Glycobiology vol. 7 no 6 pp. 769-776, 1997
N-Glycosylation is requisite for the enzyme activity and Golgi retention of
N-acetylglucosaminyltransferase DI
Kaoru Nagai1"2, Yoshito Ihara1, Yoshinao Wada2 and
Naoyuki Taniguchi1*3
'Department of Biochemistry, Osaka University Medical School, 2-2
Yamadaoka, Suita, Osaka 565, Japan and 2Department of Molecular
Medicine, Osaka Medical Center and Research Institute for Maternal and
Child Health, 840 Murodo-cho, Izumi, Osaka 590-02, Japan
^To whom correspondence should be addressed
UDP-N-acetylglucosamine: P-D-mannoside P-1,4Nacetylglucosaminyltransferase III (GnT-m, EC 2.4.1.144)
is a glycoprotein involved in the biosynthesis of N-linked
oligosaccharides. Rat GnT-IJl contains three potential Nglycosylation sites, which have been predicted to be
Asn243, Asn261, and Asn399. To study the roles of Nglycosylation in the GnT-III function, rat GnT-III was expressed in COS-1 cells under tunicamycin or castanospermine treatment. The tunicamycin-treated GnT-EQ,
which was not N-glycosylated, had almost no activity. The
castanospermine-treated GnT-EQ was not localized in the
Golgi, but glucosylatdon did not affect its activity. To clarify
the role of individual N-glycosylations, we obtained a series
of mutant cDNAs in which some or all of the potential
glycosylation sites were eliminated by site-directed mutagenesis, and expressed them in COS-1 cells. All the mutants
exhibited lower enzyme activity than the wild-type, but deglycosylation at individual sites had different effects on the
enzyme activity. The deglycosylation at Asn243 or Asn261
was more effective on the activity than that at Asn399. The
enzyme activity decreased as the number of glycosylation
sites decreased. The null glycosylation mutant had no activity, corresponding to the case of tunicamycin-treated
wild-type GnT-EQ. Kinetic analysis revealed that the deglycosylation at Asn243 or Asn261 resulted in slightly lower
affinity for the donor substrate, but the other mutation did
not significantly change the Km value for either the donor
or acceptor. None of the mutant GnT-EQs showed perinudear localization or Golgi retention, that was observed for
the wild-type protein. This is the first demonstration that
the glycosyltransferase localized in the Golgi apparatus requires N-glycosylation for its activity and retention.
Key words: N-acetylglucosaminyltransferase m/N-glycosylation/Golgi apparatus/glycoprotein/protein folding
Introduction
Oligosaccharides of glycoproteins play crucial roles in a variety of biological and physical properties, such as folding, secretion, solubility, stability, and clearance from the blood
stream, of glycoproteins (for reviews, see Rademacher et al,
1988; Varki, 1993). These sugar chains, and those of glycolipids, are synthesized by glycosidases and glycosyltransferO Oxford University Press
ases localized in the ER and Golgi apparatus (Field and Wainwright, 1995). All known glycosyltransferases except for
3-1,2-N-acetylglucosaminyltransferase I (Kumar et al, 1990;
Sarkar et al, 1991) have potential N-glycosylation sites. However, the roles of these N-linked oligosaccharides have not
been evaluated, except in a few cases, such as ot-2,6sialyltransferase and p-l,4-N-acetylgalactosaminyltransferase
(Fast et al, 1993; Haraguchi et al, 1995).
UDP-N-Acetylglucosamine: P-D-mannoside (J-1,4Nacetylglucosaminyltransferase III (GnT-III, EC 2.4.1.144)
catalyzes the attachment of a GlcNAc residue to (31-4 mannose
in the core region of N-glycans and forms bisecting GlcNAc
(Figure 1). GnT-III is presumed to be involved in pathological
conditions, because increased expression is accompanied by
malignant transformation or oncofetal changes (Koenderman et
al, 1987; Koenderman et al, 1989; Miyoshi et al, 1993;
Yoshimura et al, 1995). Rat GnT-HI is a type II membrane
glycoprotein composed of 536 amino acids and has a domain
structure typical of glycosyltransferases (Nishikawa et al,
1992). The domain structure comprises a short amino terminal
cytoplasmic tail, transmembrane and neck regions, and a long
carboxyl terminal catalytic domain protruding into the Golgi
lumen. The catalytic domain has three potential Nglycosylation sites at Asn243, Asn261, and Asn399. The recombinant GnT-III expressed in E.coli is almost inactive, and
the activity is not always correlated with the mRNA levels in
various tissues (Yoshimura et al, 1995). These findings suggest that GnT-III activity is regulated by posttranslational processes, such as phosphorylation and glycosylation. N-linked
oligosaccharide structures are sequentially synthesized through
the removal and addition of specific sugar residues by specific
glycosidases and glycosyltransferases. It must, therefore, be
important to clarify the mechanisms by which these processing
enzymes, including GnT-III, are retained in the Golgi apparatus as well as their activation.
