Glycosylation of a neoglycoprotein by using glycosynthase and thioglycoligase approaches: The generation of a thioglycoprotein

Article abstract of DOI:10.1002/anie.200503900

(Chemical Equation Presented) Remodeling: Glycosylation of a neoglycoprotein was achieved in quantitative yields by using a glycosynthase or a thioglycoligase. The resulting glycoproteins function as good acceptors for glycosyl transferases, and the thiog

Full text of DOI:10.1002/anie.200503900

Angewandte
Chemie
Glycoproteins
sugar and a synthetic glycopeptide containing one N-linked b-
N-acetylglycosaminyl residue as an acceptor.[11]
DOI: 10.1002/anie.200503900
Glycosynthases and thioglycoligases are two new classes
of mutant glycosidases recently developed for oligosaccharide
synthesis, but they have not yet been applied to glycoprotein
synthesis.[12,13] Indeed, the relatively high KM values for their
acceptors has caused concern that they would not be useful in
this role. The goal of this study was to explore the applicability
of these two classes of enzymes for the synthesis of
glycoproteins by first using a neoglycoprotein acceptor as a
model system to simplify the analysis. A particular goal was to
explore the generation of glycoproteins in which the terminal
sugar is linked through a glycosidase-resistant thioglycosidic
bond and determine if such thioglycoproteins are recognized
by other glycan-modifying enzymes. Through appropriate
applications of such technology, it should be possible to
generate therapeutic glycoproteins in which the terminal
sugar is sulfur-linked. Such proteins should enjoy consider-
ably longer serum half-lives.
The endo xylanase from Bacillus circulans (Bcx) was
chosen as the protein base for a model neoglycoprotein for
several reasons. First, it is a small, highly characterized
protein of approximately the same size as cytokines; for some
cytokines glycosylation is extremely important both for
activity and serum half-life.[14] Second, as it is an enzyme,
the effects of chemical and enzymatic manipulations on
overall structural integrity can easily be followed by activity-
based assays. Third, the three-dimensional structure of Bcx
has been solved by X-ray crystallography and extensive NMR
spectroscopy assignments.[15,16] Furthermore, the native pro-
tein contains no cysteines, thus unique thiol groups can be
introduced at the surface by in vitro mutagenesis. This allows
site-selective modification of Bcx with sugar entities through
thiol-reactive reagents.[17] The sugar structures selected for
conjugation were cellobiose and its 4’-thio analogue because
these sugars function as good acceptors for Agrobacterium sp.
(Abg) glycosynthase and Abg thioglycoligase, respectively.
These are the enzymes that were used as glycosylation
catalysts.
Thiol-reactive cellobioside 1 was synthesized by reduction
and bromoacetylation of p-nitrophenyl cellobioside (see
Supporting Information). After reaction with Bcx S22C for
1 h, analysis with MALDI-TOF MS showed complete label-
ing with a mass increase equivalent to one incorporated label
(+ 474.2 Da) resulting in neoglycoprotein G2-Bcx (Fig-
ure 1A, B). This neoglycoprotein was then tested as an
acceptor for Abg 2F6 glycosynthase[18] by using a-galactopyr-
anosyl fluoride (a-GalF) as a donor and monitoring the
reaction with MALDI-TOF MS. Complete conversion of the
G2-Bcx into a glycoprotein bearing three sugar units (named
LacG-Bcx) was observed after overnight incubation (Fig-
ure 1C, Figure 2G, Scheme 1A). No incorporation was
observed in control reactions that used Bcx wild-type (wt)
or the unlabeled mutant Bcx S22C as acceptors. Likewise, no
reaction occurred upon incubation of G2-Bcx with only a-
GalF over extended periods of time. The time course for this
reaction showed conversion of ꢀ 80% in 200 minutes (see
Supporting Information) when a molar ratio of 19:1 for
acceptor/glycosynthase was used.
Glycosylation of a Neoglycoprotein by Using
Glycosynthase and Thioglycoligase Approaches:
The Generation of a Thioglycoprotein**
Johannes Müllegger, Hong Ming Chen,
R. Antony J. Warren, and Stephen G. Withers*
Glycan structures, both intracellular and exposed on the
surface of cells, play important roles in a large number of
biological processes, such as cell–cell recognition, ligand–
receptor binding, immunomodulation, and protein folding.[1,2]
Elucidation of the biological function of glycoproteins is
currently hampered mainly by the fact that they are not easily
available as homogenous isoforms but exist as complex
mixtures of different glycoforms.[3,4] As of yet, there are no
generally useful strategies for the direct glycosylation of a
protein to generate a natural glycoprotein: the strategies of
Wong, Schultz, and co-workers are technically impressive but
not yet applicable to the general production of glycoproteins
on any scale.[5,6] Strategies therefore involve partial, specific
deglycosylation of existing glycoproteins to create more
homogenous populations of the glycoprotein, followed by
chemoenzymatic remodeling.[7,8] These steps are required as
the synthetic chemical assembly of larger glycostructures is
essentially impossible on native proteins owing to the harsh
conditions required. Two general strategies are to use glycosyl
transferases[9] or the transglycosylation activity of glycosi-
dases.[10] Only modest yields of shorter glycopeptides have
generally been obtained by using the transglycosylation
activity of endo glycosidases, although one very recent
publication reports the high-yielding transglycosylation of a
core N-glycan structure by using an activated oxazoline-donor
[*] Dr. J. Müllegger, Dr. H. M. Chen, Prof. Dr. S. G. Withers
Protein Engineering Network ofCentres ofExcellence
Department ofChemistry
University ofBritish Columbia
Vancouver, B. C. V6T 1Z1 (Canada)
Fax: (+1)604-822-8869
E-mail: withers@chem.ubc.ca
Dr. R. A. J. Warren
Protein Engineering Network ofCentres ofExcellence
Department ofMicrobiology and Immunology
University ofBritish Columbia
Vancouver, B. C. V6T 1Z1 (Canada)
[**] The authors would like to thank Prof. L. P. McIntosh and Prof. W. W.
Wakarchuk for Bcx clones, Dr. Y. W. Kim for Abg 2F6 glycosynthase,
and L. Lairson for glycosyl transferases, as well as the proteomics
core facility of UBC for providing assistance with, and access to, the
mass spectrometer. The project was funded by the Natural Sciences
and Engineering Research Council ofCanada (NSERC) and the
Protein Engineering Network ofCentres ofExcellence ofCanada
(PENCE). J.M. was financed by a Schrödinger Fellowship from the
Austrian FWF.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2006, 45, 2585 –2588
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2585

