Báo cáo khoa học: Saccharomyces cerevisiae Ybr004c and its human homologue are required for addition of the second mannose during glycosylphosphatidylinositol precursor assembly ppt

Saccharomyces cerevisiae Ybr004c and its human
homologue are required for addition of the second
mannose during glycosylphosphatidylinositol precursor
assembly
Anne-Lise Fabre
1
, Peter Orlean
2
and Christopher H. Taron
1
1 New England Biolabs, Beverly, MA, USA
2 Department of Microbiology, University of Illinois, Urbana, IL, USA
Glycosylphosphatidylinositols (GPIs) are key glyco-
lipids produced by all eukaryotes. GPIs become cova-
lently attached to the C-termini of certain secretory
proteins and act as anchors to attach such proteins to
the outer face of the plasma membrane [1,2]. Synthesis
of GPIs is essential for cell wall formation and viabil-
ity of yeast cells [3–5], for embryonic development in
mammalian cells [6], and for viability of the parasites
Leishmania mexicana [7] and the bloodstream form of
Trypanosoma brucei [8].
GPIs are assembled in the membranes of the endo-
plasmic reticulum (ER) by sequential addition of
components to phosphatidylinositol. GPIs from all
organisms have a conserved core structure of NH
2
-
CH
2
-CH

2
-PO
4
-6Mana1,2Mana1,6Mana1,4-GlcNa1,6-
myo-inositol-PO
4
-lipid. The three core mannoses may
be further modified with side-branching groups that
vary between species. For example, a fourth mannose
(Man4) is side-branched to the third core mannose
(Man3) of all yeast GPIs [9] and of certain human
GPIs [10–12], and additional side-branching phospho-
ethanolamines (EthN-Ps) may be added to the first
and second mannoses of yeast [13–15] and mammalian
GPIs [16,17].
S. cerevisiae genes implicated in addition of man-
noses and EthN-P residues during GPI precursor
assembly have been identified following characteriza-
tion of the glycolipids that accumulate in conditional
mutant strains. The three a-linked mannoses compri-
sing the GPI core are individually transferred from
Keywords
cell wall; glycosylphosphatidylinositol;
mannosyltransferase; Saccharomyces
cerevisiae
Correspondence
Christopher H. Taron, New England Biolabs,
32 Tozer Road, Beverly, MA 01915, USA
Fax: +978 9211350
Tel: +978 9275054

E-mail:
(Received 28 October 2004, revised 21
December 2004, accepted 4 January 2005)
doi:10.1111/j.1742-4658.2005.04551.x
Addition of the second mannose is the only obvious step in glycosylphos-
phatidylinositol (GPI) precursor assembly for which a responsible gene has
not been discovered. A bioinformatics-based strategy identified the essential
Saccharomyces cerevisiae Ybr004c protein as a candidate for the second
GPI a-mannosyltransferase (GPI-MT-II). S. cerevisiae cells depleted of
Ybr004cp have weakened cell walls and abnormal morphology, are unable
to incorporate [
3
H]inositol into proteins, and accumulate a GPI intermedi-
ate having a single mannose that is likely modified with ethanolamine
phosphate. These data indicate that Ybr004cp-depleted yeast cells are
defective in second mannose addition to GPIs, and suggest that Ybr004cp
is GPI-MT-II or an essential subunit of that enzyme. Ybr004cp homo-
logues are encoded in all sequenced eukaryotic genomes, and are predicted
to have 8 transmembrane domains, but show no obvious resemblance to
members of established glycosyltransferase families. The human Ybr004cp
homologue can substitute for its S. cerevisiae counterpart in vivo.
Abbreviations
CFW, Calcofluor white; Dol-P-Man, dolichol monophosphate mannose; ER, endoplasmic reticulum; EthN, ethanolamine; EthN-P,
ethanolamine phosphate; 5-FOA, 5-fluoro-orotic acid; GPI, glycosylphosphatidylinositol; GPI-MT, GPI a-mannosyltransferase; JbaM, jack bean
a-mannosidase; Man
1
-GPI, mannosyl GPI; Man
2
-GPI, dimannosyl GPI; Man
3

-GPI, trimannosyl GPI; Man
4
-GPI, tetramannosyl GPI; PI-PLC,
phospholipase C.
1160 FEBS Journal 272 (2005) 1160–1168 ª 2005 FEBS
Dol-P-Man [18] to GPI biosynthetic intermediates by
separate candidate GPI mannosyltransferases (GPI-
MT). The mammalian Pig-M protein and its yeast
ortholog Yjr013wp are required for addition of Man1
to GPI precursors [19], and the PIG-B ⁄ Gpi10 proteins
for addition of Man3 [20–22]. However, a candidate
GPI-MT-II has not yet been identified from any
organism and the transfer of Man2 to GPI glycans
remains the only obvious step of GPI precursor syn-
thesis or side-chain decoration for which a candidate
gene has not yet been discovered.
We report here the identification of the yeast gene
encoding a novel 433 amino acid membrane protein
(Ybr004cp) required for addition of the second man-
nose to GPI precursors. Yeast cells depleted of
Ybr004cp exhibit cell wall and morphological abnor-
malities, are defective in the incorporation of [
3
H]ino-
sitol into protein, and accumulate a GPI precursor
whose glycan contains a single mannose modified with
a substituent, probably EthN-P, that makes it a-man-
nosidase resistant. Additionally, the human homologue
of Ybr004cp is able to substitute for its S. cerevisiae
counterpart in vivo.

