(Received for publication, February 28, 1997, and in revised form, April 14, 1997)
From the Department of Biology, McGill University,
Montréal, Québec, Canada, H3A 1B1 and the
§ Institut de Génétique et Microbiologie,
Université Paris-Sud, Orsay, France
We have determined a role for Ktr1p and Ktr3p as
mannosyltransferases in the synthesis of the carbohydrate chains
attached to Saccharomyces cerevisiae O- and
N-modified proteins. KTR1 and KTR3
encode related proteins that are highly similar to the Kre2p/Mnt1p Golgi 1,2-mannosyltransferase (Lussier, M., Camirand, A., Sdicu, A.-M., and Bussey, H. (1993) Yeast 9, 1057-1063; Mallet,
L., Bussereau, F., and Jacquet, M. (1994) Yeast 10, 819-831). Examination of the electrophoretic mobility of a
specifically O-linked protein from mutants and an analysis
of their total O-linked mannosyl chains demonstrates that
Ktr1p, Ktr3p, and Kre2p/Mnt1p have overlapping roles and collectively
add most of the second and the third
1,2-linked mannose residues on
O-linked oligosaccharides. Determination of the mobility of
the specifically N-linked glycoprotein invertase in
different null strains indicates that Ktr1p, Ktr3p, and Kre2p are also
jointly involved in N-linked glycosylation, possibly in
establishing some of the outer chain
1,2-linkages.
Nascent proteins synthesized by membrane-bound ribosomes are
translocated across the ER1 membrane and
acquire carbohydrate chains on specific serine, threonine, and
asparagine residues. These glycoproteins then proceed through the Golgi
complex where their oligosaccharides are further modified. The
O-modified proteins of Saccharomyces cerevisiae possess a linear carbohydrate chain of up to 5 mannose residues (1, 2).
O-Glycosylation is initiated in the ER with dolichyl phosphate-D-mannose as the immediate sugar donor for the
mannosyl residue transferred to the hydroxy-amino acids serine and
threonine (2, 3). GDP-Man is utilized as the sugar donor in the
subsequent elongation of O-linked carbohydrate chains
resulting in a linear glycan in which the second and third mannoses
possess 1,2-linkages, whereas the terminal fourth and fifth mannosyl
residues have
1,3-linkages (1, 2).
Yeast N-linked modified proteins can acquire two different
types of glycan chains after the addition of a
Man8GlcNAc2 core in the endoplasmic reticulum,
a process common to the majority of eukaryotes (1, 2). This primary
oligosaccharide can undergo maturation in the Golgi resulting in a
Man8-13GlcNAc2 carbohydrate structure, or it
may be extended by an outer chain of variable size (up to 200 mannose
residues) that is made up of a backbone of 1,6-mannosyl residues to
which are attached branched
1,2- and
1,3-mannosyl side
chains.
Some of the enzymes involved in the elaboration of O-linked
oligosaccharides and in the synthesis of N-linked outer
chains have been identified, and their structural genes have been
isolated. At least four different genes of the seven-membered
PMT1-7 gene family encoding dolichyl
phosphate-D-mannose:protein
O-D-mannosyltransferases are responsible for
initiating the O-linked glycans (4). Protein O-glycosylation is essential for cell function because
mutants of S. cerevisiae lacking different combinations of
three of the PMT genes are not viable (4). The
OCH1 gene is responsible for adding the first 1,6-mannose
residue involved in initiating N-linked outer chain
elaboration (5, 6). Two enzymes in particular have been shown to
participate in both O- and N-linked glycosylation. The KRE2/MNT1 gene encodes a
medial Golgi
1,2-mannosyltransferase required for the addition of
the third mannose residue on O-linked chains (7, 8) and is
also implicated in N-linked outer chain oligosaccharide
synthesis (9, 10). The O-linked trisaccharide can be further
extended by one or two
1,3-linked mannoses. The fourth mannose
residue is added by the Golgi localized
1,3-mannosyltransferase Mnn1p, which also terminally
mannosylates core and outer chain modified N-linked glycans
(8, 11-13).
