Cycloaddition of cyclopentadiene to 3-deoxy-1,2:5,6-di-O-isopropylidene-α-D-erythro-hex-3-enofuranose. Synthesis and representative chemistry of 1,6-anhydro-2,3-dideoxy-β-D-glycero-hex-2-enopyran-4-ulose ("isolevoglucosenone")

Article abstract of DOI:10.1021/jo960032y

Treatment of D-glucose-derived alkene 4 with cyclopentadiene in the presence of a Lewis acid results in the formation of cycloaddition products 8-11. Evidence is presented to show that these 1,6-anhydro sugar-cyclopentadiene adducts do not arise from rearrangement of 4 to isolevoglucosenone (5) followed by cycloaddition but are the result of Lewis acid-catalyzed rearrangement of alkene 4 to acyclic dienophile 12 followed by addition of cyclopentadiene. Major cycloadduct 8 has been utilized as a source of the enantiomerically pure carbocycles 14-25 by manipulation of the alkene and ketone functions and cleavage of the 1,6-anhydro bridge. In the absence of diene, alkene 4 undergoes rearrangement to enone 5 in 32% yield. Reaction of 5 with several dienes results only in the formation of "bottom-face" adducts 10, 11, 28, and 29, and conjugate addition of either HN₃ or Me₃COOH is found to be completely stereoselective to afford 30 and 31, respectively. Subsequent manipulation of azide 30 leads to precursors of several naturally occurring 2-amino-2,3-dideoxy sugars.

Full text of DOI:10.1021/jo960032y

J . Org. Chem. 1996, 61, 3783-3793
3783
Cycloa d d ition of Cyclop en ta d ien e to
3-Deoxy-1,2:5,6-d i-O-isop r op ylid en e-r-D-er yth r o-h ex-3-en ofu r a n ose.
Syn th esis a n d Rep r esen ta tive Ch em istr y of
1,6-An h yd r o-2,3-d id eoxy-â-D-glycer o-h ex-2-en op yr a n -4-u lose
(“Isolevoglu cosen on e”)
Derek Horton,* J ames P. Roski,† and Peter Norris‡
Department of Chemistry, American University, 4400 Massachusetts Avenue, NW,
Washington, D.C. 20016-8014
Received J anuary 3, 1996X
Treatment of D-glucose-derived alkene 4 with cyclopentadiene in the presence of a Lewis acid results
in the formation of cycloaddition products 8-11. Evidence is presented to show that these 1,6-
anhydro sugar-cyclopentadiene adducts do not arise from rearrangement of 4 to isolevoglucosenone
(5) followed by cycloaddition but are the result of Lewis acid-catalyzed rearrangement of alkene 4
to acyclic dienophile 12 followed by addition of cyclopentadiene. Major cycloadduct 8 has been
utilized as a source of the enantiomerically pure carbocycles 14-25 by manipulation of the alkene
and ketone functions and cleavage of the 1,6-anhydro bridge. In the absence of diene, alkene 4
undergoes rearrangement to enone 5 in 32% yield. Reaction of 5 with several dienes results only
in the formation of “bottom-face” adducts 10, 11, 28, and 29, and conjugate addition of either HN3
or Me3COOH is found to be completely stereoselective to afford 30 and 31, respectively. Subsequent
manipulation of azide 30 leads to precursors of several naturally occurring 2-amino-2,3-dideoxy
sugars.
The use of natural monosaccharides for the synthesis
of enantiomerically pure compounds has been of sus-
tained interest1 since many sugars are inexpensive, are
available with a variety of relative and absolute stereo-
chemistries, and are capable of undergoing a wide range
of synthetic transformations.2 Of recent interest has
been the development of radical methodologies3 and
cycloaddition chemistry4 in the carbohydrate field as
routes to configurationally pure carbo- and heterocycles.
Many examples of cycloaddition reactions of sugar-
derived dienophiles to dienes, and conversely of sugar-
derived dienes to dienophiles, have been reported in
recent years.5 We have shown6 that [4 + 2] cycloaddition
of cyclopentadiene to sugar-derived acyclic enoates is a
predictable and useful method for the synthesis of
configurationally well-defined cyclopentane derivatives.7
In this paper, we detail the unusual rearrangement-
cycloaddition chemistry of a D-glucose-derived dienophile
with cyclopentadiene, which results in the formation of
1,6-anhydrohexopyranose-cyclopentadiene adducts, a
practical preparation of “isolevoglucosenone”, and several
aspects of the chemistry of this versatile sugar enone.
* Author to whom correspondence should be addressed.
† Present address: Lubrizol Corporation, 29400 Lakeland Blvd,
Wickliffe, OH 44092.
‡ Present address: Department of Chemistry, Youngstown State
University, Youngstown, OH 44555.
Resu lts a n d Discu ssion
X Abstract published in Advance ACS Abstracts, May 1, 1996.
(1) For general references, see: Hanessian, S. Total Synthesis of
Natural Products: The Chiron Approach; Pergammon: New York,
1983. Inch, T. D. Tetrahedron 1984, 40, 3161. Fraser-Reid, B.;
Anderson, R. C. Carbohydrate Derivatives in the Asymmetric Synthesis
of Natural Products. Fortschr. Chem. Org. Naturst. 1980, 1. Licht-
enthaler, F. W. Pure Appl. Chem. 1978, 50, 1342. For selected
applications in the total synthesis of natural products, see: Wu, W.-
L.; Wu, Y.-L. J . Org. Chem. 1993, 58, 2760. McIntosh, M. C.; Weinreb,
S. M. J . Org. Chem. 1993, 58, 4823. Hanessian, S.; Sahoo, S. P.; Botta,
M. Tetrahedron Lett. 1987, 28, 1143, 1147, 1151. Li, W.-R.; Han, S.-
Y.; J oullie´, M. M. Tetrahedron Lett. 1992, 33, 3595. Bruns, W.; Horns,
F.; Redlich, H. Synthesis 1995, 335. Chen, S.-H.; Horvath, R. F.; J oglar,
J .; Fisher, M. J .; Danishefsky, S. J . J . Org. Chem. 1991, 56, 5834.
Nicolaou, K. C.; Nugiel, D. A.; Couladouros, E.; Hwang, C.-K. Tetra-
hedron 1990, 46, 4517. Nicolaou, K. C.; Tiebes, J .; Theodorakis, E. A.;
Rutjes, F. P. J . T.; Sato, M.; Unterstellar, E. J . Am. Chem. Soc. 1994,
116, 9371.
(2) Lichtenthaler, F. W. In Pyranoid Chiral Building Blocks: Prepa-
ration from Monosaccharides and Utilization for Natural Product
Synthesis; Rahman, A. U., Ed.; Springer-Verlag: Berlin-Heidelberg,
1986, p 227.
(3) Giese, B.; Dupuis, J . Angew. Chem., Int. Ed. Engl. 1983, 22, 622.
Addlington, R. M.; Baldwin, J . E.; Basak, A.; Kozyrod, R. P. J . Chem.
Soc., Chem. Commun. 1983, 944. Baumberger, F.; Vasella, A. Helv.
Chim. Acta 1983, 66, 2210. Keck, G. E.; Enholm, E. J . Tetrahedron
1985, 41, 4074. Kobertz, W. R.; Bertozzi, C. R.; Bednarski, M. D.
Tetrahedron Lett. 1992, 33, 737. Tsang, R.; Fraser-Reid, B. J . Am.
Chem. Soc. 1986, 108, 2116. Rajan Babu, T. V.; Fukunaga, T.; Reddy,
G. S. J . Am. Chem. Soc. 1989, 111, 1759.
1,2:5,6-Di-O-isopropylidene-R-D-glucofuranose (1) has
enjoyed myriad applications in carbohydrate chemistry.
Oxidation of the 3-hydroxyl group and reaction of the
resultant 3-keto sugar with nucleophiles can lead to a
wide variety of products having the D-gluco or D-allo
configurations.8 The 3-hydroxyl group is also readily
converted into a potential leaving group through reaction
(5) Kunz, H.; Mu¨ller, B.; Schwanzenbach Angew. Chem., Int. Ed.
Engl. 1987, 26, 267. Gupta, R. C.; Raynor, P. A. C. M.; Stoodley, R. J .
J . Chem. Soc., Perkin Trans. 1 1988, 1773. Horton, D.; Roski, J . P. J .
Chem. Soc., Chem. Commun. 1992, 759. Bhate´, P.; Horton, D. Carbo-
hydr. Res. 1983, 122, 189. Sun, K.-M.; Giuliano, R. M.; Fraser-Reid,
B. J . Org. Chem. 1985, 50, 4774. Giuliano, R. M.; Buzby, J . H.;
Marcopulos, N. J . Org. Chem. 1990, 55, 3555. Lubineau, A.; Queneau,
Y. Tetrahedron 1989, 45, 6697. Bernet, B.; Vasella, A. Helv. Chim. Acta
1979, 62, 1990, 2400, 2411. Lopez, J . C.; Lameignere, E.; Lukacs, G.
J . Chem. Soc., Chem. Commun. 1988, 514.
(6) Horton, D.; Machinami, T. J . Chem. Soc., Chem. Commun. 1981,
88. Horton, D.; Machinami, T.; Takagi, Y.; Bergmann, C.; Cristoph,
G. G. J . Chem. Soc., Chem. Commun. 1983, 1164. Horton, D.;
Machinami, T.; Takagi, Y. Carbohydr. Res. 1983, 121, 135. Horton,
D.; Usui, T. Carbohydr. Res. 1991, 216, 33.
(7) Horton, D.; Koh, D. Carbohydr. Res. 1993, 250, 231, 249. Horton,
D.; Koh, D.; Takagi, Y. Carbohydr. Res. 1993, 250, 261.
(8) Baker, D. C.; Horton, D.; Tindall , C. G., J r. In Methods in
Carbohydrate Chemistry; Whistler, R. L., BeMiller, J . N., Eds.;
(4) Horton, D.; Koh, D.; Takagi, Y.; Usui, T. ACS Symp. Ser. 1992,
494, 66.
Academic Press: New York, 1976; Vol. VII, p
contained therein.
3 and references
S0022-3263(96)00032-1 CCC: $12.00 © 1996 American Chemical Society

3784 J . Org. Chem., Vol. 61, No. 11, 1996
Horton et al.
with, for example, triflic anhydride9 or p-toluenesulfonyl
chloride (see compound 2);10 however, displacement of
these sulfonic esters with nucleophiles such as azide11
or ammonia12 is seldom a clean process.13 On account of
steric (and possibly electronic) interference from the
adjacent 1,2-O-isopropylidene ring, bimolecular displace-
ment at C-3 of such derivatives as 2 is retarded, and
reactions with nucleophiles generally lead to varying
amounts of displacement product 3 as well as the product
of bimolecular elimination, namely, 3-deoxy-1,2:5,6-di-
O-isopropylidene-R-D-erythro-hex-3-enofuranose (4). We
were intrigued by the possibility of inverse-demand
Diels-Alder cycloaddition chemistry14 involving 4 as a
source of unusual carbocycles.
