Henry condensations with 4,6-O-benzylidenylated and non-protected D-glucose and L-fucose via DBU-catalysis

Article abstract of DOI:10.1016/S0040-4039(03)01328-5

Mechanistic intermediates, and thermodynamically favored side products, in the Henry condensations of partially protected and non-protected pyranoses with a free anomeric hemiacetal function with nitromethane in various solvents for the kinetically controlled syntheses of C-glycopyranosides in the presence of DBU/molecular sieve catalyst system were identified.

Full text of DOI:10.1016/S0040-4039(03)01328-5

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 44 (2003) 5495–5498
Henry condensations with 4,6-O-benzylidenylated and
non-protected -glucose and -fucose via DBU-catalysis
D
L
Pasit Phiasivongsa, Vyacheslav V. Samoshin and Paul H. Gross*
Department of Chemistry, University of the Pacific, 3601 Pacific Avenue, Stockton, CA 95211, USA
Received 18 March 2003; revised 20 May 2003; accepted 20 May 2003
Abstract—Mechanistic intermediates, and thermodynamically favored side products, in the Henry condensations of partially
protected and non-protected pyranoses with a free anomeric hemiacetal function with nitromethane in various solvents for the
kinetically controlled syntheses of C-glycopyranosides in the presence of DBU/molecular sieve catalyst system were identified.
© 2003 Elsevier Science Ltd. All rights reserved.
Carbohydrates play an important role in the biological
activity of glycoconjugates such as glycolipids, blood-
group antigens and complex glycoconjugates of plant,
bacterial or viral origin.1 A particular class of carbohy-
drate derivatives, C-glycoside carbohydrate analogs are
important enzyme inhibitors.2 Being structurally, chem-
ically, and conformationally similar to O-glycosides3
but acid- and enzymatically stable, aminomethyl C-gly-
cosides are potentially more stable as mimetics and
pseudo-substrates.4 They are obtainable from
nitromethyl C-glycosides.4b
We had synthesized7 directly cyclic compounds in
aprotic solvents with a highly active bifunctional8a cata-
lyst 2-hydroxypyridine8b and DBU/molecular sieve7a
with an improved7b yield of 77%. We avoided protic
solvents in view of the quoted (Scheme 3) pKa of DBU,
which would provide just another way of creating RO−,
in ROH. The cyclic 3b was produced in a single step
along with two minor byproducts7a 2b and 4b.
One of the most common methods for the chain exten-
sion of carbohydrates is the Henry condensation of
pentoses and hexoses with nitromethane in the presence
of a strong base. In 1946 Sowden and Fisher condensed
4,6-O-benzylidene-
the presence of methoxide to give acyclic
D
-glucose 1b with nitromethane in
-heptitol 2b
D
and cyclic nitromethyl b- -glucopyranoside 3b in 21
D
and 5% yield, respectively.5 Later, Petrus et al.
improved the total yields of unprotected nitromethyl
Scheme 1.
D
-hexopyranosides, but their approach is lengthy and
complicated, requires careful workup, with adjustment
of pH, and produces isomeric mixtures.6a Strong bases
(HO−, MeO−) in these procedures in protic solvents5
lead to acyclic compounds (i.e. 2b, Scheme 1). Their hot
dehydration6b (DS>0; DG=DH−TDS) gives cyclic
forms, reverting only very slowly to acyclics, at low
temperatures. Presumably, acyclics, having more
hydroxyl groups, are stabilized by solvation in protic
solvents.
Figure 1. TLC analysis (SiO2; 6:1 CHCl3:MeOH) of reaction
of 4,6-O-benzylidene-D-glucopyranose with CH3NO2 and
DBU/molecular sieve, in THF (Due to weaker spotting, 4b is
not visible on the 50 h plate).
Keywords: nitromethane; C-glycoside; DBU.
* Corresponding author. Fax: +1-209-946-2607; e-mail: vsamoshi@
uop.edu
0040-4039/03/$ - see front matter © 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0040-4039(03)01328-5

5496
P. Phiasi6ongsa et al. / Tetrahedron Letters 44 (2003) 5495–5498
The actual catalytic molecule, whether it was 2-HP, a
known 1,3-proton transfer catalyst for pyranose ring
opening in benzene,8b the anionic form of 2-HP, DBU,
or the protonated form of DBU, had not been unequiv-
ocally determined. The effects of aprotic solvents and
of 4,6-O-acetalation with respect to the system’s cata-
lytic efficiency had yet to be clarified. Also, this conden-
sation procedure had not been extended to simple
unprotected sugars are typically not very soluble in
THF or dioxane. But when acetonitrile, pyridine or a
mixture of these was used as the solvent and the
reaction temperature kept at 50°C, the condensations
were fairly clean.10 The condensation progressed to
completion within one day, and only nitromethyl b-D-
glucopyranoside 3a was detected and isolated when 5:1
acetonitrile/pyridine (38% yield) or pyridine (57% yield)
were used as solvents (Scheme 1). For completion and
maximal yield, a stoichiometric amount of DBU was
needed.
sugars such as
L-fucose. Since few 4,6-O-alkylidene
glycopyranoses are easily accessible, advantages of such
extensions had not been fully realized.