Results
Identification of a candidate yeast GPI-MT-II
sequence
Although GPI MT-II might be expected to show
amino acid sequence similarity to known Dol-P-Man-
utilizing transferases such as GPI-MT-I, III, or IV, or
to protein: O-mannosyltransferases [23], searches of
protein sequence databases failed to identify any
sequences with statistically significant homology to the
above query sequences, suggesting that the yeast GPI-
MT-II has little resemblance to known mannosyl-
transferases at the primary sequence level.
We therefore pursued an alternative, bioinformatics-
based strategy to identify candidate GPI MT-II
sequences. We relied on a recent analysis of the prote-
ome of the pathogenic yeast Candida albicans in which
495 proteins with N-terminal signal sequences and that
likely localize to various compartments of the secretory
pathway were identified [24]. We reasoned that this
subset of C. albicans sequences likely included GPI-
MT-II. We next eliminated sequences that failed to
meet the following criteria. First, because GPI-MT I,
III, and IV are integral membrane proteins having at
least eight transmembrane domains and overall lengths
between 403 and 678 amino acids [25], we eliminated
sequences that had less than two predicted transmem-
brane domains or that had lengths greater than 1000
amino acids. Second, we expected that the gene enco-
ding GPI-MT-II would be essential, and we there-
fore cross-referenced the remaining sequences to the

S. cerevisiae Genome Database (yeastgenome.org) keep-
ing only sequences with obvious S. cerevisiae homo-
logues whose systematic gene deletions were lethal.
Third, we eliminated proteins with well-characterized
functions, leaving only three sequences. Finally, we
expected GPI-MT-II to be encoded in every eukaryotic
genome. BLAST searches [26] against the GenBank
database demonstrated that two of the three candidate
proteins have homologues only in fungi, whereas the
third, Ybr004cp, has homologues in fungi, mammals,
plants, insects, nematodes and protozoa (Table 1).
Thus, we considered Ybr004cp to be the most plaus-
ible candidate S. cerevisiae GPI-MT-II.
Table 1. The Ybr004c protein sequence family. % Identity ⁄ similar-
ity calculated relative to S. cerevisiae sequence.
Organism
Length
(amino
acids)
% Identity ⁄
similarity
GenBank
accession
number
Fungi
Candida albicans 394 37 ⁄ 70 EAK94563
Candida glabrata 433 48 ⁄ 82 CAG58266
Cryptosporidium parvum 436 23 ⁄ 71 CAD98327
Debaryomyces hansenii 428 33 ⁄ 65 CAG87189
Encephalitozoon cuniculi 393 20 ⁄ 60 NP_596980

Eremothecium gossypii 427 47 ⁄ 77 NP_984865
Gibberella zeae 423 27 ⁄ 63 EAA76896
Kluyveromyces lactis 417 47 ⁄ 81 CAG99810
Saccharomyces cerevisiae 433 – NP_009558
Schizosaccharomyces pombe 426 24 ⁄ 60 NP_592878
Magnaporthe grisea 441 23 ⁄ 59 XP_368608
Neurospora crassa 593 28 ⁄ 67 XP_325815
Ustilago maydis 485 16 ⁄ 55 EAK86320
Yarrowia lipolytica 357 31 ⁄ 70 CAG78103
Mammals
Homo sapiens 493 18 ⁄ 53 NP_060307
Mus musculus 493 22 ⁄ 55 NP_848813
Rattus norvegicus 491 23 ⁄ 56 XP_345587
Plants
Arabidopsis thaliana 492 22 ⁄ 52 NP_172652
Oryza sativa 486 20 ⁄ 58 AK073582
Insects
Drosophila melanogaster 449 17 ⁄ 51 AAF23239
Fish
Tetraodon nigroviridis 494 22 ⁄ 56 CAG00037
Nematodes
Caenorhabditis briggsae 673 20 ⁄ 59 CAE67131
Caenorhabditis elegans 672 19 ⁄ 61 NP_491783
Protozoa
Giardia lamblia 381 24 ⁄ 57 EAA41032
Plasmodium falciparum 503 16 ⁄ 53 NP_701814
A L. Fabre et al. Second mannose addition to GPI precursors
FEBS Journal 272 (2005) 1160–1168 ª 2005 FEBS 1161
Growth and GPI anchoring defects of Ybr004cp-
depleted strains

To establish whether Ybr004cp is involved in GPI
assembly, we tested whether depletion of this protein in
YBR004c-disrupted haploid cells leads to a GPI assembly
defect. We constructed a YBR004c-disrupted haploid
strain in which expression of a plasmid-borne wild-type
allele of YBR004c is regulated by the glucose-repressible
GAL10 promoter (ybr004cD-pGAL-YBR004c). When
grown in medium containing glucose, expression of
YBR004c is repressed, uncovering recessive phenotypes
associated with depleting cells of Ybr004cp. We tested
this strain for growth and biochemical defects charac-
teristic of a GPI anchoring deficiency.
Strains defective in GPI anchoring are typically
hypersensitive to the fluorescent dye Calcofluor white
(CFW) and have weakened cell walls [27]. This was
the case for ybr004cD-pGAL-YBR004c cells which
showed impaired growth compared to a wild-type
strain on medium containing glucose and 16 lg CFW
per mL (Fig. 1A). Furthermore, glucose-grown
ybr004cD-pGAL-YBR004c cells examined by phase-
contrast microscopy were generally large, misshapen,
and clumpy (Fig. 1B), phenotypes indicating a loss of
cell wall integrity and seen with other gpi mutants [28].
Because GPI-anchored proteins are the only known
proteins covalently linked to inositol in yeast [29,30], we
examined the ability of Ybr004cp-depleted cells to incor-
porate [
3
H]inositol into proteins. The ybr004cD-pGAL-
YBR004c strain was grown and labeled with [