Treatment of a dichloromethane solution of 4 and freshly
distilled cyclopentadiene with a catalytic amount of this
oxidant for 15 min indeed resulted in consumption of 4;
however, TLC analysis showed a complex mixture of
products, along with much polymeric material near the
base line. Such complexity suggested examination of the
more common methods employed in Diels-Alder chem-
istry, namely, thermal and Lewis acid catalysis, to learn
whether a simplification of the reaction outcome would
be possible.
Thermal reaction of 4 with cyclopentadiene in refluxing
toluene was slow, and TLC analysis revealed only
extensive degradation and production of highly polar
products. Reaction under Lewis acid catalysis was,
however, more fruitful. Compound 4 was rapidly con-
sumed in the presence of cyclopentadiene and BF3‚OEt2
catalyst at -20 °C to give, after extensive chromatogra-
phy, two products. After much experimentation with
various Lewis acids, the reaction of 4 with cyclopenta-
diene in the presence of ZnCl2 in benzene at 47 °C was
found to give optimal conversion and afforded, after
chromatographic separation, three pure products, includ-
ing the two isolated from the cycloaddition of 4 in the
presence of BF3‚OEt2. It should be noted that the TLC
profile from the ZnCl2-catalyzed reaction was very similar
to that of the “radical-cation catalyzed” reaction described
earlier; however, the ZnCl2-catalyzed reaction proved to
be much cleaner and the products easier to separate.
The three compounds isolated, in 35, 16, and 3% yields,
were isomeric, as judged by high-resolution mass-spectral
analysis, with the molecular formula C11H12O3. As 1:1
cycloadducts of 4 and cyclopentadiene should have the
formula C17H25O5, it was obvious that direct cycloaddition
had not taken place. The 13C spectra of the products gave
key structural information on the molecular framework
of these products. Thus, the major isomer showed 11
distinct 13C signals, including a carbonyl singlet, an
“anomeric region” doublet, two alkene doublets, a high-
field methylene triplet, and a triplet for a CH2O function.
These signals, along with five additional doublets, could
be accommodated into a molecular framework structure
theoretically resulting from cycloaddition of cyclopenta-
diene with isolevoglucosenone (1,6-anhydro-2,3-dideoxy-
â-D-glycero-hex-2-enopyran-4-ulose, 5).17 The 13C spectra
of the other two products were similar. All three products
were optically active, and all showed carbonyl absorption
in the infrared.
Several preparations of 4 from 3-sulfonate derivatives
of 1 have been reported in the literature;15 however, none
of them proved amenable to large-scale preparation of
the alkene, requiring either expensive or inconvenient
reagents. The method of Srivastava et al.,15c in which
tosylate 2 was dissolved in cold dimethyl sulfoxide and
then treated with KOtBu, proved to be promising; how-
ever, preventing the solution from freezing was opera-
tionally difficult. A modification of this method, in which
THF was used in place of dimethyl sulfoxide, was found
to be convenient. Simply cooling a dilute THF solution
of 4 to 0 °C, then adding the KOtBu in portions, and then
refrigerating the resultant deep-red mixture overnight
allowed the isolation of reasonably pure 4 in 80-90%
yield. The major contaminant was 1,2:5,6-di-O-isopro-
pylidene-R-D-glucofuranose (1), the product of ester hy-
drolysis, but this could readily be removed by either flash
chromatography16 or Kugelrohr distillation of the crude
reaction mixture. However, the crude product proved to
be sufficiently pure for the subsequent cycloaddition
chemistry described here, and compound 4 could be
conveniently prepared in 25 g batches in three simple
steps from D-glucose in 60-65% overall yield.
With convenient access to large amounts of 4, we
investigated the possiblity of cycloaddition chemistry
with cyclopentadiene. Acceleration of inverse-demand
Diels-Alder reactions has been shown to be possible by
“radical-cation catalysis” 14 in which one of the reactants
(presumably the dienophile) is oxidized by a reagent such
as tris(p-bromophenyl)aminium hexachloroantimonate.14a
This reaction might be expected to yield up to four
isomeric products, depicted generally as “exo” adducts 6
and “endo” adducts 7. Stereochemical assignments were
made possible by use of 1H NMR spin-coupling and NOE
data.
(9) Flechtner, T. W. Carbohydr. Res. 1979, 77, 262.
(10) Zinner, H.; Wulf, G.; Heinatz, R. Chem. Ber. 1964, 96, 3536.
Freudenberg, K.; Brauns, F. Chem. Ber. 1922, 55, 3233.
(11) Nayak, U. G.; Whistler, R. L. J . Org. Chem. 1969, 34, 3819.
(12) Freudenberg, K.; Burkhart, G.; Braun, E. Chem. Ber. 1926, 59,
714.
(13) Horton, D.; J ewell, J . S.; Prihar, H. S. Can. J . Chem. 1968, 46,
1580.
(14) (a) Belville, D. J .; Wirth, D. D.; Bauld, N. L. J . Am. Chem. Soc.
1981, 103, 718. (b) Pabon, R. A.; Belville, D. J .; Bauld, N. L. J . Am.
Chem. Soc. 1983, 105, 5158. (c) Reynolds, D. W.; Lorenz, K. T.; Chiou,
H. S.; Belville, D. J .; Pabon, R. A.; Bauld, N. L. J . Am. Chem. Soc.
1987, 109, 4960.
The major isomer (the most polar of the three products)
was isolated in ∼35% yield and identified as the “down-
endo” derivative 8 (Scheme 1). That the annelated
(15) (a) Chiu, T.; Watanabe, K.; Fox, J . Carbohydr. Res. 1974, 32,
211. (b) Srivastava, H.; Srivastava, V. Carbohydr. Res. 1978, 60, 210.
(16) Still, W. C.; Kahn, M.; Mitra, A. J . Org. Chem. 1978, 43, 2923.
(17) (a) Ko¨ll, P.; Schultek, T.; Rennecke, R. W. Chem. Ber. 1976,
109, 337. (b)Furneaux, R.; Gainsford, G.; Shafizadeh, F.; Stevenson,
T. Carbohydr. Res. 1986, 146, 113.

Synthesis and Chemistry of Isolevoglucosenone
J . Org. Chem., Vol. 61, No. 11, 1996 3785
Sch em e 1
imity between the alkene portion of the cyclopentene ring
and the endo proton at H-6 of the 1,6-anhydro bridge.
The third (least polar) of the three products proved to
be the “down-exo” adduct 10, isolated in ∼3% yield.
Again, as with compound 8, a coupling constant of 0 Hz
between H-1 and H-2 suggested that the cyclopentene
ring resides below the plane of the anhydro sugar
skeleton in 10. The NOE measurements on compound
10 were inconclusive for establishing the exo disposition
of the cyclopentene ring, but the signals for H-2 and H-3
in the 1H NMR spectra of 10 are significantly shifted
upfield (relative to compound 8), suggesting that these
protons are affected by the anistropy of the nearby CdC
bond. Such an effect would be less significant if the
cyclopentene ring were in an endo orientation.
Traces of a fourth isomer were indictated in the 300
MHz H NMR spectrum of the crude reaction mixture,
Sch em e 2
1
with the most significant signal being a doublet at 5.86
ppm (J 1,2 ) 4.3 Hz). This signal was tentatively assigned
to the “up-exo” isomer 11, although this compound could
not be isolated in pure form. The two “down” isomers 8
and 10 were strongly dextrorotatory (+121o and +178°,
respectively), whereas the “up” isomer 9 was strongly
levorotatory (-200°).
The formation of compounds 8-11 is obviously the
consequence of a rearrangement of alkene 4. The timing
of the cycloaddition step is not, however, immediately
apparent, and two plausible mechanisms may be postu-
lated. Since the products are those which might be
expected to arise from the cycloaddition of isolevoglu-
cosenone (5) with cyclopentadiene, it is conceivable that
5 is formed by Lewis acid-catalyzed rearrangement of
alkene 4 and 5 which then undergoes cycloaddition. It
is also possible that 4 could rearrange to some other
reactive dienophile under the reaction conditions. Such
a species could then react with cyclopentadiene and the
subsequent Diels-Alder adduct rearrange to the ob-
served products. To clarify this point, the behavior of
isolevoglucosenone (5) on direct reaction with cyclopen-
tadiene was investigated.
Isolevoglucosenone (5) is available in low to moderate
yields by several multistep sequences,17 and a sample was
prepared by the method described by Shafizadeh and co-
workers,17b beginning with “levoglucosenone” obtained by
the pyrolysis of microcrystalline cellulose.18 Enone 5 was
found to react rapidly and essentially completely with
cyclopentadiene in the presence of ZnCl2 and also ther-
mally without Lewis acid catalysis, and the reaction
afforded two products only. Under both sets of reaction
conditions, the sole products isolated were adducts 8 and
10, the result of “bottom”-face addition to 5. The 1,6-
anhydro bridge appears to present a steric block to attack
from the “top” face. In the case of ZnCl2 catalysis,
adducts 8 and 10 were formed in 83% combined yield,
with the down-endo isomer 8 being favored by a factor
of 4.5:1. Thermal conditions afforded 8 and 10 in 88%
yield with 8 again being the major product, this time by
a factor of ∼11:1. Under both thermal and Lewis acid
conditions, enone 5 reacts much faster with cyclopenta-
diene than does alkene 4 and affords the cycloaddition
products in higher yields.
cyclopentene ring lies below the plane of the 1,6-anhydro
sugar is indicated from the H-1-H-2 coupling constant
1
of 0 Hz in the H NMR spectrum, a conclusion that is
reinforced by examination of models which clearly show
a dihedral angle between H-1 and H-2 in such a structure
as being ∼90°. The depiction of compound 8 as having
the 1C4(D) conformation stems from the large (∼9 Hz)
coupling constant between H-2 and H-3 in the proton
NMR. This value suggests a dihedral angle close to 0°
and the eclipsing of H-2 and H-3, which is attributable
to the pyranose ring being distorted by the annelated
cyclopentene ring. The NOE enhancements shown in
Scheme 2 support this structural assignment, and the
remaining 1H signals, as well as the 13C absorbances, can
be further reconciled with the structural assignment of
8. These data are detailed in Tables 1 and 2, respec-
tively.
The compound of intermediate polarity, isolated in
∼16% yield, was identified as the isomeric “up-endo”
cycloadduct 9. A coupling constant of 4.3 Hz between
These experiments show that cycloaddition to the top
face of 5 does not occur, strongly suggesting that enone
1
H-1 and H-2 in the H NMR spectrum of 9 corresponded
(18) Shafizadeh, F.; Chin, P. Carbohydr. Res. 1977, 58, 79. Brima-
combe, J . S.; Hunedy, F.; Mathers, A.; Tucker, L. C. N. Carbohydr.