We reinvestigated our original condensation7 of
Condensations with L
-fucose10 were monitored by TLC
and flash chromatography, and gave sequential prod-
ucts as outlined in Scheme 2, summarized in Table 1.
Under procedure A and B, within 12 h more than half
nitromethane with 4,6-O-benzylidene-
1b to nitromethyl 4,6-O-benzylidene b-
D
-glucopyranose
-glucopyran-
D
oside 3b and discovered that 2-HP was unnecessary. A
time study, with careful TLC analysis and flash chro-
matography of the reaction mixture with DBU and
molecular sieve revealed sequential products, which are
conveniently explained as outlined in the mechanisti-
cally simplified Scheme 1.
of the starting
sponding 1,6-dideoxy-1-nitro-
tol 6. By the end of day 3, most of
transformed into the acyclic heptitol 6, and the subse-
-fucopyran-
L
-fucose 5 was converted into its corre-
-glycero- -manno-hepti-
-fucose had been
L
D
L
quent conversion into cyclic nitromethyl b-
L
oside 7 was followed for another 5 days, during which
time a significant portion of the desired cyclic product
had also been slowly converted into 1,1,6-trideoxy-1-
The condensation progressed similarly in THF (Fig. 1)9
and dioxane, with 4,6-O-benzylidene-
via the acyclic 5,7-O-benzylidene-1-deoxy-1-nitro-
glycero- -gulo-heptitol 2b, which subsequently changed
into nitromethyl b- -glucopyranoside 3b. The former
D-glucopyranose
D
-
nitro-2-nitromethyl-
L
-galactoheptitol 8. Thus, the maxi-
required
D
mal yield for desired cyclic product
7
D
minimizing decomposition and the yields of 6 and 8.
The best system discovered is shown for procedure D
(Table 1). Under this condition, only 7 was isolated as
the product. In addition, there were decomposition
products that did not move on the column or on TLC.
was observed accumulating before being further con-
verted during at least 50 h at room temperature to the
cyclic product 3b, along with some 5,7-O-benzylidene-
1,2-dideoxy-1-nitro-2-nitromethyl- -glucoheptitol 4b, in
D
yields similar to those already reported.7 In THF or
dioxane, accumulation of the final acyclic product 4b
was dependent on temperature, time, and solvent: less
decomposition was observed with dioxane. In these
condensations, we could not determine if the presence
of 2-HP had any effect at all, at any stage of the
reaction. In any case, the first acyclic heptitol 2b and
the desired cyclic 3b were kinetically controlled interme-
diates, which, given enough time, were transformed into
the ‘thermodynamic’ product 4b, by a still slower over-
all rate. Compound 2b,5 3a,6 3b, and 4b7 have already
been fully characterized.
For the condensation of 4,6-O-benzylidene-D-glucopy-
ranose, its slow dissolution rate in THF and acetonitrile
was likely rate-determining step. Acyclic intermediate
2b was observed accumulating, but most of the starting
When similar conditions were employed for unpro-
tected
D
-glucose 1a, there was no reaction for pro-
longed periods of time. Heating only resulted in almost
complete decomposition. This was expected since
Scheme 2.
Table 1. Reaction parameters and yields for Henry condensations of 510
Procedure c
Solvent
Ratio of 5:DBU:2-HP
%/Yielda 6:7:8
tb
Tc (°C)
A
B
C
D
E
F
THF
THF
1.0:1.2:0.5
1.0:1.2:0
1.0:1.2:0.5
1.0:1.2:0
1.0:1.2:0
1.0:1.2:0
1.0:1.2:0
5:35:11
5:51:5
0:49:8
0:62:0
0:30:25
0:27:31
0:5:50
9
9
1.5
1.5
1
70
70
50
50
25
25
25
Dioxane
Dioxane
MeCN
Pyridine
DMF
1
0.5
G
a Isolated yields.
b Time in days.
c Temperature.

