O
-Linked Chains of Mucin 273
273
23
O
-Linked Chain Glycosyltransferases
Inka Brockhausen
1. Introduction
The complex O-linked oligosaccharide chains (O-glycans) attached to the
polypeptide backbone of mucins are assembled by glycosyltransferases. These
enzymes act in the Golgi apparatus in a controlled sequence that is determined by
their substrate specificities, their localization in Golgi compartments, and their
relative catalytic activities (1). Activities are controlled by many factors, includ-
ing the membrane environment, metal ions, concentrations of donor and acceptor
substrates, cofactors, and, in some cases, posttranslational modifications of
enzymes. Cloning of glycosyltransferases has revealed the existence of families of
homologous glycosyltransferases with similar actions but encoded by different
genes. Thus, many steps in the pathways of O-glycosylation appear to be cata-
lyzed by several related glycosyltransferases that may show slight differences in
properties and substrate specificities. The relative expression levels of these
enzymes is cell typespecific and appears to be regulated during the growth and
differentiation of cells and, during tissue development, and is altered in many
disease states (2,3).
Figure 1 shows the biosynthetic pathways of O-glycans with the common
mucin O-glycans core structures 1–4. The biosynthesis of other less common core
structures (1) has not been studied in detail. Core structures can be elongated
by repeating GlcNAcβ1-3Galβ1-4 or GlcNAc1-3Galβ1-3 structures (poly-N-
acetyllactosamine chains, i antigens). Poly-N-acetyllactosamine chains may contain
branches of GlcNAcβ1-6 residues linked to Gal- (I antigen), and may be terminated
by blood group or tissue antigens (blood group ABO and Lewis antigens) or by
sialic acid and sulfate. Many of the enzymes involved in these elongation and

termination reactions also act on N-linked oligosaccharides of glycoproteins and
on glycolipids.
From:
Methods in Molecular Biology, Vol. 125: Glycoprotein Methods and Protocols: The Mucins
Edited by: A. Corfield © Humana Press Inc., Totowa, NJ
274 Brockhausen
Fig. 1. Pathways of O-glycan biosynthesis. The first step of O-glycosylation is cata-
lyzed by polypeptide α-GalNAc-transferase (path a), which acts preferably on Thr in
vitro. The occurrence of various O-glycan structures is cell type specific and varies with
cellular activation and differentiation, and in disease states. Core 1 and 2 structures are
the most common core structures in mucins, and are synthesized by core 1 β3-Gal-trans-
ferase (path b) and core 2 β6-GlcNAc-transferase (path d). Core 3 is synthesized by core 3
β3-GlcNAc-transferase (path c) and core 4 by core 4 β6-GlcNAc-transferase (path e).
GalNAc- may be sialylated by α6-sialyltransferases (path j) to form sialylα2-6GalNAc-,
which cannot be converted to any of the core structures. After the synthesis of cores,
chains may be elongated, sulfated, fucosylated, or sialylated, and blood group and other
antigenic determinants may be added. Core 1 is sialylated by α3-sialyltransferase (path l),
and this reaction blocks core 1 branching and elongation with the exception of α6-
sialylation by α6-sialyltransferase (path m), which may differ from those catalyzing
path j. The α3-sialyltransferase can also act on the Gal residue of core 2. The Gal residue
of core 1 may be sulfated by Gal 3-sulfotransferase (path k). Sulfation will also block
core 1 elongation and branching. Cores 1 and 2 are elongated by elongation β3-GlcNAc-
transferase (path h). On galactosylation of the GlcNAc residue of core 2 by β4-Gal-trans-
ferase (path f), the poly-N-acetyllactosamine chains can be assembled by the repeated
actions of β4-Gal-transferase and i β3-GlcNAc-transferase (paths f and g, respectively).
GlcNAcβ1-6 Gal branches (I antigen) may be introduced into these chains by I β6-GlcNAc-
transferase (path i).
O
-Linked Chains of Mucin 275
Golgi glycosyltransferases have a domain structure characteristic of type II