Res. 1979, 68, 231.
well to that expected from models, and the NOE en-
hancements indicated in Scheme 2 established the prox-

3786 J . Org. Chem., Vol. 61, No. 11, 1996
Horton et al.
Ta ble 1. 1H NMR Ch em ica l Sh ifts (δ) a n d Cou p lin g Con sta n ts (J in Hz) for 1,6-An h yd r o Su ga r Der iva tives (a t 500 MHz
in CDCl3 r ela tive to Me4Si)
H-1
H-2
H-3
H-4
H-5
H-6
H-6′
H-7
H-8
H-9
H-10
H-11′
H-11
compd
J 1,2
J
2,3, J 2,7
J
3,4, J 3,10
J 4,5
J 5,6
J 6,6′
J 5,6′
J 7,8
J 8,9
J 9,10
J 10,11
J
10,11′, J 7,11′
J
11,11′, J 7,11
8
5.54 s 2.54dd
0.0 8.9, 3.3
5.73 d 2.86 m
4.0
5.69 s 1.79 m
0.0 8.1, a
5.31 s 2.36 dd
2.94 dd
s, 4.4
2.91 m
a
2.21 d
s, 0.0
4.20 d
5.8
3.85 dd 3.77 dd 3.01 bs 6.15 dd 6.06 dd 3.38 bs
8.3 1.6 2.9 5.6 2.9 1.8
1.31 d
0.0, 0.0
1.33 d
0.0, 0.0
1.21 ddd
1.5, 1.5
1.33 d
0.0, 0.0
1.25 d
0.0, 0.0
1.26 d
0.0, 0.0
1.23 d
0.0, 0.0
0.69 d
0.0, 0.0
1.38 dd
1.2, 0.0
1.33 d
1.49 ddd
8.5,1.8
1.47 dd
8.0, 1.0
1.79 m
9.0, a
1.44 dd
8.3, 1.8
1.40 dd
9.9, 1.8
1.40 d
8.0, 0.0
1.36 d
8.2, 0.0
1.43 dt
9.9, 0.0
2.03 d
9
10
14
15
16
17
18
19
20
21
24
25
20
21
24
25
4.31 dt 3.62 dd 3.32 dd 2.96 bs 5.96 dd 6.10 dd 3.26 bs
5.0
4.43 d
6.0
a
7.4
3.86 d
8.2
1.0
3.82 dd 2.83 d 6.27 dd 6.19 d
1.3 3.1 5.6 3.0
3.0
5.6
3.7
1.0
3.24 bs
a
2.26 ddd 3.62 m 4.17 td 3.62 m 3.91 dd 2.86 t
10.0, 3.3 5.2, 4.8 6.0 6.0 7.9 1.9
3.84 m 4.18 dd 3.84 m 3.56 dd 2.90 bs 6.15 dd 6.36 dd 3.08 bs
6.18 m 6.18 m 3.06 bs
0.0
a
a
a
1.8
5.29 s 2.21 dd
2.75 dt
9.6, 3.8
0.0
5.36 s 2.38 dd
0.0 9.6, 3.0
5.32 s 2.19 dd
0.0 9.5, 3.7
5.48 s 2.16 dd
0.0
5.54 s 1.84 dd
0.0 9.9, 2.3
5.24 d 3.01 dd
7.9 8.3, 3.7
5.62 s 2.23 dd
0.0
5.42 s 1.91 m
0.0
5.43 s 1.83 m
0.0
5.24 d 3.01 dd
7.9 8.3, 3.7
5.62 s 2.23 dd
0.0 1.8, 4.3
5.42 s 1.91 m
0.0
5.43 s 1.83 m
0.0
9.9, 3.8
0.0
8.0
4.30 t
5.3
8.0
2.9
3.0
5.6
3.0
1.8
2.33 ddd 4.63 t
3.63 t
7.8
3.89 d
2.0
2.89 bs 6.22 m 6.31 m 3.09 bs
9.5, 4.5
2.81 dd
9.6, 3.4
2.08 dt
5.3
4.93 d
0.0
4.17 t
5.4
a
a
a
a
4.21 dd 3.64 dd 3.81 t
7.1 7.1 1.9
4.35 td 3.68 dd 4.00 dd 2.59 d 3.36 d
6.2 7.9 2.0 7.0 3.5
2.88 bs 6,11 m 6.21 m 2.93 bs
a
a
a
a
3.33 d
7.0
2.70 d
1.9
10.9, 3.7 4.7, 4.7
2.47 dd
5.3, 5.3
3.04 dd
s, 3.5
3.93 dd 4.49 m 3.73 m 3.73 m 2.15 s
4.49 m 4.10 d
0.0 4.9
2.77 t
0.0
2.6
a
a
a
a
10.7, 0.0
1.44 d
8.8, 0.0
3.94 t
a
4.26 dd 4.33 dd 2.99bs 3.37 bs 6.26 m 6.12m
a
a
a
a
a
a
0.0 0.0
2.58 dd
4.47 dd 3.88 t
8.3 8.3
5.16 dd 4.56 m 3.96 dd 3.70 t
3.81 dd 2.49 bs 2.74 bs
1.26-1.47
10.8, 4.3 s, 4.8
1.0
a
a
a
2.01 m
6.2, a
2.33 bs 2.40 bs
1.26-1.76
a
4.1
4.91 d
0.0
1.5
6.0
7.8
a
a
a
1.95 m
10.3, a
3.04 dd
s, 3.5
2.58 dd
s, 4.3
2.01 m
6.2, a
4.44 dd 3.81 t
3.74 d
0.0
2.18 bs 2.39 bs
1.27-1.78
a
1.0
3.94 t
a
7.3
a
a
a
1.33 d
0.0, 0.0
1.26-1.47
a
4.26 dd 4.33 dd 2.99 bs 3.37 bs 6.26 m 6.12 m
1.44 d
8.8, 0.0
a
a
a
a
a
a
4.47 dd 3.88 t
8.3 8.3
5.16 dd 4.56 m 3.96 dd 3.70 t
3.81 dd 2.49 bs 2.74 bs
1.0
a
a
2.33 bs 2.40 bs
1.26-1.76
a
a
4.1
4.91 d
0.0
1.5
4.44 dd 3.81 t
1.0 7.3
6.0
7.8
a
a
1.95 m
10.3, a
3.74 d
0.0
2.18 bs 2.39 bs
1.27-1.78
a
a
a
a
a
Overlapping signals.
Ta ble 2. 13C NMR Da ta for 1,6-An h yd r o Su ga r -Cyclop en ta d ien e Ad d u cts (125 MHz in CDCl3 r ela tive to Me4Si)
chemical shift (δ)
compd
C-1
C-2
C-3
C-4
C-5
C-6
C-7
C-8
C-9
C-10
C-11
CdO
CH3
8
9
102.5
101.9
104.0
101.7
102.6
102.4
103.0
100.9
101.6
95.6
44.3
42.8
43.5
48.2
48.5
47.3
205.3
210.8
206.6
61.4
69.4
71.0
71.5
65.0
78.0
78.0
76.8
78.1
73.2
76.4
71.1
74.3
73.0
89.9
79.8
79.2
79.7
75.9
65.9
66.0
65.7
68.4
64.3
62.9
65.9
61.4
64.3
63.9
65.8
69.1
67.7
45.2
44.5
45.3
134.6
131.6
139.2
135.8
133.0
136.0
132.2
50.1
75.3
134.7
23.9
23.0
24.7
134.7
136.4
136.8
135.5
137.4
136.5
137.6
49.3
42.3
137.9
25.1
24.1
25.1
47.1
46.0
47.2
50.0
49.5
43.9
50.6
50.7
50.9
50.2
27.1
33.4
50.7
48.2
39.4
36.2
10
14
15
16
17
18
19
20
21
24
25
45.8a
40.3a
41.7a
38.1a
45.9a
47.0a
43.6a
42.0a
42.3a
41.4a
43.9a
44.3a
46.6a
43.5a
41.9a
42.8a
45.1a
44.4a
44.9a
41.5a
46.5a
46.1a
47.2a
46.6a
40.1a
34.8a
50.7a
40.0a
40.4a
40.1a
48.6a
46.8a
49.1a
47.4a
38.7a
33.4a
48.7a
40.0a
41.0a
40.3a
170.6
21.8
170.9
21.9
209.0
207.8
71.0
169.1, 170.3
20.0, 21.2
103.1
102.8
103.6
170.9
171.1
21.8
21.7
73.6
a
Interchangeable.
5 is not an actual intermediate in the observed rear-
rangement-cycloaddition reaction of 4. It is most likely
that alkene 4 undergoes Lewis acid-catalyzed rearrange-
ment before cycloaddition takes place. A possible course
that would explain the reaction outcome is outlined in
Scheme 2; thus, complexation of ZnCl2 with alkene 4
could promote loss of the 1,2-O-isopropylidene group to
create an enal (the acyclic dienophile 12) which could
then undergo cycloaddition to afford generalized adduct
13. Loss of a second molecule of acetone from 13 would
lead to 1,6-anhydro bridge formation and the production
of all four isomeric products 8-11.
We have previously shown such derivatives to be useful
intermediates in the asymmetric synthesis of chirally
substituted carbocycles related to such natural products
as prostaglandins.7 The potential for diastereoselective
conversions of 8 relies mainly on the relative steric
encumbrance of the 1,6-anhydro bridge above the plane
of the sugar ring and that of the cyclopentene ring below.
Simple borohydride reduction of the 4-ketone in 8 led
to approximately equal amounts of two diastereomeric
alcohols (14 and 15). Isolation of 14 and 15 in 45.3 and
43% yields, respectively, suggested that the steric con-
tributions of the 1,6-anhydro bridge and the annelated
cyclopentene ring play an approximately equal role in
deciding the reaction outcome. Separation of compounds
14 and 15 required MPLC due to their similar migrations
on silica gel; however, conversion into their respective
Ch em istr y of Nor bor n en e 8. The major cycloadduct
8 isolated in the rearrangement-Diels-Alder reaction
was next studied in detail as a potential source of
stereochemically defined norbornenes and norbornanes.

Synthesis and Chemistry of Isolevoglucosenone
J . Org. Chem., Vol. 61, No. 11, 1996 3787
epoxidation followed by intramolecular epoxide opening
with the axial OH group acting as the nucleophile, and
therefore alcohols 14 and 15 can be assigned as having
the D-gulo and D-allo configurations, respectively. The
two oxidation products 18 and 19 gave analytical and
spectral data completely consistent with the proposed
1
structures (for H and 13C NMR data see Tables 1 and 2,
respectively).
