(δP 17.2)† and UMP 3 (δP 1.6). The nucleotide 3 then reacted
with the imidazolide 4 (δP Ϫ8.7), which is in excess, to give sym-
metrical diuridine diphosphate (δP Ϫ9.6, d, JP, P 21.5 Hz). After
the work-up, UDP-Galf 1 (a dipotassium salt) was isolated in
23% yield using gel-chromatography on a column of Sephadex
G-15 followed by reversed-phase C-18 HPLC.
It is noteworthy that coupling of the galactosyl phosphate
2 and UMP-morpholidate in pyridine in the presence of
1H-tetrazole, which was reported18 to be an efficient catalyst in
phospho-morpholidate condensations, produced the nucleotide
sugar 1 in 12% yield. The highest proportion of the product
(corresponding to 25–30% yield) in the reaction mixture was
observed at 15–17 h using 31P NMR spectroscopy.
minϪ1. Reversed-phase HPLC was done on a semipreparative
C-18 column (Chromsphere 5, 25 × 1 cm) using 0.02 mol dmϪ3
aq. KH2PO4 as eluent at 3 cm3 minϪ1 with UV-monitoring at
270 nm. Anhydrous pyridine and DMF were purchased from
Aldrich. Dichloromethane, toluene and pyridine were freshly
distilled from CaH2. All the solutions containing UDP-Galf
1 were concentrated in vacuo (oil pump, ≈0.01 mmHg) at
18–20 ЊC.
Uridine 5Ј-monophosphate disodium salt (Sigma) was con-
verted to the triethylammonium salt 3 (containing 1.42 equiv.
1
of Et3N, detected by H NMR spectrometry) by passing its
aq. solution through a column (15 × 2.5 cm) of Dowex
50W × 8 (triethylammonium form); the column was washed
with water, then appropriate fractions were pooled and
freeze-dried.
The structure of compound 1 was unambiguously confirmed
by NMR and mass spectrometry. The 31P NMR spectrum
exhibited two doublet signals (δP Ϫ10.1, d, Pα and Ϫ11.4, d, Pβ,
JP, P 20.9 Hz), which are characteristic for nucleoside diphospho-
sugars.9,14–16,18 The presence of the (1→5)-pyrophosphate bridge
was confirmed by the C-1, C-2, H-1 and H-2 signals of the
-galactofuranose residue and C-4 and C-5 signals of the
-ribofuranose residue (see Experimental section). These
signals were shifted as a result of the α- and β-effects of phos-
phorylation and coupled with phosphorus. The α-configuration
of the -galactofuranosyl phosphate fragment was evident
from the value of JH1,H2 4.2 Hz, which was close to that pub-
lished11 for the α-phosphate 2. The coupling constant value
for the corresponding isomeric β-phosphate was reported11 to
Dibenzyl 2,3,5,6-tetra-O-benzoyl-ꢀ-D-galactofuranosyl
phosphate 9
1-O-Acetyl-2,3,5,6-tetra-O-benzoyl-α,β--galactofuranose
6
{[α]D25 ϩ9.5 (c 1, CHCl3); δH (CDCl3) 2.10 (s, Acα), 2.22 (s, Acβ),
4.69–4.86 (3 H, m, 4-H and 6-H2), 5.64 (d, J2,3 1.4, 2-Hβ), 5.75
(dd, J3,4 4.8, 3-Hβ), 5.79 (dd, J2,3 7.1, 2-Hα), 5.91 (dt, J4,5
=
J5,6a = 6.0, J5,6b 3.8, 5-Hα), 6.12 (dt, J4,5 = J5,6a = 4.0, J5,6b 7.0,
5-Hβ), 6.27 (dd, J3,4 6.3, 3-Hα), 6.55 (s, 1-Hβ), 6.70 (d, J1,2 4.8,
1-Hα) and 7.15–8.20 (20 H, m, 4 × Ph); α:β = 1:3.7} was
prepared from methyl α,β--galactofuranoside 510 by conven-
tional benzoylation with benzoyl chloride in pyridine followed
by acetolysis with 1.4% H2SO4 in Ac2O (v/v).
1
be about 1.7 Hz. In general, our H, 13C and 13P NMR data
were in agreement with those published9 for UDP-Galf 1,
apart from distinctive assignments of signals for C-2, C-3 and
C-4 of the -ribofuranose residue, which were amended in the
light of the 13C NMR data of uridine 5Ј-monophosphate 3.19
The molecular mass of compound 1 was confirmed by electro-
spray mass spectrometry. The signals in the mass spectrum (see
Experimental section) corresponded to the pseudo-molecular
ions for the uridine diphosphohexose (M, 566.055 for the free
acid 1).