Communications
Figure 1. MALDI-TOF MS spectra ofBcx S22C beof re chemical glyco-
sylation (A), G2-Bcx (B), LacG-Bcx (C), 4SG2-Bcx (D), and GalSG-Bcx
(E). Average measured masses for the protein and calculated masses
(in brackets) are shown. Asterisks (*) denote artifact peaks from
sinapinic acid (SPA) matrix addition to the protein (ꢁ206–224 Da).
I=intensity.
Two sets of experiments were undertaken to confirm the
type of linkage formed by the glycosynthase on the surface of
the neoglycoprotein. Cleavage of the terminal galactose by
jack-bean b-galactosidase (b-Gal) and both the galactose and
the glucose by b-glucosidase from Abg (bifunctional b-
galactosidase/glucosidase), as shown by MALDI-TOF MS
(see Supporting Information), clearly shows formation of the
expected b linkage.[19] Confirmation of the terminal lactose
structure on the neoglycoprotein was provided by the action
of two glycosyl transferases with high specificities for lactose
as an acceptor: LgtC, an a-1,4-galactosyl transferase from
Neisseria meningitides,[20] and Cst1, an a-2,3-sialyltransferase
from Campylobacter jejuni.[21] Both enzymes used LacG-Bcx
as an acceptor, resulting in the formation of the Pk antigen
(Gal a-1,4-lactose) and sialyllactose epitopes, respectively, on
the surface of Bcx (Figure 2B, C). Finally, to confirm that all
manipulations were carried out on a properly folded protein
and that the glycosylation did not affect this conformation,
Figure 2. MALDI-TOF MS spectra ofLacG-Bcx (A), and the products
ofglycosylation by Cst1 (B) and LgtC (C). Panels (D, E, F) show the
respective spectra for GalSG2-Bcx. Asterisks (*) denote artifact peaks
from SPA matrix addition to the protein (ꢁ206–224 Da). Average
masses measured for the protein and calculated masses (in brackets)
are shown. (G) SDS-PAGE analysis ofall Bcx neoglycoprotein variants.
Small shifts in migration are observed, but more importantly the
complete conversions achieved in the glycosynthase and thioglycoli-
gase reactions are demonstrated. ° indicates protein contamination
introduced with LgtC. I=intensity.
the xylanase activities of Bcx wt, Bcx S22C, and LacG-Bcx
were measured with 2,5-dinitrophenyl xylobioside as the
2586
www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 2585 –2588