membrane proteins; the amino terminus extends into the cytoplasm, followed by a
membrane anchor domain, and a catalytic domain at the carboxy terminus, which
extends into the lumen of the Golgi. Mainly the membrane anchor and adjacent
amino acid sequences, but also other protein determinants of these enzymes, as
well as the membrane structure, determine the localization of transferases in vari-
ous Golgi compartments (4). The donor substrates for mucin glycosyltransferases
are nucleotide sugars:
nucleotide-sugar + acceptor → product + nucleotide (1)
Mucin sulfotransferases transfer sulfate from 3'-phosphoadenosine 5'-phosphosulfate
(PAPS) to the hydroxyls of sugars:
PAPS + acceptor → product + PAP (2)
These donor substrates are synthesized in the cytosol, with the exception of
CMP-sialic acid, which is made in the nuclear compartment, and are transported into
the Golgi by specific transporter systems (5).
In the sequences of glycosylation reactions, glycosyltransferases often compete
for a common substrate. For example, the enzymes that synthesize cores 1 and 3
(Fig. 1, paths b and c) compete for GalNAc-R substrates. Conversely, certain products
formed may block further reactions; for example, no glycosyl transferase acts on
sialylα2-6GalNAc (Fig. 1, path j), which therefore blocks extension of chains. Alter-
natively, certain reactions may be required prior to further conversions. For example,
core 1 has to be formed before a GlcNAcβ1-6 residue can be added to GalNAc in the
synthesis of core 2 (Fig. 1, path d). The distinct specificities of glycosyltransferases
therefore regulate the pathways, and thus the relative amounts of final O-glycan struc-
tures found in mucins. The peptide backbones as well as existing glycosylation of
substrates near the O-glycosylation sites also have an important function in regulating
O-glycosylation (6). Thus, primary O-glycosylation as well as the synthesis of various
O-glycan core structures appear to be sitedirected by peptide sequences and their
glycosylation patterns.
Based on many different studies, O-glycosylation appears to be initiated mainly
in early Golgi compartments. The first enzyme in the O-glycosylation pathways,

polypeptide α-GalNAc-transferase (Fig. 1, path a), has been localized to the cis
Golgi compartment in porcine submaxillary gland (7) but can be more broadly dis-
tributed throughout the Golgi in other cell types (8). The various members of this
glycosyltransferase family have slightly different specificities toward their peptide
substrates, have different cell type–specific expression patterns, and may be local-
ized to different subcellular compartments (9,10). Polypeptide α-GalNAc-trans-
ferase does not require a specific peptide sequence in the substrate; however,
particular charged amino acids (11) as well as existing glycosylation (12) influence
the activity.
Most mucins and other glycoproteins contain O-glycans with the core 1 structure,
and the enzyme synthesizing core 1, core 1 β3-Gal-transferase (Fig. 1, path b), is a
276 Brockhausen
ubiquitous enzyme (1,13). The activity is a prerequisite for the synthesis of T-antigens
and sialylated core 1 structures, as well as core 2 structures (14). The peptide sequences
of glycopeptide substrates and the existing glycosylation determine the activity of core 1
β3-Gal-transferase (15). Erythrocytes from patients with permanent mixed-field
polyagglutinability (16), human T-lymphoblastoid Jurkat cells (17), and human colon
cancer cells LSC (18) lack the enzyme and therefore cannot make cores 1 and 2 struc-
tures. A similar effect can be introduced by the use of GalNAcα-benzyl, which is an
alternative substrate for enzymes extending O-glycan chains, and which can penetrate
cell membranes to compete with endogenous substrates of mucin. Thus, cells treated
with this O-glycosylation inhibitor exhibit truncated O-glycans, terminating mainly in
GalNAc (19). These truncated chains no longer carry ligands for cell-cell interactions,
and the binding of colon cancer cells to the endothelium via E-selectin is significantly
reduced (20).
The synthesis of core 2 (Fig. 1, path d) is catalyzed by core 2 β6-GlcNAc-trans-
ferase (21). Several apparently related β6-GlcNAc-transferases exist that synthesize
GlcNAcβ1-6 branches on Gal or GalNAc (1,22,23). The L-type core 2 β6-GlcNAc-
transferase occurs in leukocytes and other cells and only synthesizes core 2. The
M-type enzyme is found in most mucin-secreting cell types and can synthesize the

