Full text of DOI:10.1093/glycob/11.5.413

Glycobiology vol. 11 no. 5 pp. 413–422, 2001
The UDPase activity of the Kluyveromyces lactis Golgi GDPase has a role in uridine
nucleotide sugar transport into Golgi vesicles
Maria Dolores Lopez-Avalos, Daniela Uccelletti,
Claudia Abeijon, and Carlos B. Hirschberg1
Department of Molecular and Cell Biology, Boston University Goldman
School of Dental Medicine, 700 Albany Street Boston, MA 02118, USA
lumen, where mannose is transferred to mannans by specific
mannosyltransferases (Hong et al., 1994; Berninsone and
Hirschberg, 2000). The other reaction product, GDP, is then
hydrolyzed by a GDPase to GMP, which then exits the Golgi
lumen in a coupled, equimolar exchange with cytosolic GDP-
mannose (Hong et al., 1994; Berninsone and Hirschberg,
2000). This antiport cycle was originally described in vitro
with Golgi vesicles of mammalian cells. Evidence of its
physiological relevance has been obtained in vivo and in vitro
in Saccharomyces cerevisiae (Abeijon et al., 1993; Berninsone
et al., 1994). Following identification and purification of a
Golgi lumenal, membrane-bound GDPase, the gene encoding
this type II membrane protein (GDA1) was cloned and
disrupted (Abeijon et al., 1993). gda1 null mutants have a
block in O- and N- mannosylation of proteins, such as chitinase
and carboxypeptidase Y, undermannosylation of invertase, and
drastically reduced levels of mannosylphosphorylceramides
(Abeijon et al., 1993). These pleiotropic defects in mannosyla-
tion are most likely the result of reduced availability of GDP
mannose in the Golgi lumen; indeed, transport of GDP-mannose
into Golgi vesicles is impaired in gda1 null mutants due to
reduced levels of GMP in the Golgi lumen (Berninsone et al.,
1994). Ectoapyrases (E-ATPases) are enzymes that hydrolyze
nucleoside tri- or diphosphates. An E-ATPase was recently
cloned from S. cerevisiae, Ynd1p (Guillen et al., 1999). This
protein also plays a role in Golgi mannosylation, and gda1ynd1
double mutants have a more severe phenotype than the single
ones, suggesting functional redundance. Considered together,
these results are consistent with a model in which nucleotide
sugar transporter activities are regulated by the availability of
the antiporter molecule in the Golgi lumen (Berninsone et al.,
1994). The antiport generation cycle by nucleoside diphos-
phatases constitutes a novel mechanism of regulation of post-
translational modifications that occur in the Golgi lumen and
affect most secreted and membrane proteins (Hong et al.,
1994; Berninsone and Hirschberg, 2000).
Received on November 15, 2000; revised on January 17, 2001; accepted on
January 17, 2001
In Saccharomyces cerevisiae a Golgi lumenal GDPase
(ScGda1p) generates GMP, the antiporter required for
entry of GDP-mannose, from the cytosol, into the Golgi
lumen. Scgda1 deletion strains have severe defects in N-
and O-mannosylation of proteins and glycosphingolipids.
ScGda1p has also significant UDPase activity even though
S. cerevisiae does not utilize uridine nucleotide sugars in its
Golgi lumen. Kluyveromyces lactis, a species closely related
to S. cerevisiae, transports UDP-N-acetylglucosamine into
its Golgi lumen, where it is the sugar donor for terminal
N-acetylglucosamine of the mannan chains. We have iden-
tified and cloned a K. lactis orthologue of ScGda1p.
KlGda1p is 65% identical to ScGda1p and shares four
apyrase conserved regions with other nucleoside diphos-
phatases. KlGda1p has UDPase activity as ScGda1p.
Transport of both GDP-mannose, and UDP-GlcNAc was
decreased into Golgi vesicles from Klgda1 null mutants,
demonstrating that KlGda1p generates both GMP and
UMP required as antiporters for guanosine and uridine
nucleotide sugar transport into the Golgi lumen.
Membranes from Klgda1 null mutants showed inhibition of
glycosyltransferases utilizing uridine- and guanosine-
nucleotide sugars, presumably due to accumulation of
nucleoside diphosphates because the inhibition could be
relieved by addition of apyrase to the incubations. KlGDA1
and ScGDA1 restore the wild-type phenotype of the other
yeast gda1 deletion mutant. Surprisingly, KlGDA1 has only
a role in O-glycosylation in K. lactis but also complements
N-glycosylation defects in S. cerevisiae. Deletion mutants of
both genes have altered cell wall stability and composition,
demonstrating a broader role for the above enzymes.
Nucleoside diphosphatase activities have also been
described in mammalian membranes of the Golgi apparatus
(Brandan and Fleischer, 1982; Wang and Guidotti, 1998) and
endoplasmic reticulum (ER; Trombetta and Helenius, 1999;
Ohkubo et al., 1980). Two related nucleoside diphosphatases,
hypothesized to have a role in generating nucleoside mono-
phosphates in the lumen of the ER and Golgi apparatus, have
recently been cloned. An ecto (E)-ATPase from human brain
was shown to localize to the Golgi apparatus of transfected
COS 7 cells (Wang and Guidotti, 1998). This protein has high
activity toward UDP; lower toward GDP, CDP, and TDP; and
lowest toward the corresponding nucleoside triphosphates.
Trombetta and Helenius (1999) recently purified and cloned a
rat liver ER nucleoside diphosphatase that hydrolyzes UDP,
Key words: yeast/Golgi/glycosylation/nucleoside diphosphatases
Introduction
Terminal mannosylation of yeast proteins and lipids occurs in
the lumen of the Golgi apparatus. GDP-mannose, the sugar
donor of these reactions, must first be transported by a specific
transporter from the cytosol, its site of synthesis, into the Golgi
1To whom correspondence should be addressed
© 2001 Oxford University Press
413

M.D. Lopez-Avalos et al.
GDP, and IDP but no other nucleoside phosphates. These
mammalian nucleoside diphosphatases share high sequence
similarity in four conserved apyrase regions with Gdalp and
other proteins of still unidentified function. A Golgi GDPase
has also been detected in plants and hypothesized to yield
GMP, the putative antiporter for GDP-fucose transport into the
Golgi lumen (Wulff et al., 2000).