In the present study, a full-length rat GnT-EQ was expressed
in COS-1 cells. The roles of core glycosylation and subsequent
glucose trimming in the activity and Golgi retention of GnT-HI
were examined by using processing inhibitors, tunicamycin
and castanospermine, and a series of mutants that lack potential
N-glycosylation sites. We confirmed that all three Nglycosylation sites are necessary for Golgi retention and full
expression of the enzyme activity.
Results
Characterization of the anti-rat GnT-III antibody
To check the immunoreactivity of antibody against the active
GnT-m expressed in COS-1 cells, the lysate of transfected
COS-1 cells was incubated with increasing volumes of antiGnT-HI antiserum. And then, the immune complex was precipitated with protein-G Sepharose. The residual GnT-III ac769
K.Nagai
UDP-GlcNAc
(a) WT-GnT-ll
GlcNAcpi-2Mana1
GtcNAc£1-2Mana\
6
243 281
NH2Mana1-4R
GlcNAcpi-2Mam1 /
acf4Ac$1-2Mana1/
Fig. 1. The reaction catalyzed by GnT-III. GlcNAc and Man denote
N-acetylglucosamine and mannose, respectively. R was GlcNAcfJl4GlcNAc-2-aminopyridine in the assay system used in the present study.
tivity in the supernatant was reduced in a dose-dependent manner (Figure 2).
Effects of tunicamycin and castanospermine on GnT-III in
COS-1 cells
The Asn residues at 243, 261, and 399 of GnT-m are potential
N-glycosylation sites (Asn-X-Thr) (Figure 3a). Tunicamycin
blocks core glycosylation, and castanospermine, an inhibitor of
glucosidase I and n, blocks the initial step of N-glycan processing. The Western blotting pattern of GnT-III with or without treatment is presented in Figure 4. The wild-type GnT-IH
migrated on SDS-PAGE to 68.1 kDa. Compared with intact
GnT-EQ, the tunicamycin-treated enzyme migrated faster, to
60.3 kDa, due to the lack of N-glycans, and the castanospermine-treated form migrated slowly, to 69.7 kDa, because
of the remaining glucose residues. The enzyme activities of
these structurally modified GnT-HIs were measured and are
presented in Figure 5. GnT-EQ activity was completely abolished by tunicamycin, whereas it was not affected by castanospermine. Intracellular localization was then analyzed by fluorescence microscopy. As shown in Figure 6, wild-type GnT-HI
displayed a Golgi staining pattern similar to Gal-T and WGA.
In contrast, the inhibitor-treated enzymes were not localized in
the Golgi apparatus.
These results, taken together, indicated that the normal enzyme activity requires core glycosylation but not glucose trimming, although these steps are necessary for GnT-KI to be
retained in the Golgi apparatus.
120 -,
•
antf-GnT-lll serum
•
pre-immune serum
399
—)—
-COOH
(b) mutarrt-GnT-ll
MT1
UDP
4
—H—
GlcNAcpi-4Mana1-4R
•Q-N-
MT2
-N-Q-
MT3
-N-N-
MT1.2
•Q-Q-
MT1.3
-Q-N-
MT2.3
-N-Q-
MT1.2.3
-Q-Q-
Fig. 3. Structure of rat GnT-fJI and the mutant constructs, (a) Three
potential N-glycosylation sites, Asn243, Asn261, and Asn399, of rat
GnT-III are localized in the catalytic domain, (b) Mutant GnT-IUs
generated in this study.