GlcNAcβ1-6 branch of core 2, core 4 (Fig. 1, path e), and the I antigen (Fig. 1,
path i) (24). The L-enzyme activity increases during cellular activation and differ-
entiation (25,26). The M enzyme may be affected in cancer cells (3,27). Core 2
β6-GlcNAc-transferase appears to be localized to cis and medial Golgi com-
partments (27a,28), which is in agreement with its role in synthesizing a central
O-glycan core structure.
The enzymes synthesizing O-glycan cores 3 and 4 (Fig. 1, paths c and e, respec-
tively) appear to occur exclusively in mucin-secreting tissues since these cores
have not been found in nonmucin molecules (1). Core 3 is synthesized by core 3
β3-GlcNAc-transferase (29). The enzyme is enriched in colonic tissues but reduced in
colon cancer tissue (30,31) and is lacking in many other tissues. The activity appar-
ently is lacking in colon cancer cell lines (27). The enzyme activity synthesizing
core 4, core 4 β6-GlcNAc-transferase, resides in the M-type core 2 β6-GlcNAc-
transferase (24,29).
Poly-N-acetyllactosamine chains of mucins are assembled by the repeating actions
of β4-Gal-transferase (Fig. 1, path f) (32) and i β3-GlcNAc-transferase (Fig. 1, path g)
(33). These enzymes are ubiquitous, and may be considered as housekeeping enzymes.
However, their expression is often up- or downregulated in healthy tissues as well as
in a number of disease states (2,3,34). The reaction catalyzed by β4-Gal-transferase
occurs mainly in the trans-Golgi (35). Yet another elongation β3-GlcNAc-trans-
ferase elongates core 1 and 2 structures, also by a GlcNAcβ1-3Gal linkage Fig. 1,
path h) (36).
Poly-N-acetyllactosamine chains may acquire GlcNAcβ1-6 (GlcNAcβ1-3) Gal
branches in a developmentally regulated fashion, which leads to a change from the i
to the I antigenicity. Some of the I β6-GlcNAc-transferases (Fig. 1, path i) synthe-
sizing the I branch act on terminal Gal residues whereas others recognize internal
O
-Linked Chains of Mucin 277
Gal residues (1,37,38). Most of the enzymes synthesizing blood group ABO, Lewis,
and other antigenic determinants act on O- and N-glycans as well as glycolipids. By

contrast, sialyltransferases often prefer one type of glycoconjugate (39). Two
sialyltransferase families, α3- and α6-sialyltransferase, act preferably on mucin-
type O-glycans.
α3-Sialyltransferase acts on Gal residues of cores 1 and 2 (Fig. 1, path l) (24,40,41).
The enzyme is developmentally regulated in thymocytes (42) and increased in leuke-
mia cells (43) and several cancer models (3,30). The α3-sialyltransferase has been
localized to medial and trans-Golgi compartments (44). The sialylation reaction cata-
lyzed by this enzyme has an important role in keeping O-glycan chains short and
sialylated. Since the enzyme acts relatively early in the O-glycan extension pathways
(Fig. 1, path l), it has the ability to compete with branching and elongation reactions.
Once core 1 is α3-sialylated, it is no longer a substrate for extension reactions
although it can still be converted to the disialylated core 1 by α6-sialyltransferase
(Fig. 1, path m).
The α6-sialyltransferase (Fig. 1, path j) that acts on GalNAc-R to form the
sialyl-Tn antigen, sialylα2-6GalNAc-Th/Ser (45), requires glycoproteins as substrate
and cannot act on GalNAc-benzyl or nitrophenyl substrates (46,47). However,
another type of α6-sialyltransferase (α6-sialyltransferase III) does not have a pep-
tide requirement, but is specific for the α3-sialylated core 1 structure (48). The
disialylated core structure can probably be synthesized by α6-sialyltransferase III
and other α6-sialyltransferases (Fig. 1, path m). Modifications of the sialic acid
residues of mucins include O-acetylation, catalyzed by specific O-acetyltransferases
acting in the Golgi (49).
The common sulfate ester linkages in mucins are SO
4
-6-GlcNAc and SO
4
-3-Gal.
Several types of sulfotransferases have been described that act on the 6-position of
GlcNAc (50) or the 3-position of Gal of core 1 (Fig. 1, path k) and N-acetyllactosamine
structures (51,52). Sulfated oligosaccharides appear to play an important role in cell