Hydrolysis of the 1,6-anhydro bridge in cycloadduct 8
proved to be somewhat more difficult than expected. It
is known20 that 1,6-anhydrohexopyranoses are generally
hydrolyzed faster than simple glycosides, presumably
through release of strain in the fused-ring system;
however, 1,6-anhydropyranoses possessing electronega-
tive or basic substituents at C-2 are hydrolyzed with more
difficulty.21 Initial attempts at hydrolysis utilizing either
90% aqueous trifluoroacetic acid at room temperature or
standard acetolysis conditions (sulfuric acid-acetic acid-
acetic anhydride mixture at room temperature) resulted
only in the recovery of starting material. Success was
met, however, when 8 was dissolved in trifluoroacetic acid
and then treated with acetic anhydride. This afforded a
single compound isolated in 63% yield after chromatog-
F igu r e 1. ORTEP representation of polycyclic alcohol 15.
monoacetates (16 and 17) permitted facile separation by
flash chromatography.
The formulation of 14 and 15 as products having the
D-gulo and D-allo configurations, respectively, could not
be made with confidence on the basis of 1H NMR coupling
1
raphy and identified as keto diacetate 20. The H NMR
spectrum of 20 revealed a doublet at 5.23 ppm with a
coupling constant J 1,2 ) 8.3 Hz. Such a large value for
J 1,2 can only be explained by the formation of the â
anomer 20. The relief of strain after the 1,6-anhydro
bridge is hydrolyzed allows for more flexibility in the
pyranose ring and adoption of a flattened chairlike
conformation.
1
constants alone. The distortion created in the C4 pyra-
nose ring by the attached cyclopentene ring causes
enough perturbance around C-3, C-4, and C-5 of the
sugar ring that little information regarding the stereo-
chemistry at C-4 can be gathered from the values of J 3,4
and J 4,5. Firm assignments of 14 and 15 could, however,
be made from subsequent chemical transformations on
both of the epimeric alcohols and independently from the
single-crystal X-ray structural analysis of 15.19,29 The
ORTEP representation of alcohol 15 (Figure 1) clearly
shows the OH group at C-4 occupying the axial position
at C-4 of the sugar ring and provides independent
verification of the structural assignments for 14 and 15
by chemical means.
Reduction of the cyclopentene portion of norbornene 8
was achieved with hydrogen in the presence of a plati-
num catalyst and yielded norbornane 21, isolated as an
oil in 92% yield. The 1H and 13C NMR spectra of 21
(Tables 1 and 2, respectively) clearly showed the absence
of alkene signals and an increase in the complexity of
the aliphatic region between 1.2 and 2.8 ppm. The IR
spectrum of 21 showed an absorption at 1725 cm-1
,
revealing that the ketone function remained intact.
Reduction of the ketone group in compound 21 with
sodium borohydride, in a manner similar to that em-
ployed for norbornene 8, again resulted in the formation
of two epimeric alcohols. The infrared spectrum of the
crude mixture of products showed a strong OH absorption
at 3500 cm-1 but no ketone absorption at 1725 cm-1. TLC
analysis revealed two closely migrating products, and the
The proximity of the alkene portion of the cyclopentene
ring to the hydroxyl group in 14 and 15 suggested that
chemical transformation of the double bond might be
affected by the axial hydroxyl group in 15 but not by the
equatorial hydroxyl considered to be present in 14. Thus,
treatment of either alcohol with m-CPBA led to consump-
tion of starting material and the formation of different
crystalline products. Alcohol 14 underwent smooth ep-
oxidation to afford epoxy alcohol 18 in 88% yield after
recrystallization. Alcohol 15 also gave a single product,
formulated as the polycyclic alcohol 19, in 95% yield.
Compound 19 is clearly the product of double-bond
1
300 MHz H NMR spectrum showed two resonances for
anomeric signals at 5.40 and 5.43 ppm in approximately
1:2 ratio, arising from the epimeric alcohols 22 and 23,
respectively. The products again proved difficult to
separate, and so the reaction mixture was immediately
acetylated (acetic anhydride-pyridine) to afford the two
epimeric monoacetates 24 and 25 in 21 and 47% yields,
respectively, after flash chromatography. In the reduc-
(19) The crystal structure analysis of compound 15 was performed
by Dr. J udith Gallucci of the Department of Chemistry at The Ohio
State University using the following programs: TEXSAN, TEXRAY
Structural Analysis Package, version 2.1, by Molecular Structure
Corporation, College Station, Texas, 1987; MITHRIL, A Computer
Program for the Automatic Solution of Crystal Structures from X-ray
Data, by C. J . Gilmore, University of Glasgow, Scotland, 1983.
(20) Hall, H. K.; DeBlauwe, F. J . Am. Chem. Soc. 1975, 97, 655.
(21) Carlson, L. J . J . Org. Chem. 1965, 30, 3953.

3788 J . Org. Chem., Vol. 61, No. 11, 1996
Horton et al.
tion of keto norbornene 8, a 1/1 mixture of C-4 epimers
was formed, and the preferential formation of D-allo
isomer 25 from norbornane 21 is presumably caused by
the more flexible cyclopentane ring in 21 blocking the
bottom face of the carbonyl group.
preparation of 5. All three acid catalysts (TsOH‚H2O,
ZnCl2, or AlCl3) led to 5 (TLC), but extensive experimen-
tation revealed AlCl3 to be the most efficient. A variety
of solvents were examined (THF, toluene, CH2Cl2), and
a pentane-Et2O mixture gave the maximum yield.
Isolation of 5 from the reaction mixture necessitated
anhydrous conditions and rapid removal of catalyst to
avoid extensive degradation of the sensitive enone. This
was best achieved by complexing the catalyst with Et3N
and rapidly eluting the product through a short column
of silica gel. This procedure gave a yellow syrup that
contained 5 as the major component, and subsequent
chromatography afforded 5 in 32% isolated yield. This
convenient new access to substantial quantities of enone
5 in only four steps from D-glucose and in ∼20% overall
yield opened the possibility of a more extensive study of
this potentially useful synthon.
Rep r esen ta tive Ch em istr y of Isolevoglu cosen on e
(5). Only two reactions of the previously inaccessible
enone 5 have been reported: its stereospecific reduction
with sodium borohydride17a to give allylic alcohol 26 and
pyrolysis17b to afford cycloaddition product 27. Both
reactions take place as expected exclusively from the
bottom face of the enone; the 1,6-anhydro bridge ef-
fectively blocks the top face of the enone. This selectivity
augments the great potential of compound 5 as a versatile
synthon. The chemistry detailed next shows a variety
of reactions involving stereospecific addition of reagents
to the bottom face of 5, providing entry into a range of
diastereomerically defined products.
Acid -Ca ta lyzed Rea r r a n gem en t of 4 in th e a b-
sen ce of Dien e: F a cile P r ep a r a tion of Isolevoglu -
cosen on e. Although the intermediacy of isolevoglucose-
none (5) in the Lewis acid-catalyzed rearrangement-
cycloaddition reaction of alkene 4 with cyclopentadiene
had been discounted, it was considered possible that 5
might arise from rearrangement of compound 4 if a diene
were not present to trap the presumed acyclic dieneophile
intermediate 12 (Scheme 2). Accordingly, experiments
were performed to study the possibility of converting
alkene 4 into isolevoglucosenone (5).
Treating a benzene solution of 4 with ZnCl2, as
described earlier for the preparation of norbornene ad-
ducts 8-11 but without the addition of cyclopentadiene,
and monitoring the reaction by TLC over several hours
against an authentic sample of 5 revealed the consump-
tion of alkene 4 and the formation of numerous products,
including one comparable in Rf and UV activity to 5.
However, a clean separation of compound 5 from this
mixture could not be effected, and a detailed search was
made to find reaction conditions that would permit facile
isolation of 5 in good yield. Three general methods (A,
B, and C) were examined.
As described earlier, compound 5 reacts efficiently with
cyclopentadiene to afford two crystalline products identi-
fied as the down-endo adduct 8 and down-exo adduct 10
(Scheme 1), the 1,6-anhydro bridge blocking the top face
of the dienophile and preventing the formation of “top-
side” adducts 9 and 11. Similar reaction occurs in the
presence of the symmetrical dienes cyclohexadiene and
anthracene to afford crystalline cycloaddition products
28 and 29, in 59 and 62% yields, respectively. In each
case, only one product is formed because the symmetrical
diene adds exclusively from the bottom face of the
Method A was related to the preparation of levoglu-
cosenone from cellulose as employed by Brimacombe and
co-workers:18b alkene 4 was treated with a variety of acid
catalysts under high vacuum and compound 5 was
collected as a pale-yellow distillate. By using the cata-
lysts KHSO4, TsOH‚H2O, and ZnCl2 and a variety of
reaction temperatures and catalyst concentrations, the
maximum yield of reasonably pure 5 isolated was ∼5%,
with TsOH‚H2O as catalyst.
Method B employed a heterogeneous catalyst system
in which the catalyst (ZnCl2 or H2SO4) was adsorbed onto
silica gel on the top of a flash chromatography column.
By eluting a solution of alkene 4 through the column,
the compound first came into contact with the catalyst,
causing reaction, and the resultant products were sepa-
rated on the bottom half of the column. Method B proved
most efficient when H2SO4 was employed as catalyst, but
the maximum yield of crude 5 attained in this study was
only 10%.
1
dienophile. The H NMR resonances of H-1 in 28 and
29 both appear as singlets, indicating zero spin-coupling
with H-2 as observed with the down-endo and down-exo
adducts 9 and 11 formed from reaction of 5 with cyclo-
pentadiene.