We showed previously that the S. cerevisiae Golgi Gda1p
could hydrolyze UDP, in addition to GDP, albeit at consider-
ably slower rate. In the presence of Mn2+ ions, the UDPase
activity was 33% of GDPase activity, whereas in the presence
of Ca2+ ions it was 11% (Yanagisawa et al., 1990). The role of
this UDPase activity is not clear because S. cerevisiae solely
utilizes GDP-mannose as sugar donor in the Golgi lumen and
no reaction utilizing uridine nucleotide sugars has been
described in the Golgi apparatus of this organism.
We then constructed a null mutant of the KlGDA1 gene to
study the physiological function of the KlGda1p. The
disrupted strain lost most (95%) of its ability to hydrolyze both
GDP and UDP in vitro; hydrolysis of both GDP and UDP was
restored on transformation of the deletion strain with a plasmid
carrying either the K. lactis (not shown) or the S. cerevisiae
gene (Table I). Golgi membrane extracts from K. lactis wild
type showed preferential hydrolytic activity towards GDP
compared to UDP in the presence of either Ca2+ or Mn2+
(Table I), although the latter ion stimulated the UDPase
activity by approximately fivefold compared to calcium ions.
Membranes from disrupted strains showed virtually no
UDPase activity in vitro (Table I).
Translocation of uridine and guanosine nucleotide sugars into
Golgi-enriched vesicles is reduced in Klgda1∆
Kluyveromyces lactis is a yeast species closely related to
S. cerevisiae; its outer mannan chains differ from those of
S. cerevisiae by having terminal N-acetylglucosamine and no
mannose phosphate (Raschke and Ballou, 1972). Previous
studies from our laboratory characterized a mutant of K. lactis
that lacked terminal N-acetylglucosamine as a consequence of
a deficiency in the Golgi UDP-N-acetylglucosamine trans-
porter (Abeijon et al., 1996a). Phenotypic correction of this
mutant allowed the molecular cloning of the first nucleotide
sugar transporter gene (Abeijon et al., 1996b) which was later
found to be contiguous to the N- acetylglucosaminyttrans-
ferase that catalyzes the addition of the terminal N- acetyl-
glucosamine to the K. lactis outer mannan chains (Hirschberg
et al., 1998).
K. lactis uses both guanosine and uridine nucleotide sugars
for synthesis of its sugar polymers in the lumen of its Golgi
apparatus. We were therefore interested in determining
whether it has a nucleoside diphosphatase with both GDPase
and UDPase activities in the Golgi lumen, and if so, its role in
transport of GDP-mannose and UDP-GlcNAc into Golgi
vesicles.
We had previously shown that S. cerevisiae Gda1p can
hydrolize UDP in vitro, in addition to its principal substrate,
GDP. No other nucleoside phosphates were substrates. To
determine the function of this UDPase activity, we studied
transport of both GDP-mannose- and uridine-containing
nucleotide sugars into Golgi vesicles from wild-type and
Klgda1 null mutants. K. lactis, in contrast to S. cerevisiae, has
N-acetylglucosamine bound to its mannan chains and UDP-
GlcNAc must first be transported into the K. lactis Golgi
lumen before the sugar can be transferred to the mannans,
yielding also UDP as reaction product. This UDP must be
converted to UMP before it can exit the Golgi lumen in a
coupled exchange with the entry of (additional) UDP-GlcNAc.
If KlGda1p is also a UDPase in vivo, one would predict that the
null mutant would have a reduction in transport of UDP-GlcNAc
into K. lactis Golgi vesicles, in addition to that of GDP-mannose.
As can be seen in Figure 3, transport of UDP-GlcNAc into the
lumen of Golgi-enriched vesicles from K. lactis Klgda1 null
mutants was approximately half compared to that into wild-
type vesicles. This reduction was seen in both solutes within
vesicles as well as in sugars bound to macromolecules within
the vesicles. Transport of GDP-mannose and UDP-glucose
was also reduced (Figure 3). Reduction of UDP-glucose trans-
port had not been observed in S. cerevisiae gda1 null mutants
(not shown), suggesting that K. lactis may have a UDP-glucose
transport activity in both the Golgi apparatus and endoplasmic
reticulum. In S. cerevisiae UDP-glucose enters primarily into
the endoplasmic reticulum (Castro et al., 1999), whereas
Gda1p resides in the Golgi.
Results
The K. lactis Golgi GDPase gene (KlGDA1) is a new member
of the apyrase family
We cloned the K. lactis counterpart of the S. cerevisiae GDA1
gene using the latter as a probe in a low-stringency screen of a
K. lactis genomic library. The K. lactis GDPase gene (KlGDA1;
GenBank accession no. AJ401304) encodes a protein of 522
amino acids that is 65% identical to the S. cerevisiae GDA1. The
Kyte-Doolittle algorithm revealed a single hydrophobic
domain close to the amino terminus (not shown), consistent
with a type II membrane protein like the S. cerevisiae counter-
part. Alignment of the KlGda1p with other proteins in the
DataBank shows its homology to apyrases. All these proteins
share four apyrase conserved motifs as shown in Figure 1.
Sucrose gradient fractionation of wild-type cells showed that
the GDPase activity comigrated with Golgi membrane markers
and was separated from an ER marker (Figure 2). The catalytic
domain appears to be luminal because low concentrations of
Triton X-100 are required for detecting its activity (not
shown).
Inhibition of α1,2 mannosyltransferase and α1,2 N-
acetylglucosaminyltransferase activities in Klgda1∆ protein
extracts containing permeabilized Golgi vesicles
Once nucleotide sugars enter the lumen of the Golgi apparatus,
glycosyltransferases catalyze the transfer of the sugars to
endogenous acceptors; the other reaction products, nucleoside
diphosphates, are converted by the Golgi GDPase/UDPase to
the corresponding nucleoside monophosphates. These then
exit the lumen in an antiport with uridine- and guanosine-
nucleotide sugars. GDP and UDP are inhibitors of glycosyl-
transferases that use guanosine- and uridine-diphosphate
sugars as substrates. One would therefore predict that in
Klgda1∆ permeabilized Golgi vesicles, glycosyltransferases
are inhibited compared to the corresponding activities in wild-type
414

Yeast Golgi nucleoside diphosphatases
Fig. 1. Multiple alignment of proteins related to Klgda1p. The BOXSHADE program was used to compare sequences of proteins, from fungi to mammals. Black
background shows identical residues and gray background shows conservative substitutions. Asterisk indicates conserved domains with putative apyrase activity.