To further characterize protein glycosylation in yeast and identify
some of the responsible glycosyltransferases, we have studied a gene
family encoding proteins that are highly homologous to Kre2p/Mnt1p.
This nine-membered KTR mannosyltransferase gene family
contains the KRE2/MNT1, YUR1, KTR1,
KTR2, KTR3, KTR4, KTR5, KTR6, and KTR7 genes (14-16). As with other
known glycosyltransferases (17), all genes in this family are
predicted, and some have been shown to encode type-II membrane proteins
with a short cytoplasmic N terminus, a membrane-spanning region, and a
highly conserved catalytic lumenal domain (8, 10, 14-16). Similarity
between different family members is variable (14-16). For example,
Yur1p and Ktr2p are most similar sharing 62% identity, followed by
Ktr1p and Ktr3p with 54% identity. Ktr6p and Ktr5p with 24% identity constitute the two most divergent enzymes in the family. Initial functional characterization of Yur1p, Ktr1p, and Ktr2p has revealed that they are Golgi mannosyltransferases involved in
N-linked glycosylation, possibly as redundant enzymes, but
when inactivated they show no defects in O-linked
glycosylation (9, 10, 14). Here, we report that KTR1,
KTR3, and KRE2, which form a related subfamily
encode 1,2-mannosyltransferases that together add the second and the
third
1,2-linked mannose residues on O-linked carbohydrate chains and that also participate in N-linked
outer chain elaboration.
All yeast
strains used were based on SEY6210 (MATa, leu2-3, ura3-52,
his3-200, lys2-801, trp1-
901,
suc2-
9) and were grown under standard conditions
(Yeast extract, Peptone, Dextrose, Yeast Nitrogen Base) as described
previously (18).
Deletional disruptions of KRE2
and KTR1 are described elsewhere (8, 10). Disruption of the
KTR3 locus was made by a single-step gene replacement
procedure (19). First, the KTR3 gene was synthesized in vitro by the polymerase chain reaction (20) using the
pfu DNA polymerase (Stratagene, La Jolla, CA) and
CCACCACTTTCAGCATGG and GAACCCAAGAAGGCACTAG as 5 and 3
primers, with
yeast genomic DNA as a template. The 2-kilobase fragment obtained was
then subcloned directly in the pTZ/PC vector as described previously
(14). A 294-base pair EcoRV coding sequence fragment was
removed and replaced by a 1.8-kilobase EcoRI fragment
containing the complete HIS3 gene. A linear 3.5-kilobase
NcoI fragment containing the complete HIS3 gene
as well as the flanking regions plus the coding sequences of
KTR3 gene was gel purified and integrated into the isogenic
diploid SEY6210 strain. Integrants were verified by Southern analysis
(data not shown).
The extracellular proteins Kre9p and invertase were expressed at high levels from a 2µ-based multicopy vector in yeast strains bearing different disruptions and concentrated from cultures growing exponentially in YNB selective medium containing 5% (v/v) glycerol and 2% glucose using Amicon centriprep concentrators (W. R. Grace & Co. Danvers, MA). Immunoblots were carried out as described (10, 21) using anti-Kre9p (22) and anti-invertase antisera (10), and detection was performed by an enhanced chemiluminescence procedure (Amersham, Oakville, ON, Canada).
Mannose Labeling andYeast cells were
labeled using [3H]mannose, and -eliminations were
performed as described previously (10, 14, 21).
The possible roles of members of the KTR family in O- and N-linked glycosylation remain to be determined. A testable assumption is that those proteins in the family that are most closely related are most likely to possess functional similarities. A relational homology tree constructed using the catalytic domains of each protein clarifies such relationships and has permitted an attempt to functionally group the different enzymes (16). One conclusion drawn from this analysis was that Kre2p is most related to Ktr1p and Ktr3p, indicating that these three proteins form a subfamily with possible related functions (16).