Method C, in which a dilute solution of alkene 4 and
catalyst was stirred and the reaction monitored by TLC,
proved to be the most convenient procedure for one-step
Enone 5 offers promise as a valuable synthon in the
stereocontrolled synthesis of 2-amino-2,3-dideoxy sugars,

Synthesis and Chemistry of Isolevoglucosenone
J . Org. Chem., Vol. 61, No. 11, 1996 3789
Ta ble 3. 1H NMR Ch em ica l Sh ifts (p p m ) a n d Mu ltip licities for Isolevoglu cosen on e (5) a n d Der iva tives
chemical shift (δ) and coupling constants (Hz)
H-1
H-2
H-3
H-3′
H-4
H-4′(OH)
H-5
H-6endo
H-6exo
compd
J 1,2
J 2,3
J 2,3
′
J 3,3′
J
3,4, J 4,5
J 3',4
J 5,6exo
J 5,6endo
J 6exo,6endo
5
5.80 d
3.3
7.11 d
9.8
6.10 ddd
4.77 dt
6.3
3.64 dd
1.5
4.10 dd
8.2
6.3
1.5
30
31
32
33
5.56 s
0.0
5.87 s
0.0
5.41 s
0.0
5.33 s
0.0
3.74 d
7.6
3.33 dd
3.6
3.57 d
4.8
3.45 d
5.3
2.80 dd
0.0
3.46 dd
2.56 ddd
17.9
4.54 d
5.2
4.64 dt
6.5
4.53 d
2.4
4.42 t
4.0
3.96 d
0.0
3.96 dd
1.3
3.83 m
0.0
4.13 d
0.0
3.87 d
8.3
3.79 dd
8.4
3.83 m
a
3.73 dd
a
2.14 d
1.7
1.86 ddd
1.5
1.91 dt
15.6
2.14 dddd
14.4
3.62 dt
4.8, 0.0
4.19 dd
2.83 d
1.7
2.06 bs
6.1
10.6, 0.0
a
Overlapping signals.
common structural components of many important an-
tibiotics.22 Conjugate addition to the enone may be
expected to be stereospecific to afford D-erythro analogs,
with attack by the incoming nucleophile at the alkene
face opposite the 1,6-anhydro bridge. Subsequent reduc-
tion of the ketone function should likewise be stereose-
lective. Treatment of 5 with an acidic solution of sodium
azide (HOAc, NaN3) by analogy with the procedure
reported by Gero et al.23 resulted only in degradation of
the enone, but replacing HOAc with the less nucleophilic
CF3CO2H resulted in almost complete conversion into the
D-erythro azido sugar 30. Rapid workup afforded an
almost quantitative yield of crude 30, which proved to
be unstable and was therefore used without further
purification for subsequent transformations. The 1H
NMR spectrum of this crude product did not show signals
corresponding to the potential D-threo epimer of 30,
indicating that the stereochemistry of azide addition to
compound 5 is completely governed by the steric bulk of
the 1,6-anhydro bridge. Although complete characteriza-
tion of compound 30 could not be made by elemental
analysis and mass spectrometry, the IR spectrum showed
a strong azide band at 2100 cm-1 and the 13C NMR
spectrum showed no alkene signals; the C-3 signal
appeared upfield at 37.6 ppm. Proton-proton couplings
Azide 30 offers potential in the synthesis of precursors
of certain amino sugars.22 Accessible in only five steps
from D-glucose and formed in a completely stereospecific
manner, azide 30 is potentially useful for syntheses of
amino sugars having three chiral centers in the D-ribo
configuration, such as nebrosamine, purpurosamine, and
sisosamine.23,25 Essential to this idea was the stereo-
chemical course of the reduction of the C-4 ketone
function in 30. Unlike isolevoglucosenone (5), where the
ketone reduction was predictably controlled by the 1,6-
anhydro bridge, the analogous reduction in 30 was
expected to be governed by the relative steric contribu-
tions of the axial â azide group at C-2 as well as the 1,6-
anhydro bridge.
Reduction of 30 with sodium borohydride was expected
to exhibit little face selectivity, and such reduction at -78
°C in 95% EtOH provided two major products identified
as the epimeric azido alcohols 32 and 33 isolated in 24
and 46% yields, respectively, after chromatographic
1
in the H NMR spectrum of 30 confirmed the D-erythro
1
separation. The H and 13C NMR spectra (see Tables 3
stereochemistry with J 1,2 ) 0 Hz. As with the cycload-
ducts described earlier, this coupling strongly suggests
the substituent at C-2 to be axial with a quasiequatorial
relationship between H-1 and H-2, as seen in compound
30.
and 4, respectively) of 32 and 33 firmly support their
assignments as the D-ribo and D-xylo derivatives, respec-
tively. The signal for the axial H-4 proton in compound
33 showed a diaxial coupling constant of 10.6 Hz through
interaction with the neighboring axial H-3 proton whereas
the corresponding H-4 proton in 32 showed a much
smaller coupling because of its equatorial orientation.
Also detected as a very minor component of the reaction
mixture was allylic alcohol 34. It is possible that this
compound is formed by base-promoted loss of HN3
between C-2 and C-3 to regenerate enone 5, which is then
reduced selectively to compound 34.
It was thought that the bulkier reductants lithium
trisamylborohydride (LS Selectride) or sodium bis(2-
methoxyethoxy)aluminum hydride (Red Al) might pro-
vide greater stereodifferentiation.26 However, treatment
of azido ketone 30 with LS Selectride at -78 °C in THF
solution gave a complex mixture from which axial alcohol
32 could be isolated with some difficulty and in only 11%
yield. Similar reaction with Red Al also yielded numer-
ous products, not including axial alcohol 32 but including
A second reaction of enone 5 considered to be syntheti-
cally important was that with reagents capable of epoxi-
dizing the double bond. tert-Butyl hydroperoxide was
chosen for this transformation, since its steric bulk would
presumably contribute to stereodifferentiation upon ad-
dition to the enone double bond and the reagent can be
used in nonaqueous medium. It was found that enone 5
reacted readily with t-BuOOH, in the presence of a
catalytic amount of benzyltrimethylammonium hydrox-
ide, to afford the crystalline epoxide 31 in 58% yield. The
physical constants reported for 31 were in excellent
agreement with those previously reported,24 and none of
the corresponding epoxide that would have resulted from
top-face addition to compound 5 could be detected in the
reaction mixture.
(22) Tanaka, N. in Antibiotics; Corcoran, J . W., Ed.; Springer-
Verlag: Berlin, 1975; p 340.
(23) J egou, E.; Cle´ophax, J .; Leboul, J .; Gero, S. D. Carbohydr. Res.
1975, 45, 323.
(25) Leboul, L.; Cle´ophax, J .; Gero, S. D.; Rolland, A.; Forchiune,
A. Tetrahedron 1977, 33, 965.
(26) Krishnamurthy, S.; Brown, H. C. J . Am. Chem. Soc. 1976, 98,
3383.
(24) Rolf, K. H.; Rennecke, W.; Ko¨ll, P. Chem. Ber. 1975, 108, 3645.

3790 J . Org. Chem., Vol. 61, No. 11, 1996
Horton et al.
Ta ble 4. 13C NMR Ch em ica l Sh ifts (p p m ) for
Isolevoglu cosen on e Der iva tives
chemical shift (δ)
compd
C-1
C-2
C-3
C-4
C-5
C-6
5
30
31
32
33
95.9
101.5
99.7
99.9
99.9
147.0
59.4
49.2
59.0
59.1
127.0
37.6t
52.7
30.0
30.0
194.4
201.5
198.6
63.2
79.5
79.1
79.1
75.7
75.7
62.6
66.9
66.5
63.4
63.4
63.2
equatorial alcohol 33 as only a minor component. The
outcome of these reactions can probably be attributed to
the relatively high basicities of these reagents,27 which
causes extensive degradation of the base-sensitive 30.
Reaction with the electrophilic aluminum reagents
diisobutylaluminum hydride (DIBAH) and triisobutyl-
aluminum (TIBA) proved much more informative. When
azido ketone 30 was treated with DIBAH in dichlo-
romethane solution at -78 °C, both D-ribo and D-xylo
epimers 32 and 33 were isolated in 42 and 34% yields,
respectively. Raising the reaction temperature did not
affect the isomer ratio, but at 25 °C increased amounts
of side products were observed. This greater propensity
for side-product formation was also noticed when the
solvent was changed to pentane at -78 °C, although
there was no noticeable change in the ratio of 32 and
33. A dramatic change in the course of the DIBAH
reduction was observed, however, when THF was em-
F igu r e 2. Possible steric interactions in the reduction of azido
ketone 30 with DIBAH and TIBA.
Sch em e 3
1
ployed as solvent. Analysis of the crude mixture by H
NMR showed it to contain mainly equatorial alcohol 33,
with axial alcohol 32 and allylic alcohol 34 being present
in only trace amounts. Ultimately, compound 33 could
be isolated in 46% yield after chromatography.
anhydro bridge and the azide substituent and thus
attacks from the more accessible endo face of the ketone
in compound 30 resulting in almost exclusive formation
of D-xylo alcohol 33.
Reduction of 30 with TIBA also proved to be stereose-
lective when CH2Cl2 was employed as solvent at room
temperature. Under these conditions, D-xylo alcohol 33
was again isolated as the major product in 56% yield,
and epimeric alcohol 32 could not be detected upon 1H
NMR analysis of the crude mixture. The change in
selectivity on moving from DIBAH in CH2Cl2 to TIBA in
CH2Cl2 is probably a consequence of the different mech-
anisms operating with these two reductants. Although
both reagents possess considerable steric bulk, the mode
of hydride delivery is significantly different in each. As
seen in Scheme 3, DIBAH can coordinate to the ketonic
oxygen in either an exo (A) or endo (B) manner, with the
bulk of the reducing agent residing away from the sugar
skeleton and thereby allowing the formation of both
epimeric alcohols 32 and 33. When TIBA is employed,
the hydride is transferred from an isobutyl group. In
situation C, when the reducing agent has to align itself
to deliver hydride to the exo face of 30, an isobutyl group
has to interfere with the axial 1,6-anhydro bridge, a
situation that should be unfavorable. The alternative
situation, where TIBA coordinates to the endo face of the
ketone (D), is likely to be much more favorable with the
azide group, creating less steric interference. The ex-
clusive formation of alcohol 33 suggests that D is the
favored reaction path in this reduction.
The change in stereospecificity upon changing from
CH2Cl2 to THF was surprising, but may be explained by
considering the nature of the reducing agent in the two
solvents. Previous studies have shown28 that DIBAH
exists as a trimeric species (35, Figure 2) in such
noncoordinating solvents as CH2Cl2; whereas, in solvents
capable of donating lone pairs such as THF, the reducing
agent forms reasonably stable 1:1 adducts (36, Figure 2).
The difference in steric bulk between the trimeric species
35 and the 1:1 adduct 36 may well be the cause of the
differing reaction outcome. In CH2Cl2 solution, the attack
of species 35 from either the exo or endo face of the ketone
would probably be hindered by both the axial R substitu-
ent (the 1,6-anhydro bridge) and the axial â substituent
(the azide group), thereby leading to formation of both
alcohols 32 and 33. When THF is employed as solvent,
the much smaller species 36 is probably able to dif-
ferentiate better between the relative sizes of the 1,6-
(27) Bazant, B.; Capka, M.; Cerny, M.; Chvalovsky, V.; Kochloefl,
K.; Kraus, M.; Malek, J . Tetrahedron Lett. 1968, 3303.
(28) Noyori, R.; Baba, Y.; Hayakawa, Y. J . Am. Chem. Soc. 1974,
96, 3336.
(29) The author has deposited atomic coordinates for compound 15
with the Cambridge Crystallographic Data Centre. The coordinates
can be obtained, on request, from the Director, Cambridge Crystal-
lographic Data Centre, 12 Union Road, Cambridge, CB2 1EZ, U.K.
Synthetic conversion of axial alcohol 32 into several
amino sugars has already been demonstrated,23,25 and the
convenient access to 32 in six steps from D-glucose will
facilitate synthesis of this class of amino sugars.

Synthesis and Chemistry of Isolevoglucosenone
J . Org. Chem., Vol. 61, No. 11, 1996 3791
CHCl3); IR (CHCl3) 1725 cm-1; m/ z 192 (M+)]. For 1H and
13C NMR, see Tables 1 and 2, respectively. Anal. Calcd for
C11H12O3 (8): C, 68.73; H, 6.30. Found: C, 69.24; H, 6.53.