3
H]inositol
in medium containing galactose or glucose to promote
or repress YBR004c expression, respectively. Radiolabe-
led cells were lysed in detergent and extracted proteins
were separated by SDS ⁄ PAGE, after which [
3
H]inositol-
labeled proteins were detected by fluorography. Wild-
type cells were capable of forming [
3
H]inositol-labeled
GPI anchored pr oteins in medium containing either galac-
tose or glucose (Fig. 1C, lanes 1, 2), whereas ybr004cD-
pGAL-YBR004c cells incorporated significantly less
[
3
H]inositol into proteins in glucose-containing medium
(Fig. 1C, lane 4), where YBR004c expression is
repressed. Thus, Ybr004cp-depleted cells exhibit a
global defect in formation of GPI-anchored proteins.
Ybr004cp-depleted cells accumulate a novel
GPI precursor
Yeast strains with conditional defects in mannosylation
and EthN-P addition to the GPI precursor or in GPI
transfer to protein accumulate GPI assembly inter-
mediates that can be detected by pulse-radiolabeling
such strains under nonpermissive conditions [14,15,
31,32]. The step in GPI assembly affected in such
mutants can be inferred from the structure of the accu-

mulating GPI. Therefore, we looked for evidence
of lipid accumulation in glucose-repressed ybr004cD-
pGAL-YBR004c cells. The strain was metabolically
labeled with [
3
H]inositol, after which lipids were
extracted from cells, separated by TLC, and [
3
H]inosi-
tol-labeled lipids were detected by fluorography. Cells
radiolabeled under repressing conditions accumulated
an aberrant [
3
H]inositol-containing lipid (lipid 004–1;
Fig. 2A, lane 4) that was nearly absent from lipids iso-
lated from cells grown in medium containing galactose
(Fig. 2A, lane 3). Lipid 004–1 was susceptible to treat-
ment with mild-base (Fig. 2B, lane 2) and resistant to
cleavage by PI-PLC (Fig. 2B, lane 4), indicating that it
contained ester-linked fatty acids and an inositol acyl
chain, respectively. This combination of traits is a
characteristic of lipid intermediates in GPI precursor
synthesis. Finally, lipid 004–1 migrated as a less polar
species than the previously characterized Man
2
- and
Man
3
-GPIs that accumulate in cells defective in
A

BC
Fig. 1. ybr004c mutants have defects in cell wall synthesis, mor-
phogenesis, and GPI anchoring. (A) Ten-fold serial dilutions of wild-
type (wt) or ybr004cD-pGAL-YBR004c cells were spotted onto YPD
agar-containing medium with or without 16 lg CFW per mL and
grown 3 days at 30 °C. (B) ybr004cD-pGAL-YBR004c cells were
grown either in galactose- (Gal) or glucose-containing (Glc) medium.
Cellular phenotypes were observed by phase contrast microscopy.
(C) Proteins from wt and ybr004cD -pGAL-YBR004c strains were
metabolically labeled with [
3
H]inositol in medium containing either
galactose or glucose for 60 min at 30 °C. Proteins were extracted
from cells, separated by SDS ⁄ PAGE and radiolabeled GPI anchored
proteins were visualized by fluorography.
Second mannose addition to GPI precursors A L. Fabre et al.
1162 FEBS Journal 272 (2005) 1160–1168 ª 2005 FEBS
addition of the third [21,22] and fourth [31] mannoses
to GPI precursors (Fig. 3B, and data not shown), sug-
gesting that it is a GPI intermediate that forms prior
to addition of Man3 and -4 to yeast GPI precursors.
A yeast strain defective in GPI-MT-II would be pre-
dicted to accumulate a GPI intermediate bearing a sin-
gle mannose that may or may not be substituted with
a side-branching EthN-P residue. Phosphatidylethanol-
amine, the donor of EthN-P residues to Man1 and -3
of GPIs [33,34], can be synthesized either de novo from
exogenous ethanolamine (EthN), or by decarboxylation
of phosphatidylserine. Metabolic labeling experiments
using [

14
C]EthN or [
3
H]serine were therefore carried
out to determine if lipid 004–1 contains an EthN-P moi-
ety. To enhance [
14
C]EthN incorporation into lipids,
radiolabeling was carried out in a ybr004cD-pGAL-
YBR004c ⁄ psd1D⁄psd2D strain, which lacks phosphati-
dylserine decarboxylase activity (see Experimental
procedures). This strain accumulated lipid 004–1 upon
labeling with [
14
C]EthN in medium containing glucose
(Fig. 2C, lane 3), but not in galactose-containing
medium (Fig. 2C, lane 2). Similarly, ybr004cD-pGAL-
YBR004c cells accumulated lipid 004–1 upon labeling
with [
3
H]serine in the presence of glucose (Fig. 2C, lane
5). Taken together, these results are strong evidence
that lipid 004–1 contains EthN-P, and therefore that
004–1 contains at least one mannose residue.
We next compared the TLC mobility of lipid 004–1
to that of a Man
1
(EthN-P)-GPI mobility standard
derived from the previously characterized GPI interme-
diate that accumulates upon depletion of Gpi13p, the

GPI EthN-P transferase that adds EthN-P to Man3
[14,15]. The GPI that accumulates in gpi13D-pGAL-
GPI13 cells is a Man
4
-GPI, much of which is modified
by a single EthN-P on Man1, but lesser amounts of
which bear their EthN-P on Man2 [15]. Treatment of
the major Man
4
-GPI isoform with JbaM would there-
fore yield a GPI with a single mannose bearing EthN-
P [a Man
1
(EthN-P)-GPI], whereas the minor isoform
would be converted to a Man
2
(EthN-P)Man
1
-GPI.
The Man
1
(EthN-P)-GPI comigrated with lipid 004–1
on TLC (Fig. 2D, lanes 1 and 4) suggesting the two
share the same structure. A GPI precursor with the
thin layer chromatographic mobility of Man
1
(EthN-P)-
GPI has not previously been reported to accumulate in
any yeast GPI assembly mutant. In addition, lipid
004–1 was resistant to treatment with JbaM, indicating

