd, 1H, J ) 3.1 and 12.0 Hz), 3.96 (m, 1H), 4.10
(m, 2H), 4.90 (d, 1H, J ) 5.3 Hz), 6.38 (s, 1H), 6.93 (t, 1H, J
) 7.3 Hz), 7.02 (t, 1H, J ) 7.4 Hz), 7.30 (d, 1H, J ) 7.9 Hz),
11.8, 6.5, and 4.7 Hz), 4.32 (m, 2H), 4.43 (m, 1H), 5.02 (d, 1H,
J ) 7.5 Hz), 7.50 (s, 2H). 13C NMR (CD3OD): δ 61.72, 72.75,
75.99, 78.56, 83.93, 120.41, 148.31. MS (ES): m/z 201 [M +
H]+. Anal. Calcd for C8H12N2O4: C, 48.00; H, 6.04; N, 13.99.
Found: C, 47.76; H, 6.31; N, 13.82.
Ack n ow led gm en t. We are grateful to the CNRS for
a doctoral fellowship (BDI) to D.G., to M. T. Adeline for
HPLC analyses, and to Dr. P. Vierling for stimulating
discussions.
Su p p or tin g In for m a tion Ava ila ble: Molecular modeling
results (Tables 4-9), copies of MS and NMR spectra (1H, 13C,
1
2D COSY H-1H and 1H-13C) of compounds 3e, 4a , 7a , 7b,
9a , 10a , 11a , 14a r, 15a r, 18d r, 18d â, 20r, and 20â, and text
giving experimental procedures and characterization data for
2-5, 6b, 7a ,b,e, 8a , 9a , 10b, 11a ,b,e, 12a , 13a , 14br, 15a r,
15br, 15er, 16a r, and 17a r. This material is available free
of charge via the Internet at http://pubs.acs.org.
J O016345X