Enzyme activity of mutant GnT-IIIs
From the wild-type GnT-III cDNA constructed in an expression vector, pSVK3, we generated seven types of GnT-IH mutant by site-directed mutagenesis, replacing potential Nglycosylation site Asn residues with Gin (Figure 3b). Mutants
1-3 (MT1-3) were missing one of the N-glycosylation sites,
mutants 1,2, 1,3, and 2,3 (MT1.2, MT1,3, and MT2,3) lacked
two sites, and mutant 1,2,3 (MT 1,2,3) had no N-glycosylation
sites. As shown in Figure 7, the mutant GnT-IIIs migrated
differently from one another on SDS-PAGE, according to the
number of oligosaccharide chains attached. A 68.1 kDa protein
(kDa)
203
6
Anttxrtnn (\il)
Fig. 2. Adsorption of the GnT-IH activity with anti-GnT-III serum. Lysates
of the COS-1 cells transfected with pSVK3-(rat GnT-III) was incubated
with increasing amounts of anti-GnT-in serum or preimmune rabbit serum,
and then precipitated with protein G-Sepharose. GnT-III activity in the
supernatant was measured. The graph shows the percent activity relative to
the corresponding blank sample without serum. The mean values from three
experiments are presented.
770
Fig. 4. SDS-PAGE and Western blotting analysis of tunicamycin and
castanospermine-treated GnT-DIs. Lysates of COS-1 cells transfected with
pSVK3-wild-type GnT-III were separated on a 5-15% gradient SDS-PAGE
gel under reducing conditions, and the GnT-III bands were detected
utilizing rabbit polyclonal antibodies. Lanes from the left; mock-transfected
COS-1 cells, wild-type enzyme in the absence of inhibitors,
tunicamycin-treated cells (TM), and castanospermine-treated cells (CS).
Role of N-glycosylation in GnT-HI function
140
and MT3) transfectants. On the other hand, the apparent K™
values for the wild-type and MT3 transfectants were similar to
each other for either the donor or acceptor, while MT1 and
MT2 exhibited a little higher K,,, value for the donor. These
findings indicated that the MT3 mutation scarcely affected the
affinity to the donor or acceptor, whereas the MT1 and MT2
mutations decreased the affinity to the donor but not that to the
acceptor. The K,,, values for the wild-type transfectant in the
present study were similar to those in rat tissue reported by
Nishikawa et al. (1990).
Discussion
Fig. 5. GnT-III activity in the COS-1 cells treated with inhibitors or
overexpressing wild-type and mutant GnT-UIs. GnT-III activities in the
COS-1 cell lysates were measured using a fluorescence-labeled
oligosaccharide acceptor and a donor UDP-GlcNAc as substrates. TM and
CS denote the wild-type GnT-III treated with tunicamycin and
castanospermine, respectively. MT1-MT1,2,3 are GnT-m mutants.
Normalization of the GnT-III activity is based on the cotransfected
(3-galactosidase activity. The mean values from three experiments are
presented.
was generated from wild-type GnT-III, whereas carbohydrate
removal by site-directed mutagenesis at a single site resulted in
a decrease in molecular mass of 2.0-3.0 kDa, a value corresponding to a single sugar chain. The molecular mass of the
triple mutant, MT1,2,3, was almost the same as that of tunicamycin-treated GnT-III (Figures 4, 7). The GnT-III activities
of these mutants are shown in Figure 5. MT1 or MT2 transfectants exhibited 40-50% of the activity of wild-type transfectants, while MT3 exhibited 70-80%. MT1.2 transfectants
showed a decrease of 90%, and MT1.3 or MT2.3 expressed
only a little less activity compared with MT1 and MT2.
MT 1,2,3 transfectants exhibited almost no activity, corresponding to the case of tunicamycin-treated GnT-HI described
above. These results suggested that glycosylation at Asn243
and Asn261 is important for the enzyme to express GnT-HI
activity, while that at Asn399 is not essential.
Intracellular localization of mutant GnT-UIs
The distributions of the wild-type and mutant GnT-UIs in
transfected COS-1 cells were analyzed by the immunohistochemical technique (Figure 8). Anti-rat GnT-HI antibody revealed perinuclear staining for wild-type GnT-ITJ, and the
stained compartment turned out to be the Golgi apparatus, as
judged by the WGA lectin binding. On the other hand, all
mutants were diffusely distributed in the cells, indicating that
N-glycosylation all the three potential sites are indispensable
for GnT-HI to be retained in the Golgi apparatus.
Kinetics of mutant GnT-IIIs
In order to examine the enzymological effect of the mutations
in detail, kinetic analysis of GnT-III as to the donor and acceptor was performed. As shown in Figure 9, the V ^ values
were different for the wild-type and three mutant (MT1, MT2,
In eukaryotes, the sugar moieties of glycoproteins are synthesized by specific glycosidases and glycosyltransferases, and
most of the glycosyltransferases with known structures are
themselves glycoproteins (Field and Wainwright, 1995). However, the significance of N-glycosylation in these glycosyltransferases has been examined in only a few studies so far
(Fast et al., 1993; Haraguchi et al., 1995). In the present study,
we examined whether or not three potential sites on GnT-III
are actually glycosylated. We found that all three potential sites
are N-glycosylated in COS-1 cells.