Anal. Calcd for C11H12O3 (9): C, 68.73; H, 6.30. Found: C,
68.70; H, 6.33.
With the controlled reduction of azide 30 affording 1,6-
anhydro-2-azido-2,3-dideoxy-â-D-ribo-hexopyranose (32),
we examined the cleavage of the 1,6-anhydro bridge in
32 as a route to known precursors of 2-amino-2,3-dideoxy
sugars. Treatment of compound 32 with dilute (2%) H2-
SO4 in refluxing methanol resulted in slow degradation
and eventual recovery of most of the starting material.
In contrast, acetolysis using trifluoroacetic acid and acetic
anhydride afforded in high yield (97%) an anomeric
mixture of the azido triacetates 37. Separation of the
anomers proved impossible because of their almost
identical Rf values. However, the 1H and 13C NMR
spectra, as well as IR and microanalytical data for the
mixture, firmly established their identity.
B. Zn Cl2-Ca ta lyzed . A solution of 4 (0.99 g, 0.409 mmol)
in benzene (5 mL) was warmed to 47 °C, and then freshly
distilled cyclopentadiene (0.07 mL, 0.84 mmol) and ZnCl2 (1
M in ether, 0.20 mL, 0.84 mmol) were added successively. After
30 min of stirring, Et3N (0.10 mL, 0.72 mmol) was added and
the mixture immediately passed, with pressure, through a
short plug of silica gel which was washed through with
additional benzene (10 mL). Evaporation afforded a yellow
syrup which was initially purified by flash chromatography
(3:1 hexane-EtOAc). Fractions that contained a mixture of
isomers were further purified by MPLC (9:1 hexane-EtOAc).
Combination of pure fractions gave 1,6-anhydro-2,3-C-[(3S,5R)-
1-cyclopentene-3,5-diyl]-2,3-dideoxy-â-D-ribo-hexopyranos-4-ul-
ose (8, 274 mg, 34.8%), 1,6-anhydro-2,3-C-[(3R,5S)-1-cyclopen-
tene-3,5-diyl]-2,3-dideoxy-â-D-lyxo-hexopyranos-4-ulose (9, 12.5
mg, 15.9%), and 1,6-anhydro-2,3-C-[(3R,5S)-1-cyclopentene-
3,5-diyl]-2,3-dideoxy-â-D-ribo-hexopyranos-4-ulose [10, 2.6 mg,
The formation of this anomeric mixture of 1,4,6-tri-O-
acetyl-2-azido-2,3-dideoxy-D-ribo-pyranose in eight steps
from D-glucose documents the promising synthetic utility
of isolevoglucosenone for the preparation of amino sugars,
as well as non-carbohydrate targets.
o
3.3%, mp 107-108 C, recrystallized from ether-pentane; Rf
0.58 (3:1 hexane-EtOAc); [R]D +177.8° (c 0.6, CHCl3); IR
(CHCl3) 1725 cm-1; m/ z 192 (M+)]. Anal. Calcd for C11H12O3
1
(10): C, 68.73; H, 6.30. Found: C, 68.74; H, 6.33. For H and
13C NMR data, see Tables 1 and 2, respectively.
1,6-An h yd r o-2,3-C-[(3S,5R)-1-cyclop en t en e-3,5-d iyl]-
2,3-d id eoxy-â-D-gu lop yr a n ose (14) a n d 1,6-An h yd r o-2,3-
C-[(3S,5R)-1-cyclop en t en e-3,5-d iyl]-2,3-d id eoxy-â-D-a l-
lop yr a n ose (15). Cycloadduct 8 (1.50 g, 7.81 mmol) was
dissolved in 95% EtOH at rt, and NaBH4 (0.39 g, 24.0 mmol)
was added in four portions over 10 min. After 1 h, acetone (2
mL) was added to decompose excess borohydride and the
solution made neutral by stirring for 30 min with Dowex-50W
Exp er im en ta l Section
Gen er a l P r oced u r es. Melting points are uncorrected.
Optical rotations were measured at ∼25 °C. Mass spectra
were obtained either in EI mode at 70 eV or using CI (NH3).
1H NMR spectra were recorded at either 500 or 300 MHz and
13C NMR spectra at either 125 or 75 MHz. All NMR spectra
were recorded for solutions in CDCl3. All J values are reported
in hertz. Column chromatography was performed on silica gel
60 (230-400 mesh, E. Merck) and thin layer chromatography
(TLC) on aluminum-backed plates of silica gel 60 F254 (E.
Merck). Zones were detected by spraying plates with 5% H2-
SO4 in aqueous ethanol and subsequent heating. Elemental
analyses were performed by Atlantic Microlabs, Inc., Norcross,
GA.
1,2:5,6-Di-O-isop r op ylid en e-3-d eoxy-r-D-er yth r o-h ex-3-
en ofu r a n ose (4). To an ice-cold solution of 1,2:5,6-di-O-
isopropylidene-3-O-tosyl-R-D-glucofuranoses10 (10.0 g, 24.2
mmol) in dry THF (400 mL) was added KOtBu (8.14 g, 72.4
mmol) in four portions over 10 min. After stirring for 2 h at
0 °C, the solution was partitioned between hexane (300 mL)
and water (300 mL). The aqueous layer was extracted with
hexane (200 mL), and the combined organic layers were
washed with 5% H2SO4, saturated NaHCO3, and saturated
NaCl and then dried. Evaporation afforded 4 as a yellow solid
(5.96 g) which proved sufficiently pure for the next step;
however, chromatography (3:1 hexane-EtOAc) yielded the
pure alkene as a colorless solid (4.9 g, 84%): mp 48-49 °C,
lit.15 mp 48-50 °C; [R]D +25.4° (c 1.5, CHCl3), lit.15 [R]D
+21.2° (c 1, CHCl3). Anal. Calcd for C12H18O5: C, 59.49; H,
7.49. Found: C, 59.53; H, 7.49.
Cycloa d d ition of 4 w ith Cyclop en ta d ien e. P r ep a r a -
tion of Nor bor n en e Der iva tives 8, 9, a n d 10. A. BF 3-
Ca ta lyzed . To a solution of 4 (7.0 g, 28.9 mmol) in dry toluene
(15 mL) kept at -20 °C under an argon atmosphere was added
freshly distilled cyclopentadiene (6.8 mL, 28.7 mmol), and
BF3‚OEt2 (3.54 mL, 28.7 mmol) was added slowly. The
mixture was stirred at -20 °C for 3 h. After being warmed to
rt, the solution was washed twice with saturated NaHCO3,
once with water, and then dried and evaporated to afford a
yellow syrup. Two separations on silica gel (6:1 hexane-
EtOAc) gave 8 [549 mg, 9.9%; mp 94-95 °C, recrystallized
from EtOAc-hexane; Rf 0.47 (3:1 hexane-EtOAc); [R]D +121°
(c 1.1, CHCl3); IR (CHCl3) 1725 cm-1; m/ z 192 (M+)] and 9
[67 mg, 1.2%; mp 107-108 °C, recrystallized from EtOAc-
hexane; Rf 0.53 (3:1 hexane-EtOAc); [R]D -198.9° (c 1.6,
H
+ resin (1.35 g). The solids were filtered off, and the filtrate
evaporated to afford a pale-yellow syrup (1.55 g). A portion
of this mixture (1.07 g) was separated by MPLC on silica gel
(6:1 hexane-EtOAc) to afford the epimeric alcohols 14 and 15.
Compound 14 was isolated as colorless solid [474 mg, 45.3%
corrected; mp 111-113 oC, recrystallized from hexane-EtOAc;
Rf 0.26 (2:1 hexane-EtOAc); [R]D +18.8° (c 1.4, CHCl3); IR
(KBr) 3484 cm-1; m/ z 194 (M+)]. Compound 15 was also a
colorless solid [450 mg, 43.0% corrected; mp 90-92 oC,
recrystallized from hexane-EtOAc; Rf 0.23 (2:1 hexane-
EtOAc); [R]D -61.1° (c 1.2, CHCl3); IR (KBr) 3446 cm-1; m/ z
194 (M+)]. Anal. Calcd for C11H14O3 (14): C, 68.00; H, 7.27.
Found: C, 67.86; H, 7.31. Anal. Calcd for C11H14O3 (15): C,
68.00; H, 7.27. Found: C, 67.90; H, 7.30. The 1H and 13C
NMR data for 14 and 15 are detailed in Tables 1 and 2,
respectively.
The separation of 14 and 15 could be greatly facilitated by
derivatization to the monoacetates 16 and 17. Reduction of
compound 8 was carried out as above and after neutralization
and evaporation of the reaction mixture, the residue was
dissolved in pyridine, acetic anhydride was added, and the
mixture was stirred at rt overnight. After standard aqueous
workup, the residue was separated by flash chromatography
to afford the two monoacetates. Compound 16 was a colorless
oil [Rf 0.31 (6:1 hexane-EtOAc); [R]D +6.5° (c 0.89, CHCl3);
IR (neat) 1730 cm-1; m/ z 237 (M+ + 1)]. Anal. Calcd for
C13H16O3: C, 66.10; H, 6.78. Found: C, 65.80; H, 6.92.
o
Compound 17 was a colorless solid [mp 79-80 C, recrystal-
lized from EtOAc-hexane; Rf 0.27 (6:1 hexane-EtOAc); [R]D
+16.2° (c 0.91, CHCl3); IR (CHCl3) 1735 cm-1; m/ z 237 (M+
+
1)]. Anal. Calcd for C13H16O3: C, 66.10; H, 6.78. Found: C,
66.0; H, 6.85. For 1H and 13C NMR data, see Tables 1 and 2,
respectively.
1,6-An h ydr o-2,3-C-[(1R,3S,4S,5R)-4,5-epoxycyclopen tan -
1,3-d iyl]-2,3-d id eoxy-â-D-a llop yr a n ose (18). m-Chloroper-
oxybenzoic acid (80% purity, 159 mg, 0.737 mmol) dissolved
in CHCl3 (5 mL) was added dropwise to a solution of 14 (136
mg, 0.702 mmol) in CHCl3 (10 mL). After the solution was
stirred for 2h at rt, saturated NaHCO3 (15 mL) was added
and the biphasic mixture stirred an additional 0.5 h. The
aqueous layer was extracted with additional CHCl3 (3 × 10

3792 J . Org. Chem., Vol. 61, No. 11, 1996
Horton et al.
mL), and the organic layers were dried and evaporated to yield
18 as a white solid [131 mg, 88.6%; mp 154-156 C, recrystal-
lized from hexane-ether; Rf 0.40 (3:1 hexane-EtOAc); [R]D
+11.2° (c 0.8, CHCl3); IR (KBr) 3472 cm-1; m/ z 210 (M+)].
