QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

Vol. 18, Issue 11, 4304-4316, November 2007

This Article Abstract Full Text (PDF) Supplemental Materials All Versions of this Article: E07-05-0482v1 18/11/4304    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Umemura, M. Articles by Jigami, Y. Search for Related Content PubMed PubMed Citation Articles by Umemura, M. Articles by Jigami, Y.

Saccharomyces cerevisiae CWH43 Is Involved in the Remodeling of the Lipid Moiety of GPI Anchors to Ceramides

Mariko Umemura^*,, Morihisa Fujita^*, Takehiko Yoko-o^*, Akiyoshi Fukamizu, and Yoshifumi Jigami^*,

*Research Institute for Cell Engineering, National Institute of Advanced Industrial Science and Technology, Ibaraki 305-8566, Japan; Graduate School of Life and Environmental Science, University of Tsukuba, Ibaraki 305-8572, Japan; and Center for Tsukuba Advanced Research Alliance, Graduate School of Life and Environmental Science, University of Tsukuba, Ibaraki 305-8577, Japan

Submitted May 22, 2007; Revised August 9, 2007; Accepted August 17, 2007
Monitoring Editor: Sean Munro

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The glycosylphosphatidylinositol (GPI)-anchored proteins are subjected to lipid remodeling during their biosynthesis. In the yeast Saccharomyces cerevisiae, the mature GPI-anchored proteins contain mainly ceramide or diacylglycerol with a saturated long-fatty acid, whereas conventional phosphatidylinositol (PI) used for GPI biosynthesis contains an unsaturated fatty acid. Here, we report that S. cerevisiae Cwh43p, whose N-terminal region contains a sequence homologous to mammalian PGAP2, is involved in the remodeling of the lipid moiety of GPI anchors to ceramides. In cwh43 disruptant cells, the PI moiety of the GPI-anchored protein contains a saturated long fatty acid and lyso-PI but not inositolphosphorylceramides, which are the main lipid moieties of GPI-anchored proteins from wild-type cells. Moreover, the C-terminal region of Cwh43p (Cwh43-C), which is not present in PGAP2, is essential for the ability to remodel GPI lipids to ceramides. The N-terminal region of Cwh43p (Cwh43-N) is associated with Cwh43-C, and it enhanced the lipid remodeling to ceramides by Cwh43-C. Our results also indicate that mouse FRAG1 and C130090K23, which are homologous to Cwh43-N and -C, respectively, share these activities.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Many proteins are attached to the cell surface through glycosylphosphatidylinositol (GPI). GPI modification is conserved among mammalian cells, yeast, and protozoa (Ferguson, 1999; Kinoshita and Inoue, 2000; Pittet and Conzelmann, 2007). In the budding yeast Saccharomyces cerevisiae, 60 proteins have been predicted to be GPI-anchored (Caro et al., 1997). Some of these proteins, such as Gas1p, seem to remain anchored to the plasma membrane, whereas the others become mainly cell wall-linked (Caro et al., 1997; Hamada et al., 1999; Kapteyn et al., 1999). Thus, yeast GPI-anchored proteins are mainly involved in cell wall integrity.

The biosynthesis of GPI and its attachment to the target protein are carried out on the endoplasmic reticulum (ER) membrane. GPIs, which have conserved core structures, are preassembled in the ER in a multistep pathway before their transfer to target proteins. GPIs are assembled by the sequential addition to PI of glucosamine, an acyl chain, mannoses, and ethanolamine phosphates. The complete GPIs are then attached to ER-translocated proproteins bearing a C-terminal signal sequence for GPI attachment. More than 20 genes involved in GPI biosynthesis and attachment to proteins have been identified, and, in most cases, these genes are conserved in yeast and mammalian cells (Kinoshita and Inoue, 2000; Orlean and Menon, 2007; Pittet and Conzelmann, 2007).

After the transfer of GPI to a protein, the acyl chain is removed from the inositol portion of GPI-anchored proteins (Tanaka et al., 2004; Fujita et al., 2006b). The inositol-deacylated GPI-anchored proteins are then transported to the plasma membrane by vesicular trafficking (Mayor and Riezman, 2004). GPI-anchored proteins are concentrated at the microdomains of plasma membrane, generally called lipid rafts, that are enriched with sterols and sphingolipids, and they can be isolated biochemically as detergent-resistant membrane (DRM) fractions (Simons and Ikonen, 1997). The lipid rafts are thought to play an important role in signal transduction and membrane trafficking.

