O-benzoyl-R-D-Galf 1-phosphate triethylammonium salt us-
ing a reported protocol.5a The benzoyl groups were subse-
quently removed using a triethylamine-methanol-water
solution to afford phosphorylated monosaccharide 3. Puri-
fication was accomplished by ion-exchange chromatography
to provide 3 as the bisammonium salt.
Both the model coupling and the key reaction were
followed using 31P NMR spectroscopy. Each component of
the reaction mixture, including intermediates and products,
could be distinguished.11 We found that when the coupling
reactions were conducted at 0 °C, only a slight excess of
electrophile 4 (1.2 equiv) was needed for complete consump-
tion of the sugar 1-phosphate. Under these conditions, little
or no degradation of the product to the cyclic phosphodiester
and 5′-UMP was observed.
Our UDP-Galf 2 coupling procedure consists of several
steps. The electrophilic UMP-N-methylimidazolide 4 is
formed by the reaction of the triethylammonium salt of 5′-
UMP with an excess of trifluoroacetic anhydride in the
presence of triethylamine and N,N-dimethylaniline in aceto-
nitrile.8 Excess trifluoroacetic anhydride is added to increase
the solubility of the electrophile in the reaction medium; its
addition results in acylation of the 2′- and 3′-hydroxyl groups
as well as the phosphoryl group. The nucleophilic catalyst,
N-methylimidazole, is added to provide the electrophilic
derivative 4. This compound can be detected by 31P NMR
spectroscopy, HPLC, and TLC. An excess of activated 4 is
treated with the tributylammonium salt of R-D-Galf 1-phos-
phate (3). To minimize side reactions of the product, the
reaction is quenched with aqueous ammonium acetate buffer
(pH 7), which also promotes the removal of the O-
trifluoroacetyl groups. After workup, the product 2 is isolated
by HPLC purification.
A number of parameters were found to affect the yield of
the coupling reaction. Various reaction conditions were tested
in the model coupling of commercially available R-D-Galp
1-phosphate to form UDP-Galp 1. We altered the counterion
for R-D-Galp 1-phosphate, the number of equivalents of
activated 5′-UMP 4, the reaction duration, and the reaction
temperature.
With regard to counterion, bisammonium salts of both 3
and R-D-Galp 1-phosphate were found to be insoluble in
acetonitrile. The tributylammonium salt of R-D-Galp 1-phos-
phate was completely soluble in acetonitrile, but the corre-
sponding triethyl- and tetraethylammonium derivatives were
not. Thus, we employed the tributylammonium salt of 3 in
the key coupling reaction.
Under optimized conditions, the tributylammonium salt
of R-D-Galf 1-phosphate (3) was treated with UMP-N-
methylimidazolide 4 (1.2 equiv) at 0 °C. The sugar 1-phos-
phate 3 was completely consumed after 2 h. According to
the 31P NMR spectrum, the proportion of UDP-Galf 2 at this
stage was 83% and the amount of cyclic phosphodiester
produced was low (17%). This ratio compares well with that
observed using other methods; the highest yield of UDP-
Galf 2 reported previously was 50% after 18-19 h. With
the less reactive UMP-imidazolide, degradation of the
product to the cyclic phosphodiester and 5′-UMP is a more
significant side reaction.5a We found that ammonium acetate
buffer (pH 7, 250 mM) is the most effective solution for
quenching the reaction. The product 2 was purified by ion-
exchange HPLC. Slight degradation (loss of ca. 4%) of the
product to R-D-Galf 1,2-cyclic phosphate and 5′-UMP (1:1
ratio) occurs during purification. This protocol affords UDP-
Galf 2 reproducibly in 35% yield.
In conclusion, we devised an effective and reproducible
coupling method to prepare UDP-Galf 2. Our protocol
involves reaction of a sugar 1-phosphate and the activated
5′-N-methyl phosphorylimidazolide nucleoside. The in-
creased reactivity of the electrophilic phosphoryl group used
here eliminates the long reactions times typically required
for nucleoside diphospho sugar synthesis. We found that this
method is effective for preparing a highly labile biosynthetic
intermediate. We anticipate that it may also facilitate the
preparation of diphosphoryl bonds embedded in other reac-
tive or unstable products.
The extent of completion of the coupling reaction de-
pended on the number of equivalents of activated UMP-N-
methylimidazolide 4 added. For nucleoside triphosphate
synthesis, the pyrophosphate nucleophile is used in excess
(2-fold) relative to the electrophile.8 This ratio results in a
suppression of the reaction of the product nucleoside
triphosphate with the activated NMP-N-methylimidazolide,
a process that produces the symmetrical dinucleoside tetra-
phosphate. In the formation of nucleoside diphospho sugars,
however, it is desirable to use the electrophile in excess,
rather than the valuable sugar 1-phosphate. The disadvantage
of this approach is that any hydrolysis of UMP-N-meth-
ylimidazolide 4 produces 5′-UMP, which can react with the
electrophile 4 to form the dimer, UPPU. The presence of
this species complicates purification of the desired product.
Thus, conditions were sought that would maximize the
formation of the product while minimizing formation of
UPPU.
Acknowledgment. This research was supported by the
NSF and the Dreyfus Foundation. A.L.M. thanks the NIH
for a postdoctoral fellowship (GM20582). We thank D.
Perreault and C. Sherrill for conducting preliminary experi-
ments.
Supporting Information Available: Experimental pro-
cedure for the coupling to produce UDP-Galf 2. This material
OL016170D
(11) Relative to external aqueous 85% H3PO4: R-D-galactose 1,2-cyclic
phosphate (10.0 ppm for Galp and 16.9 ppm for Galf2a), 5′-UMP (0.5-1.1
ppm), galactose 1-phosphate (-0.2 ppm for Galp 11 and 0.3 ppm for Galf
3), UMP-N-Me-imidazolide (-10.9 ppm), product UDP-galactose (-11.1
(d), -12.6 (d) ppm for Galp 1 and -11.3 (d), -12.6 (d) ppm for Galf 2),
and side product UPPU (-11.2 ppm).
Org. Lett., Vol. 3, No. 16, 2001
2519