Anal. Calcd for C11H14O4: C, 62.81; H, 6.71. Found: C, 62.69;
H, 6.77. The 1H and 13C NMR spectra of 18 are listed in Tables
1 and 2, respectively.
high vacuum (0.03 mmHg), and after the solvent had been
removed, the resultant solid was heated to 80 °C for 1 h. The
vacuum was released, and the distillate, warmed to rt, was
then evaporated (12 mm) to afford a brown syrup. Chroma-
tography (3:1 hexane-EtOAc) gave a yellow syrup (12.0 mg,
4.6% yield) which contained 5 as the major component, as
o
1
judged by H NMR analysis.
1,6-An h yd r o-5′,4-ep oxy-2,3-C-[(1R,3S,4R,5R)-4-h yd r ox-
ycyclop en ta n -1,3-d iyl]-2,3-d id eoxy-â-D-a llop yr a n ose (19).
A solution of m-CPBA (80% purity, 113 mg, 0.534 mmol) in
CHCl3 (5 mL) was added dropwise to 15 (96.6 mg, 0.498 mmol)
in CHCl3 (10 mL). After 4 h, additional m-CPBA (20 mg, 0.116
mmol) was added and the mixture stirred an additional 3 h
at rt. Saturated NaHCO3 (15 mL) was added, the mixture was
stirred for 0.5 h, and the layers were separated. Workup was
then performed exactly as for compound 18 to yield, after
evaporation, 19 as a white solid [99.5 mg, 95.2%; mp 143-
145 oC, recrystallized from hexane-ether; Rf 0.24 (3:1 hexane-
EtOAc); [R]D -104.0° (c 1.0, CHCl3); IR (KBr) 3456 cm-1; m/ z
210 (M+)]. Anal. Calcd for C11H14O4: C, 62.81; H, 6.71.
Found: C, 62.73; H, 6.73. The 1H and 13C NMR spectra for
19 are detailed in Tables 1 and 2, respectively.
1,6-Di-O-a cetyl-2,3-C-[(3S,5R)-1-cyclop en ten e-3,5-d iyl]-
2,3-dideoxy-â-D-r ibo-h exopyr an os-4-u lose (20). Compound
8 (50.0 mg, 0.260 mmol) was dissolved in a mixture of
trifluoroacetic acid (0.1 mL) and acetic anhydride (1.5 mL).
The mixture was stirred at rt for 3 h and then evaporated to
a pale-brown syrup. Chromatography on silica gel (6:1 hex-
ane-EtOAc) provided 20 as a colorless oil [48 mg, 63%; Rf 0.25
(6:1 hexane-EtOAc); [R]D +99.4° (c 0.79, CHCl3); IR (neat)
1750 cm-1; m/ z 294 (M+)]. Anal. Calcd for C15H18O6: C, 61.22;
H, 6.12. Found: C, 61.12; H, 6.20. 1H and 13C NMR spectra
are detailed in Tables 1 and 2, respectively.
1,6-An h yd r o-2,3-C-[(3S,5R)-1-cyclop en t a n e-3,5-d iyl]-
2,3-dideoxy-â-D-r ibo-h exopyr an os-4-u lose (21). Compound
8 (1.15 g, 5.99 mmol) was dissolved in EtOAc (30 mL) and
hydrogenated under 42 psi in the presence of 10% Pt/C (600
mg) for 2 h. After removal of catalyst, the solution was
evaporated to afford 21 as a chromatographically homogenous
syrup [1.07 g, 92%; Rf 0.41 (6:1 hexane-EtOAc); [R]D +58.2°
(c 0.68, CHCl3); IR (neat) 1720 cm-1; m/ z 194]. Anal. Calcd
for C11H14O3: C, 68.04; H, 7.22. Found: C, 67.99; H, 7.35. 1H
and 13C NMR data can be found in Tables 1 and 2, respectively.
1,6-An h yd r o-2,3-C-[(3S,5R)-1-cyclop en t a n e-3,5-d iyl]-
2,3,-d id eoxy-â-D-gu lop yr a n ose (22), 1,6-An h yd r o-2,3-C-
[(3S,5R)-1-cyclop en ta n e-3,5-d iyl]-2,3-d id eoxy-â-D-a llop y-
r a n ose (23), a n d Th eir Resp ective Mon oa ceta tes (24 a n d
25). To a solution of norbornane 21 (1.297 g, 6.69 mmol) in
95% EtOH (20 mL) was added NaBH4 (0.40 g, 10.5 mmol) in
four portions over 10 min. After the reaction was stirred at
rt for 2 h, acetone (3 mL) was added to decompose the excess
of NaBH4 and then the solution was made neutral by stirring
with Dowex-50W H+ resin (1.6 g) for 0.5 h. After filtration,
the solids were washed with ethanol (10 mL) and the filtrate
was evaporated to afford a colorless syrup (1.14 g) that
contained two products by 1H NMR and TLC (Rf 0.35 and 0.40,
6:1 hexane-EtOAc). The crude mixture (1.14 g) was dissolved
in pyridine (15 mL) and acetic anhydride (10 mL), and the
mixture was stirred at rt overnight. After evaporation, the
residue was chromatographed (10:1 hexane-EtOAc) to afford
24 as a colorless solid [0.27 g, 17% yield; mp 85-87 oC,
recrystallized from EtOAc-hexane; Rf 0.42, 6:1 hexane-
EtOAc; [R]D -3.4° (c 1.77, CHCl3); IR (CH2Cl2) 1730 cm-1; m/ z
238]. Anal. Calcd for C13H18O4: C, 65.55; H, 7.56. Found:
C, 65.63; H, 7.62.
Meth od B. A glass chromatography column (12 mm i.d.)
was filled with silica gel to a depth of 12.5 mm and then
equilibrated with 2:1 hexane-EtOAc. The solvent layer was
lowered to that of the silica gel, a 10% solution of H2SO4 in
acetone (0.5 mL) was added, and the level was again lowered
to the top of the silica gel. Compound 4 (50 mg, 2.1 mmol)
was adsorbed onto silica gel (0.5 g). This was added to the
top of the column and then hexane-EtOAc was passed
through the silica gel at rt at a rate of ∼12 mL/min. After
200 mL of eluent had been collected, no further UV-active
material could be detected by TLC. The solvent was evapo-
rated to afford a yellow syrup (2.6 mg, 10%), which again
1
contained 5 as the major component by H NMR.
Meth od C (P r ep a r a tive Meth od ). Anhydrous aluminum
chloride (260 mg, 1.95 mmol) was added to an efficiently
stirred mixture of dry Et2O (30 mL) and pentane (22 mL)
which had been cooled to 0 °C under an argon atmosphere. A
solution of alkene 4 (1.07 g, 4.42 mmol) in Et2O (6 mL) was
added dropwise over a 2 min period. The resultant solution
was stirred for an additional 7 min at 0 °C and then quickly
transferred to a flash chromatography column (25 mm i.d.)
containing silica gel (4 cm depth). The solution was rapidly
passed through the silica gel with air pressure, followed by a
1:1 mixture of Et2O-pentane until no more UV-active material
could be detected by TLC (∼300 mL). Evaporation afforded a
brown syrup which was flash chromatographed using 5:1
pentane-Et2O to give 5 as a pale-yellow syrup (179 mg,
32.1%). Compound 5 could be obtained as an analytically pure
colorless syrup after a second chromatographic purification
using 5:1 pentane-Et2O ([R]D +412° (c 0.9, CHCl3), lit.17 [R]D
+331°; IR (neat) 1725 cm-1; m/ z 126 (M+)). Anal. Calcd for
C6H6O3: C, 57.14; H, 4.80. Found: C, 57.41; H, 4.88.
Rea ction of Isolevoglu cosen on e (5) w ith Cyclop en ta -
d ien e. A. Th er m a l Con d ition s. A mixture of freshly
distilled cyclopentadiene (0.20 mL, 2.3 mmol) and compound
5 (30.0 mg, 0.238 mmol) in benzene (3 mL) was boiled under
reflux for 1.5 h, additional cyclopentadiene (0.20 mL, 2.3 mL)
was added, and refluxing was continued an additional 4.5 h.
Evaporation of solvent afforded a yellow syrup which was
chromatographed to remove the most and least polar impuri-
ties. The middle fractions afforded a white solid (40.4 mg,
88.4%) which was found by 1H NMR to contain norbornene
adducts 8 and 10 in a 11.1:1.0 ratio.
B. Lew is Acid Ca ta lysis. Compound 5 (53.0 mg, 0.421
mmol) was dissolved in benzene (5 mL) kept at 48 °C, freshly
distilled cyclopentadiene (0.07 mL, 0.85 mmol) and 1 M ZnCl2
in Et2O (0.20 mL, 0.20 mmol) were added, and the mixture
was stirred for 45 min. The mixture was cooled to rt, Et3N
(0.10 mL) was added, and the mixture was passed through a
plug of silica gel. Evaporation of the solvent afforded a white
solid (66.8 mg, 82.7%) which contained compounds 8 and 10
1
in a 4.5:1 ratio by H NMR.
Rea ction of Isolevoglu cosen on e (5) w ith Cycloh exa -
d ien e: F or m a tion of Cycloa d d u ct 28. To a solution of 5
(71 mg, 0.57 mmol) in benzene (1 mL) were added ZnCl2 (0.57
mL, 0.57 mmol, 1M in Et2O) and cyclohexadiene (0.15 mL, 2.8
mmol). After 4 h at rt, additional cyclohexadiene (0.15 mL,
2.8 mmol) was added and stirring continued a further 20 h.
Et3N (0.56 mL, 1.1 mmol) was added, and the mixture was
passed through a short plug of silica gel with pressure.
Evaporation of solvent afforded a colorless syrup which was
chromatographed (3:1 hexane-EtOAc) to give 1,6-anhydro-2,3-
C-[3S,5R)-1-cyclohexene-3,6-diyl]-2,3-dideoxy-â-D-ribo-hexopy-
ranos-4-ulose (28) as a colorless solid [69.7 mg, 59% yield; mp
78-79 °C; Rf 0.56 (3:1 hexane-EtOAc); [R]D +67.2° (c 0.5,
CHCl3); IR (KBr) 1708 cm-1; m/ z 206 (M+)]. For 28: 1H NMR
(500 MHz, CDCl3) δ 1.25 (dt, J ) 11.8, 1 H), 1.30 (dt, J ) 2.0,
11.3, 1 H), 1.35 (dt, J ) 11.8, 3.3, 1 H), 1.60 (dt, J ) 11.8, 1
H), 2.23 (d, J ) 9.2, 1 H), 2.61 (m, J ) 7.4, 3.0, 2.8, 2 H), 3.19
Compound 25 was a colorless oil [0.61 g, 38% yield; Rf 0.33,
6:1 hexane-EtOAc; [R]D +24.2° (c 1.95, CHCl3); IR (neat) 1728
cm-1; m/ z 238]. Anal. Calcd for C13H18O4: C, 65.55; H, 7.56.