that it lacks an unsubstituted terminal mannose
(Fig. 2D, lane 2). JbaM treatment of lipids from
Ybr004cp-depleted cells also generated some very non-
polar material whose mobility is consistent with that
of GlcN [acyl-Ins]PI, which may have originated from
an unsubstituted Man
1
-GPI that may comigrate with
AB
CD
Fig. 2. ybr004cD-pGAL-YBR004c cells accumulate a putative Man
1
-
(EthN-P)-GPI. (A) Wild-type and ybr004cD-pGAL-YBR004c cells
were grown and [
3
H]inositol-labeled in galactose- (lanes 1 and 3) or
glucose-containing medium (lanes 2 and 4) to induce or repress
YBR004c expression, respectively. Extracted lipids were separated
by TLC. (B) ybr004cD-pGAL-YBR004c cells were grown and [
3
H]ino-
sitol-labeled in glucose-containing medium. Lipids were extracted
from cells and incubated either with or without mild-base (lanes 1
and 2) and with or without PI-PLC (lanes 3 and 4). (C) Lipids were
extracted from [
3
H]inositol- (lane 1) or [
3
H]serine-labeled (lanes 4

and 5) ybr004cD-pGAL-YBR004c cells or from [
14
C]EthN-labeled
ybr004cD-pGAL-YBR004c ⁄ psd1D⁄psd2D cells (lanes 2 and 3) and
separated by TLC. Lane 1 is from a 3-day film exposure that was
digitally cropped and precisely re-aligned with adjacent lanes 2–5,
which were exposed to film for 10 days. (D) Lipids were extracted
from ybr004cD-pGAL-YBR004c or gpi13D-pGAL-GPI13 cells grown
and [
3
H]inositol labeled in glucose-containing medium and incuba-
ted with or without JbaM (lanes 1–4) prior to their separation by
TLC. The lipid that accumulates in gpi13D-pGAL-GPI13 cells is a
mixture of two Man
4
-GPI isoforms that each bear a single EthN-P
on either Man1 or Man2 [15]. JbaM treatment digests Man
4
-GPI
(lane 3) into a Man
2
(EthN-P)Man
1
-GPI and a Man
1
(EthN-P)-
GPI (lane 4). Lipid 004–1 (lanes 1 and 2) comigrates with the
Man
1
(EthN-P)-GPI (lane 4). M1, M2 and M3 represent GPI man-

noses in the order of their addition to GPIs; PE, phosphoethanol-
amine; G, glucosamine; PI, phosphatidylinositol.
A L. Fabre et al. Second mannose addition to GPI precursors
FEBS Journal 272 (2005) 1160–1168 ª 2005 FEBS 1163
[
3
H]inositol-labeled non-GPIs in this chromatographic
solvent system, obscuring its detection.
Taken together, these data strongly suggest that lipid
004–1 is a GPI intermediate containing a single mannose
substituted with a side-branching EthN-P residue, and
corresponds to GPI species H5 in mammalian cells [35],
which can be generated by JbaM treatment of mamma-
lian Man
3
(EthN-P)-GPI [36]. The accumulation of this
GPI suggests that ybr004cD-pGAL-YBR004c cells have
a defect in addition of Man2 to GPI precursors.
Epistasis tests place Ybr004cp in the GPI
biosynthetic pathway
To obtain genetic evidence that YBR004c functions in
the GPI biosynthetic pathway, the epistasis relation-
ships to genes upstream and downstream of Man2
addition to GPIs were tested. Two double mutant
strains were created by mating haploids harboring
either smp3–2 or Dgpi1 temperature-sensitive alleles
with the ybr004cD-pGAL-YBR004c strain and the
[
3
H]inositol-labeled lipids they accumulate at 37 °C

under repressing conditions were examined.
At 37 °C, the Dgpi1 mutation, which blocks the
transfer of GlcNAc to phosphatidylinositol (PI),
the first step of GPI precursor assembly [28], blocks
the accumulation of lipid 004–1. gpi1D⁄ybr004cD-
pGAL-YBR004c cells grown and labeled at 25 °Cin
medium containing glucose showed prominent accu-
mulation of lipid 004–1 (Fig. 3A, lane 5). However,
the same cells grown in glucose-containing medium
at 37 °C showed no accumulation of lipid 004–1
(Fig. 3A, lane 6) indicating that formation of 004–1 is
dependent upon GlcNAc-PI synthesis.
An analogous experiment was performed with an
smp3–2 ⁄ ybr004cD-pGAL-YBR004c double mutant.
smp3–2 mutants are defective in addition of Man4 to
GPI precursors and accumulate a Man
3
-GPI inter-
mediate [31]. smp3–2 ⁄ ybr004cD-pGAL-YBR004c cells
grown and [
3
H]inositol-labeled in medium containing
galactose prominently accumulate the Man
3
-GPI at
25 °C (Fig. 3B, lane 3) and to a lesser degree at 37 °C
(Fig. 3B, lane 4). However, lipids from double mutant
cells labeled in glucose medium at 25 °C contain
predominantly lipid 004–1 and significantly less Man
3

-
GPI (Fig. 3B, lane 5), indicating that Ybr004cp func-
tions upstream of Smp3p. Together, these data further
support the conclusion that Ybr004cp functions in the
yeast GPI assembly pathway.
Sequence analysis of the Ybr004cp protein family
Database searches using the S. cerevisiae Ybr004cp
protein sequence and the Psi-BLAST algorithm
revealed 25 similar sequences in various eukaryotes,
including Homo sapiens (Table 1). No significant
homology was observed between Ybr004cp and
proteins from prokaryotes, and no eukaryotic genome
encoded obvious additional Ybr004cp-like sequences.
The consensus membrane topology predictive
algorithm of Persson and Argos [37] suggests that
Ybr004c proteins typically have eight transmembrane
domains with four intraluminally oriented loops
(Fig. 4). Alignment of all members of the Ybr004cp
AB
Fig. 3. ybr004c acts downstream of gpi1
and upstream of smp3 in the GPI biosyn-
thetic pathway. (A) A gpi1D⁄ybr004cD-
pGAL-YBR004c double mutant strain was
radiolabeled with [
3
H]inositol in SGalYE
medium at 25 °Cor37°C (lanes 3 and 4),
or in SGlcYE at 25 °Cor37°C (lanes 5 and
6). Lipids were extracted from cells and sep-
arated by TLC. Lipid 004-1 accumulates in