Accession numbers for the sequences used in the alignment are: K. lactis GDPase, AJ401304; S. cerevisiae GDPase, NP_010872; pea NTPase, BAA89275; potato
ATPase, P80595; human CD39, NP_001767; mouse CD39, NP_033978; C. elegans, Q18411; S. cerevisiae YND1, NP_010920; human Golgi UDPase AF016032.
permeabilized vesicles. This is because the GDP generated as by-
product following the transfer of mannose (from GDP-mannose) to
the exogenous acceptor is accumulated in the incubation medium.
Inhibition would not be seen in the wild-type extracts because
there Gda1p hydrolyzes GDP, which is inhibitory, to GMP,
which is not. As shown in Figure 4A, α1,2 mannosyltransferse
activity was indeed reduced in Golgi membranes from null
mutants compared to those from wild-type cells (column 1
415

M.D. Lopez-Avalos et al.
Table I. Guanosine and uridine-diphosphatase activities in membranes of K. lactis
KlGDA1
Ca2+
Klgda1∆
Ca2+
Klgda1∆ + [pScGDA1]
Substrate
GDP
Mn2+
Mn2+
Ca2+
Mn2+
100 0.7
97 0.6
21 0.2
4
0
1.5
1.8
1
0
0.7
0.9
100 0.5
87 1.9
UDP
4
0.0
2
0.0
9
1.1
Membranes of K. lactis wild type, gda1∆ and null mutants transformed with an S. cerevisiae GDA1 multicopy plasmid were assayed for
hydrolysis of GDP and UDP in the presence of Ca2+ and Mn2+ as described in Materials and methods. Results are expressed as percentage of
GDP hydrolysis of wild-type membranes in the presence of Ca2+ and are the mean of three independent experiments SE. The value for
hydrolysis of GDP/Ca2+ in the wild type was 3.4 nmol phosphate/µg prot/20 min and 23.7 nmol phosphate/µg prot/20 min for Klgda1∆ +
[pScGDA1].
Fig. 2. The GDPase and UDPase Activities are in the Golgi apparatus. Enzyme
activities were assayed after velocity sucrose gradient centrifugation. Cells were
converted to spheroplasts and the 10,000 × g supernatant was loaded on to a
25–50% sucrose gradient. Fractions were collected from the top, diluted, and
centrifuged at 100,000 × g for 40 min. Pellets were resuspended and used for the
assays indicated. Black circles: endoplasmic reticulum marker: NADPH
cytochrome C reductase. Black squares: Golgi marker: α1,2-mannosyltransferase.
Gray triangles: GDPase activity. Dashed line: sucrose concentration.
Fig. 3. Translocation of nucleotide sugars into Golgi vesicles. Vesicles from
wild-type and Klgda1 mutant cells were assayed for transport of GDP-Man,
versus 5). Moreover, this inhibition could be specifically
overcome by addition of apyrase to the assay of extracts from
Klgda1∆ (columns 6–8). As expected, addition of apyrase had
no effect on wild type extracts (columns 2–4). These results
demonstrate that the inhibition was caused by accumulation of
GDP in the incubation medium of extracts from the mutant. One
would also postulate that the above inhibition of glycosyl-
transferases by nucleoside diphosphates in Klgda1∆ Golgi
membranes would be higher when increasing the concentration
of nucleotide sugar substrates in the assay, because this would
result in higher concentrations of nucleoside diphosphate
by-products. This inhibition should, in turn, be overcome by
membranes from Klgda1∆ cells expressing a plasmid copy of
the ScGDA1 gene. Figures 4B and 5 show that indeed this
occurred and that the mannosyltransferase activity appeared to
be more sensitive to inhibition by GDP than the α1,2 N-acetyl-
glucosaminyltransferase activity to inhibition by UDP. These
results suggest that in the absence of Golgi GDPase/UDPase
activities, GDP and UDP generated from the glycosyltrans-
ferase reactions inhibit such enzymes and probably impair the
correct glycosylation of macromolecules.
UDP-GlcNAc, and UDP-Glc at 30°C for 3 min. Black bars: solutes within
vesicles; gray bars: sugars bound to macromolecules. Results are mean SE of
three independent experiments.
S. cerevisiae gda1 mutants. Using N-glycosylation reporters,
such as invertase and carboxypeptidase Y, we found, to our
surprise, no differences between the K. lactis wild type and
null mutant, even though drastic differences in the behavior of
the above proteins had been previously observed between S.
cerevisiae wild type and gda1 null mutants (Abeijon et al.,
1993). No differences in binding of GS II lectin to cell surfaces
of wild-type and Klgda1 cells were detected. Although
Klgda1∆ cells did not display any defect in N-glycosylation,
the KlGDA1 gene complemented all the alterations in N-glyco-
sylation of S. cerevisiae gda1 null mutants, including CPY
glycosylation, synthesis of outer mannan chains of cell wall
mannoproteins and also lipid mannosylation (not shown).
We then used chitinase, a heavily O-glycosylated protein that is
secreted into the media, as a probe for changes in O-glyco-
sylation. As can be seen in Figure 6 lanes 4 and 5, chitinase
from the null mutant migrated faster than that of wild-type
cells. We had previously shown that in S. cerevisiae gda1 null
mutants this faster migration of chitinase on SDS gels was the
result of decreased O-mannosylation (Abeijon et al., 1993).
KlGDA1 has a role in O-glycosylation
To determine whether KIGDA1 is the functional homolog of
GDA1 from S. cerevisiae, we compared the glycosylation
phenotypes of Klgda1∆ cells with those already reported for
416

Yeast Golgi nucleoside diphosphatases
Fig. 5. Effect of UDP-GlcNAc concentration and cross-species
complementation on α1,2-N-acetylglucosaminyltransferase activity. Transfer
of GlcNac to the exogenous acceptor Man α1,3 Man was measured as described
in Materials and methods. Results are mean SE of three independent
experiments.
Fig. 6. Chitinase analysis of K. lactis and S. cerevisiae cells. Secreted chitinase
was isolated by binding to chitin, separated by SDS–PAGE, and visualized by
Western blots with antibodies against S. cerevisiae chitinase.
The same behavior was observed here (Figure 6 lanes 1 and 2).
We also observed a slight increase of shorter mannose chains
on β-elimination of total glycoproteins from the null mutant
compared to wild-type cells (not shown). Finally, we tested the
ability of a plasmid containing KlGDA1 to correct the manno-
sylation defect of chitinase expressed by the S. cerevisiae null
mutants. As can be seen in Figure 6, lane 3, S. cerevisiae
chitinase regained its wild-type mobility, demonstrating
functional homology between the S. cerevisiae and K. lactis
GDA1 genes.
