Chemoenzymatic Iterative Synthesis of Difficult Linkages of Oligosaccharides on Soluble Polymeric Supports

Full text of DOI:10.1021/ol016466j

ORGANIC
LETTERS
2001
Vol. 3, No. 21
3265-3268
Chemoenzymatic Iterative Synthesis of
Difficult Linkages of Oligosaccharides
on Soluble Polymeric Supports
Fengyang Yan, Michel Gilbert, Warren W. Wakarchuk, Jean-Robert Brisson, and
Dennis M. Whitfield*
Institute for Biological Sciences, National Research Council of Canada,
100 Sussex DriVe, Room 3024, Ottawa, Ontario K1A 0R6, Canada
dennis.whitfield@nrc.ca
Received July 20, 2001
ABSTRACT
A trisaccharide donor containing a cis-Galpr(1f4)Galp linkage was prepared using a synthetic strategy based on chemoenzymatic
oligosaccharide synthesis on a soluble polymeric support. Significantly, only retaining glycosyltransferases gave complete reactions, whereas
inverting enzymes showed little or no activity with poly(ethylene glycol) (MPEG)-bound lactose as an acceptor. The MPEG-attached trisaccharide
was shown to bind to Verotoxin-1 by transfer NOE studies through the Galpr(1f4)Galp portion of the molecule.
Our research activities are directed toward the simplification
of the synthesis of oligosaccharides. Toward this goal, we
have developed methodological improvements to polymer-
supported and chemoenzymatic syntheses. One of the major
unsolved problems in polymer-supported oligosaccharide
synthesis is how to incorporate “difficult” linkages such as
cis linkages.1 One strategy is to prepare oligosaccharide
building blocks containing the “difficult” linkage and use
the building block in the preparation of larger oligosaccha-
rides. We demonstrate that oligosaccharide building blocks
containing cis linkages can be synthesized on polymeric
supports chemoenzymatically. This strategy combines the
advantage of the absolute stereochemical control of cis-
glycoside bond formation afforded by glycosyltransferases2
and the simple purifications from using polymeric supports.3
These efforts culminated in the synthesis of the Pk antigen
as a trisaccharide and a tetrasaccharide attached to the
polymer poly(ethylene glycol) (MPEG), CH3O(CH2CH2O)nH
via the linker dioxyxylene [DOX, -(O)CH2PhCH2(O)-].4
These oligosaccharides contain a cis-GalpR(1f4)Galp link-
age and are of interest because they could bind to the
Verotoxin from Escherichia coli O157.5
Initially, lactose was glycosidically bound to (MPEG)-
(DOX)OH 1 by glycosylation with the perbenzoylated
(2) (a) Endo, T.; Koizumi, S. Curr. Opin. Struct. Biol. 2000, 10, 536.
(b) Windmuller, R.; Schmidt, R. R. Tetrahedron Lett. 1994, 35, 7927.
(3) (a)Krepinsky, J. J.; Douglas, S. P. Carbohydrates in Chemistry and
Biology, Part I: Chemistry of Saccharides; Ernst, B., Hart, G. W., Sinay,
P., Ed.; Wiley-VCH: Weinheim, 2000. (b) Haase, W.-C.; Seeberger, P. H.
Curr. Org. Chem. 2000, 4, 481.
* To whom correspondence should be addressed. Phone: (613) 993-
5265. Fax: (613) 952-9092.
(1) (a) Belogi, G.; Zhu, T.; Boons, G.-J. Tetrahedron Lett. 2000, 41,
6969. (b) Flitsch, S. L.; Watt, G. M. Glycopeptides and Related Compounds;
Large, D. G., Warren, C. D., Ed.; Marcel Dekker: New York, 1997; p
207.
(4) Douglas, S. P.; Whitfield, D. M.; Krepinsky, J. J. J. Am. Chem. Soc.
1995, 117, 2116.
(5) (a) Kitov, P. I.; Sadowska, J. M.; Mulvey, G.; Armstrong, G. D.;
Ling, H.; Pannu, N. S.; Read, R. J.; Bundle, D. R. Nature 2000, 403, 669.
(b) Fan, E.; Merritt, E. A.; Verlinde, C. L. M. J.; Hol, W. G. J. Curr. Opin.
Struct. Biol. 2000, 10, 680.