P. Phiasi6ongsa et al. / Tetrahedron Letters 44 (2003) 5495–5498
5497
trast to the Fischer–Sowden scheme, we could not find
by NMR or TLC any evidence for the 2-epimers of 6 or
2b. Also, replacement of DBU by bases not capable of
forming bifunctional catalysts such as TEA, Hunig’s
base, DABCO or Proton-Sponge resulted in no reac-
tion. In the presence of excess nitromethane (pKa=10),
DBU (pKA=24) exists predominantly in its protonated
form together with nitromethyl anion, which has only
the basicity of phenolate. DBU-H+ is an even weaker
1b also remained unconverted. In pyridine, complete
dissolution of non-protected -glucose (1a, Scheme 1)
was observed within a few minutes. These results sug-
gest that solubility might be the most important factor
that affected Henry condensations of unprotected sug-
ars. Instructive in this respect were the condensations of
D
L
-fucose under procedure A and B (Table 1): Complete
dissolution of -fucose occurred in about 5 h, after
L
which viscosity of the reaction mixture increased and
acyclic heptitol 6 was precipitating. This precipitation
from THF explained why conversion of 6 into 7 was
slow. When more polar solvents were employed to
increase the solubility of acyclic heptitol to accelerate
this conversion, another difficulty appeared: The ratios
of products significantly shifted all the way towards the
thermodynamically favored 1,1,6-trideoxy-1-nitro-2-
base, but it may complex with L-fucose to catalyze the
glycosidic ring opening and formation of its aldehyde
form (5i) in one step. After Henry addition of
nitromethyl anion and stabilization of the resultant
alkoxide 6i by an adjacent 2-hydroxyl proton, forma-
tion of the intermediate acyclic heptitol 6, DBU-H+
could catalyze its dehydration with assistance of molec-
ular sieve into intermediate 7i and subsequent cycliza-
tion into 7. From this perspective, under procedure A
and C, addition of 2-HP might just mean a better
1,3-proton transfer, which also skewed the reaction
towards 7i and allowed for the increased addition of
another mole of nitromethane into acyclic 8. So, not
only was 2-HP not required for the initial ring opening
in our Henry condensations, it actually had an adverse
effect on the yield of the desired cyclic product (proce-
dure A and C, Table 1).
nitromethyl- -galactoheptitol 8 (as seen in procedure E
L
to G). This shift was also observed when condensations
were performed under procedure A and C for a pro-
longed period of time, in the presence of 2-HP,
although not as efficiently.
A possible mechanism of these Henry condensations
must account for the isolated intermediates and prod-
ucts outlined in Schemes 1 and 2. The most probable
mechanism, for the example of L-fucose, is illustrated in
The conversion 7 to 7i is formally analogous to 5 to 5i.
DBU-H+ may also serve as a dehydration catalyst
(6“7i). The reverse reaction (7i“6) is trimolecular, and
in presence of molecular sieves irreversible. Reaction
(5i“5) would be fast, but is omitted for clarity, since 5i
is depleted as shown. The dehydration, and the CꢀC
bond forming reactions (5i“6; 7i“8) provide the driv-
ing force.
Scheme 3 (mechanistic intermediates are labeled ‘i’).
This scheme accounts for all facts known to us. The
condensations required a 2-hydroxyl proton for the
Henry adduct alkoxide. 2,3,4-Protected glycosides that
lack the required stabilizations failed to undergo Henry
condensation with our system.7
Several aspects of the described condensations make a
‘classical’ base catalyzed mechanism unlikely. In con-
It is also very likely that solvents play an important role
in the stabilization of alkoxides 6i (Scheme 3): The
most polar solvent, DMF, showed the fastest reaction.
In good solvents, where solubility was not a problem,
with molecular sieve, dehydration of acyclic heptitol 6
into nitroalkene 7i and cyclization into the desired
cyclic nitromethyl b- -fucopyranoside 7 occurred read-
L
ily. Good solvents, such as acetonitrile, pyridine and
DMF (procedure E to G, Table 1), may have skewed
the reaction towards the intermediate nitroalkene 7i,
which then reacted with another mole of nitromethane
to form almost exclusively the thermodynamically
favored 1,1,6-trideoxy-1-nitro-2-nitromethyl-L-galacto-
heptitol 8. High reaction rates are generally associated
with low selectivity and thermodynamic control of
products. The discovery of the acyclic 8 as thermody-
namic end product with excess MeNO2 at low tempera-
tures was actually to be expected for reasons discussed
in the introduction of this article.