To study the role of N-glycosylation in GnT-HI, we completely blocked the N-glycosylation of GnT-in with tunicamycin. The GnT-IH activity was almost completely lost by tunicamycin-treatment, as was expected from the recombinant
GnT-EU expressed in E.coli (our unpublished data).
Subsequently, we used castanospermine, an inhibitor of glucosidase I and n. Castanospermine-treated GnT-UI migrated
on SDS-PAGE to 69.7 kDa, the size corresponding to that of
glucosylated GnT-UI (Figure 4). Although the activity of castanospermine-treated GnT-UI was almost the same as that of
the wild-type, the transfected enzyme was not retained in the
Golgi. In some glycoproteins, the inhibition of glucosidase
activity was found to block secretion or cell surface expression
in a line of studies (Gross et al., 1983; Lodish and Kong, 1984;
Schlesinger et al, 1984; Moore and Spiro, 1993). These results
suggested that the glucoses on N-glycans function as specific
signals for glycoprotein retention and/or degradation in the ER.
Conversely, in the case of other glycoproteins, a glucosidase
inhibitor was found not to inhibit the secretion or cell surface
expression in another line of studies (Collet and Fielding,
1991; Ennsetal, 1991). They suggested that glucose trimming
is involved in the protein folding of particular glycoproteins.
Indeed, in a soybean lectin, the high-mannose sugar chain acquires the protein folding function on glucose trimming (Nagai
et al., 1993). In the present study, however, castanosperminetreated GnT-III exhibited sufficient activity suggesting that
GnT-III was completely folded into an active form. This suggests a possibility that the completely folded form recognized
by cells is not the same as the completely active form.
Although removal of any one of the three N-linked oligosaccharides resulted in a significant, but not marked, decrease
in the enzyme activity, deglycosylation at individual sites
showed different effects on the enzyme activity. While the
MT3 mutation resulted in a -30% decrease, the MT1 and MT2
ones showed 50-60% less activity than that of the wild-type.
The double mutation, MT1,2, showed enzyme activity reduced
to -10% of the wild-type level, but the MT1,3 and MT2,3
mutations showed only a little less activity than single ones,
MT1 and MT2. Complete deglycosylation of GnT-UI resulted
in negligible activity, corresponding to the loss of activity of
771
K.Nagai et al.
WT(198)
TM(135)
CS (242)
Fig. 6. The effects of tunicamycin and castanospermine on the subcellular localization of GnT-III in COS-1 cells, (a) Transfected COS-1 cells were stained
with anti-rat GnT-III polyclonal antibody (wild-type, TM, and CS). Typical ER staining or Golgi staining pattern was shown by anti-calreticulin antibody
(ER) or anti-Gal-T serum (Gal-T), respectively. The Golgi staining pattern of Gal-T and wild-type was verified by double staining with FITC-labeled WGA
(Golgi) (panels on the left). TM, tunicamycin; CS, castanospermine. (b) The relative proportions of staining patterns, ER versus Golgi, in the transfected
cells. The number of cells counted is given in parentheses.
tunicamycin-treated GnT-III. These results indicate that Nglycosylation at any of the three potential sites significantly
contributes to the expression of GnT-III activity, but the glycosylations at Asn243 and Asn261 are more important than
that at Asn399. To determine the mechanism by which the
removal of single or multiple N-glycosylations in GnT-HI resulted in an additive decrease in enzyme activity, kinetic analysis was then performed. Though the V max values apparently
varied between the wild-type and mutant transfectants, glycosylations at Asn243 and Asn261 affect the Km value for the
donor but not for the acceptor sugar. On the other hand, gly772
cosylation at Asn399 does not contribute to the affinity for
either the donor or acceptor sugar. These results suggested that
glycosylations at Asn243 and Asn261 participate in the binding
to donor, and the glycosylation at Asn399 is not directly involved in the interaction of the enzyme with donor or acceptor.
The glycosylation at the Asn399 may contribute to the stability
of the enzyme structure and is thus required for the full enzyme
activity.