Some mature GPI-anchored proteins have saturated fatty acid chains in the lipid moiety (McConville et al., 1993; Brewis et al., 1995; Benting et al., 1999), whereas GPI is synthesized from conventional PI, which usually has an unsaturated fatty acid chain. This conversion of the lipid moiety is called "lipid remodeling" of GPI. In mammalian cells, after exchanging an unsaturated chain at the sn-2 position of the lipid moieties with a saturated chain, the lipid moieties of mature GPI-anchored proteins contain PI with two saturated fatty chains (Maeda et al., 2007). In contrast, mature GPI-anchored proteins in yeast contain two different types of lipid moieties (Conzelmann et al., 1992; Fankhauser et al., 1993; Sipos et al., 1994, 1997; Reggiori et al., 1997). One is diacylglycerol with a C26 fatty acid at the sn-2 position. The other type of lipid moiety is ceramide consisting mainly of C18:0 phytosphingosine (PHS) and C26:0 fatty acid (Sipos et al., 1997). In both types of lipid moieties, the C26 fatty acid may be hydroxylated (Sipos et al., 1997). However, which GPI-anchored proteins contain a ceramide-type lipid moiety in the GPI anchor remains unclear, although the ceramide type represents a major fraction of the total cells and Gas1p, a major GPI-anchored protein, is an example of protein containing diacylglycerol type of lipid moiety (Fankhauser et al., 1993).

The remodeling of the yeast lipid starts in the ER with the removal of the unsaturated acyl chain at the sn-2 position of diacylglycerol to form lyso-GPI. This GPI-phospholipase A₂ (GPI-PLA₂) activity requires Per1p (Fujita et al., 2006a). Next, a C26 saturated acyl chain is transferred to the sn-2 position by the O-acyltransferase Gup1p, generating diacylglycerol with a C26 fatty acid (Bosson et al., 2006). Recent studies suggest that these lipid remodeling events are required for both the efficient transport of GPI-anchored proteins from the ER to the Golgi and their association with lipid rafts (Bosson et al., 2006; Fujita et al., 2006a; Maeda et al., 2007). The diacylglycerol of the lipid moiety is replaced by ceramide consisting of PHS with a C26 fatty acid, a process that seems to occur in the ER (Sipos et al., 1997). In the Golgi, the ceramide of the lipid moiety is further changed to ceramide consisting of PHS with a hydroxy-C26 fatty acid (Reggiori et al., 1997; Sipos et al., 1997). The pathway for replacement by ceramide has not been elucidated, and proteins involved in this event have not been identified. Gpi7p and Mcd4p, which transfer ethanolamine phosphate to mannose during GPI biosynthesis, are reported to be indirectly involved in the remodeling of diacylglycerol to ceramides (Benachour et al., 1999; Zhu et al., 2006), but it remains unclear what proteins are directly responsible for this.

In this study, we found that yeast CWH43 is required for the remodeling of the lipid moiety of GPI anchors to ceramides and that its C-terminal domain is especially important for this function. The N-terminal region of Cwh43p associated with C-terminal region is able to enhance the lipid remodeling to ceramides by C-terminal region. Furthermore, in yeast cwh43-disuruptant cells, the mouse FRAG1 and C130090K23 gene product, which are homologues of the N- and C-terminal regions of Cwh43p, respectively, can share these activities, indicating that the GPI lipid is remodeled to ceramides by mouse C130090K23 protein, whose function is enhanced when FRAG1 protein is introduced.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Media and Strains
YPAD and synthetic complete (SC) media were described previously (Sherman, 1991). For [³H]dihydrosphingosine (DHS) labeling, cells were cultured at 30°C in SC medium. The yeast strains used in this study included the following: cwh43 (MATa his31 leu20 met150 ura30 cwh43::KanMX4; EUROSCARF), BY4741 (MATa his31 leu20 met150 ura30; wild-type strain), and gpi7 (MATa his31 leu20 met150 ura30 gpi7::KanMX4; EUROSCARF).

Plasmids
Plasmids used in this report are listed in Table 1.