Found: C, 65.44; H, 7.66. See Tables 1 and 2, respectively,
for 1H and 13C NMR data.
P r ep a r a tion of 1,6-An h yd r o-2,3-d id eoxy-â-D-glycer o-
h ex-2-en opyr an -4-u lose (Isolevoglu cosen on e, 5). Meth od
A. Alkene 4 (500 mg, 2.07 mmol) and TsOH‚H2O (8.0 mg,
0.042 mmol) were dissolved in CH2Cl2 (5 mL) in a 50 mL round
bottom flask which was then connected to a trap held at -78
°C in an acetone-dry ice bath. The system was placed under

Synthesis and Chemistry of Isolevoglucosenone
J . Org. Chem., Vol. 61, No. 11, 1996 3793
(bd, J ) 3.3, 2.9, 1 H), 3.79 (m, 2 H), 4.24 (t, J ) 3.5, 1 H),
1,6-Anhydro-2-azido-2,3-dideoxy-R-D-ribo-hexopyranose (32)
was a colorless syrup [21.0 mg, 24% yield; Rf 0.18 (1:1 hexane-
5.44 (s, 1 H), 6.12 (t, J ) 7.1, 1 H), 6.30 (dt, J ) 7.4, 1 H); 13
C
NMR (125 MHz, CDCl3) δ 22.9, 26.6, 32.6, 33.1, 44.5, 48.4,
EtOAc); [R]D -75° (c 3.1, CHCl3); IR (CHCl3) 3500, 2100 cm-1
;
1
66.5, 78.0, 104.7, 131.6, 133.9, 206.4. Anal. Calcd for C12H14
O3: C, 69.88; H, 6.84. Found: C, 70.33; H, 6.88.
-
m/ z 101 (M+ - C2H3N, H)]. The H and 13C NMR constants
for compound 32 are given in Tables 3 and 4, respectively.
Anal. Calcd for C6H9N3O3: C, 42.10; H, 5.30; N, 24.56.
Found: C, 42.31; H, 5.34; N, 24.56.
R ea ct ion of Isolevoglu cosen on e (5) w it h 9,10-
Dim eth yla n th r a cen e: F or m a tion of Cycloa d d u ct 29. A
solution of 5 (100 mg, 0.794 mmol) and 9,10-dimethylan-
thracene (233 mg, 1.13 mmol) in benzene (10 mL) was refluxed
for 72 h. The solvent was evaporated and the pale-yellow solid
dissolved in hot ether and then cooled to crystallize excess 9,10-
dimethylanthracene, which was removed by filtration. Chro-
matography of the filtrate (3:1 hexane-EtOAc) gave 1,6-
anhydro-2,3-C-[9,10-dimethyl-9,10-dihydroanthracene-9,10-
diyl]-2,3-dideoxy-â-D-ribo-hexopyranos-4-ulose (29) as a white
solid [170 mg, 62% yield; mp 149.5-150.5 °C, recrystallized
from pentane-Et2O; Rf 0.53 (3:1 hexane-EtOAc); [R]D +61.7°
(c 1.7, CHCl3); IR (KBr) 1725 cm-1; m/ z 332 (M+)]. For 29:
1H NMR (500 MHz, CDCl3) δ 2.10 (s, 3 H), 2.17 (s, 3 H), 2.18
(d, J ) 9.4, 1 H), 2.57 (dd, 1 H), 3.52 (dd, J ) 7.6, 1 H), 3.71
(d, J ) 7.6, 1 H), 3.97 (d, J ) 5.3, 1 H), 5.70 (d, 1 H), 7.15-
7.35 (m, 8 H). 13C NMR (125 MHz, CDCl3) δ 15.9, 16.9, 42.7,
45.0, 49.5, 54.7, 68.3, 78.7, 101.6, 120.6-146.1, 207.4. Anal.
Calcd for C28H26O3: C, 79.49; H, 6.07. Found: C, 79.43; H,
6.08.
1,6-An h yd r o-2-a zid o-2,3-d id eoxy-r-D-er yth r o-h exop y-
r a n os-4-u lose (30). A solution of compound 5 (760 mg, 6.03
mmol), sodium azide (1.18 g, 18.15 mmol), and trifluoroacetic
acid (0.466 mL, 6.05 mmol) in dry THF (46 mL) was stirred
at rt for 10 h. The solvent was evaporated at rt to a thick
paste which was then extracted with successive portions (2 ×
50 mL) of pentane. After filtration, the organic extracts were
evaporated under reduced pressure at rt to afford crude 30 as
an unstable syrup [IR (neat) 2100, 1740 cm-1] which was used
without further purification.
B. With LS Selectr id e. Reduction of 30 (167 mg, 0.998
mmol) in THF (3 mL) with 1 M LS Selectride in THF (1.0 mL,
1.0 mmol) at -78 °C for 3 h afforded, after aqueous workup,
a yellow syrup (77 mg) which contained 32 as the major
component by 1H NMR. Crystallization from 1:1 Et2O-
pentane afforded 32 as a pale-yellow solid (18.6 mg, 11%).
C. With Diisobu tyla lu m in u m Hyd r id e in CH2Cl2.
A
solution of 30 (1.11 g, 6.58 mmol) in dry CH2Cl2 (75 mL) under
an Ar atmosphere was cooled to -78 °C, and then a 1.5 M
solution of DIBAH in toluene (5.20 mL, 7.80 mmol) was added
slowly. After 3 h of stirring at -78 °C, the mixture was
brought to rt and then water (1.20 mL, 66.0 mmol) was added
cautiously. After the solution was stirred for 10 min, MgSO4
was added, the resultant paste filtered, and the solid extracted
with EtOAc (2 × 10 mL). The combined filtrate and extracts
were evaporated, and the crude syrup chromatographed (3:2
hexane-EtOAc) to afford alcohol 33 (475 mg, 42%) as a
colorless syrup and alcohol 32 (336 mg, 30%) as a colorless
solid.
D. With Diisobu tyla lu m in u m Hyd r id e in THF .
A
solution of 30 (59.4 mg, 0.351 mmol) in dry THF (3 mL) was
cooled to -78 °C under Ar, a solution of 1.5 M DIBAH in
toluene (0.47 mL, 0.71 mmol) was added slowly, and the
resultant mixture stirred at -78 °C for 20 h. After the solution
was warmed to rt and stirred for 1.5 h, water (0.84 mL) was
added carefully and the mixture worked up as described in
the previous experiment to afford, after flash chromatography,
alcohol 33 as a colorless syrup (27.5 mg, 45.7%).
1,6:2,3-Dia n h yd r o-â-D-r ibo-h exop yr a n os-4-u lose (31).
A solution of compound 5 (169 mg, 1.34 mmol) and 3 M tert-
butyl hydroperoxide in isooctane (0.671 mL, 20.1 mmol) in
benzene (1 mL) was cooled to 0 °C, and then 3 M benzyltri-
methylammonium hydroxide (Triton B) in MeOH (0.01 mL,
0.03 mmol) was added. After the solution was stirred for 1 h
at 0 °C, the solvent was evaporated and the resultant syrup
purified twice by chromatography (1:2 hexane-EtOAc) to
afford 31 as a colorless solid [110 mg, 58%; mp 63-64 °C, lit.24
mp 64.5-65.5 °C; [R]D -7.4o (c 1.3, dioxane), lit.24 [R]D -4.5°
(c 1.0, dioxane); m/ z 142 (M+)]. The 1H and 13C NMR data
(Tables 3 and 4, respectively) were in excellent agreement with
those previously reported.24
E. With Tr iisobu tyla lu m in u m Hyd r id e in CH2Cl2. A
solution of freshly prepared 30 (99.7 mg, 0.59 mmol) in dry
CH2Cl2 (7.5 mL) under Ar was treated with 1 M TIBA in
toluene (1.18 mL, 1.18 mmol). After the solution was stirred
for 1 h, water (0.10 mL) was added carefully, and the mixture
was stirred for 2 h, and then worked up as described in the
previous experiments to afford, after chromatography, alcohol
33 as a colorless syrup (56.5 mg, 56%).
Acetolysis of 32 To Give 1,4,6-Tr i-O-a cetyl-2-a zid o-2,3-
d id eoxy-r,â-D-r ibo-h exop yr a n osid es 37. Azido alcohol 32
(361 mg, 2.11 mmol) was dissolved in Ac2O (7.46 mL) and CF3-
CO2H (5.1 mL), and the mixture was stirred at rt for 8 h. After
the solvent was evaporated, the resultant brown syrup (741
mg) was chromatographed (2:1 hexane-EtOAc) to afford
anomeric triacetates 37 as a colorless syrup (640 mg, 97%); Rf
0.30 (3:1 hexane-EtOAc); [R]D +84° (c 2.2, CHCl3); IR (neat)
2100, 1750 cm-1; m/ z 256 (M+ - CH3CO2). Anal. Calcd for
C12H17N3O7: C, 45.71; H, 5.44; N, 13.33. Found: C, 45.61; H,
5.45; N, 13.23.
Red u ction of Azid o Keton e 30: F or m a tion of Azid o
Alcoh ols 32 a n d 33. A. With Sod iu m Bor oh yd r id e.
A
solution of freshly prepared 30 (85.5 mg, 0.506 mmol) in 95%
EtOH (5 mL) was cooled to -78 °C, and NaBH4 (38 mg, 1.0
mmol) in EtOH (2 mL) was added dropwise. After being
warmed to rt and stirred overnight, the solution was treated
with Dowex-50W H+ resin (0.20 g) for 15 min, the resin filtered
off, and the filtrate evaporated to afford a colorless syrup
1
Ack n ow led gm en t. We thank the members of the
Laboratory of Analytical Chemistry (NIDDK) at the
National Institutes of Health, Bethesda, MD (Messrs.
Wesley White and Noel Whittaker and Dr. Herman
Yeh), and Miss J une Yun of NIDDK for their invaluable
assistance in obtaining mass and NMR spectra during
some of this work.
which, by H NMR analysis, contained 32 and 33 in a ratio of
0.64:1.00 and a trace of allylic alcohol 34. Separation on silica
gel (1:2 hexane-EtOAc) afforded two products. 1,6-Anhydro-
2-azido-2,3-dideoxy-R-D-xylo-hexopyranose (33) was a syrup
[40.4 mg, 46% yield; Rf 0.37 (1:1 hexane-EtOAc); [R]D -15.4°
(c 3.4, CHCl3); IR (neat) 3400, 2100 cm-1; m/ z 101 (M+
-
C2H3N3, H)]. 1H and 13C NMR data are given in Tables 3 and
4, respectively. Anal. Calcd for C6H9N3O3: C, 42.10; H, 5.30;
N, 24.56. Found: C, 42.97; H, 5.24; N, 24.27.
J O960032Y