adhesion through binding to selectins and in the control of bacterial binding (53,54).
Sulfation also functions in directing the biosynthetic pathways of complex O-glycans
by blocking certain reactions. For example, sulfation of core 1 prevents the branching
reaction to form core 2 (51).
The enzymes catalyzing the reactions depicted in Fig. 1 assemble mucin-type
O-linked carbohydrate chains and are listed in Table 1, together with their substrates,
enzyme products, and the high performance liquid chromatography (HPLC) condi-
tions of product separation. Probably none of these enzymes are specific for mucins,
but also act on other glycoproteins that carry O-glycans, and can act on various glyco-
peptides with O-linkages. In vitro, many of these enzymes utilize synthetic compounds
as substrates in which the peptide chain is replaced by a hydrophobic group. The
substrate should be clean, specific, and easy to isolate in order to determine the
enzyme activity and specificity accurately. For a few enzymes, purified mucins with
defined glycosylation are available as substrates. However, mucins are usually too
heterogeneous in their carbohydrate structures, and therefore the use of synthetic com-
pounds with defined structure is preferred. In addition, it is much easier to determine
278 Brockhausen
the product structure of synthetic substrates as a proof of the assayed activity. When a
compound is a potential substrate for several glycosyltransferases present in the
enzyme preparation, or several reactions occur in sequence, the various products
have to be separated and identified. This can usually be achieved by HPLC. With the
exception of β1,6-GlcNAc-transferases, UDP-sugar binding enzymes require the pres-
ence of divalent metal ion for optimal activity. Thus, measuring a β6-GlcNAc-trans-
ferase activity in the presence of EDTA will eliminate the activity of other
GlcNAc-transferases potentially acting on the same substrate. Enzymes utilizing
CMP-sialic acid may be stimulated by metal ions but usually can act in their absence.
Unless they are released and secreted, or are produced as soluble recombinant
enzymes, glycosyltransferases are membrane-bound enzymes, and their activities are
stimulated by detergents.
A convenient way of identifying and quantifying glycosyltransferase products is by

the use of nucleotide-sugar donors that contain
14
C or
3
H-labeled radioactive sugar.
Similarly, the sulfate moiety of PAPS can be labeled with
35
S. Calculations of
sulfotransferase activities must take into account the relatively short half-life of
35
S (about 87 d).
2. Materials
2.1. Preparation of Enzymes
1. 0.25 M Sucrose.
2. 0.9% NaCl.
3. Potter-Elvehjem hand homogenizer.
4. Low-speed centrifuge (10,000g).
5. Ultracentrifuge (100,000g).
6. Small pieces of tissue, or cells.
2.2. Preparation of Substrates and Standard Compounds
1. Commercially available oligosaccharides: GlcNAc, GalNAcα-benzyl, Galβ1-3
GalNAcα-benzyl, GlcNAcβ1-3 GalNAcα-p-nitrophenyl [pnp], Galβ1-4 GlcNAc,
GlcNAcβ1-3 Galβ-methyl (Sigma, St. Louis MO); Galβ1-3 GalNAcα-pnp (Toronto
Research Chemicals, Toronto, Canada).
2. Thr-peptides, synthesized by Hans Paulsen, University of Hamburg, Germany
(15,55).
3. Frozen sheep submaxillary glands (Pel-Freez, Rogers, AR) to isolate ovine submaxillary
mucin (OSM), 0.1 N H
2
SO

4
, bovine testicular β-galactosidase (Boehringer, Laval,
Canada), 0.1 M Na-citrate buffer, Sephadex G25 column.
4. Components of enzyme assays to prepare product standards enzymatically GlcNAcβ1-3
GalNAcα-benzyl, GlcNAcβ1-6 (GlcNAcβ1-3) GalNAcα-pnp, GlcNAcβ1-3 Galβ1-4
GlcNAc, GlcNAcβ1-6 (GlcNAcβ1-3 Galβ1-3) GalNAcα-benzyl, GlcNAcβ1-6
(GlcNAcβ1-3) Galβ-methyl, sialylα2-3 Galβ1-3 GalNAcα-pnp, SO
4
-3 Galβ1-3
GalNAcα-benzyl, SO
4
-3 Galβ1-4 GlcNAc.
5. Enzymatically prepared substrates: GlcNAcβ1-6 (Galβ1-3) GalNAcα-benzyl, sialylα2-3
Galβ1-3 GalNAcα-pnp
O
-Linked Chains of Mucin 279
6. Nuclear magnetic resonance (NMR) and mass spectrometers, reagents for methylation
analysis.
2.3. Separation and Identification
of Glycosyltransferase and Sulfotransferase Products
1. HPLC apparatus.
2. HPLC columns C18, NH
2
(amine), PAC (cyano-amine).
3. Acetonitrile/water mixtures.
4. Dionex system for high-performance anion-exchange chromatography (HPAEC).
5. Bio-Gel P4 or P2 column (80 × 1.6 cm) (Bio-Rad, Hercules, CA).
6. Ion-exchange columns (AG1 × 8, 100–200 mesh, Bio-Rad).
7. High-voltage electrophoresis apparatus, 1% Na-tetraborate, Whatman No. 1 paper.
8. C18 Sep-Pak columns, methanol.