Fig. 4. αl, 2-mannosyltransferase activity is reduced in Klgda1∆ protein extracts
containing permeabilized Golgi vesicles (A) Protein extracts containing wild-
type or mutant permeabilized Golgi vesicles were assayed for mannose transfer
to methyl αD- mannopyranoside in the presence of different concentrations of
apyrase added to the incubation medium. (B) Effect of GDP-mannose
concentration and cross-species gene complementation on α1,2
mannosyltransferase activity. Results are mean SE of three independent
experiment.
KlGda1p has a role in cell wall morphogenesis
The yeast cell wall contains glucans and mannoproteins as
major components, and chitin, a polymer of N-acetyl-
glucosamine, is a minor one. The enzymes involved in the
synthesis of these polymers, as those of mannoproteins, either
417

M.D. Lopez-Avalos et al.
travel to the plasma membrane via secretory vesicles or are
localized in organelles of the secretory pathway, such as the
Golgi apparatus. We therefore determined whether deletion of
KlGDA1 had an effect of cell wall strength by measuring
sensitivity toward cell lysis following β-1,3 glucanase (Zymo-
lyase™) treatment. As can be seen in Figure 7A and B, null
mutant cells were more sensitive toward lysis following Zymo-
lyase™ compared to wild-type cells; this sensitivity was even
more pronounced in cultures grown at 37°C than those grown
at 23°C. When cells were grown at 37°C the sensitivity toward
lysis of the mutant increased slightly, whereas that of the wild-
type cells decreased. Indeed, after 2 min of incubation with
Zymolyase™ only 60% of the mutant cells grown at 37°C
were resistant to lysis compared to 80% of those grown at
23°C. Cell walls from wild-type cells were considerably more
resistant to Zymolyase™ treatment than mutants grown at
either temperature (Figure 7A,B).
of cell wall β-1,3 glucans, relative to wild type, in cells grown
at both 23°C and 37°C. A more robust change was seen at the
latter temperature (Figure 8, top). An increase of β-1,6 glucans
was also observed in K. lactis gda1 mutants, but only when
grown at 37°C. In S. cerevisiae, contrary to K. lactis, null
mutants did not show an increase of β-1,3 glucans but a
decrease; this was more pronounced in cells grown at 23°C
than 37°C and was also observed to a lesser extent for β-1,6
glucans. Chitin, another cell wall component, was not changed
in the K. lactis gda1 null mutant but showed a twofold increase
in the corresponding mutant of S. cerevisiae (Figure 9).
The above-described cell wall fragility had not been
examined in the S. cerevisiae gda1 null mutants. We therefore
compared the strength of the cell wall in S. cerevisiae wild-
type and gda1 cells. Surprisingly, we found a behavior
opposite to that of K. lactis. The cell wall of S. cerevisiae gda1
mutants was more resistant to lysis following Zymolyase™
treatment than that of wild-type cells grown at either 23°C or
37°C. Mutant cells grown at 37°C showed an even higher
resistance toward lysis than those grown at 23°C (not shown).
The different lysis behavior of cells following Zymolyase™
treatment suggested differences in cell wall composition. To test
this hypothesis, alkali-insoluble glucans were prepared from cell
walls of wild-type and gda1 null mutants of S. cerevisiae and
K. lactis. Deletion of GDA1 in K. lactis resulted in an increase
Fig. 7. K. lactis and S. cerevisiae cell lysis sensitivity following treatment with
Zymolyase™. Cells were treated with zymolyase for different times and lysis
was measured by the decrease in OD660 after osmotic shock. Results are mean
SE of three independent experiments.
Fig. 8. Cell wall glucans of K. lactis and S. cerevisiae. Cell wall β1,3- and
β1,6-glucans were prepared and measured as described in Materials and
methods. Results are mean SE of three independent experiments.
418

Yeast Golgi nucleoside diphosphatases
results presented here: (1) K. lactis has a GDPase/UDPase,
KlGda1p, that corrects S. cerevisiae N-and O-mannosylation
defects in vivo; in K. lactis it only affects O-mannosylation in
vivo. (2) The UDPase activity of KlGda1p has a role in trans-
port of uridine nucleotide sugars into Golgi vesicles in vitro
and in generating UMP; it thereby overcomes the inhibition by
UDP of those glycosyltransferases that use uridine nucleotide
sugars as substrates. These studies could not be performed in
S. cerevisiae because no reactions using uridine nucleotide
sugar substrates are known to occur in its Golgi apparatus. (3)
The GDPase activity of K. lactis Gda1p is required, as that
from S.cerevisiae, for efficient transport of GDP-mannose into
the Golgi lumen.
Deletion of KlGDA1 resulted in some residual GDPase
activity in the K. lactis membranes, while almost no remaining
activity had been detected in membranes of S. cerevisiae gda1
null mutants (Abeijon et al., 1993). Even though a very low
nucleoside diphosphatase activity can be measured in vitro in
S. cerevisiae gda1 null mutants, another apyrase, Ynd1p, has
recently been identified and cloned in this organism (Guillen et
al., 1999). The substrate specificity of Ynd1p is broader than
that of Gda1p, but their function is apparently partially
redundant because the double mutant gda1∆ ynd1∆ has a more
severe glycosylation phenotype than any of the individual
mutants (Guillen et al., 1999). Ynd1p apyrase activity in the
Golgi lumen has been shown to have a very narrow optimal
pH, between 7.5 and 8, and to be down-regulated by direct
binding of its cytosolic domain to the activator subunit
Vma13p of the vacuolar H+ ATPase (Zhong et al., 2000). Most
likely, K. lactis has at least one other nucleoside diphosphatase
or apyrase that generates nucleoside monophosphates in its
Golgi lumen. This could explain why N-glycosylation was not
affected in KlGDA1 mutants.
The possibility of multiple proteins in the Golgi lumen
giving rise to nucleoside monophosphates is particularly rele-
vant to mammalian Golgi membranes where evidence already
suggests that more that one protein with nucleoside diphos-
phatase activity may reside (Wang and Guidotti, 1998). These
observations, in addition to results of this study, strongly
suggest that there is regulation of glycosylation also at the level
of the production of the antiporter molecules involved in the
nucleotide sugar transport cycle in higher and lower eukaryotes.