The actual mechanism is probably governed by a com-
bination of several factors. In any case, some intermedi-
ates and sequences have been established, and the
conditions required for kinetically controlled synthesis
of the desired cyclic products from 4,6-O-benzylidene
D
-glucose, unprotected D-glucose and L-fucose have
been determined. This methodology should be extend-
able to other sugars.
Scheme 3.

5498
P. Phiasi6ongsa et al. / Tetrahedron Letters 44 (2003) 5495–5498
Acknowledgements
(MeOH); TLC system (EtOAc): Rf 0.13; IR 3200–3400,
2940 (OH), 1550, 1650 (NO2); 1H NMR (CD3OD) 4.88
(dd, CHaNO2), 4.48 (dd, CHbNO2), 4.34–4.41 (m, J1,2
=
We would like to thank the University of the Pacific for
financial support, and Dr. Xiaoling Li for helpful
discussion.
8.4 Hz, H1), 3.76 (dd, H2), 3.86 (dd, H3), 3.41 (dd, H4),
4.06 (dq, H5), 1.24 (d, CH3); 13C NMR (CD3OD) 80.8
(CH2NO2), 74.5 (C1), 67.5 (C2), 70.1 (C3), 72.2 (C4), 70.5
(C5), 20.0 (CH3); MS-ESI+ (in MeOH ionization) m/z
250.2 [5M+4Na+MeOH]5+
, 291.0 [7M+3Na+MeOH+
3H]6+. Anal. calcd for C7H15O7N (225.26): C, 37.32; H,
6.71; N, 6.22. Found; C, 36.83; H, 6.69; N, 6.12.
References
Fractions containing the major product at 425–600 ml of
procedure A and B were collected and were concd in
vacuo. The residual crystals were recrystallized from
THF/(iPr)2O to give 7: [0.44 g, 2.1 mmol (35%)] (A); and
[0.64 g, 3.1 mmol (51%)] (B); mp 182°C; [h]2D0 −26.0°,
c=1 (MeOH); TLC system (EtOAc): Rf 0.18; IR 3200–
1. Flowers, H. M. Adv. Carbohydr. Chem. Biochem. 1981,
39, 279.
2. (a) Bimwala, R. M.; Vogel, P. Helv. Chim. Acta 1989, 72,
1825; (b) Hanessian, S.; Pernet, A. G. Adv. Carbohydr.
Chem. Biochem. 1976, 33, 111.
3. (a) Kishi, Y. Pure Appl. Chem. 1989, 61, 313; (b) Levy,
D. E.; Tang, C. The Chemistry of C-Glycosides; Perga-
mon, Elsevier Science Ltd: Oxford, UK, 1995; p. 4.
4. (a) Fritz, H.; Lehman, J.; Schlesselman, P. Carbohydr.
Res. 1983, 113, 71; (b) Gross, P. H. Carbohydr. Poly.
1998, 37, 215.
1
3400, 2940 (OH), 1550, 1650 (NO2); H NMR (CD3OD)
4.82 (dd, CHaNO2), 4.49 (dd, CHbNO2), 3.90 (dt, J1,2
=
9.3 Hz, H1), 3.44 (dd, H2), 3.50 (dd, H3), 3.63 (dd, H4),
3.90 (dq, H5), 1.20 (d, CH3); 13C NMR (CD3OD) 78.4
(CH2NO2), 78.2 (C1), 69.2 (C2), 73.4 (C3), 76.2 (C4), 75.7
(C5), 16.9 (CH3); MS-ESI+ (in MeOH ionization) m/z
230.0 [M+Na]+. Anal. calcd for C7H13O6N (207.18): C,
40.58; H, 6.32; N, 6.76. Found; C, 40.40; H, 6.32; N,
6.69.
5. Sowden, J. C.; Fischer, H. O. L. J. Am. Chem. Soc. 1946,
68, 1511.
6. (a) Petrus, L.; Bystricky, S.; Sticzay, T.; Bilik, V.
Chemicke Zvesti 1982, 36, 103; (b) Ko¨ll, P.; Petrusova,
M.; Petrus, L.; Kopf, J. Carbohydr. Res. 1993, 248, 349.
7. (a) Drew, K.; Gross, P. H. Tetrahedron 1991, 47, 6113;
(b) Wang, X.; Gross, P. H. J. Org. Chem. 1995, 60, 1201.
8. (a) Swain, C. G.; Brown, J. F. J. Am. Chem. Soc. 1952,
74, 2538; (b) Li, J. B. Aldichim. Acta 1972, 5, 5.