Construction of Plasmid for Overexpression of CWH43-HA
The open reading frame of CWH43 with a sequence encoding three copies of the HA epitope at the C terminus of the protein was amplified from pRS316T-CWH43-HA by using primers Kpn-CWH43-F18 (5'-GGGGTACCATGCTGATCATCAATGGGAAG-3') and HANheStpXba-R17 (5'-GCTCTAGATTAGCTAGCGTAATCCGGTAC-3'). The amplified fragment was digested with KpnI and XbaI. The fragment was inserted into KpnI/XbaI-digested YEp352GAPII (Abe et al., 2003) to generate YEp352GAPII-CWH43-HA.

Construction of CWH43-N and CWH43-N-HA Plasmids
The short promoter region and the open reading frame encoding the N-terminal region (229 amino acids) of Cwh43p (CWH43-N) were amplified from pRS316T-CWH43 by using primers Xho-CWH43-P-N-F14 (5'-GATTTCTCGAGGAATAAGTAAC-3') and CWH43-N-NheStpXho-R15 (5'-CCG- CTCGAGGTCGACTTAGCTAGCACTAGAGTCTCTAATTTGGAAAA-3'). The amplified fragment was digested with XhoI. The fragment was inserted into pRS316T-CWH43, which contains the entire CWH43 upper promoter region to generate pRS316T-CWH43-N. Three copies of the HA epitope were amplified and inserted into the NheI site of pRS316T-CWH43-N to generate pRS316T-CWH43-N-HA. The open reading frame of CWH43-N and the sequence encoding three copies of the HA epitope (CWH43-N-HA) were amplified from pRS316T-CWH43-N-HA by using primers Kpn-CWH43-F18 (5'-GGGGTACCATGCTGATCATCAATGGGAAG-3') and HANheStpXba-R17 (5'-GCTCTAGATTAGCTAGCGTAATCCGGTAC-3'). The amplified fragment was digested with KpnI/XbaI. The fragment was inserted into KpnI/XbaI-digested YEp352GAPII to generate YEp352GAPII-CWH43-N-3HA.

Construction of CWH43-C and CWH43-C-HA Plasmids
The short promoter region and open reading frame encoding the C-terminal region (736 amino acids) of Cwh43p (CWH43-C) were amplified from pRS316T-CWH43 by using primers Xho-CWH43-P-C1-F11 (5'-TTTCTCGA- GGAATAAGTAACCAGGAATACAGAAGGTATCCACCGCCAGTTATGG- GTAATACCAGTTTTTTCCAAATT-3') and NheStpXho-R12 (5'-CCGCT- CGAGGTCGACTTAGCTAG-3'). The amplified fragment was digested with XhoI. The fragment was inserted into pRS316T-CWH43, which contains the entire CWH43 promoter region to generate pRS316T-CWH43-C. A sequence encoding three copies of the HA epitope were amplified and inserted into the NheI site of pRS316T-CWH43-C to generate pRS316T-CWH43-C-HA. The sequence encoding the open reading frame of CWH43-C and three copies of the HA epitope (CWH43-C-HA) were amplified from pRS316T-CWH43-C-HA by using primers Kpn-CWH43-C1-F19 (5'-GGGGTACCATGGGTAATACCAGTTTTTTCCAA-3') and HANheStpXba-R17 (5'-GCTCTAGATTAGC- TAGCGTAATCCGGTAC-3'). The amplified fragment was digested with KpnI/XbaI. The fragment was inserted into KpnI/XbaI-digested YEp352GAPII to generate YEp352GAPII-CWH43-C-HA.

Construction of Mutated cwh43-HA and cwh43-C-HA Plasmids
We substituted G57 with arginine by using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) and pRS316T-CWH43-HA, generating the plasmid pRS316T-CWH43-G57R-HA. We also substituted H472, D693, H770/H771, E807, D862, and R882 with alanine by using pRS316T-CWH43-C-HA, generating plasmids pRS316T-CWH43-C-H472A-HA, pRS316T-CWH43-C-D693A-HA, pRS316T-CWH43-C-H770AH771A-HA, pRS316T-CWH43-C-E807A-HA, pRS316T-CWH43-C-D862A-HA, and pRS316T-CWH43-C-R882A-HA, respectively.