9. 0.05 M KOH/1 M NaBH
4
for β-elimination.
10. Scintillation fluid, scintillation counter.
2.4. Polypeptide
α
-GalNAc-Transferase Assays
1. 5% Triton X-100.
2. 0.5 M N-morpholino ethanesufonate (MES) buffer, pH 7.
3. 0.05 M Adenosine 5'-monophosphate (AMP) to inhibit pyrophosphatases.
4. 0.5 M MnCl
2
.
5. 10 mM UDP-GalNAc (2000 dpm/nmol) donor substrate.
6. 5 mM Acceptor substrate solution: Thr-containing peptide.
7. Enzyme homogenate or solution.
2.5.
β
3- and
β
6-GlcNAc-Transferase Assays
1. 5% Triton X-100.
2. 0.5 M MnCl
2
(for β3-GlcNAc-transferases only).
3. 0.5 M MES buffer, pH 7.
4. 0.05 M AMP.
5. 0.5 M GlcNAc to inhibit N-acetylglucosaminidases.
6. 50 mM γ-galactonolactone (if substrate with terminal Gal is used) to inhibit
galactosidases.

7. 10 mM UDP-GlcNAc (2000 dpm/nmol).
8. 5 mM Acceptor substrate solution: GalNAcα-benzyl, Galβ1-3 GalNAcα-benzyl, Galβ1-4
GlcNAc, GlcNAcβ1-3Galβ-methyl, or GlcNAcβ1-6 (Galβ1-3) GalNAcα-benzyl.
9. Enzyme homogenate or solution.
2.6. Core 1
β
3-Gal- and
β
4-Gal-Transferase Assays
1. 5% Triton X-100.
2. 0.5 M MnCl
2
.
3. 0.5 M MES buffer, pH 7.
4. 0.05 M AMP.
5. 0.05 M γ-galactonolactone.
6. 10 mM UDP-Gal (2000 dpm/nmol).
280 Brockhausen
7. 5 mM Acceptor substrate solution: GalNAcα-benzyl or GlcNAc.
8. Enzyme homogenate or solution.
2.7.
α
3- and
α
6-Sialyltransferase Assays
1. 5% Triton X-100.
2. 0.5 M Tri-HCl buffer, pH 7.
3. 0.05 M AMP.
4. 10 mM CMP-sialic acid (2000 dpm/nmol).
5. 5 mM Acceptor substrate solution: DS-OSM. with 3 mM GalNAc concentration, Galβ1-3

GalNAcα-pnp, or sialylα2-3 Galβ1-3 GalNAcα-pnp.
6. Enzyme homogenate or solution.
7. High-voltage electrophoresis apparatus.
8. 20 mM EDTA/2 % Na-tetraborate.
9. 1% Na-tetraborate.
10. Whatman No. 1 paper.
11. HPLC apparatus.
2.8. Sulfotransferase Assays
1. 5% Triton X-100.
2. 0.1 M magnesium-acetate.
3. 0.1 M NaF to inhibit sulfatases.
4. 0.5 M Tris-HCl buffer, pH 7
5. 0.05 M adenosine triphosphate (ATP).
6. 0.1 M 2,3-Mercaptopropanol to inhibit PAPS degradation.
7. 0.3 mM PAPS (2000 dpm/nmol).
8. 5 mM Acceptor substrate solution: Galβ1-3GalNAcα-benzyl, Galβ1-4 GlcNAc, or
GlcNAcβ1-3Galβ-methyl.
9. Enzyme-homogenate or solution.
10. High-voltage electrophoresis apparatus.
11. 20 mM EDTA/2% Na-tetraborate.
12. 1% Na-tetraborate.
13. Whatman No. 1 paper.
14. HPLC apparatus.
15. Dionex system for HPAEC.
3. Methods
3.1. Preparation of Enzymes
Ideally, enzymes are present in the highly purified state, and soluble in the assay
mixture. A number of enzymes have been purified. However, these procedures
depend on the specific enzyme and tissue and may take several months or years.
Therefore, purification protocols are not described here. Purified enzymes may be

stable at 4°C for months but are usually more stable at lower temperatures. With
tissue homogenates or microsomes, however, this is rarely the case. The enzyme
preparations inevitably contain interfering substances and degradative enzymes. For
example pyrophosphatases and phosphatases that degrade nucleotide sugar donors,