Stereoselective enzymatic galactosylation of C-glucosides

Article abstract of DOI:10.1039/a701747b

The enzyme β-1,4-galactosyl transferase from bovine colostrum (GalT) is able stereoselectively to galactosylate C-glucosides (i.e. 1 and 4), precursors of stable glycoconjugate analogues, and a systematic investigation of the structural modifications at C

Full text of DOI:10.1039/a701747b

View Article Online / Journal Homepage / Table of Contents for this issue
Stereoselective enzymatic galactosylation of C-glucosides
Luigi Panza,*,a Pietro L. Chiappini,a Giovanni Russo,a Daniela Montib and
Sergio Riva *,b
a
Dipartimento di Chimica Organica e Industriale, Università di Milano, via Venezian,
21 I-20133 Milano, Italy
b Istituto di Chimica degli Ormoni, CNR, via Mario Bianco, 9 I-20131 Milano, Italy
The enzyme â-1,4-galactosyl transferase from bovine
colostrum (G alT) is able stereoselectively to galactosylate
C-glucosides (i.e. 1 and 4), precursors of stable glyco-
conjugate analogues, and a systematic investigation of the
structural modifications at C-1 and/or C-5 of the glycosides
that can be accepted by this enzyme has been undertaken,
adding information to the currently accepted model of
substrate binding into the G alT active site.
C-Glycosides have recently become one of the major topics in
carbohydrate chemistry since they are challenging synthetic
targets and can be used as stable chemical analogues of natur-
ally occurring bioactive compounds for metabolic studies.
There is an increasing demand for these compounds,1 especially
in the fast growing area of glycobiology, in which they could
offer easy access to stable analogues of glycoconjugates, pro-
vided that the carbohydrate part can be easily manipulated to
give more complex sugar structures.
In the context of our general interest in biocatalysis,2 we have
recently become involved in the isolation and synthetic applica-
tion of the β-1,4-galactosyl transferase from bovine colostrum
(GalT).3 This enzyme catalyses the transfer of -galactose from
Scheme 1
its uridine 5Ј-diphosphate (UDP) derivative to the hydroxy
group at C-4 of -glucose and N-acetyl--glucosamine.
Recently it has been shown that this enzyme is also able to
transfer -galactose to various substrate acceptors with some
degree of structural variability.4 However, despite the large
number of reports, to our knowledge there are no examples
of galactosylation of C-glycosides. We initially studied the
behaviour of GalT with the simple hydroxymethyl derivative 1,
prepared by catalytic hydrogenolysis of the known C-(2,3,4,6-
tetra-O-benzyl-β--glucopyranosyl)methanol5 with Pd/C in
MeOH, as this transformation could offer access to glycosyl-
ated C-glycosides and also offer additional valuable information
on the substrate specificity and stereoselectivity of this bovine
transferase. Even though 1 is a meso-compound, it was reason-
able to expect, as a consequence of the stereochemical require-
ments of GalT described so far,4,6 that this enzyme should be
able to differentiate between the two enantiotopic hydroxy
groups at the C-3 and C-5 positions of 1 (formerly C-2 and C-4
of the parent glucose), to give 1a as the only product.
dithiothreitol (DTT) (1 m) and NaN3 (0.01%). The usual
work-up 3 isolated 26 mg (37% isolated yield) of a single prod-
uct derived from galactose and 1. The presence of galactose is
1
clearly shown by a broad doublet at δ 5.73 in the H NMR
spectrum of the fully benzoylated 1a, attributable to H-4Ј.
Although in some cases the in situ generation of UDP-
galactose from UDP-glucose with UDP-glucose epimerase
resulted in a transfer of glucose, we never observed the form-
ation of glucose-containing products. However, the symmetry
of the molecule of the acceptor C-glucoside made it difficult
to determine the position of attachment of galactose to 1.
To solve this problem, we synthesized compound 2 (from 14
via 15, see Scheme 2),7 the C-2 deuteriated analogue of 1, to
As shown in Table 1, the meso-C-glucoside 1 was a good
substrate for GalT, its relative rate of transformation being
similar to that of propyl β--glucopyranoside -9 and higher
than that of methyl -8 or phenyl β--glucopyranoside -10.
Compound 1 (40 mg, 0.205 mmol) and UDP-glucose (158
mg, 0.26 mmol) were reacted for 4 days at 30 ЊC in the presence
of UDP-glucose epimerase (10 units), GalT (1.75 units),† alka-
line phosphatase (35 units) and 5 mg of α-lactalbumine in 4 ml
of 50 m TRIS buffer (pH 7.4) containing MnCl2 (1 m),
Scheme 2
introduce an asymmetric element without modifying the
reactivity of the acceptor (see Table 1). Enzymatic galactosyl-
ation gave a deuteriated pseudodisaccharide (26% isolated yield)
which was fully benzoylated (BzCl, py, 60 ЊC, overnight) and
analysed by a 1H–1H COSY experiment. Complete proton
assignment allowed us to assign the signal at relatively high field
(δ 4.22) to H-5 and to identify the product as the expected
† GalT was purified from bovine colostrum as described in ref. 3 to give
an enzymatic sample with a specific activity of 67 U mgϪ1 (see ref. 3 and
references cited therein).
J. Chem. Soc., Perkin Trans. 1, 1997
1255