During the process of biosynthesis of N-linked oligosaccharides in the ER and Golgi, certain sugar residues are removed
and others are added through a defined order of reactions cata-
Role of N-glycosylation in GnT-III function
CO
CM CO CO C\T
i-" 1-" cvf 1-"
Materials and methods
Materials
Materials were obtained from the following sources: pSVK3, protem-GSepharose (Pharmacia, Uppsala, Sweden), pET3d and pCHHO (Novagen),
tunicamycin, a-fucosidase, UDP-GlcNAc and bovine -y-globulin (Sigma
Chemical Co.), castanospermine (Wako Chem. Co., Osaka, Japan), B-galactosidase (Seikagaku-Kogyo, Tokyo, Japan), ODS-80TM column (Tosoh, Tokyo, Japan), Lab-Tek chamber slide (Nunc), rhodamine-conjugated anti-rabbit
immunoglobulin (Tago), FTTC-labeled wheat germ agglutinin (Honen, Tokyo,
Japan), Vectashield mounting medium (Vector Lab.), FTTC-labeled anti-goat
immunoglobulins and horseradish peroxidase-conjugated anti-rabbit immunoglobulin (Dako, Glostrup, Denmark), ECL chemiluminesence detection kit
(Amersham), goat anti-calrericulin serum and rabbit anti-galactosyltransferase
serum were gifts from Dr. MacLennan at University of Toronto and Dr. Shaper
at Johns Hopkins University, respectively.
Preparation of antiserum against rat GnT-III
A GnT-IJJ protein lacking the N-tenninal 22 amino acid residues was produced
in E.coli by using a pb'l'Jd expression vector. The recombinant GnT-III isolated on SDS-PAGE, and recovered directly from the gel after staining, was
used to raise antibody in rabbits. One of the rabbits produced an antiserum that
reacted strongly with the GnT-in expressed in COS-1 cells as judged by
Ouchterlony memods and immunoblotting.
Construction of a rat GnT-III expression vector and site-directed
mutagenesis
Fig. 7. SDS-PAGE and Western blotting analysis of GnT-III mutants.
Lysates of the COS-1 cells transfected with mock, wild-type enzyme, or
various mutants were separated on an 8% polyacrylamide gel under
reducing conditions. The GnT-III bands were detected with rabbit
polyclonal antibody.
lyzed by specific enzymes. Therefore, the glycosylating enzymes (glycosyltransferases and glycosidases) including GnTIII should have a targeting mechanism by which precise glycosylation steps should be organized and completed. GnT-III is
one of the medial Golgi enzymes. Although the mechanism
underlying Golgi retention has not been resolved, two hypothetical models have been reported. One is that the length of
the transmembrane domain determines the behavior of proteins
in the secretory pathway (Munro, 1991, 1995), and the other is
the "kin recognition" model (Nilsson et aL, 1994). On the
basis of the latter hypothesis, a protein that should be localized
in the medial Golgi is retained in the proper compartment
through the formation of "hetero-oligomers," which are too
large to enter forward-moving transported vesicles, in situ,
with other medial Golgi proteins. Elimination of any one of the
three N-glycosylation sites might result in failure to form such
"hetero-oligomers" and thus in loss of retention in the Golgi.
Although the activity of MT3 was 70-80% of the wild-type
level, MT3 was not localized in the Golgi (Figure 8). This
finding, as well as the localization of castanospermine-treated
GnT-III, suggests that the native folding of the domains, which
contain these glycosylation sites but are not so important for
enzyme activity, is required for Golgi retention. Otherwise,
glycosylation and glucose trimming of GnT-EQ may participate
in the "kin recognition."
In conclusion, this study demonstrated that all three potential
sites are glycosylated in GnT-m, and that they are required for
Golgi retention and for full expression of the catalytic activity
of this enzyme. This study also suggested that the active form
in vitro is not always the correctly folded one. However, further studies are required to determine the precise roles of individual N-linked carbohydrate chains, and the mechanism by
which they affect enzyme activity and Golgi retention.
The construction of plasmid pSVK3-(rat GnT-uTl was described previously
(Nishikawa et at, 1992). Site-directed mutagenesis was carried out using
synthetic oligonucleotide primers according to the method of Kunkel (1985).
The uracil-substituted single-stranded DNA was prepared from an E.coli
strain CJ236 transformed with the plasmid. The uracil template was used
together with oligonucleotide primers to generate mutant sequences. The synthetic oligonucleotide primers used to replace Asn by Gin were as follows:
5'-CTCCCCGTACGCGGTGAACTGGGATTCGCAG-3' for N243Q; 5'CTCGAAGGTACCCTGGGTCAGCATC-3' for N261Q; and 5 ' GATGTGGCCCGTACGCTGCTCATACTGTCG-3' for N399Q. The mutations in the pSVK3-(GnT-IIIs) were verified by sequence analysis.