glucose-containing medium at 25 °C (lane 5)
but does not when the temperature-sensi-
tive gpi1 allele is suppressed at 37 °C (lane
6). (B) An smp3–2 ⁄ ybr004cD-pGAL-YBR004c
double mutant strain was [
3
H]inositol-labeled
as described above after which lipids were
extracted and separated by TLC.
Second mannose addition to GPI precursors A L. Fabre et al.
1164 FEBS Journal 272 (2005) 1160–1168 ª 2005 FEBS
family (Supplementary Fig. S1) revealed three invariably
conserved residues (Glu, Gln, and Trp) that each are
predicted to reside within an intraluminal loop (Fig. 4).
Expression of human YBR004c restores viability
to Dybr004c yeast
We tested if the human Ybr004c homologue (GenBank
NP_060307) could complement the lethal ybr004c::
Kan
R
null mutation in vivo in S. cerevisiae. Heterozy-
gous ybr004c::Kan
R
⁄ YBR004c ura3 ⁄ ura3 diploids were
transformed with pGAL-hYBR004c. Transformants
were sporulated and asci were dissected onto YPGal
agar medium to assess the viability of the individual
haploid spores. Asci from diploids harboring pGAL-
hYBR004c gave rise to four viable haploid progeny.
Additionally, two haploids from each tetrad were

resistant to G418 (Fig. 5A) and sensitive to 5-FOA
(Fig. 5B), indicating that they harbored the ybr004c::
Kan
R
allele and that their viability was dependent upon
the complementing URA3-containing plasmid. Addition-
ally, neither pGAL-hYBR004c nor pGAL-YBR004c
were able to complement lethal null mutations of
YJR013w, GPI10,orSMP3, genes encoding the
mannosyltransferases that add Man1 [19], Man3 [22]
and Man4 [31] to yeast GPI precursors, respectively.
Therefore, hYBR004c expression specifically restores
viability to yeast defective in Man2 addition to GPIs.
We conclude that human Ybr004cp is the functional
equivalent of S. cerevisiae Ybr004cp.
Discussion
The majority of the steps in assembly and decoration
of the GPI precursor glycolipid have been defined
genetically in that at least one gene’s product has been
implicated in all but one of the predicted reactions in
the GPI pathway. The exception is the addition of the
second mannose to the GPI core. We show here that
depletion of the essential, multispanning membrane
protein Ybr004cp from yeast cells leads to the bio-
chemical defects expected if addition of the second,
a-1,6-linked mannose to GPI precursors is prevented.
These defects are a block in the incorporation of
[
3
H]inositol into protein, consistent with abolition of

GPI anchoring, and the accumulation of a PI-PLC-
resistant, base-labile [
3
H]inositol-labeled glycolipid
whose glycan headgroup likely contains a single man-
nose that is modified with an EthN-P residue.
Our epistasis tests with known GPI assembly
mutants indicate that Ybr004cp functions in the GPI
assembly pathway, and further, Ybr004cp-depletion
gives rise to cell wall and morphological defects charac-
teristic of GPI assembly mutants. We therefore propose
that Ybr004cp is an excellent candidate for GPI-MT-II
itself or an essential subunit of that enzyme.
Our results also shed light on the first EthN-P addi-
tion step in yeast. Because the GPI precursor that
accumulates when addition of Man2 is blocked is
modified with phosphoethanolamine, EthN-P can be
Fig. 4. Predicted membrane topology of Ybr004c proteins. The fig-
ure was drawn using data predicted by alignment of 25 Ybr004c
protein sequences (Fig. S1) using the
CLUSTAL W program [42] fol-
lowed by analysis of the aligned sequences using the TMAP algo-
rithm [37] to predict conserved membrane topology as described in
Experimental procedures. Black circles represent the position of
strictly conserved amino acids, whereas gray circles indicate amino
acids conserved in > 85% (22 of 25) of the aligned sequences. Pre-
dicted loop lengths range from the shortest to the longest size
observed in all 25 sequences.
A
B

Fig. 5. Human YBR004c expression restores viability to Dybr004c
S. cerevisiae cells. A heterozygous ybr004c::Kan
R
⁄ YBR004c diploid
yeast strain harboring the pGAL-hYBR004c expression vector was
sporulated and tetrads microdissected onto YPGal agar medium.
For tetrads giving rise to four viable progeny, each haploid segre-
gant was streaked on YPGal agar medium containing either 200 lg
G418 per mL (A) or 1 mg 5-FOA per mL (B) and grown for 3 days
at 25 °C.
A L. Fabre et al. Second mannose addition to GPI precursors
FEBS Journal 272 (2005) 1160–1168 ª 2005 FEBS 1165
added to Man1 of GPI precursors as early as the
Man
1
-GPI stage.
To date, no biochemical function has been described
for any Ybr004c protein, although its Drosophila homo-
logue (termed ‘vegetable’) was identified in a screen
for genes implicated in formation of the peripheral ner-
vous system [38]. These findings, and our assignment of
function to Ybr004c proteins, suggest the importance of
efficient GPI anchoring in this developmental process.
Our identification of a novel, conserved protein
essential for Man2 addition to GPIs will allow us to
carry out detailed biochemical and genetic analyses of
this uncharacterized step in GPI biosynthesis.
Experimental procedures
Materials
[2-