View Article Online
Table 1 Relative rate of galactosylation of C-glucosides and - and
-glucose derivatives by action of GalT
make the reaction much too slow to be detected under our
conditions and, consequently, are useless for synthesis.
For this reason, despite the fact that they are so similar to our
meso-C-glucoside 1, in some cases differing only in the substitu-
ent RЉ, none of the -sugars considered was a substrate for
GalT. On the other hand, in addition to the natural substrate
-glucose and the previously described -xylose,11 1,5--
anhydroglucitol -7 was also accepted by this enzyme and the
corresponding pseudodisaccharide -7a was isolated (15%
isolated yield) and characterized.
In conclusion, in addition to their interest for synthesis, our
results add information to the currently accepted model of sub-
strate binding into the active site of GalT proposed by
Wiemann et al.11 and Yu et al.,12 especially with regard to the
possible modifications at C-1 and at C-5. As it has recently been
shown that this GalT can be used for galactosylation of N- and
O-linked glycopeptides,10,14 work is in progress to apply the
results obtained so far with 1 and 4 to the preparation of stable
analogues of glycopeptides.
Entry
Compound
Relative rate (%)a
Product
1
2
3
4
5
1
10
2.5
10
3.5
10
0
β-1,5
β-1,4
β-1,4
β-1,4
β-1,5
—
-8
-9
-10
2
6
3
7
8
9
10
11
12
13
4
10
100
0
0.8
0
β-1,5
β-1,4
—
-5
-5
-7
-7
-6
-6
β-1,4
—
2.5
0
β-1,4 and β-1,1b
a
Relative to -glucose (entry 8) and determined according to literature
methods.6b,13 b The initial rate refers to the whole reactivity of the
substrate.11
compound 2a. Comparison with the NMR spectra of the pre-
References
viously synthesized non-deuteriated analogue, after benzoyl-
1 C. Bertozzi and M. Bednarsky, in Modern Methods in Carbohydrate
Synthesis, ed. S. H. Khan and R. A. O’Neill, Harwood Academic
Publisher, Amsterdam, NL, 1996, p. 316.
2 (a) G. Carrea, G. Ottolina and S. Riva, Trends Biotechnol., 1995, 13,
63; (b) B. Danieli and S. Riva, Pure Appl. Chem., 1994, 66, 2215; (c)
K. Faber and S. Riva, Synthesis, 1992, 895; (d) L. Panza, S. Brasca,
S. Riva and G. Russo, Tetrahedron: Asymmetry, 1993, 4, 931; (e)
L. Panza, M. Luisetti, E. Crociati and S. Riva, J. Carbohydr. Chem.,
1993, 12, 125; (f) S. Riva, J. Chopineau, A. P. G. Kieboom and
A. M. Klibanov, J. Am. Chem. Soc., 1988, 110, 584.
ation, allowed us to assign to it the structure 1a. In fact, the 13
C
1
1
NMR and H NMR spectra of 2a and 1a as well as the H
NMR of their fully benzoylated derivatives were very similar.
As expected, the only differences were a consequence of the
presence of deuterium at C-2 of 2a. Specifically, the 13C NMR
spectrum of 2a lacks the signal at δ 79.5 (C-2) that is present in
1
the spectrum of 1a, while the H NMR spectrum of benzoyl-
ated 2a lacks a signal at δ 4.03 (ddd, H-2) and the coupling
constants of the correlated protons.
3 D. Monti, E. Giosuè, S. Riva and L. Panza, Gazz. Chim. Ital., 1996,
126, 303.
To extend the scope of this preliminary work to the synthesis
of precursors of stable glycopeptides we considered the amino
C-glucoside 3 in which the amino group could allow the linkage
with an aspartic acid to give analogues of N-linked glycosyl
amino acids. Not only was this compound not a substrate, it
was an inhibitor of the catalytic activity of the enzyme.‡ In a pre-
vious investigation we had found that propyl 4-amino-4-deoxy-
β--glucopyranoside (11) was also an inhibitor of GalT (Ki ≅
90 m). In addition, it is known from the literature6 that two
other amino derivatives, 12 and 13, are both poor substrates for
the enzyme. Therefore, we might speculate that a free amino
group on the sugar acceptor does not interact productively with
the active site of GalT.8 On the other hand, the N-protected
benzyloxycarbonyl derivative 4 was a good substrate (KM ≅ 20