Angewandte
Chemie
GSG2-Bcx), whereas only Abg E170G was active when using
DNPGal as a donor (creating GalSG2-Bcx; Figure 1E and
Figure 2G). This specificity is in full accordance with earlier
observations.[23] Control reactions carried out with Bcx wt and
the “internal” control Bcx S22C showed no signs of glyco-
sylation, suggesting specific attachment of the hexose unit to
the thioglycoside part of the thio-neoglycoprotein. The
presence of a thioglycosidic linkage was established in three
ways. First, the mixture of GalSG2-Bcx and Bcx S22C was
treated with the thiol-reactive compound (+)-biotinyl-iodo-
A
N
tion mixture analyzed by MALDI-TOF MS. The mass of
Bcx S22C, but not that of GalSG2-Bcx, increased by the mass
of the biotin label (+ 415.5 Da), confirming the absence of a
free thiol in the latter case and thus selective glycosylation of
the 4-thio residue on the thio-neoglycoprotein. Further,
GalSG2-Bcx was then easily separated from the contaminat-
ing Bcx S22C by capturing the biotin-labeled protein on
streptavidin beads (Pierce). Second, GalSG2-Bcx was glyco-
sylated by LgtC and Cst1, with the thio-neoglycoprotein
proving to be a good acceptor for both transferases, as shown
by MALDI-TOF MS (Figure 2D,E,F). This confirms that a
thiolactose structure is present and that the presence of sulfur
does not affect the action of the transferase. Third, the
formation of a glycosidase-resistant thioglycoside linkage by
the thioglycoligase was confirmed by subjecting the samples
of GalSG2-Bcx to digestion with jack-bean b-Gal and
Abg.[24,25] In both cases cleavage was not observed by
MALDI-TOF MS. Thus, not only is a thioglycoprotein
produced for the first time, but it has also proven to be
resistant to glycosidase digestion.
In summary, we demonstrate unambiguously the utility of
the glycosynthase and thioglycoligase approaches for the
remodeling of glycoproteins, and describe the first generation
of a thioglycosylated glycoprotein. This work thereby com-
plements the recent demonstration of the synthesis of
thioglycoside-containing glycolipids by using the very low
inherent activities of a few glycosyltransferases to catalyze
thioglycoside formation.[26,27] The resistance of this thiogly-
coprotein, but not of its oxygen analogue, to glycosidase
digestion and the further glycosylation of both structures by
glycosyl transferases directly demonstrates the potential of
this approach for generating metabolically stable glycopro-
tein structures. Work is currently underway to extend this
strategy with other glycosynthases and thioglycoligases to
other glycosylated structures, particularly those which termi-
nate in a thiosialoside, and to install these in natural
glycoproteins.
Scheme 1. Reaction scheme for the chemical glycosylation and subse-
quent transfer of a galactose by Abg glycosynthase (A) or Abg
thioglycoligase (B).
substrate. No significant differences in kcat or KM from
literature values were observed,[22] confirming the integrity
of the protein (see Supporting Information).
Parallel studies to test the thioglycoligase approach
required incorporation of
a
4’-deoxy-4’-thiocellobiosyl
moiety onto the neoglycoprotein. This was achieved by
incubating Bcx S22C with an excess of 2 (used in its disulfide
form to minimize self-polymerization) overnight, followed by
reduction with dithiothreitol (DTT). MALDI-TOF MS
analysis revealed about 50% incorporation to create thio-
neoglycoprotein Bcx-4SG2 (Figure 1D). This mixture of
labeled and unlabeled protein was directly used as an
acceptor for the Abg thioglycoligases, with the unreacted
Bcx S22C acting as a useful internal control. Three different
thioglycoligase mutants, namely Abg E170A, Abg E170G,
and Abg E170Q, were tested by using 2,4-dinitrophenyl b-d-
glucopyranoside (DNPGlc) and 2,5-dinitrophenyl b-d-galac-
topyranoside (DNPGal) as donors. Glycosylation of Bcx-
4SG2 was observed by MALDI-TOF MS when each of the
thioglycoligase mutants were used with DNPGlc (creating
Received: November 4, 2005
Revised: January 21, 2006
Published online: March 10, 2006
Keywords: enzyme models · glycoproteins · glycosylation ·
.
thioglycoligases · thioglycoproteins
[1] P. Van den Steen, P. M. Rudd, R. A. Dwek, G. Opdenakker, Crit.
Rev. Biochem. Mol. Biol. 1998, 33, 151 – 208.
Angew. Chem. Int. Ed. 2006, 45, 2585 –2588
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
2587