10.1021/ol016466j CCC: $20.00 © 2001 American Chemical Society
Published on Web 09/18/2001

Scheme 1
Figure 1. Partial one-dimensional transient NOE spectra of 2 mM
polymer-bound trisaccharide 5 and 5 with Verotoxin-1 (200/1) in
D2O: (a) 1H NMR spectra of 5 in D2O; (b) 1D NOESY spectra of
5 and VT-1 in D2O; (c) 1D NOESY spectra of 5 in D2O.
to yield 10 and after cleavage to yield tetrasaccharide 11.9
Thus, the cis-GalpR(1f4)Galp linkage of 5 is incorporated
into a polymer-bound oligosaccharide. This sequence is a
thioglycoside donor 2 (see Scheme 1). All attempts to use
the more accessible peracetylated lactose donors were
frustrated by concomitant acetyl transfer to the polymer-
bound construct.6 Reactions with 2 proceed in essentially
quantitative yield to 3 free of this side reaction. The soluble
polymer-bound compounds can be purified by precipitation
with tert-butyl methyl ether (TBME) and reprecipitation from
absolute ethanol. After debenzoylation, the acceptor 4 can
be dissolved in aqueous buffers since (MPEG) is also readily
water soluble. Thus, the well-characterized enzyme R(1f4)-
galactosyltransferase (LgtC)7 and the nucleotide donor UDP-
Gal can be incubated with 4. The reaction proceeds to near
completion to yield polymer-bound Pk trisaccharide 5.
Purification involves centrifugation to remove insoluble
precipitates, lyophilization followed by extraction into
dichloromethane, filtration, and evaporation. This material
is nearly pure and can be further purified by reprecipitation
from ethanol. In some cases, treatment with acidic ion-
exchange resin is recommended to remove traces of para-
magnetic metals (e.g., Mn2+), which can interfere with NMR
studies. NMR studies are conveniently performed in CDCl3
or D2O solutions. Signals for 4 were not detectable, and only
those consistent with 5 were observed (see Figure 1a).
The trisaccharide can be cleaved from the (MPEG)(DOX)
by Sc(OTf)3 in the presence of acetic anhydride to yield
easily purifiable peracetylated sugar-DOXOAc products.
These procedures are easily scaled up to yield hundreds of
milligrams of trisaccharide 6 (see Scheme 2).8 A sequence
of hydrogenation to yield 7 followed by treatment with CCl3-
CN and DBU leads to the trichloroacetimidyl donor 8. This
donor can be used to glycosylate polymer bound acceptor 9
Scheme 2
(6) Nukada, T.; Berces, A.; Whitfield, D. M. J. Org. Chem. 1999, 64,
9030.
(7) Wakarchuk, W. W.; Cunningham, A.; Watson D. C.; Young, N. M.
Protein Eng. 1998, 11, 295.
3266
Org. Lett., Vol. 3, No. 21, 2001

prototype of a powerful strategy to overcome one of the
inherent problems in synthesizing oligosaccharides by polymer-
supported methods, namely absolute stereochemical control
of cis-glycoside formation. This new iterative protocol
extends our previous approaches to this problem based on
chemoenzymatic synthesis of oligosaccharide donors10 and
complements a recent entirely chemical approach, which
relies on developing a 100% stereoselective glycosylation
reaction.11
This chemoenzymatic protocol relies on finding enzymes
that are compatible with MPEG-bound acceptors. We have
examined a number of glycosyltransferases, which include
so-called retaining and inverting glycosyltransferases, for this
activity (see Table 1). Experiments with the inverting
enzymes, R(2f3)sialyltransferases and â(1f3)-N-acetyl-
glucosaminyltransferase (LgtA) from Neisseria meningitidis
or Campylobacter jejuni gave extremely low yields or no
reaction.12 Typically, the reactions were treated with more
enzyme and excesses of nucleotide sugar donors. However,
the glycosylation reaction yields were not increased. LgtC
is a retaining enzyme, so we tested 4 as a substrate for
another retaining enzyme, the bovine R(1f3)galactosyl
transferase.13 As expected, near-quantitative galactosylation
of the MPEG-bound acceptor 4 was achieved in the presence
Table 1. Glycosylation Reaction of
GalpR(1f4)Glcpâ-O-(DOX)(PEGM) 4 with a Number of
Retaining and Inverting Glycosyltransferases
yield (%)
(config (R/â))
enzyme (gene)
donor (R/â)
R(2f3) sialyltransferase
(NST-27)a
(CST-04)b
CMP-NeuNAc (â) ∼0
inversion (R)
∼0-2
inversion (R)
∼0-5
(CST-06)b
inversion (R)
â(1f3)-N-acetylglucosaminyl UPD-GlcNAc (R) ∼0-5
transferase (LgtA)c
R(1f3)galactosyltransferase
(bovine)
inversion (â)
>95
retention (R)
>95
retention (R)
UDP-Gal (R)
R(1f4)galactosyltransferase
(LgtC)
a For reaction conditions, see ref 20. b Acceptor 4 (1 mM), HEPES (50
mM), MgCl2 (20 mM), CMP-Neu5Ac (2 mM), 0.08 U of enzyme, 37 °C,
18 h. c For reaction conditions, see ref 20.