These studies are the first to show a role for GDA1 in yeast
cell wall morphogenesis. Surprisingly, despite of the amino
acid sequence similarity between ScGDA1 and KlGDA1 and
their functional homology, deletion of each gene had very
different effects on osmotic stability and cell wall polymer
composition of both null mutant strains compared to each other
and to their respective wild types. These differences were even
more accentuated when the cells were grown at 37°C instead of
23°C. Thus, K. lactis gda1 null mutants were more sensitive
toward β-1,3 glucanase–induced cell lysis than their wild type
possibly because they had increased cell wall β-1,3 glucans at
both temperatures. β-1,6 glucans were increased only when
mutants were grown at 37°C, and chitin was unchanged at both
growth temperatures. In marked contrast, S. cerevisiae deletion
mutants were more resistant toward β-1,3 glucanase–induced
cell lysis than wild type at both temperatures. This behavior
may be a consequence of increased cell wall chitin content
found in S. cerevisiae mutants grown at both temperatures.
Chitin fibers may give strength to the wall and make it more
Fig. 9. Chitin content of K. lactis and S. cerevisiae. Chitin content of cells was
measured as described in Materials and methods. Results are mean SE of
three independent experiments.
Discussion
The previous studies of the S. cerevisiae Golgi GDPase have
provided strong evidence for the role of this enzyme in the
nucleotide sugar transporter/antiport cycle. Evidence that this
cycle has a pivotal function in regulating those postranslational
modifications that occur in the lumen of the Golgi apparatus
and affects approximately 80% of all secreted and membrane
proteins in eukaryotes is beginning to emerge. Several
important and novel conclusions were obtained from the
419

M.D. Lopez-Avalos et al.
resistant to osmotic shock, even in the presence of a decrease
in β-1,3 glucans. The changes found in the cell wall of
S. cerevisiae gda1∆ mutants are very similar to those reported
on inactivation of the KNR4 gene (Kapteyn et al., 1999). This
gene appears to encode a cytoplasmic protein that is part of a
regulatory complex involved in the assembly of the cell wall
(Martin et al., 1999).
GDA1 gene. The whole ORF was located within a 1.9-kb
HindIII fragment that was subcloned into the HindIII site of
Kep6 to obtain pDL-13. The Kep6 vector has a segment of
pKD1 that carries the replication origin of the 1.6-µm plasmid
(Bianchi et al., 1987). Stable replication of this vector requires
the presence of resident pKD1 in the recipient yeast cell. The
copy number and stability of this plasmid is comparable to the
2-µm plasmid–derived vectors of S. cerevisiae. Kep6 also has
the URA3 gene of S. cerevisiae that complements the uraA
mutation of K. lactis; it also contains pBR322 sequences to
allow selection and amplification in E. coli. The complete
sequence of the 1.9-kb genomic HindIII fragment containing
the KlGDA1 gene was deposited in the EMBL Nucleotide
Sequence Database with the accession number AJ401304.
Although the mechanism underlying the above changes in
cell wall polymers in GDA1 null mutants from both yeast
species are not clearly understood, they may be related to
changes in the O-glycosylation pathway. Recently, a highly
O-glycosylated cell wall protein family that is not anchored to
the wall by the clasical glycosylphosphatidylinositol anchor
PIR-CWP was characterized (Toh-e et al., 1993; Mrsa and
Tanner, 1999). These proteins are thought to play a compensatory
role when cell walls are weakened; O-linked sugars may be
necessary to link these proteins to β-1,3-glucans (Ketela et al.,
1999).Ccw11p, an abundant PIR-related protein, is absent in
cell walls of a pmt4∆ strain, which is known to be defective in
O-glycosylation (Mrsa et al., 1997). Two potential sensors of
cell wall stress, Hkr1p and Mid2p, are exclusively O-glyco-
sylated (Yabe et al., 1996; Kuranda and Robbins , 1991) and
may also play a role in regulating the different compensatory
changes occurring in the wall of K. lactis and S. cerevisiae. The
direct targets of these modifications will be subject of further
studies and may provide insights into the molecular basis of the
organization and dynamics of the yeast cell wall.
Construction of Klgda1 null mutant strain, KL8∆
Inactivation of the KlGDA1 gene was achieved by the URA
blaster protocol (Alani et al., 1987). To obtain the disruption
plasmid, pBlueKlhUh, the 1.9-kb HindIII fragment containing
the ORF was cloned into the HindIII site of pBluescript
(Stratagene, La Jolla, CA) whose BamHI site had been previ-
ously removed. An internal 0.9-kb BamHI fragment from the
ORF was replaced with a 3.8-kb BamHI–BamHl fragment
from plasmid pNKY51 (kindly provided by N. Kleckner) that
contained the URA3 gene flanked by hisG direct repeats.
Recombination between the flanking direct repeats results in
the elimination of the URA3 gene and recovery of uracil
auxotrophy. Digestion of pBlueKlhUh with Xhol and EcoRl
released a linear fragment containing the disruption cassette
that was introduced in K. lactis strain K18 by electroporation.
DNA from transformants was used to confirm the disruption of
the KlGDA1 gene by polymerase chain reaction (PCR)
analysis. The following primers were generated for this
purpose: Forward primer 5′-CAAAAGCAAACAGCCAAA-
GACC-3′ (located at position + 198). Reverse primer 5′-CGTC-
TTTGCTGTTCTCAACTCTC-3′ (located at position +1475).
These primers amplified a 1.2-kb PCR product in wild-type
cells; in mutant cells a product of 4.3 kb was obtained. PCR of
genomic DNA from mutant cells after the pop-out of the hisG–
URA–hisG sequence resulted in a product of 1.6 kb. Correct
integration was found in 1% of the clones and was further
confirmed by assaying for reduced GDPase/UDPase activity in
membrane protein extracts.