Communications
[2] A. Helenius, M. Aebi, Science 2001, 291, 2364– 2369.
[3] J. F. Nemeth, G. P. Hochgesang, Jr., L. J. Marnett, R. M. Cap-
rioli, Biochemistry 2001, 40, 3109 – 3116.
[4] C. Neususs, U. Demelbauer, M. Pelzing, Electrophoresis 2005,
26, 1442 – 1450.
[5] R. Xu, S. R. Hanson, Z. Zhang, Y. Y. Yang, P. G. Schultz, C. H.
Wong, J. Am. Chem. Soc. 2004, 126, 15654– 15655.
[6] Z. Zhang, J. Gildersleeve, Y. Y. Yang, R. Xu, J. A. Loo, S. Uryu,
C. H. Wong, P. G. Schultz, Science 2004, 303, 371 – 373.
[7] B. G. Davis, Chem. Rev. 2002, 102, 579 – 602.
[8] M. J. Grogan, M. R. Pratt, L. A. Marcaurelle, C. R. Bertozzi,
Annu. Rev. Biochem. 2002, 71, 593 – 634.
[9] K. Witte, P. Sears, R. Martin, C. H. Wong, J. Am. Chem. Soc.
1997, 119, 2114– 2118.
[10] K. Yamamoto, S. Kadowaki, J. Watanabe, H. Kumagai, Biochem.
Biophys. Res. Commun. 1994, 203, 244 – 252.
[11] B. Li, Y. Zeng, S. Hauser, H. Song, L. X. Wang, J. Am. Chem.
Soc. 2005, 127, 9692 – 9693.
[12] M. Jahn, S. G. Withers, Biocatal. Biotransform. 2003, 21, 159 –
166.
[13] G. Perugino, A. Trincone, M. Rossi, M. Moracci, Trends
Biotechnol. 2004, 22, 31 – 37.
[14] L. C. Wasley, G. Timony, P. Murtha, J. Stoudemire, A. J. Dorner,
J. Caro, M. Krieger, R. Kaufman, Blood 1991, 77, 2624– 2632.
[15] W. W. Wakarchuk, R. L. Campbell, W. L. Sung, J. Davoodi, M.
Yaguchi, Protein Sci. 1994, 3, 467 – 475.
[16] L. A. Plesniak, W. W. Wakarchuk, L. P. McIntosh, Protein Sci.
1996, 5, 1118 – 1135.
[17] D. Macmillan, R. M. Bill, K. A. Sage, D. Fern, S. L. Flitsch,
Chem. Biol. 2001, 8, 133 – 145.
[18] Y. W. Kim, S. S. Lee, R. A. Warren, S. G. Withers, J. Biol. Chem.
2004, 279, 42787 – 42793.
[19] L. F. Mackenzie, Q. P. Wang, R. A. J. Warren, S. G. Withers, J.
Am. Chem. Soc. 1998, 120, 5583 – 5584.
[20] L. L. Lairson, C. P. Chiu, H. D. Ly, S. He, W. W. Wakarchuk,
N. C. Strynadka, S. G. Withers, J. Biol. Chem. 2004, 279, 28339 –
28344.
[21] M. Gilbert, J. R. Brisson, M. F. Karwaski, J. Michniewicz, A. M.
Cunningham, Y. Wu, N. Young, W. Wakarchuk, J. Biol. Chem.
2000, 275, 3896 – 3906.
[22] L. Ziser, I. Setyawati, S. G. Withers, Carbohydr. Res. 1995, 274,
137 – 153.
[23] J. Mꢀllegger, M. Jahn, H. M. Chen, R. A. J. Warren, S. G.
Withers, Protein Eng. Des. Sel. 2005, 18, 33 – 40.
[24] M. Jahn, J. Marles, R. A. J. Warren, S. G. Withers, Angew. Chem.
2003, 115, 366 – 368; Angew. Chem. Int. Ed. 2003, 42, 352 – 354; .
[25] C. Orgeret, E. Seillier, C. Gautier, J. Defaye, H. Driguez,
Carbohydr. Res. 1992, 224, 29 – 40.
[26] J. Rich, W. Wakarchuk, D. Bundle, Chemistry 2005, 12, 845 – 848.
[27] J. Rich, A. Szpacenko, M. Palcic, D. Bundle, Angew. Chem. 2004,
116, 623 – 625; Angew. Chem. Int. Ed. 2004, 43, 613 – 615.
2588
www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 2585 –2588