of only a small excess of uridine 5′-diphospho galactose
(UDP-Gal, 1.5 equiv). The 1D and 2D NMR spectroscopy
(gCOSY, gHSQC, and HMBC) of 14 indicated that the
glycosidic linkage is a cis-GalpR(1f3)Galp linkage (chemi-
cal shift of terminal Gal anomeric proton: δ ) 5.23 ppm, J
) 2.0 Hz). This allows for the synthesis of the so-called
xenotransplantation antigen GalpR(1f3)Galp14 (see Scheme
3).
(8) The stereo- and regiochemistry of R(1f4) linkage 6 was confirmed
by 1D and 2D NMR spectroscopy (gCOSY, gHSQC, and HMBC). Selected
1H and 13C NMR data for compound 6: 1H NMR (500 MHz, CDCl3) δ
3.58-3.64 (m, 1H, H-5B), 3.75 (t, 1H J ) 7.0 Hz, H-5C), 3.82 (t, 1H, J )
9.0 Hz, H-4B), 4.00 (d, 1H, J ) 2.0 Hz, H-4C), 4.08-4.18 (m, 4H, H-6B,
H-6C, 2 × H-6D), 4.46-4.51 (m, 2H, H-5D, H-6B), 4.51 (d, 1H, J ) 8.0
Hz, H-1C), 4.52 (d, 1H J ) 8.0 Hz, H-1B), 4.60 (d, 1H, J ) 12.5 Hz,
CHDOX), 4.73 (dd, 1H, J ) 3.0, 10.5 Hz, H-3C), 4.86 (d, 1H, J ) 12.0
Hz, CHDOX), 4.96 (dd, 1H, J ) 8.5, 10.0 Hz, H-2B), 4.98 (d, 1H, J ) 3.5
Hz, H-1D), 5.07 (dd, 1H, J ) 8.0, 9.2 Hz, H-2C), 5.09 (s, 2H, DOXCH2),
5.16 (t, 1H, J ) 9.0 Hz, H-3B), 5.17 (dd, 1H, J ) 3.5, 11.0 Hz, H-2D), 5.38
(dd, 1H, J ) 3.5, 11.0 Hz, H-3D), 5.58 (d, 1H, J ) 2.5 Hz, H-4D); 13C
NMR (50.32 MHz, CDCl3,) δ 60.24 (C-6D), 61.26 (C-6C), 62.12 (C-6B),
65.90 (CH2), 67.04 (C-5D), 67.10 (C-3D), 67.86 (C-4D), 68.80 (C-2D), 68.95
(C-2C), 70.26 (CH2), 71.71 (C-5C), 71.78 (C-2B), 72.57 (C-5B), 72.76 (C-
3C), 73.08 (C-3B), 76.37 (C-4B), 76.88 (C-4C), 99.00 (C-1B), 99.59 (C-1D),
101.03 (C-1C).
Scheme 3
(9) The â(1f6) regio- and stereochemistry of 11 was confirmed by
measuring 1D and 2D NMR spectroscopy (gCOSY, gHSQC, and HMBC).