Transient expression of rat GnT-III in COS-I cells and tunicamycin and
castanospermine treatments
Transfection of pSVK3-(rat GnT-DT) into COS-1 cells was performed by electroporarion as described previously (Nishikawa et at, 1992). A (3-galactosidase
expression vector pCHl 10 was cotransfected in order to normalize the transfection efficiency (Kang et aL, 1996).
Tunicamycin or castanospermine was added to the cells at a concentration of
1 u,g/ml or 50 M.g/ml, respectively, 6 h after transfection. Two days later, the
total 5 x 1 0 * cells were harvested and lysed by sonication in 0.3 ml of phosphate-buffered saline (PBS).
Immunoadsorption of GnT-III activity with anti-GnT-III serum
Lysates (-0.1 mg protein) of the transfected COS-1 cells were incubated in 50
\ii PBS with increasing volumes (0-10 |xl) of anti-GnT-m serum or preimmune rabbit serum at 4°C for 3 h with gentle agitation. The solution was
then incubated with 10 u.1 of protein-G-Sepharose Fast Flow for 30 min, and
the immune complex was precipitated by centrifugation at 2000 x g for 5 min.
GnT-JH activity in the supernatant was measured.
Measurement of GnT-III activity
A fluorescence-labeled acceptor sugar chain, G l c N A c B l - 2 M a n a l 6(GlcNAc61-2Manal-3)Manpi^GlcNAc31^GlcNAc-2-aminopyridine,
was prepared from bovine -y-globulin. Briefly, a pronase digest of -/-globulin
was subjected to hydrazinolysis and N-acetylation, and the resulting oligosaccharides were pyridylaminated according to the method of Hase et aL (1984).
After consecutive digestions with B-galactosidase and a-fucosidase, the acceptor sugar chain was purified using a reverse-phase ODS-80TM column.
The standard GnT-JH assay was carried out in a 50 u,l solution containing 40
\>M of the fluorescence-labeled acceptor sugar chain, 20 mM of the donor
substrate UDP-GlcNAc, and 10 uJ of cell lysate at 37°C for 1 h (Nishikawa et
113
K.Nagai et aL
a
WT(198)
N10(244)
N2Q(192)
N3Q (225)
Fig. 8. Subcellular localization of mutant GnT-Hls in the transfected COS-1 cells, (a) The COS-1 cells transfected with wild-type, or mutant, enzymes were
stained with anti-rat GnT-HI polyclonal antibody, (b) Stained cells were counted as described in Figure 6.
aL, 1990). The entire solution was directly injected to the ODS column after
being boiled.
The GnT-ITJ activity in the transfected cells were normalized based on the
cotransfected B-galactosidase activity (Kang et al, 1996).
directly. Alternatively, the latter was obtained by rabbit anti-Gal-T serum
followed by rhodamine-conjugated anti-rabbit immunoglobulin. After washing, the stained cells were mounted in the Vectashield medium. A Zeiss Axiophot, with epi-illumination for fluorescence, was used for fluorescent microscopy.
Fluorescence microscopy
Transfected COS-1 cells (6 x 10V0.3 ml) were grown on a Lab-Tek chamber
slide for two days. The cells were then fixed with 4% paraformaldehyde in
PBS and permeabilized for 10 min with 0.1% Triton X-100 in PBS. The cells
were blocked with 1% BSA in PBS, and incubated with the rabbit anti-rat
GnT-m antibody for 1 h. After washing with the PBS containing 1% BSA, the
immunoreacted primary antibody was visualized with rhodamine-conjugated
anti-rabbit immunoglobulin for 1 h. The typical pattern of ER staining was
obtained by goat anti-calreticulin serum and FITC-labeled anti-goat immunoglobulin, and that of Golgi staining was obtained by FITC-labeled WGA
774
SDS-PACE and Western blot analysis
Approximately 20 u-g of the COS-1 cell lysate were electrophoresed on a
5—159b gradient or an 8% homogeneous SDS-polyacrylamide gel under reducing conditions, and the separated proteins were electrophoretically transferred to a nitrocellulose membrane. GnT-III was detected utilizing the rabbit
polyclonal antibody described above and horseradish peroxidase-conjugated
goat anti-rabbit immunoglobulin. The immunoreactive bands were visualized
with an ECL chemiluminesence detection kit.