Materials and methods
Strains, media, and reagents
K. lactis strain KL8 (MG 1/2, MAT α, ura A, arg-, lys-, K+,
pKD1+) was obtained from C. Falcone. S. cerevisiae strains,
G2-10 (MAT α, ura3-52, lys2-801 am, ade2-101 oc,trp1-∆1,
his3-∆200, leu2-∆1) and G2-11 (MAT α, ura3-52, lys2-801
am, ade2-101 oc, trp1-∆1, his3-∆200, leu2-∆1, gda1::LEU2)
were described previously (Abeijon et al., 1993). Cells were
grown in YPD or SD medium supplemented with 1%
casamino acids. Solid media were made by adding 2% agar to
the liquid stock. 5-Fluoro-orotic acid was added when
indicated to a final concentration of 0.1%. Yeast cells were
grown at 30°C. Transformations with plasmids were done by
electroporation (Abeijon et al., 1996b). Escherichia coli strain
DH5α (Gibco BRL) was grown in LB medium with 50 µg/ml
ampicillin when needed. Reagents for media were from Difco
Laboratories (Detroit, MI). Unless otherwise stated, all other
reagents were from Sigma-Aldrich (St. Louis, MO). GDP-
[3H]mannose (15 Ci/mmol), UDP-[3H]glucose (11.5 Ci/mmol),
and UDP- [3H]N-acetyl-D-glucosamine (30 Ci/mmol) were
purchased from NEN Life Science Products (Boston, MA).
Construction of plasmids pKEpSC and pYEpKl for
cross-species complementation
To express GDA1 and KlGDA1 in K. lactis and S. cerevisiae,
we constructed the following plasmids. A 2.2-kb HindIII–NheI
fragment containing the S. cerevisiae gene was cloned into
Kep6 previously opened with HindIII and NheI to generate the
K. lactis plasmid pKEpSc. Similarly, a 1.9-kb HindIII fragment
containing the KlGDA1 gene was cloned into the HindIII site
of the S. cerevisiae plasmid Yep 352 to generate plasmid
pYEpKl. K. lactis and S. cerevisiae null mutants were then
transformed with pKEpSc and pYEpKl, respectively.
Isolation of KlGDA1 gene
A K. lactis genomic library previously made in our laboratory
(Abeijon et al., 1996a) was screened at low stringency
(Maniatis et al., 1997) using the open reading frame (ORF) of
the S. cerevisiae GDA1 gene as probe (Abeijon et al., 1993). A
positive clone, pDL-12, caused threefold overexpression of
GDPase activity when transformed into K. lactis. Restriction
analysis and sequencing of the 11.5-kb insert demonstrated a
complete ORF of 1.5 kb that was homologous to the S. cerevisiae
Preparation of vesicle fractions and sucrose gradient
fractionation
Vesicle fractions were prepared from cells (6 L), grown to a
density of 0.8 OD600. After washing with cold 10 mM sodium
azide as previously described (Abeijon et al., 1993), cells were
420

Yeast Golgi nucleoside diphosphatases
resuspended in spheroplast buffer (50 mM potassium
phosphate, pH 7.5, 1.4 M sorbitol, 10 mM sodium azide, 0.3%
β-mercaptoethanol) containing 0.5 mg/ml of Zymolyase 100T
(Seikagaku Co., Tokyo) and incubated at 37°C for 40 min. The
spheroplast suspension was centrifuged and the pellet was
washed with membrane buffer (10 mM triethanolamine acetic
acid, pH 7.2, 0.8 M sorbitol, 1 mM EDTA, 1 µg/ml leupeptin,
0.5 mM PMSF) and resuspended in the same buffer. Spheroplasts
were then broken by passage through a narrow-bore 10-ml glass
pipette, diluted in membrane buffer, and centrifuged at 1000 × g
for 10 min to obtain P1. The supernatant was centrifuged to
10,000 × g for 20 min to obtain P2, and the resulting supernatant
was again centrifuged at 100,000 × g for 30 min to obtain P3.
This fraction is enriched in Golgi markers, such as GDPase and
α1,2 mannosyltransferase, and was used for nucleotide sugar
transport assays. Vesicles were kept frozen at –70°C. Protein
was measured using the BCA method (Bio-Rad, Hercules,
CA).
For sucrose velocity gradient fractionation experiments, 1-L
cultures were used. Speroplasts were prepared as described
above and suspended in membrane buffer. The P2 supernatant
fraction was prepared and the final concentration of MgCl2 was
adjusted to 1 mM. Aliquots (9 ml) were placed on top of two
30-ml preformed, 25–50% continuous sucrose gradients in
Beckman SW28 centrifuge tubes. The sucrose solutions
contained 1 mM MgCl2 and 10 mM triethanolamine acetic
acid, pH 7.2. Gradients were centrifuged at 4°C for 90 min, at
25,000 r.p.m. in an L8-90 Beckman Ultracentrifuge as
described previously (Abeijon et al., 1993). Twelve 2.5-ml
fractions were collected from the top of each gradient, diluted,
and centrifuged at 100,000 × g for 30 min; pellets were resus-
pended in membrane buffer.
acceptor, was measured. Between 10 and 60 units of potato
apyrase (Grade VII, Sigma) were added to the reaction mixture
when needed.
Nucleotide sugar translocation assay
The theoretical basis for the translocation assay of nucleotide
derivatives into vesicles has been described in detail (Perez
and Hirschberg, 1986). Briefly, Golgi-enriched vesicles (P3
fraction, 0.25–0.5 mg protein) were incubated at 30°C or 0°C
for 3 min with the corresponding radioactive nucleotide sugar
in the following 1-ml reaction mixture: 10 mM Tris–HCl
(pH 7.4), 0.25 M sucrose, 0.15 KCl, 1 mM CaC12, and 5 mM
MgC12. Parallel incubations were done with the standard
vesicle nonpenetrator [3 H] acetate. The reaction was stopped
with 2 ml of ice-cold 0.25 M sucrose, 1 mM EDTA. Vesicles
were separated from the incubation medium by centrifugation
at 100,000 × g for 40 min. Total acid-soluble radioactivity, St,
associated with the washed vesicle pellets was determined. The
amount of radioactivity within vesicles in the pellet, Si, was
calculated by multiplying the concentration of solutes in the
incubation medium times the volume outside the vesicles in
the pellet, Vo, of the nonpenetrator [3H] acetate (1.9 µl/mg
protein). Transport activity is defined as Si after incubations at
30°C minus Si after incubations at 0°C. Latency of the vesicles
was determined by measuring the lumenal Golgi GDPase
activity, as described above, in the presence and absence of
Triton X-100; it was 90% or higher in every instance. For
gda1∆ strains, latency of N-actetylglucosamiyltransferase was
measured.
Zymolyase sensitivity assay
Cells were grown to exponential phase at 23°C or 37°C. 5 ×
108 cells were resuspended in 4 ml of buffered sorbitol (20 mM
Tris–HCl, pH 7.2, 1.2 M sorbitol, 10 mM MgCl2) with 3% 2-
mercaptoethanol. After 10 min, 1 ml of 1 mg/ml Zymolyase
100T was added, and the cells were incubated at 23°C. Lysis
was determined by measurements of A660 after 1:10 dilution in
water of samples taken at 1.5-min intervals for K. lactis and 3-min
intervals for S. cerevisiae, due to their different cell wall
fragility.