Selected 1H NMR and MS data for compound 11: 1H and 13C NMR (500
MHz, CDCl3) δ 3.62-3.66 (m, 1H, H-5B), 3.73-3.83 (m, 3H, H-5C, H-6A,
H-4B), 3.89 (dd, 1H, J ) 4.0, 10.5 Hz, H-6A), 3.96-4.05 (m, 2H, H-5A,
H-4C), 4.09-4.19 (m, 4H, H-6B, 2 × H-6D, H-6C), 4.41-4.46 (m, 1H, H-6C),
4.46-4.54 (m, 2H, H-6B, H-5D), 4.52 (d, 1H, J ) 7.5 Hz, H-1C), 4.59 (d,
1H, J ) 8.0 Hz, H-1B), 4.69 (d, 1H, J ) 8.5 Hz, H-1A), 4.69 (d, 1H, J )
11.5 Hz, CHDOX), 4.73 (dd, 1H, J ) 2.0, 10.5 Hz, H-3C), 4.92 (dd, 1H,
J ) 8.5, 10.0 Hz, H-2B), 4.93 (d, 1H, J ) 12.0 Hz, CHDOX), 4.98 (1H, J
) 3.0 Hz, H-1D), 5.04 (s, 2H, DOXCH2), 5.11 (dd, 1H, J ) 7.5, 10.5 Hz,
H-2C), 5.18 (dd, 1H, J ) 3.5, 11.0 Hz, H-2D), 5.20 (t, 1H, J ) 9.5 Hz,
H-3B), 5.33 (dd, 1H, J ) 3.0, 10.5 Hz, H-3A), 5.39 (dd, 1H, J ) 3.0, 10.5
Hz, H-3D), 5.59 (2H, H-4A, H-4D), 5.72 (dd, 1H, J ) 8.0, 10.0 Hz, H-2A);
13C NMR (50.32 MHz, CDCl3) δ 60.29(C-6D), 61.37 (C-6C), 62.09 (C-6B),
65.96 (CH2), 67.09 (C-3D), 67.17 (C-5D), 67.90 (C-6A), 67.89 (C-4D), 68.98
(C-4A), 69.52 (C-2D), 69.62 (C-2C), 69.95 (C-2A), 70.00 (CH2), 71.66 (C-
2B), 71.79 (C-3A), 71.88 (C-5C), 72.73 (C-5B), 72.74 (C-5A), 72.85 (C-3C),
73.05 (C-3B), 76.32 (C-4B), 77.23 (C-4C), 99.48 (C-1A), 99.67 (C-1D), 100.27
(C-1B), 101.03 (C-1C); MS (MALDI) calcd for C70H82O36Na 1521.45, found
m/z 1521.12 (M + Na+).
(10) Mehta, S.; Gilbert, M.; Wakarchuk, W. W.; Whitfield, D. M. Org.
Lett. 2000, 2, 751.
(11) Zhu, T.; Boons, G.-J. J. Am. Chem. Soc. 2000, 122, 10222.
(12) (a) Wakarchuk, W. W.; Martin, A.; Jennings, M. P.; Moxon, E. R.;
Richards, J. C. J. Biol. Chem. 1996, 271, 19166, (b) R(2f3)Sialyltrans-
ferases from C. jejuni: Gilbert, M.; Brisson, J.-R.; Karwaski, M.-F.;
Michniewicz, J.; Cunningham, A.-M.; Wu, Y.; Young, N. M.; Wakarchuk,
W. W. J. Biol. Chem. 2000, 275, 3896.
A recent report successfully used glycosidases with our
(MPEG)(DOX) system, but the maximum reported yields
were 24%, which would compromise our new iterative
(13) Blanken, W. M.; Van den Eijnden, D. H. J. Biol. Chem. 1985, 260,
12927.
Org. Lett., Vol. 3, No. 21, 2001
3267

protocol.15 A recent report using a dendritic linker system
attached to a PEG derivative also encountered difficulties
in obtaining quantitative yields.16 The origin of this inhibition
of the transferases by MPEG-bound acceptors is not clear.
We have done a number of control experiments that rule
out simple explanations. For example, the deacetylated
lactoseDOX glycoside 12a is an acceptor for some of the
inactive transferases, ruling out a simple conformational
explanation. Also, incubating a standard lactose-FCHASE17
acceptor 12b in the presence of (MPEG)(DOX)OH 1 does
not inhibit the inverting transferase of R(2f3)sialyltrans-
ferase from N. meningitidis. Similarly, incubating 12a in the
presence of 4 and the R(2f3)sialyltransferase from C. jejuni
(CST-06) did not inhibit the sialylation of 12a, while 4 was
not sialylated. These direct competition experiments rule out
nonspecific inhibition by PEG.
small molecule with ωτc < 1 and, hence, positive NOEs;
see Figure 1c. Once bound to the protein, the NOEs across
the GalpR(1f4)Galp linkage become negative (ωτc > 1),
whereas those in the DOX linker remain positive; see Figure
1b. This clearly indicates that the major binding is between
the sugar and the protein and not the linker or the PEG.