3
H]-myo-Inositol (sp. act. 30 CiÆmmol
)1
), [1,2–
14
C]-etha-
nolamine hydrocloride and L-[
3
H(G)]-serine were obtained
from American Radiolabeled Chemicals. Calcofluor white
(fluorescent brightener 28), Geneticin (G418), Jack bean
a-mannosidase (JbaM), phospholipase C (PI-PLC) and
5-fluoroorotic acid (5-FOA) were from Sigma.
Yeast strains and media
SD (SGlc) and YPD media were made as described [39].
YPGal medium has the same composition as YPD but with
2% (w ⁄ v) galactose instead of glucose. Inositol-free syn-
thetic medium and synthetic medium containing 0.2% yeast
extract (w ⁄ v) and glycerol (SGlyYE), galactose (SGalYE)
or glucose (SGlcYE) were prepared as described [15]. Cal-
cofluor white hypersensitivity was tested on YPD agar con-
taining 16 lg Calcofluor white per mL. Sensitivity of yeast
to 5-FOA was determined on YPGal medium containing
1 mg 5-FOA per mL.
Diploid heterozygous YBR004c ⁄ ybr004c::Kan
R
, YJR013-
w ⁄ yjr013w::Kan
R
, GPI10 ⁄ gpi10::Kan

R
and SMP3 ⁄ smp3::
Kan
R
strains were purchased from Research Genetics. To
construct a glucose-repressible allele of YBR004c, the
YBR004c ⁄ ybr004c::Kan
R
heterozygous diploid was trans-
formed with pGAL-YBR004c (see below). Transformants
were sporulated and tetrads dissected. Haploid progeny
harboring a ybr004c::Kan
R
allele complemented by pGAL-
YBR004c were identified by growth on YPGal plates con-
taining 200 lg G418 per mL. The double mutant strains
gpi1D⁄ybr004cD-pGAL-YBR004c and smp3–2 ⁄ ybr004cD-
pGAL-YBR004c were created by mating ybr004cD-pGAL-
YBR004c (MAT a, his3D1, leu2D1, ura3D0, met15D0,
ybr004c::Kan
R
) with Dgpi1 [28] and smp3–2 [31] strains,
respectively. A ybr004cD-pGAL-YBR004c strain back-
ground harboring an ethanolamine auxotrophy was created
by mating ybr004cD-pGAL-YBR004c with RYY51 (MAT
a, trp1–1, ura3–1, leu2–3,112, his3–11, suc2, rho
+
, lys2,
psd1::TRP1, psd2 ::HIS3) [40].
Construction of YBR004c yeast expression

plasmids
The human (GenBank NP_060307) and S. cerevisiae
YBR004c genes were PCR-amplified from human liver
cDNA or S. cerevisiae genomic DNA, respectively. Each
was cloned as a EcoRI-BamHI fragment downstream of the
galactose-inducible ⁄ glucose-repressible GAL10-1 promoter
in vector pMW20 [41] to produce the pGAL-hYBR004c
(human) and pGAL-YBR004c (yeast) S. cerevisiae expres-
sion plasmids.
In vivo radiolabeling of S. cerevisiae lipids and
thin layer chromatography
[
3
H]Inositol labeling of lipids in temperature-sensitive yeast
strains was performed as previously described [15]. For
[
3
H]inositol or [
3
H]serine labeling of the ybr004cD-pGAL-
YBR004c strain, cells were first grown in SGlyYE medium,
then shifted to SGlcYE or SGalYE medium for 16 h and
labeled for 2 h at 30 °C with 15 lCi [
3
H]inositol or 50 lCi
[
3
H]serine. [
14
C]Ethanolamine labeling of the Dpsd1 ⁄Dpsd2 ⁄

ybr004cD-pGAL-YBR004c strain was performed in the same
manner except that each growth medium was supplemented
with 5 mm ethanolamine and 5 mm choline, and metabolic
labeling was performed with 20 lCi [
14
C]ethanolamine for
 23 h at 25 °C. For radiolabeling of double mutant strains,
cells were grown in SGlyYE medium for 2 days at 25 °C,
then grown in SGalYE or SGlcYE medium for 16 h. Cells
were shifted to 25 °Cor37°C for 20 min and radiolabeled
with 15 lCi [
3
H]inositol for 2 h. Radiolabeled lipids were
extracted from cells and treated with mild-base, phospho-
lipase C, or JbaM as described [12,15].
Isolated lipids were separated by TLC on silica 60 plates
(VWR). TLC plates were prerun in chloroform ⁄ meth-
anol ⁄ water (65 : 25 : 4, v ⁄ v ⁄ v), after which lipids were
applied and separated in chloroform–methanol–water
(5:5:1, v⁄ v ⁄ v). TLC-separated lipids were exposed to
BioMax MS film (Eastman Kodak) for 1–4 days using a
BioMax Transcreen LE intensifier screen.
[
3
H]Inositol labeling of proteins in ybr004cD-pGAL-
YBR004c cells was performed as described [15]. [
3
H]ino-
sitol-labeled proteins were separated on a 10–20%
SDS ⁄ PAGE (Daichii) and detected by fluorography as des-

cribed above.
Protein sequence analysis
Consensus topology prediction for 25 Ybr004c proteins
(Table 1 and supplementary Fig. S1) was performed using
the program clustal w [42] to align the primary amino
Second mannose addition to GPI precursors A L. Fabre et al.
1166 FEBS Journal 272 (2005) 1160–1168 ª 2005 FEBS
acid sequences (parameters: protein weight matrix, BLO-
SUM series; gap open penalty, 10; gap extension penalty,
0.1). The aligned sequences were submitted as input to the
tmap program [37] to predict conserved membrane topol-
ogy using default parameters.
Acknowledgements
CHT thanks Dr Donald Comb of New England Bio-
labs for financial support. PO is supported by National
Institutes of Health Grant GM46220. The authors
thank B. Taron and P. Colussi for advice and technical
assistance.
References
1 McConville MJ & Ferguson MA (1993) The structure,
biosynthesis and function of glycosylated phosphatidyl-
inositols in the parasitic protozoa and higher eukar-
yotes. Biochem J 294, 305–324.
2 Tiede A, Bastisch I, Schubert J, Orlean P & Schmidt
RE (1999) Biosynthesis of glycosylphosphatidylinositols
in mammals and unicellular microbes. Biol Chem 380,
503–523.
3 Leidich SD, Drapp DA & Orlean P (1994) A condition-
ally lethal yeast mutant blocked at the first step in gly-
cosyl phosphatidylinositol anchor synthesis. J Biol