Nucleotide diphosphate hydrolysis assay
Hydrolysis of GDP, UDP, and ADP was measured in vesicle
fractions as previously described (Yanagisawa et al., 1990) in
buffer containing 0.2 M imidazole, pH 7.5, 10 mM CaC12 or
10 mM MnC12, 0.1 % Triton X-100, and 2 mM GDP, UDP, or
ADP. One hundred microliters of this solution, containing 5–10 µg
of sample protein, were incubated at 30°C for 20–30 min. The
reaction was stopped by transferring the tubes to ice and
adding 10 µl of 10% SDS. To determine the amount of
phosphate released during hydrolysis, 200 µl of water and
700 µl of AMES reagent (1:6 mixture of 10% ascorbic acid
and 0.42% ammonium molybdate in 1 N sulfuric acid) were
added to each tube; following incubation at 40°C for 20 min,
absorbance was measured at 660 nm.
Cell wall analysis
Cells were harvested and walls were isolated after mechanical
disruption with glass beads (0.45–0.55 mm diameter) (as
described in Perez and Hirschberg, 1986). Total alkali-insoluble
fractions were obtained (Nguyen et al., 1998) and β-1,6/β-1,3
glucan content was measured as described (Boone et al., 1990).
Hexose content was determined before and after dialysis using
the borosulphuric acid method (Roemer and Bussey, 1991).
Chitin was determined according to Bulawa et al. (1986).
α1,2-N-acetylglucosaminyltransferase and α1,2-
mannosyltransferase assays
Reactions were performed in a 50 µl final volume containing
30 µg of membrane protein, 50 mM HEPES, pH 7.2, 0.1%
Triton X-100, 10 mM MnCl2, and 10–200 µM GDP-[3H]mannose
(0.1 µCi) or UDP-[3H]GlcNAc (0.1 µCi). As exogenous acceptors,
10 mM methyl αD-mannopyranoside or 0.5 mM 3-O-α-D-
mannopyranosyl-D-mannopyranoside were added for manno-
syltransferase and GlcNAc transferase reactions, respectively.
After, incubations at 30°C for 25 min, reactions were stopped
by adding 0.4 ml of 10 mM EDTA. Radioactive substrates
were separated from acceptors by binding of the former to a 1-ml
Dowex-1 column; the radioactivity in the eluate, containing the
Chitinase analysis
Native chitinase was purified from late exponential cultures of
K. lactis and S. cerevisiae mutant and wild-type cells grown in
YPD. The S. cerevisiae mutant strain carrying the KlGDA1
gene was grown overnight in minimal medium and shifted to
YPD for three generations. Chitinase was isolated by its
binding to chitin as described (Maniatis, 1997). Proteins were
then subjected to 6% SDS–PAGE and detected by Western
blot using a rabbit antiserum against chitinase (kindly provided
by Dr. Charles Specht) and anti-rabbit IgG conjugated to
421

M.D. Lopez-Avalos et al.
Hirschberg, C.B., Robbins, P.W., and Abeijon, C. (1998) Transporters of
nucleotide sugars, ATP, and nucleotide sulfate in the endoplasmic
reticulum and Golgi apparatus. Annu. Rev. Biochem., 67, 49–69.
Hong, Z., Mann, P., Brown, N.H., Tran, L.E. Shaw, K.J., Hare, P.S., and
Di Domenico, B. (1994) Cloning and characterization of KNR4, a yeast
gene involved in (1,3)-beta-glucan synthesis. Mol. Cell. Mol. Cell. Biol.,
14, 1017–1025.
peroxidase (Promega, Madison, WI). Final visualization was
done with the Renaissance chemiluminescence detection kit
(NEN Life Science, Boston, MA) according to the manufac-
turer’s instructions.
Kapteyn, J.C., van Egmond, P., Sievi, E., van Den Ende, H., Makarow, M., and
Klis, F.M. (1999) The contribution of the O-glycosylated protein Pir2p/Hsp150
to the construction of the yeast cell wall in wild-type cells and beta 1,6-glucan-
deficient mutants. Mol. Microbiol., 31, 1835–1844.
Ketela, T., Green, R. and Bussey, H. (1999) Saccharomyces cerevisiae Mid2p is a
potential cell wall stress sensor and upstream activator of the PKC1-MPK1
cell integrity pathway. J. Bact., 181, 3330–3340.
Kuranda, M., and Robbins, P.W. (1991) Chitinase is required for cell separation
during growth of Saccharomyces cerevisiae. J. Biol. Chem., 266, 19758–19767.
Maniatis, T., Fritsch, E.F., and Sambrook, J. (1997) Methods in yeast genetics:
a laboratory manual. Plainview, NY: Cold Spring Harbor Laboratory
Press.
Acknowledgments
We thank Jackie Guinard for expert typing and Patricia
Bernisone for helpful discussions. MDLA was a recipient of a
postdoctoral fellowship from the Ministerio de Educacion y
Ciencia of Spain. DU is a recipient of a fellowship from the
Pasterur Institute-Cenci Bolognetti Foundation. This work was
supported by NIH grant GM 30365.
Martin, H., Dagkessamanskia, A., Satchanska, G., Dallies, N., and Francois, J.
(1999) KNR4, a suppressor of Saccharomyces cerevisiae cwh mutants, is
involved in the transcriptional control of chitin synthase genes. Microbiology,
145, 249–258.
Abbreviations
E-ATPase, ectoapyrase; ER, endoplasmic reticulum; ORF,
open reading frame; PCR, polymerase chain reaction.
Mrsa, V., and Tanner, V. (1999) Role of NaOH-extractable cell wall proteins
Ccw5p, Ccw6p, Ccw7p and Ccw8p (members of the Pir protein family) in
stability of the Saccharomyces cerevisiae cell wall. Yeast, 15, 813–820.
Mrsa, V., Seidl, T., Gentzsch, M., and Tanner, W. (1997) Specific labelling of
cell wall proteins by biotinylation. Identification of four covalently linked
O-mannosylated proteins of Saccharomyces cerevisiae. Yeast, 13, 1145–1154.
Nguyen, T.H., Fleet, G.H., and Rogers, P.L. (1998) Composition of the cell walls
of several yeast species. Appl. Microbiol. Biotechnol., 50, 206–221.