Furthermore, without sophisticated numerical analyses of the
NOE data, one can conclude that the GalpR(1f4)Galp
portion of the molecule binds to the protein in agreement
with previous NMR studies.19 These interesting motional
properties of MPEG-bound oligosaccharides are currently
being studied in more detail to ascertain if this provides a
general method for determining binding sites of protein-
bound oligosaccharides.
Acknowledgment. We thank Ms. Melissa J. Schur for
preparation of enzymes, Dr. James L. Brunton for providing
Verotoxin-1, Ms. Suzon Larocque for NMR technical help,
and Mr. Don Krajcarski for MS analyses. We also thank
Dr. James C. Richards for discussion and critical reading of
the manuscript. This work is part of the Scientific Technical
Cooperation Program between NRC (Canada) and CNRS
(France). This is NRC paper 42446.
We currently hypothesize a “strangulation” effect in which
binding of the acceptor to the enzyme active site leads to
metastable binding of the PEG chains to the protein and
hence inhibition of transferase activity. Without the nucle-
ation by sugar-protein binding, the PEG-protein interaction
is too weak to lead to enzyme inhibition. We are currently
examining more transferases and testing different experi-
mental protocols to optimize these reactions.
Supporting Information Available: 1H and 13C NMR
spectra of 6, 8, 11, and 14, general procedure for enzymatic
reaction, preparation of donor 8, and glycosylation on MPEG.
This material is available free of charge via the Internet at
http://pubs.acs.org.
To study the mechanism(s) of MPEG-attached oligosac-
charides binding to proteins, we performed NMR studies of
5 in D2O solution. Construct 5 was shown to bind to the
Verotoxin-1 from E. coli O15718 by 1H NMR transfer NOE
measurements.19 In the absence of protein, 5 behaves like a
OL016466J
(14) (a) Fang, J.; Li, J.; Chen, X.; Zhang, X.; Wang, J.; Guo, Z.; Zhang,
W.; Yu, L.; Brew, K.; Wang, P. G. J. Am. Chem. Soc. 1998, 120, 6635. (b)
Hanessian, S.; Saavedra, O. M.; Mascitti, V.; Marterer, W.; Oehrlein, R.;
Mak, C.-P. Tetrahedron 2001, 57, 3267.
(15) Schmidt, D.; Thiem, J. J. Chem. Soc, Chem. Commun. 2000, 1919.
(16) Lubineau, A.; Malleron, A.; Le Narvor, C. Tetrahedron Lett. 2000,
41, 8887.
(17) FCHASE is a fluorescein derivative; see: Gilbert, M.; Cunningham,
A.-M.; Watson, D. C.; Martin, A.; Richards, J. C.; Wakarchuk, W. W. Eur.
J. Biochem. 1997, 249, 187.
(18) Clark, C.; Bast, D.; Sharp, A. M.; St. Hilaire, P.; Agha, R.; Stein,
P. E.; Toone, E. J.; Read, R. J.; Brunton, J. L. Mol. Microbiol. 1996, 19,
891.
(19) (a) Shimizu, H.; Field, R. A.; Homans, S. W.; Donohue-Rolfe, A.
Biochemistry 1998, 37, 11078. (b) Thompson, G. S.; Shimizu, H.; Homans,
S. W.; Donohue-Rolfe, A. Biochemistry 2000, 39, 13153. (c) Ling, H.;
Boodhoo, A.; Hazes, B.; Cummings, M. D.; Armstrong, G. D.; Brunton, J.
L.; Read, R. J. Biochemistry 1998, 37, 1777.
(20) Yan, F. Y.; Wakarchuk, W. W.; Gilbert, M.; Richards, J. C.;
Whitfield, D. M. Carbohydr. Res. 2000, 328, 3.
3268
Org. Lett., Vol. 3, No. 21, 2001