Role of N-glycosyiation in GnT-fll function
b
_
1
c
y
ED
I 20 -
/ /
E
c
E
E
z
10
//y
_
/
y
/
y
•r
s
.Ml*1
• - *
-0.6
-0.4
-0.2
0
0
-20
0.2
1 / [donor] (mM)"
1
1
1
1
1
-10
0
10
20
1
1 / [acceptor] (mM)"
1
30
Fig. 9. Kinetic analysis of the mutant GnT-IHs. The GnT-m activities in the transfected cells were measured with various concentrations of UDP-GlcNAc (a)
or an acceptor oligosaccharide (b), and the data were plotted in a Lineweaver-Burk plot wild-type (square), MT1 (circle), MT2 (triangle), MT3 (diamond).
Acknowledgments
We thank Dr. MacLennan for kindly providing anti-calreticulin serum and Dr
Shaper for kindly providing anti-Gal-T serum. This work was supported m part
by grants-in-aid for cancer research and scientific research on priority areas
from the Ministry of Education, Science, and Culture of Japan, Kanae Foundation, and the Uehara Memorial Foundation. K. Nagai received a Research
Fellowship of the Japan Society for the Promotion of Science for Young
Scientists.
Abbreviations
UDP-GlcNAc, undine 5'-diphospho-N-acetylglucosamine; GnT-UI, UDPGlcNAc: P-D-mannoside P-1,4 N-acetylglucosaminyltransferase HI; Gal-T,
UDP-galactose: GlcNAc P-1,4 galactosyltransferase; Man, mannose; FTTC,
fluorescein isothiocyanate; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; ER, endoplasmic reticulum; WGA, wheat germ
agglutinin.
References
ColleUX. and Fielding.C. J. (1991) Effects of inhibitors of N-linked oligosaccharide processing on the secretion, stability, and activity of lecithinxholesterol acyltransferase. Biochemistry, 30, 3228-3234.
Enns.C.A., CUnton.E.M., Reckhow.C.L., Root,B.J., Do,S.-I. and Cook.C.
(1991) Acquisition of the functional properties of the transferrin receptor
during its biosynthesis. / BioL Chem., 266, 13272-13277.
Fast,D.G., JamiesonJ.C. and McCaffey.G. (1993) The role of the carbohydrate
chains of GalB-l,4-GlcNAca2,6-sialyltransferase for enzyme activity. Biochim. Biophys. Acta, 1202, 325-330.
Field>l.C. and WainwrighU-J- (1995) Molecular cloning of eukaryotic glycoprotein and glycolipid glycosyltransferases: a survey. Glycobiology, 5,
Gross.V., Andus.T., Tran-Thi,T.-A., Schwarz.R.T., Decker.K. and Heinrich,P.C. (1983) 1-Deoxynojirimycin impairs oligosaccharide processing of
a 1-protease inhibitor and inhibits its secretion in primary cultures of rat
hepatocytes. J. BioL Chem.. 258, 12203-12209.
Haraguchi,M., Yamashiro.S., FurukawaJC. Takamiya,K., Shiku,H. and Furukawa.K. (1995) The effects of the site-directed removal of Nglycosylation sites from p-l,4-N-acetylgalactosaminyltransferase on its
function. Biochem. J., 312, 273-280.
Hase.S., Ibuki.T. and Ikenaka,T. (1984) Reexamination of the pyridylamination used for fluorescence labeling of oligosaccharides and its application to
glycoproteins. / Biochem., 95, 197-203.
Kangji., Saito.H., Diara,Y., Miyoshi.E., KoyamaJ^)., Sheng.Y. and Taniguchij>). (1996) Transcriptional regulation of the N-acetylglucosaminyltransferase V gene m human bile duct carcinoma cells (HuCC-Tl) is mediated by
Ets-1. J. BioL Chem., 271, 26706-26712.
Koenderman.A.H.L., Koppen,P.L., Koeleman.C.A.M. and van den EijndenJD.H. (1989) N-Acetylglucosaminyltransferase ID, IV and V activities in
Novikoff ascites tumor cells, mouse lymphoma cells and hen oviduct Eur.
J. Biochem., 181, 651-655.
Koenderman.A.H.L., Wijermans.P.W. and van den Eijnden.D.H. (1987)
Changes in the expression of N-acetylglucosaminyltransferase HI, IV, V
associated with the differentiation of HL-60 cells. FEBS Lett., 222, 42-46.