Chem 269, 10193–10196.
4 Colussi PA & Orlean P (1997) The essential Schizosac-
charomyces pombe gpil
+
gene complements a bakers’
yeast GPI anchoring mutant and is required for efficient
cell separation. Yeast 13, 139–150.
5 Grimme SJ, Colussi PA, Taron CH & Orlean P (2004)
Deficiencies in the essential Smp3 mannosyltransferase
block glycosylphosphatidylinositol assembly and lead to
defects in growth and cell wall biogenesis in Candida
albicans. Microbiology 150, 3115–3128.
6 Kawagoe K, Kitamura D, Okabe M, Taniuchi I, Ikawa
M, Watanabe T, Kinoshita T & Takeda J (1996) Glyco-
sylphosphatidylinositol-anchor-deficient mice: implica-
tions for clonal dominance of mutant cells in
paroxysmal nocturnal hemoglobinuria. Blood 87, 3600–
3606.
7 Ilgoutz SC, Zawadzki JL, Ralton JE & McConville MJ
(1999) Evidence that free GPI glycolipids are essential for
growth of Leishmania mexicana. EMBO J 18, 2746–2755.
8 Nagamune K, Nozaki T, Maeda Y, Ohishi K, Fukuma
T, Hara T, Schwarz RT, Sutterlin C, Brun R, Riezman
H & Kinoshita T (2000) Critical roles of glycosylpho-
sphatidylinositol for Trypanosoma brucei. Proc Natl
Acad Sci USA 97, 10336–10341.
9 Fankhauser C, Homans SW, Thomas-Oates JE,
McConville MJ, Desponds C, Conzelmann A & Fergu-
son MA (1993) Structures of glycosylphosphatidylinosi-
tol membrane anchors from Saccharomyces cerevisiae.

J Biol Chem 268, 26365–26374.
10 Homans SW, Ferguson MA, Dwek RA, Rademacher
TW, Anand R & Williams AF (1988) Complete struc-
ture of the glycosyl phosphatidylinositol membrane
anchor of rat brain Thy-1 glycoprotein. Nature 333,
269–272.
11 Roberts WL, Santikarn S, Reinhold VN &
Rosenberry TL (1988) Structural characterization of
the glycoinositol phospholipid membrane anchor of
human erythrocyte acetylcholinesterase by fast atom
bombardment mass spectrometry. J Biol Chem 263,
18776–18784.
12 Taron BW, Colussi PA, Wiedman JM, Orlean P &
Taron CH (2004) Human Smp3p adds a fourth man-
nose to yeast and human glycosylphosphatidylinositol
precursors in vivo. J Biol Chem 279, 36083–36092.
13 Benachour A, Sipos G, Flury I, Reggiori F, Canivenc-
Gansel E, Vionnet C, Conzelmann A & Benghezal M
(1999) Deletion of GPI7, a yeast gene required for addi-
tion of a side chain to the glycosylphosphatidylinositol
(GPI) core structure, affects GPI protein transport,
remodeling, and cell wall integrity. J Biol Chem 274,
15251–15261.
14 Flury I, Benachour A & Conzelmann A (2000)
YLL031c belongs to a novel family of membrane pro-
teins involved in the transfer of ethanolaminephosphate
onto the core structure of glycosylphosphatidylinositol
anchors in yeast. J Biol Chem 275, 24458–24465.
15 Taron CH, Wiedman JM, Grimme SJ & Orlean P
(2000) Glycosylphosphatidylinositol biosynthesis defects

in Gpi11p- and Gpi13p-deficient yeast suggest a
branched pathway and implicate gpi13p in phos-
phoethanolamine transfer to the third mannose. Mol
Biol Cell 11, 1611–1630.
16 Ueda E, Sevlever D, Prince GM, Rosenberry TL,
Hirose S & Medof ME (1993) A candidate mammalian
glycoinositol phospholipid precursor containing three
phosphoethanolamines. J Biol Chem 268, 9998–10002.
17 Hong Y, Maeda Y, Watanabe R, Ohishi K, Mishkind
M, Riezman H & Kinoshita T (1999) Pig-n, a mamma-
lian homologue of yeast Mcd4p, is involved in transfer-
ring phosphoethanolamine to the first mannose of the
glycosylphosphatidylinositol. J Biol Chem 274, 35099–
35106.
18 Menon AK, Mayor S & Schwarz RT (1990) Biosynth-
esis of glycosyl-phosphatidylinositol lipids in Trypano-
soma brucei: involvement of mannosyl-
phosphoryldolichol as the mannose donor. EMBO J 9,
4249–4258.
19 Maeda Y, Watanabe R, Harris CL, Hong Y, Ohishi K,
Kinoshita K & Kinoshita T (2001) PIG-M transfers the
first mannose to glycosylphosphatidylinositol on the
lumenal side of the ER. EMBO J 20, 250–261.
A L. Fabre et al. Second mannose addition to GPI precursors
FEBS Journal 272 (2005) 1160–1168 ª 2005 FEBS 1167
20 Takahashi M, Inoue N, Ohishi K, Maeda Y, Nakamura
N, Endo Y, Fujita T, Takeda J & Kinoshita T (1996)
PIG-B, a membrane protein of the endoplasmic reticu-
lum with a large lumenal domain, is involved in trans-
ferring the third mannose of the GPI anchor. EMBO J