Construction of FRAG1, FRAG1-HA, and FRAG1-FLAG Plasmids
The open reading frame of the mouse FRAG1 gene was amplified from the plasmid of the National Institutes of Health mammalian gene collection clone (no. 3588752) by using primers EI-mmFRAG1-F24 (5'-GGAATTCATGTACCAGGTCCCACTGAC-3') and mmFRAG1-NotI-R33 (5'-ATAAGAATGCGGCCGCGAATCTCTTTTCCTCAGGCTG-3'). The amplified fragment was digested with EcoRI and NotI. The fragment was inserted into EcoRI/NotI-digested YEp352GAPII-HA, which contains three copies of the HA epitope, to generate YEp352GAPII-FRAG1-HA. A sequence encoding three copies of the FLAG epitope was amplified and inserted into the NotI/SalI site of YEp351GAPII-FRAG1-HA to generate YEp351GAPII-FRAG1-FLAG.

Construction of C130090K23, C130090K23-HA, and C130090K23-FLAG Plasmids
The open reading frame of the mouse C130090K23 gene was amplified from the plasmid of the National Institutes of Health mammalian gene collection clone (no. 3584006) by using primers Sma-mmFLJ21511-F30 (5'-TCCCCCGGGATGCCAGGCCTGTGGAGAG-3') and mmFLJ21511-N*Sa-R27 (5'-TGCGGTCGACTCAGCTAGCAACGAAGTATTTGGGTGTATTCA-3'). The amplified fragment was digested with SmaI and SalI. The fragment was ligated into SmaI/SalI-digested YEp352GAPII to generate YEp352GAPII-C130090K23. A sequence encoding three copies of the HA or FLAG epitopes were amplified and inserted into the NheI site of YEp352GAPII-C130090K23 to generate YEp352GAPII-C130090K23-HA or YEp352GAPII-C130090K23-FLAG.

Construction of ALG13-FLAG and ALG14-HA Plasmids
The open reading frame of ALG13 was amplified by polymerase chain reaction (PCR) from genomic DNA by using primers Sac-ALG13-F (5'-CGAGCTCATGGGTATTATTGAAGAAAAGGC-3') and ALG13-Not-R (5'-ATAGTTTAGCGGCCGCGCTGTATATAGTTTCAACTAGCA-3'). The amplified fragment was digested with SacI and NotI. The fragment was inserted into SacI/NotI-digested YEp351GAPII-FRAG1-FLAG, which contains three copies of the FLAG epitope, to generate YEp351GAPII-ALG13-FLAG. The open reading frame of ALG14 was amplified by PCR from genomic DNA by using primers Sac-ALG14-F (5'-CGAGCTCATGAAAACGGCCTACTTGGC-3') and ALG14-Not-R (5'-ATAGTTTAGCGGCCGCAACAAGGATGCCGAACCACTT-3'). The amplified fragment was digested with SacI and NotI. The fragment was inserted into SacI/NotI-digested YEp352GAPII-FRAG1-HA, which contains three copies of the HA epitope, to generate YEp352GAPII-ALG14-HA.

Fluorescence Microscopy
For imaging of Cwh43p-mRFP fusion protein, cells were grown to the exponential phase at 30°C in SC–uracil medium and washed twice. The cellular localization of Cwh43p-mRFP was visualized by fluorescence microscopy by using an Olympus IX-71 fluorescence microscope system (Olympus, Tokyo, Japan) and processed using IPLab software (Scanalytics, Fairfax, VA).

Analysis of PI Moieties of GPI-anchored Protein
Isolation of the radiolabeled lipid moieties of GPI anchor and lipid analysis were performed as described previously (Guillas et al., 2000; Fujita et al., 2006a).

Lipid Analysis
The lipids were analyzed by thin-layer chromatography (TLC) on a Silica gel 60 plate (Merck, Darmstadt, Germany) by using solvent system 1 (55:45:10, vol/vol CHCl₃:CH₃OH:0.25% KCl) or solvent system 2 (10:10:3, vol/vol CHCl₃:CH₃OH:H₂O). Reaction products separated on TLC plates were detected by autoradiography and analyzed using a Molecular Imager FX (Bio-Rad, Hercules, CA).

Radiolabeling of Proteins
Radiolabeling of proteins with [³H]DHS (American Radiolabeled Chemicals, St. Louis, MO) was performed as described previously with some modifications (Guillas et al., 2000; Umemura et al., 2003).