Recent in vitro studies using O-linked type
substrates demonstrated that Ktr1p is a 1,2-mannosyltransferase with
enzymatic properties highly similar to those of Kre2p, suggesting that
these two enzymes act in similar ways (23). To explore these
relationships and examine the role of Ktr3p and reassess the role of
Ktr1p and Kre2p in the synthesis of O-linked chains, the
extent of O-glycosylation in yeast strains bearing mutations
in these genes was investigated. Deletional disruptions of
KRE2 and KTR1 were previously obtained (8, 10),
and a disruption of the KTR3 locus was made (see "Experimental Procedures"). Double and triple disruptions were subsequently constructed using standard genetic techniques. The single
and all double null mutants showed no growth defects when compared with
the wild type strain. However, the haploid kre2 ktr1 ktr3
triple null mutant had a slow growth phenotype, indicating a genetic
interaction between the three genes.2
O-glycosylation was examined in yeast
strains bearing different disruptions by following the mobility of
Kre9p (Fig. 1), a protein involved in cell wall assembly
that is highly O-mannosylated but receives no
N-linked modifications (22). Kre9p produced in a wild type
strain migrates at an apparent mass of 55 kDa (10, 21, 22). As
previously observed, Kre9p isolated from a kre2 null strain
migrated more quickly than did the wild type Kre9p, with an apparent
molecular mass of approximately 47 kDa (10, 21), whereas Kre9p produced
by ktr1 (10) or ktr32 single null
disruptants was indistinguishable from that produced by a wild type
strain. Kre9p produced in a ktr1 ktr3 double null mutant
showed a small increase in mobility, whereas Kre9p produced in a
kre2 ktr3 mutant was similar to Kre9p produced in a
kre2 single mutant. However, a ktr1 mutant
exacerbated the O-mannosylation defects of a kre2
mutant as the apparent molecular mass of Kre9p was approximately 5 kDa
smaller (42 kDa) than when produced in a kre2 mutant alone.
Kre9p was smallest (~38 kDa) when produced in a triple ktr1
ktr3 kre2 mutant, indicating a cumulative involvement of all three
proteins in O-linked modifications.
To determine the extent of this involvement, an analysis was made of
total O-linked carbohydrate chains present on glycoproteins of mutants carrying different combinations of the disrupted genes (Fig.
2). O-linked carbohydrate chains were
specifically released from the total glycoprotein fraction of in
vivo [3H]mannose-labeled yeast cells by
-elimination and resolved by chromatography (10, 14, 21; see
"Experimental Procedures"). The wild type strain (Fig.
2A) and the ktr1 and ktr3 single null
mutants2 showed the normal profile of five oligosaccharide
peaks, as was the case with the ktr1 ktr3 double null mutant
(Fig. 2B). As expected, the pattern obtained from the
kre2 null strain gave two major peaks (Man1-Man2; Fig.
2C), consistent with failure to add the third
1,2-linked
mannose residue and a minor peak indicating that a small proportion of
total glycoproteins received a third mannose, an effect previously seen
(7). Other yeast enzymes are, therefore, able to carry out this
particular reaction to a limited extent.
A significant reduction in mannosylation is evident in both the
ktr1 kre2 and ktr3 kre2 double null mutants, each
of which gave two peaks (Man1-Man2) but with a decreased proportion of glycoproteins receiving 2 mannose residues when compared with a
kre2 disruptant alone (Fig. 2, D and
E), suggesting an involvement of both Ktr1p and Ktr3p in
adding this mannose to O-modified chains. In those two
double mutants, the proportion of O-glycoproteins receiving
a third mannose is less than in a kre2 single null, suggesting that both Ktr1p and Ktr3p have a limited capacity to add a
third mannose on O-linked chains. In a ktr1 ktr3
kre2 triple mutant, the predominant O-linked
oligosaccharides assembled were composed of a single mannose (Fig.