15, 4254–4261.
21 Canivenc-Gansel E, Imhof I, Reggiori F, Burda P, Con-
zelmann A & Benachour A (1998) GPI anchor bio-
synthesis in yeast: phosphoethanolamine is attached to
the alpha1,4-linked mannose of the complete precursor
glycophospholipid. Glycobiology 8, 761–770.
22 Su
¨
tterlin C, Escribano MV, Gerold P, Maeda Y, Mazon
MJ, Kinoshita T, Schwarz RT & Riezman H (1998)
Saccharomyces cerevisiae GPI10, the functional homolo-
gue of human PIG-B, is required for glycosylphosphati-
dylinositol-anchor synthesis. Biochem J 332, 153–159.
23 Strahl-Bolsinger S, Gentzsch M & Tanner W (1999)
Protein O-mannosylation. Biochim Biophys Acta 1426,
297–307.
24 Lee SA, Wormsley S, Kamoun S, Lee AF, Joiner K &
Wong B (2003) An analysis of the Candida albicans gen-
ome database for soluble secreted proteins using compu-
ter-based prediction algorithms. Yeast 20, 595–610.
25 Oriol R, Martinez-Duncker I, Chantret I, Mollicone R
& Codogno P (2002) Common origin and evolution of
glycosyltransferases using Dol-P-monosaccharides as
donor substrate. Mol Biol Evol 19, 1451–1463.
26 Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang
Z, Miller W & Lipman DJ (1997) Gapped blast and
PSI- blast: a new generation of protein database search
programs. Nucleic Acids Res 25, 3389–3402.
27 Ram AF, Wolters A, Ten Hoopen R & Klis FM (1994)
A new approach for isolating cell wall mutants in Sac-

charomyces cerevisiae by screening for hypersensitivity
to calcofluor white. Yeast 10, 1019–1030.
28 Leidich SD & Orlean P (1996) Gpi1, a Saccharomyces
cerevisiae protein that participates in the first step in
glycosylphosphatidylinositol anchor synthesis. J Biol
Chem 271, 27829–27837.
29 Conzelmann A, Fankhauser C & Desponds C (1990)
Myoinositol gets incorporated into numerous membrane
glycoproteins of Saccharomyces cerevisiae; incorporation
is dependent on phosphomannomutase (Sec53). EMBO
J 9, 653–661.
30 Orlean P (1990) Dolichol phosphate mannose synthase
is required in vivo for glycosyl phosphatidylinositol
membrane anchoring, O mannosylation, and N glycosy-
lation of protein in Saccharomyces cerevisiae. Mol Cell
Biol 10, 5796–5805.
31 Grimme SJ, Westfall BA, Wiedman JM, Taron CH &
Orlean P (2001) The essential Smp3 protein is required
for addition of the side-branching fourth mannose
during assembly of yeast glycosylphosphatidylinositols.
J Biol Chem 276, 27731–27739.
32 Benghezal M, Lipke PN & Conzelmann A (1995) Iden-
tification of six complementation classes involved in the
biosynthesis of glycosylphosphatidylinositol anchors in
Saccharomyces cerevisiae. J Cell Biol 130, 1333–1344.
33 Menon AK & Stevens VL (1992) Phosphatidylethanol-
amine is the donor of the ethanolamine residue linking
a glycosylphosphatidylinositol anchor to protein. J Biol
Chem 267, 15277–15280.
34 Imhof I, Canivenc-Gansel E, Meyer U & Conzelmann

A (2000) Phosphatidylethanolamine is the donor of the
phosphorylethanolamine linked to the a1,4-linked man-
nose of yeast GPI structures. Glycobiology 10, 1271–
1275.
35 Hirose S, Prince GM, Sevlever D, Ravi L, Rosenberry
TL, Ueda E & Medof ME (1992) Characterization of
putative glycoinositol phospholipid anchor precursors in
mammalian cells: localization of phosphoethanolamine.
J Biol Chem 267, 16968–16974.
36 Hong Y, Maeda Y, Watanabe R, Inoue N, Ohishi K &
Kinoshita T (2000) Requirement of PIG-F and PIG-O
for transferring phosphoethanolamine to the third man-
nose in glycosylphosphatidylinositol. J Biol Chem 275,
20911–20919.
37 Persson B & Argos P (1994) Prediction of transmem-
brane segments in proteins utilising multiple sequence
alignments. J Mol Biol 237, 182–192.
38 Prokopenko SN, He Y, Lu Y & Bellen HJ (2000) Muta-
tions affecting the development of the peripheral ner-
vous system in Drosophila: a molecular screen for novel
proteins. Genetics 156, 1691–1715.
39 Sherman F (1991) Getting started with yeast. Methods
Enzymol 194, 3–21.
40 Trotter PJ & Voelker DR (1995) Identification of a
non-mitochondrial phosphatidylserine decarboxylase
activity (PSD2) in the yeast Saccharomyces cerevisiae.
J Biol Chem 270, 6062–6070.
41 Zieler HA, Walberg M & Berg P (1995) Suppression of
mutations in two Saccharomyces cerevisiae genes by
the adenovirus E1A protein. Mol Cell Biol 15, 3227–

3237.
42 Thompson JD, Higgins DG & Gibson TJ (1994)
clustal w: improving the sensitivity of progressive mul-
tiple sequence alignment through sequence weighting,
position-specific gap penalties and weight matrix choice.
Nucleic Acids Res 22, 4673–4680.
Supplementary material
The following material is available from http://www.
blackwellpublishing.com/products/journals/suppmat/EJB/
EJB4551/EJB4551sm.htm
Fig. S1. Multiple sequence alignment of 25 Ybr004c
proteins.
Second mannose addition to GPI precursors A L. Fabre et al.
1168 FEBS Journal 272 (2005) 1160–1168 ª 2005 FEBS