Labeling of ceramide-containing proteins with [³H]DHS was performed after cells were grown to the exponential phase in SC medium. Washed cells were resuspended in fresh SC medium containing 40 µg/ml myriocin (Sigma-Aldrich, St. Louis, MO), incubated at 30°C for 20 min, and incubated with [³H]DHS for 2 h. The labeling reaction was stopped by adding cycloheximide, and cells were washed with 10 mM NaN₃. Labeled proteins were extracted by shaking with glass beads in TEPI buffer (100 mM Tris-HCl, pH 7.5, containing protease inhibitor cocktail; Roche, Mannheim, Germany), and solubilized by boiling in 1% SDS. Samples were combined with concanavalin A (Con A) buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM CaCl₂, 1 mM MnCl₂, and 1% Triton X-100) and centrifuged for 13,000 x g. Con A-Sepharose (GE Healthcare, Piscataway, NJ) was added to the supernatants, and the mixture was incubated overnight at 4°C to concentrate glycoproteins. The Sepharose beads were washed with Con A buffer and resuspended in SDS sample buffer. The samples were separated by SDS-polyacrylamide gel electrophoresis (PAGE) and analyzed using a Molecular Imager FX (Bio-Rad).

Preparation of Cell Lysate and Immunoblotting
Cells were grown to the exponential phase and collected by centrifugation. The cells were washed and broken using glass beads in TNPI buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, protease inhibitor cocktail [Roche], and 1 mM phenylmethylsulfonyl fluoride) at 4°C. Cell debris was removed by centrifugation. The cell extracts were then centrifuged at 13,000 x g to sediment the ER-rich fraction. The ER-rich pellet was solubilized in 1% Triton X-100 at 4°C for 30 min, and then it was centrifuged again at 13,000 x g. The supernatant was collected, mixed with SDS sample buffer, separated by SDS-PAGE, and analyzed by immunoblotting by using one of the following antibodies: anti-HA monoclonal antibody (mAb) 16B12 (1:2000; Cell Signaling Technology, Danvers, MA), anti-FLAG mAb M2 (1:10,000; Sigma-Aldrich), anti-Gas1 peptide polyclonal antibody (1:2000; kindly provided by Dr. Katsura Hata, Eisai, Tokyo, Japan), or anti-Dpm1p mAb (1:2000; Cell Signaling Technology).

Isolation of DRMs
Cells were grown at 30°C in YPAD to the exponential phase, and 3 x 10⁸ cells were collected. DRMs for the Gas1p analysis were isolated as described previously with a slight modification (Bagnat et al., 2000; Okamoto et al., 2006). After incubation with 1% Triton X-100 for 30 min on ice, the lysates were subjected to OptiPrep density gradient floatation by centrifugation for 2 h at 40,000 rpm in a SW55Ti rotor (Beckman Coulter, Fullerton, CA). After centrifugation, six fractions of equal volume were collected starting from the top. Each fraction was mixed with sample buffer and subjected to SDS-PAGE and immunoblotting.

Immunoprecipitation
Immunoprecipitation was performed as described previously with some modifications (Rayner and Munro, 1998). The cells were grown to the exponential phase in YPAD, and then they were collected by centrifugation and washed with water. The washed cells were resuspended in Tris-Sorb buffer (1 M sorbitol, 50 mM Tris-HCl, pH 7.5) containing Zymolyase 100T (Seikagaku, Tokyo, Japan), and then they were incubated at 30°C for 30 min. The resulting spheroplasts were washed in Tris-Sorb buffer and lysed at 4°C for 30 min by gentle mixing in TNPI buffer. Next, the lysate was solubilized with 1% digitonin. Cell debris was removed by centrifugation, and the supernatant was incubated overnight with anti-FLAG M2 affinity gel (Sigma-Aldrich) or anti-HA affinity matrix (Roche). The beads were washed with TNPI buffer containing 1% digitonin and suspended into TNPI buffer containing 3xFLAG peptide (Sigma-Aldrich) or HA peptide (Roche) to elute the protein. The eluted proteins were denatured with SDS sample buffer, separated by SDS-PAGE, and analyzed by immunoblotting.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
CWH43 Is a Homologue of Mammalian PGAP2 That Is Involved in GPI Lipid Remodeling
Recently, rat PGAP2 was reported to be a Golgi/ER resident membrane protein that seems to participate in the transfer of a saturated acyl chain to the sn-2 position of lyso-PI of the GPI anchor after the removal of the unsaturated acyl chain (Tashima et al., 2006). To further investigate lipid remodeling in yeast, we searched for the homologue of PGAP2 in S. cerevisiae by using BLAST (Altschul et al., 1997). The BLAST search revealed that the yeast homologue of PGAP2 is Cwh43p, a 953-amino acid protein that contains 18 potential membrane-spanning regions according to SOSUI analysis (Hirokawa et al., 1998). The amino acid sequence of human PGAP2 shares 24% homology with yeast Cwh43p; however, Cwh43p has an extra 700-amino acid C terminus containing several membrane-spanning regions. This additional C-terminal extra part of Cwh43p is 31% identical to the product of an uncharacterized human gene, FLJ21511 (Pittet and Conzelmann, 2007). Furthermore, it has been reported that CWH43 is involved in the cell wall integrity (Martin-Yken et al., 2001). Therefore, we expected that Cwh43p participates in GPI lipid remodeling but that it also has some additional functions that are not shared by mammalian PGAP2.