Ohkubo, I., Ishibashi, T., Taniguchi, N., and Makita, A. (1980) Purification
and characterization of nucleoside diphosphatase from rat-liver micro-
somes. Evidence for metalloenzyme and glycoprotein. Eur. J. Biochem.,
112, 111–118.
Perez, M., and Hirschberg, C.B. (1986) Topography of glycosylation reactions in
the rough endoplasmic reticulum membrane. J. Biol. Chem., 261, 6822–6830.
Raschke, W.C., and Ballou, C.E. (1972) Characterization of a yeast mannan
containing N-acetyl-D-glucosamine as an immunochemical determinant.
Biochemistry, 11, 3807–3816.
Roemer, T., and Bussey, H. (1991) Yeast beta-glucan synthesis: KRE6 encodes
a predicted type II membrane protein required for glucan synthesis in vivo
and for glucan synthase activity in vitro. Proc. Natl Acad. Sci. USA, 88,
11295–11299.
Toh-e, A., Yasunaga, S., Nisogi, H., Tanaka, K., Oguchi, T., and Matsui, Y.
(1993) Three yeast genes, PIR1, PIR2 and PIR3, containing internal
tandem repeats, are related to each other, and PIR1 and PIR2 are required
for tolerance to heat shock. Yeast, 9, 481–494.
Trombetta, E.S., and Helenius, A. (1999) Glycoprotein reglucosylation and
nucleotide sugar utilization in the secretory pathway: identification of a
nucleoside diphosphatase in the endoplasmic reticulum. EMBO J., 18,
3282–3292.
Wang, T.F., and Guidotti, G. (1998) Golgi localization and functional expres-
sion of human uridine diphosphatase. J. Biol. Chem., 273, 11392–11399.
Wulff, C., Norambuena, L., and Orellana, A. (2000) GDP-fucose uptake into the
Golgi apparatus during xyloglucan biosynthesis requires the activity of a
transporter-like protein other than the UDP-glucose transporter. Plant
Physiol., 122, 867–877.
Yabe, T., Yamada-Okabe, T., Kasahara, S., Furuichi, Y., Nakajima, T.,
Ichishima, E., Arisawa, M., and Yamada-Okabe, H. (1996) HKR1 encodes
a cell surface protein that regulates both cell wall beta-glucan synthesis and
budding pattern in the yeast Saccharomyces cerevisiae. J. Bact., 178,
477–483.
Yanagisawa, K., Resnick, D., Abeijon, C., Robbins, P.W., and Hirschberg, C.B.
(1990) A guanosine diphosphatase enriched in Golgi vesicles of Saccharomyces
cerevisiae. Purification and characterization. J. Biol. Chem., 265, 19351–19355.
Zhong, X., Malhotra, R., and Guidotti (2000) Regulation of Yeast Ectoapyrase
Ynd1p Activity by Activator Subunit Vma13p of Vacuolar H+-ATPase.
J. Biol. Chem., in press.
References
Abeijon, C., Mandon, E.C., Robbins, P.W., and Hirschberg, C.B. (1996a) A mutant
yeast deficient in Golgi transport of uridine diphosphate N-acetylglucosamine.
J. Biol. Chem., 271, 8851–8854.
Abeijon, C., Robbins, P.W., and Hirschberg, C.B. (1996b) Molecular cloning
of the Golgi apparatus uridine diphosphate-N-acetylglucosamine transporter
from Kluyveromyces lactis. Proc. Natl Acad. Sci. USA, 93, 5963–5968.
Abeijon, C., Yanagisawa, K., Mandon, E., Hausler, A., Moremen, K.,
Hirschberg, C.B., and Robbins, P.W. (1993) Guanosine diphosphatase is
required for protein and sphingolipid glycosylation in the Golgi lumen of
Saccharomyces cerevisiae. J. Cell Biol., 122, 307–323.
Alani, E., Cao, L., and Kleckner, N. (1987) A method for gene disruption that
allows repeated use of URA3 selection in the construction of multiply
disrupted yeast strains. Genetics, 116, 541–545.
Berninsone, P., and Hirschberg, C.B. (2000) Nucleotide sugar transporters of the
golgi apparatus. Curr. Opin. Struct. Biol., 10, 542–547.
Berninsone, P., Miret, J.J., and Hirschberg, C.B. (1994) The Golgi guanosine
diphosphatase is required for transport of GDP-mannose into the lumen of
Saccharomyces cerevisiae Golgi vesicles. J. Biol. Chem., 269, 207–211.
Bianchi, M.M., Falcone, C., Xin Jie, C., Weslowski-Louvel, M., Frontali, L.,
and Fukuhara, H.(1987) Transformation of the yeast Kluyveromyces lactis
by new vectors derived from 1.6ÿm circular plasmid pKD1. Curr. Genet.,
12, 185–192.
Boone, C., Sommer, S.S., Hensel, A., and Bussey, H. (1990) Yeast KRE genes
provide evidence for a pathway of cell wall beta-glucan assembly. J. Cell
Biol., 110, 1833–1843.
Brandan, E., and Fleischer, B. (1982) Orientation and role of nucleo-
sidediphosphatase and 5'-nucleotidase in Golgi vesicles from rat liver.
Biochemistry, 21, 4640–4645.
Bulawa, C.E., Slater, M., Cabib, E., Au-Young, J., Sburlati, A., Adair, W.L.
Jr., and Robbins, P.W. (1986) The S. cerevisiae structural gene for chitin
synthase is not required for chitin synthesis in vivo. Cell, 46, 213–225.
Castro, O., Chen, L.Y., Parodi, A.J., and Abeijon, C. (1999) Uridine diphos-
phate-glucose transport into the endoplasmic reticulum of Saccharomyces
cerevisiae: in vivo and in vitro evidence. Mol. Biol. Cell, 10, 1019–1030.
Gao, X.-D., Kaigorodov, V., and Jigami, Y. (1999) YND1, a homologue of
GDA1, encodes membrane-bound apyrase required for Golgi N- and O-glyco-
sylation in Saccharomyces cerevisiae. J. Biol. Chem., 274, 21450–21456.
Guillen, E., Abeijon, C., and Hirschberg, C.B. (1999) Mammalian Golgi
apparatus UDP-N-acetylglucosamine transporter: molecular cloning by
phenotypic correction of a yeast mutant. J. Biol. Chem., 274, 6641–6646.
422