9. Procedure: 1,8-Diazabicyclo[5.4.0]undec-7-ene (DBU; 1.4
ml, 9.3 mmol) was slowly added to a mixture of 4,6-O-
Procedure C: procedure A was modified by substituting
dioxane (50 ml) for THF and decreasing the reaction time
to 36 h. Fractions containing the major product at 400–
700 ml were collected, and were concd in vacuo. The
residue was recrystallized to give 7: 0.62 g, 3.1 mmol
(49%).
Procedure D–G: procedure B was modified by substituting
50 ml of either dioxane (D), MeCN (E), pyr (F) or DMF
(G) for THF and changing the reaction time to 36, 24, 24,
and 12 h, respectively. Fractions containing the major
product at about 400–700 ml were collected, were concd
in vacuo, and were recrystallized to give 7 in the follow-
ing yields: [0.78 g, 3.8 mmol (62%)] (D); [0.38 g, 1.8 mmol
(30%)] (E); [0.38 g, 1.6 mmol (27%)] (F); and [0.62 g, 0.3
mmol (5.0%)] (F).
benzylidene-
D-glucopyranose 1b (5.0 g, 19.0 mmol),
MeNO2 (10 ml, 190 mmol), THF (50 ml), and molecular
,
sieve (3 A, 10 g). The slurry was stirred under N2 at rt for
50 h and was filtered. The filtrate was concd in vacuo.
The residual oil was chromatographed (SiO2, 8:1:1
CH2Cl2/EtOAc/Et2O). Fractions containing the slowest
product at 375–475 ml were collected and were concd in
vacuo. The residue was recrystallized from THF to give
2b: 0.064 g, 0.2 mmol (1.1%); mp 164–165°C. Fractions at
75–275 ml were concd in vacuo and were recrystallized
from THF/(iPr)2O to give 3b: 3.6 g, 11.6 mmol (61%);
mp 210–212°C. Fractions at 300–375 ml were concd in
vacuo and were recrystallized from THF/(iPr)2O to give
4b: 0.15 g, 0.39 mmol (2.1%); mp 124–126°C.
Fractions containing the fastest product at about 200–300
ml of procedures A, B, C, E, F, G were collected, and
were concd in vacuo. The residue was recrystallized from
(iPr)2O to give 8 in the following yields: [0.18 g, 0.67
mmol (11%)] (A); [0.08 g, 0.31 mmol (5.0%)] (B); [0.13 g,
0.49 mmol (8.0%)] (C); [0.41 g, 1.5 mmol (25%)] (E); [0.51
g, 1.9 mmol (31%)] (F); and [0.82 g, 3.0 mmol (50%)] (G);
mp 138°C; [h]2D0 +7.5°, c=1 (MeOH); TLC system (A): Rf
0.30; IR 3200–3400, 2940 (OH), 1550, 1650 (NO2); 1H
NMR (CD3OD) 4.96 (dd, CHaNO2), 4.80 (dd, CHa%NO2),
4.69 (ddd, CHbNO2, CHb%NO2), 3.31 (m, J1,2=6.6 Hz,
H1), 4.01 (dd, H2), 3.60 (dd, H3), 3.37 (dd, H4), 4.06 (dq,
H5), 1.23 (d, CH3); 13C NMR (CD3OD) 93.1 (CH2NO2),
92.4 (CH2NO2), 58.5 (C1), 89.6 (C2), 84.6 (C3), 92.1 (C4),
87.0 (C5), 37.5 (CH3); MS-ESI+ (in MeOH ionization)
10. General procedure. DBU (1.1 ml, 7.3 mmol) was slowly
added to a mixture of L-fucose 5 (1.0 g, 6.1 mmol) with
(procedure A and C, Table 1) or without (procedure B, D
to G, Table 1) 2-HP (0.6 g, 6.1 mmol), MeNO2 (10 ml,
,
190 mmol), solvent (50 ml), and molecular sieve (3 A, 5
g). The slurry was stirred under N2 at 25–70°C for 0.5 to
9 days and was filtered. The filtrate was concd in vacuo.
The residual oil was chromatographed (SiO2, EtOAc).
Fractions containing the slowest product at 725–875 ml
were collected and were concd in vacuo. The residual
crystals were recrystallized from EtOAc/THF to give 6:
0.07 g, 0.0003 mol (5.0%); mp 138°C; [h]2D0 +10.0°, c=0.7
m/z 248.0 [4M+5MeOH+5H]5+
.
Anal. calcd for
C8H16O8N2 (268.22): C, 35.82; H, 6.01; N, 10.44. Found;
C, 35.27; H, 6.15; N, 10.14.