Cwh43p Is an ER Resident Membrane Protein
We first constructed C-terminally green fluorescent protein (GFP)-tagged or mRFP-tagged CWH43 expression plasmids to determine the intracellular location of Cwh43p. Both CWH43-GFP and CWH43-mRFP rescued the Calcofluor white (CFW) sensitivity of cwh43 gene-disrupted cells (cwh43), indicating that the tagged Cwh43 proteins were fully functional. Microscopic analysis indicated that Cwh43p-mRFP was mainly present in an intracellular perinuclear region with characteristics of the ER (Figure 1). Both Cwh43p-GFP and Cwh43p-mRFP were also localized at the intracellular perinuclear region. Moreover, according to the Saccharomyces Genome Database (SGD; http://www.yeastgenome.org/), GFP-fused Cwh43 protein is localized in the ER. Collectively, these results indicate that Cwh43p is localized in the ER. This is consistent with the previous report that Per1p and Gup1p, which are involved in GPI lipid remodeling, are also localized in the ER (Fujita et al., 2006a).

View larger version (79K): [in this window] [in a new window]   Figure 1. Cwh43p is localized in the ER. The localization of mRFP-tagged Cwh43p in cwh43 cells harboring pRS316T-CWH43-mRFP plasmid. The cells were cultured overnight at 30°C, and Cwh43p-mRFP was visualized by fluorescence microscopy. DIC, differential interference contrast.

CWH43 Is Involved in Lipid Remodeling of the GPI Anchor from a Diacylglycerol- to a Ceramide Type
To address whether CWH43 is also involved in the lipid remodeling of GPI anchor, we examined PI moieties obtained from GPI-anchored proteins of cwh43 cells. Cells were labeled with [³H]inositol, and the protein pellet was delipidated. The solubilized glycoproteins were then concentrated by Con A-Sepharose and digested with pronase. The PI moieties were liberated from GPI anchors by nitrous acid deamination, and the purified PI moieties were then separated by TLC and detected by autoradiography (Figure 2A) (Guillas et al., 2000). In wild-type cells, PI moieties of GPI anchor are mainly inositolphosphorylceramide/B (IPC/B), which consists of PHS and C26:0 fatty acid, and pG1, which consists of PI containing a C26:0 fatty acid at the sn-2 position. The PI moieties of the GPI anchor in gup1, per1, and gpi7 cells were used as controls to identify the PI species. As reported previously (Bosson et al., 2006), the gup1 cells, which have a defect in acylation of lyso-PI at the sn-2 position, accumulated lyso-PI (Figure 2A, lane 2), and the per1 cells, which have a defect in GPI-PLA₂, accumulated PI containing short-chain fatty acid (Fujita et al., 2006a; Figure 2A, lane 3). In gpi7 cells, lipid remodeling in Golgi is defective, resulting in a lack of IPC/C in the GPI anchor (Benachour et al., 1999). As reported, IPC/B was decreased, whereas pG1 was increased in gpi7 cells (Figure 2A, lane 5). The cwh43 cells accumulated pG1 and lyso-PI (Figure 2A, lane 6). We detected neither IPC/B nor IPC/C species of GPI anchor in cwh43 cells, suggesting that the cwh43 cells have a defect in the lipid remodeling of the GPI anchor from a diacylglycerol type to a ceramide type.