2F), indicating that collectively Ktr1p, Ktr3p, and Kre2p
are responsible for adding the third mannose and most of the second on
O-linked glycans. At least one other enzyme, still
unidentified but possibly encoded by a member of the KTR
gene family (16), is responsible for the residual level of attachment
of a second mannose (Fig. 2F). All the enzymes now known to
be responsible for the assembly of yeast O-linked sugars are
outlined in Scheme 1.
Ktr1p, Ktr3p, and Kre2p Jointly Participate in N-Glycosylation
The Kre2p mannosyltransferase is presumably
involved in the elaboration of N-linked outer chain glycans
(9, 10), but the precise mannoses added by this enzyme remain to be
determined. In view of the cumulative role of KTR1,
KTR3, and KRE2 in O-glycosylation, the
effect on N- glycosylation of inactivating different
combination of these genes was assessed. The carbohydrate chains of
invertase, a specifically N-modified protein, were analyzed
by measuring the mobility of the secreted form of this protein (Fig.
3), which is extensively modified with core
oligosaccharides extended with outer chain glycans (24, 25).
The extracellular wild type protein, as is the case when it is synthesized in ktr1 and ktr3 single null mutants,2 has a molecular mass of around 150 kDa (Fig. 3). As seen before (10), invertase produced in a kre2 null mutant receives less N-modification (~143 kDa). The oligosaccharides attached to invertase synthesized in a ktr1 ktr3 double null mutant appear unaffected because the approximate molecular mass of the polypeptide chains made in this particular strain equals that of wild type proteins. The oligosaccharidic defects of invertase produced in a ktr1 kre2 double null mutant were similar to those observed in a kre2 single null mutant, whereas a disruption in KTR3 slightly exacerbates the N-mannosylation defects seen in a kre2 mutant because invertase synthesized in a ktr3 kre2 mutant has a molecular mass of ~137 kDa. However, invertase secreted from a ktr1 ktr3 kre2 triple null mutant was found to have a molecular mass of ~129 kDa demonstrating that, as was the case with O-glycosylation, all three proteins are collectively involved in N-linked modifications.
Ktr1p and Ktr3p are involved in the elaboration of
O-linked glycans by adding the second mannose in the linear
five mannose carbohydrate chain. This was not initially seen because
yeast strains bearing ktr1 and ktr3 single or
double disruptions possess a normal O-glycosylation pattern.
Only in a ktr1 ktr3 kre2 triple mutant is the full effect of
the absence of these enzymes apparent, where the third mannose is
absent and a severely reduced level of mannose 2 is observed. In the
absence of Ktr1p and Ktr3p, Kre2p is able to add both the second and
third 1,2-mannose residue in the linear five mannose carbohydrate
chains. Ktr1p and Ktr3p are to a limited extent also capable of
attaching the third
1,2-mannose onto O-linked chains. All
3 enzymes, therefore, appear to have overlapping roles in the addition
of both the second and third
1,2-linked mannose residues of
O-glycosyl chains in yeast (see Scheme 1). The exact
in vivo contribution of each enzyme to the synthesis of the
second and third mannose linkages in wild type cells remains to be
determined.
The results presented here demonstrate that Ktr1p, Ktr3p, and Kre2p are
also implicated in N-linked outer chain elaboration. These
transferases do not participate in N-linked core glycan synthesis because the molecular mass of the core modified
oligosaccharide received by the late Golgi protein Kex1p (26) is the
same in the ktr1 ktr3 kre2 triple null mutant and in a wild
type strain.2 Therefore, the marked reduction in the sugar
chains received by invertase in the ktr1 ktr3 kre2 triple
mutant indicates that Ktr1p, Ktr3p, and Kre2p act in the Golgi
apparatus to elaborate outer chain glycans of N-linked
oligosaccharides by making some of the branched mannosyl side chains
that are attached to the 1,6-mannosyl residues backbone (see Fig.