Kumar,R., YangJ., Larsen,R.D. and Stanley,P. (1990) Cloning and expression
of N-acetylglucosaminyltransfeTase I, the medial Golgi transferase that initiates complex N-linked carbohydrate formation. Proc. NatL Acad. Sci.
USA, 87, 9948-9952.
Kunkel.T.A. (1985) Rapid and efficient site-specific mutagenesis wimout phenotypic selection. Proc. NatL Acad. Sci. USA, 82, 488-492.
Lodish.H.F. and Kong,N. (1984) Glucose removal from N-linked oligosaccharides is required for efficient maturation of certain secretory glycoproteins
from the rough endoplasmic reticulum to the Golgi complex. J. Cell BioL,
98, 1720-1729.
MiyoshiJE., Nishikawa^A., fhara,Y., Gu,J., Sugiyama,T., Hayashi,N., Fusamoto,H., Kamada.T. and Taniguchi.N. (1993) N-Acetylglucosaminyltransferase UJ and V messenger RNA levels in LEC rats during hepatocarcinogenesis. Cancer Res., 53, 3899-3902.
Moore.S.E.H. and Spiro,R.G. (1993) Inhibition of glucose trimming by castanospermine results in rapid degradation of unassembled major histocompatibility complex class I molecules. /. BioL Chem., 268, 3809-3812.
Munro.S. (1991) Sequences within and adjacent to the transmembrane segment
of cr-2,6-sialyltransferase specify Golgi retention. EMBO J., 10,3577-3588.
Munro.S. (1995) An investigation of the role of transmembrane domains in
Golgi protein retention. EMBO J., 14, 4695-4704.
Nagai.K., Shibata,K. and Yamaguchi,H. (1993) Role of intramolecular highmannose chains in the folding and assembly of soybean {Glycine max) lectin
polypeptides: studies by the combined use of spectroscopy and gel-filtration
size analysis. /. Biochem., 114, 830-834.
Nilsson.T., Hoe.M.H., Slusarewicz,P., Rabouille.C, Watson.R., Hunte.F.,
Watzele.G., Berger.E.G. and Warren,G. (1994) Kin recognition between
medial Golgi enzymes in HeLa cells. EMBO J., 13, 562-574.
Nishikawa^k., GuJ., Fujii.S. and Taniguchi.N. (1990) Determination of Nacetylglucosaminyltransferases III, IV and V in normal and hepatoma tissues of rats. Biochim. Biophys. Acta, 1035, 313-318.
Nishikawa,A., Diara,Y., Hatakeyama.M., Kangawa,K., and Taniguchi.N.
(1992) Purification, cDNA cloning, and expression of UDP-N-
775
K-Nagai et al
acetylglucosamine: fJ-D-mannoside p-1,4 N-acetylglucosaminyltransfeTase
m from rat kidney. J. BioL Chan., 267, 18199-18204.
Rademacher.T.W., Dare.K.H. and DwekJi.A. (1988) Glycobiology. Annu.
Rev. Biochent, 57, 785-838.
Sarkar^M., HullJE., Nishikawa,Y., Simpson.R.J., MoritzJIX., Dunnjt. and
Schachter.H. (1991) Molecular cloning and expression of cDNA encoding
the enzyme that controls conversion of high-mannose to hybrid and complex
N-glycans UDP-N-acetylglucosamine: ot-3-D-mannoside P-l,2-N-acetylglucosaminyltransferasse L Proc. Natl Acad. Sci. USA, 88, 234-238.
Schlesinger.S., Malfer.C. and SchlesingeerJvIJ. (1984) The formation of vesicular stomatitis virus (San Juan strain) becomes temperature-sensitive
when glucose residues are retained on the oligosaccharides of the glycoprotein. /. Biol. Chem., 259, 7597-7601.
Varki,A. (1993) Biological roles of oligosaccharides: all of the theories are
correct. Glycobiology, 3, 97-130.
Yoshimura>l., NishikawaL.A., Diara,Y., Nishiura,T., Nakao,H., Kanayama,Y.,
Matuzawa.Y. and Taniguchi.N. (1995) High expression of UDP-Nacetylglucosamine: (3-D mannoside f3-1,4-N-acetylglucosamiryltmsferase
i n (GnT-IIT) in chronic myelogenous leukemia in blast crisis. Int. J. Cancer,
60, 443-449.
Received on November 15, 1996; revised on January 2, 1997; accepted on
March 4, 1997
776