m) for the enzyme and the corresponding pseudodisaccharide
was isolated (68% isolated yield) and characterized.9
Because of our interest in C-analogues of glycopeptides,
we also tried the enzymatic galactosylation of the C-(β--
xylopyranosyl)methanol, which corresponds to 1,5-anhydro--
glucitol (-7), since β-xylosyl glycosides of serine are present in
glycoproteins. We were encouraged by the fact that Wong et al.
succeeded in their galactosylation using GalT.10 Surprisingly,
under our conditions, derivative -7 was not a substrate for this
enzyme. This result prompted us to determine the structural
modifications at C-1 and/or at C-5 that can be accepted by the
enzyme. In some cases, structural changes at these positions (as
in β--glucopyranosides, -xylose and our meso-C-glucoside)
do not prevent enzymatic reaction, while in other cases (as
in -7 or other compounds reported in the literature11,12) no
conversion was observed.
4 (a) C.-H. Wong, L. R. Halcomb, Y. Ichikawa and T. Kajimoto,
Angew. Chem., 1995, 107, 521; Angew. Chem., Int. Ed. Engl., 1995,
34, 521; (b) M. M. Palcic, O. P. Srivastava and O. Hindsgaul,
Carbohydr. Res., 1987, 159, 315; (c) C. Augè, S. David, C. Mathieu
and C. Gautheron, Tetrahedron Lett., 1984, 25, 1467.
5 T. V. RajanBabu and G. S. Reddy, J. Org. Chem., 1986, 51, 5458.
6 (a) L. J. Berliner, M. E. Davis, K. E. Ebner, T. A. Beyer and E. Bell,
Mol. Cell. Biochem., 1984, 62, 37; (b) C.-H. Wong, Y. Ichikawa,
T. Krach, C. Gautheron-Le Narvor, P. D. Dumas and G. C. Look,
J. Am. Chem. Soc., 1991, 113, 8137.
7 Compound 14 was obtained according to L. Lay, F. Nicotra,
L. Panza and G. Russo, Synlett, 1995, 167.
8 Hindsgaul also observed an inhibitive action of an aminosugar
acceptor with a different galactosyl transferase: T. L. Lowary and
O. Hindsgaul, Carbohydr. Res., 1994, 251, 33.
9 Selected NMR data: benzoylated 1a: δH (CDCl3, J/Hz) 5.58 (t, J 9.7,
1 H, H-3), 4.88 (d, J 7.8, 1 H, H-1Ј), 4.22 (dd app. as t, J 9.5, 1 H, H-
5), 4.03 (m, 1 H, H-2); benzoylated 2a: δH (CDCl3) 5.57 (dd app. as t,
J 9.7, 1 H, H-3), 4.88 (d, J 8.0, 1 H, H-1Ј), 4.22 (dd app. as t, J 9.4, 1
H, H-5); benzoylated 4a: δH (CDCl3) 4.88 (d, J 8.0, 1 H, H-1Ј), 4.20
(dd app. as t, J 9.4, 1 H, H-5); 1a: δC(D2O) 103.2, 79.5, 79.0, 78.6,
76.2, 75.7, 72.9, 71.3, 69.9, 68.9, 61.4 (2C), 60.8; 2a: δC(D2O) 103.3,
79.0, 78.5, 76.3, 75.7, 72.9, 71.3, 69.8, 68.9, 61.4 (2C), 60.8; 4a:
δC(D2O) 159.4, 137.4, 129.7 (2C), 129.3, 128.6 (2C), 103.8, 79.5,
79.1, 78.7, 76.6, 76.2, 73.5, 71.9, 71.6, 69.5, 68.1, 67.9, 61.9, 61.3,
42.4; 5a: δC(D2O) 103.7, 79.9, 79.5, 76.9, 76.2, 73.4, 71.8, 69.9, 69.5,
69.3, 61.8, 61.1.
10 C.-H. Wong, P. Schuster, P. Wang and P. Sears, J. Am. Chem. Soc.,
1993, 115, 5893.
11 T. Wiemann, Y. Nishida, V. Sinnwell and J. Thiem, J. Org. Chem.,
1994, 59, 6744.
12 L. Yu, R. Cabrera, J. Ramirez, V. A. Malinovskii, K. Brew and
P. G. Wang, Tetrahedron Lett., 1995, 36, 2897.
13 (a) E. A. Davidson, Biochim. Biophys. Acta, 1959, 33, 238; (b) D. K.
Fitzgerald, B. Colvin, R. Mawal and K. E. Ebner, Anal. Biochem.,
1970, 43.
14 (a) M. Schultz and H. Kunz, Tetrahedron Lett., 1992, 33, 5319; (b)
J. Thiem and T. Wiemann, Angew. Chem., 1990, 102, 78; Angew.
Chem., Int. Ed. Engl., 1990, 29, 80; (c) C. Unverzagt, H. Kunz and
J. J. Paulson, J. Am. Chem. Soc., 1990, 112, 9308.
The results, summarized in Table 1, show that modifications
at either C-1 (entries 1–5, 7 and 10) or C-5 (entry 12) are
accepted by bovine GalT. However, contemporaneous modifi-
cations at C-1 and C-5 lead to no reaction or, most probably,
‡ When 20 m 3 was added to 20 m glucose (approximately twice its
KM value), a 50% drop in the reaction rate was observed. A detailed
kinetic analysis was beyond the purpose of this work and will be
reported in due course.
Paper 7/01747B
Received 12th March 1997
Accepted 13th March 1997
1256
J. Chem. Soc., Perkin Trans. 1, 1997