3). Of the five mannoses constituting this backbone sugar chain, all
are substituted by at least one
1,2-linked mannose residue, and
three are modified by at least two
1,2-linked mannoses. Because the
structure of these N-linked branched mannosyl side chains is
reminiscent of O-linked oligosaccharides, it is a reasonable
speculation that Ktr1p, Ktr3p, and Kre2p collectively participate in
establishing some of these outer chain
1,2-linkages. Recent in
vitro enzymatic studies are consistent with this (23). Both Kre2p
and Ktr1p utilize the N-glycan type oligosaccharidic
substrate, Man15-30GlcNAc, which contains the
1,6-mannose outer chain backbone attached to the core sugar but
lacks the
1,2-mannose containing branches (see Fig. 3), indicating
that both enzymes have the ability to add a first
1,2-linked mannose
residues to the outer chain backbone. Moreover, the mannosyltransferase
reaction involving Kre2p exhibited biphasic kinetics, with an increase
in mannose incorporation into product during the second part of the
reaction. Such an activity is in agreement with this enzyme
sequentially adding two mannose residues and parallels its role in
O-mannosylation.
The basis for the multiplicity of mannosyltransferases and the complexity of their involvement in the synthesis of O- and N-mannosyl chains can be addressed in the context of redundancy in gene families. The members of the mannosyltransferase family discussed here have related and overlapping functions that can take at least two forms. Firstly, multiple enzymes act at single biosynthetic steps. For example, Kre2p, Ktr1p, and Ktr3p add the second O-linked mannose residue. Individual enzymes can also participate in more than one biosynthetic step. Kre2p, Ktr1p, and Ktr3p, which can add the second mannose residue to O-linked chains, are also involved together in adding a third mannose residue, though to differing extents (see Scheme 1.).
The level of apparent redundancy observed between the enzymes can be
explained in several ways. It is possible that each enzyme possesses a
high affinity for a limited set of amino acid sequence contexts around
the Ser/Thr and Asn residues at which mannosylation occurs. By having a
number of enzymes of varying specificity, the cell is able to extend
the range of mannosylated residues on glycoproteins. There is strong
evidence for this notion with the first enzyme required in the
O-mannosylation pathway, the protein mannosyltransferase,
where some of the seven PMT encoded enzymes have been shown
to have differing specificities for the mannosylation of peptide
substrates of different sequence (4, 21, 27). Although there is no
direct evidence for this conjecture in the KTR family,
variable sequence specificities can be hinted at when comparing Kre9p
O-mannosylation patterns with those seen through
-elimination in the total O-mannoprotein fractions of the
kre2 ktr1 and kre2 ktr3 double null mutants. The
mannosylation deficiencies of Kre9p are smallest in a kre2
ktr3 mutant, whereas the defects of the bulk of
O-modified proteins are far more pronounced in this strain
(see Figs. 1 and 2, D and E). This is consistent with these enzymes having different specificities with Kre9p sites being atypical.
The redundancy observed here can also be explained by a sequential action through intracellular compartmentalization. If related enzymes have broadly similar or overlapping substrate affinities but are in a distinct Golgi compartments, then those mannosylation sites on proteins that fail to be mannosylated in one compartment have subsequent glycosylation opportunities as they move through the later mannosyltransferase-containing cisternae. If this were the case, Ktr1p and Ktr3p that have less ability to add the third mannose in O-glycosylation might be found in cisternae before Kre2p, the enzyme most responsible for Man3 addition. This would build a level of redundant function into the processive extension of mannose chains in successive Golgi compartments and increase O-mannosylation efficiency. The Golgi targeting regions of Kre2p, Ktr1p, and Ktr3p are unrelated, consistent with them having different intracellular locations (8, 10, 16).
The addition of O- and N-mannose oligosaccharides in S. cerevisiae requires the action of many related mannosyltransferases. Studying the roles of gene families such as KTR will be informative both for the analysis of protein glycosylation and to offer insights into the biological reasons that allow such diversity of related gene products to occur.
We thank the members of the Bussey laboratory and Dr. Annette Herscovics for helpful comments and suggestions and Carole Smith and Guy l'Heureux for photographic work.