To a stirred solution of the tetrabenzoate 6 (3.13 g, 4.9 mmol)
in CH2Cl2 (30 cm3) cooled to 0 ЊC was added a 33% solution (10
cm3) of HBr in AcOH. After storage for 2 h at 0 ЊC and an
additional 2 h at rt, the mixture was concentrated to dryness
and the residue was azeotroped with toluene (3 × 30 cm3) to
give the galactofuranosyl bromide 8.
To a stirred mixture of dibenzyl hydrogen phosphate (1.91 g,
6.86 mmol) and Et3N (0.97 cm3, 6.95 mmol) in toluene (10 cm3)
was added a solution of the prepared bromide 8 in the same
solvent (25 cm3) and the stirring was continued for a further
16 h. The solid (Et3NHBr) was filtered off and the solvent was
removed under reduced pressure. FCC (toluene–ethyl acetate,
9:1 v/v) of the residue produced the protected α--galacto-
furanosyl phosphate 9 (2.48 g, 59%) as a solid, mp 110–111 ЊC;
[α]D26 ϩ51.3 (c 1, CHCl3); δP Ϫ2.35 {lit.,11 mp 112–113 ЊC; [α]D
ϩ54.6 (c 1, CHCl3); δP Ϫ3.66}. The 1H and 13C NMR data were
virtually identical to those published.11
It should be noted that UDP-Galf 1 is rather unstable: it was
hydrolysed (≈95% extent), forming -galactose and uridine 5Ј-
diphosphate in 0.015 mol dmϪ3 aq. solution after 24 h at 22 ЊC.
Nevertheless, a frozen solution of the compound was stored
at Ϫ80 ЊC for several months without detectable degradation.
A purity control could be properly performed using reversed-
phase C-18 HPLC.
The nucleotide sugar 1 is currently being tested as a substrate
for the Leishmania β--galactofuranosyl transferase and the
results will be published in due course.
ꢀ-D-Galactofuranosyl bis(triethylammonium) phosphate 2
The galactosyl phosphate 9 (1.82 g, 2.13 mmol) was depro-
tected by successive hydrogenation over palladium hydroxide
(20 wt% Pd on carbon, contains ≈50% of water; 1 g, ≈0.94
mmol Pd) in ethyl acetate–triethylamine (10:1) and debenzoyl-
ation with MeOH–water–triethylamine (5:2:1 v/v/v) following
the published11 procedure. The resulting syrup was dissolved
in water and applied to a column (12 × 2.5 cm) of Amberlite
IRA-420 (HCO3Ϫ-form) resin. The column was first washed
with water and then with a gradient of 0.1→0.4 mol dmϪ3 aq.
triethylammonium hydrogen carbonate. Appropriate fractions
were freeze-dried to give the phosphate 2 (0.865 g, 88%) as a
hygroscopic amorphous solid, [α]D26 ϩ44.5 (c 1, water); δP 0.43
{lit.,11 for the bis(cyclohexylammonium) salt [α]D ϩ41 (c 1,
Experimental
General procedures
Mps were determined on a Reichter hot-plate apparatus and
are uncorrected. Optical rotations were measured with a
Perkin-Elmer 141 polarimeter; [α]D-values are given in units
of 10Ϫ1 deg cm2 gϪ1. NMR spectra (1H at 300 MHz, 13C at
75 MHz, and 31P at 121 MHz) were recorded with a Bruker
DPX-300 spectrometer for solutions in deuterium oxide, unless
otherwise indicated. Chemical shifts (δ in ppm) are given
relative to those for Me4Si (for H and 13C) and external aq.
1
85% H3PO4 (for 31P); J-values are given in Hz. ES mass spectra
were recorded with a Micromass Quattro system (Micromass
Biotech, UK). Flash-column chromatography (FCC) was
performed on Kieselgel 60 (0.040–0.063 mm) (Merck). Gel-
1
water); δP 2.50}. The H NMR data were nearly identical to
those published.11
filtration chromatography was performed on
a column
(95 × 2.5 cm) of Sephadex G-15 in water, flow rate 1.5 cm3
Uridine 5Ј-(ꢀ-D-galactofuranosyl diphosphate), dipotassium salt 1
To a solution of uridine 5Ј-monophosphate 3 (93 mg, 0.20
mmol of the triethylammonium salt containing 1.42 equiv. of
Et3N) in DMF (1.5 cm3) was added 1,1Ј-carbonyldiimidazole
(97 mg, 0.60 mmol) and the mixture was kept for 3 h at rt.
† The 13P NMR chemical-shift value is characteristic for similar cyclic
phosphates:17 for uridine 2Ј,3Ј-cyclophosphate, δP 17.6; for 4-nitro-
phenyl α--galactopyranoside 3,4-cyclophosphate, δP 16.8.
890
J. Chem. Soc., Perkin Trans. 1, 2000, 889–891