View larger version (56K): [in this window] [in a new window]   Figure 2. Lipid remodeling of lipid moieties in GPI anchors to ceramides is defective in cwh43 cells. (A) Analysis of PI moieties obtained from GPI-anchored proteins in wild-type (WT), per1, gup1, per1gup1, gpi7, and cwh43 cells. Cells were labeled with [3H]inositol at 30°C. Cell lysate was delipidated, and glycoproteins were concentrated using Con A-Sepharose. PI moieties of GPI anchors were released by deamination by using nitrous acid, separated by TLC by using solvent system 1, detected by autoradiography, and analyzed using a Molecular Imager FX. (B) Analysis of incorporation of [3H]DHS derivatives in proteins. Cells were incubated for 20 min in the presence of myriocin and then labeled with [3H]DHS at 30°C. Glycoproteins were concentrated from extracted lysate using Con A-Sepharose and separated by SDS-PAGE. Radiolabeled proteins were visualized using a Molecular Imager FX. Equal loading of total proteins was confirmed by Coomassie brilliant blue staining of the SDS-PAGE gels.

The accumulation of pG1 in cwh43 cells might be caused by a defect in sphingolipid biosynthesis. To check whether the sphingolipid biosynthesis is defective, we examined the accumulation of inositolphosphorylceramides (IPCs). The cwh43 cells have similar amounts of inositolphosphorylceramide, mannose inositolphosphorylceramide, and mannose diinositolphosphorylceramide as wild-type cells (unpublished data). This result suggests that the biosynthesis of ceramide and sphingolipid is normal in cwh43 cells and that the defect in the lipid remodeling of diacylglycerol to ceramide in cwh43 cells is not caused by a defect in ceramide biosynthesis.

We further analyzed the localization of a major GPI-anchored protein, Gas1p, to investigate whether GPI-anchored proteins have a defect in cwh43 cells. First, we compared the amount of Gas1p in the wild-type and cwh43 cells. Total cell lysates were examined by immunoblotting with an anti-Gas1 antibody. In contrast to wild-type cells, Gas1p was drastically reduced in per1 and gup1 cells, and, in turn, released to the medium (Supplemental Figure S1A), as reported previously (Fujita et al., 2006a). Alternatively, in cwh43 cells, the level of Gas1p in cell lysates was not decreased, and it was not released into the medium, indicating that Gas1p is almost retained normally in cwh43 cells. Next, we analyzed the cellular localization of mRFP-tagged Gas1p in wild-type and cwh43 cells by fluorescence microscopy. In cwh43 cells, mRFP-Gas1p was localized at the cell surface in the same manner as in wild-type cells (Supplemental Figure S1B), indicating that the localization of Gas1p is not affected in cwh43 cells.

It is reported that Gas1p is incorporated into lipid rafts at the plasma membrane (Bagnat et al., 2000). Therefore, we checked whether Gas1p is sorted to the DRM in cwh43 cells. In cwh43 cells, Pma1p, a marker for DRM-associated membrane protein, was associated with the DRMs, indicating that raft formation itself was normal (Supplemental Figure S1C). In the wild-type and cwh43 cells, Gas1p was also located predominantly in the DRM fraction, suggesting that Gas1p in cwh43 cells is located in lipid rafts as in wild-type cells, different from that in per1 or gup1 cells. It has been reported that the lipid moiety of GPI anchor of Gas1p is a diacylglycerol type containing a C26 fatty acid (Fankhauser et al., 1993). Therefore, it is possible that the lipid moiety of Gas1 could be normal and localization of Gas1p is not affected in cwh43 cells, consisting with the aforementioned results that CWH43 is involved in the lipid remodeling of GPI anchor to ceramides.

The C-Terminal Domain of Cwh43p Is Essential for Lipid Remodeling to Ceramide
Although the N-terminal region of Cwh43p (Cwh43-N) is homologous to mammalian PGAP2, Cwh43p has a large C-terminal region (Cwh43-C) that is similar to an uncharacterized gene product encoded by human FLJ21511. Therefore, Cwh43p corresponds to a fusion of PGAP2 and FLJ21511 proteins. To investigate which region of Cwh43p is important for its function, we constructed plasmids for expressing both CWH43-N (encoding amino acids 1-229; 229 amino acids) and CWH43-C (encoding amino acids 218-953; 736 amino acids). The proteins produced by these expression plasmids, Cwh43-N and Cwh43-C, are tagged with three copies of the HA epitope at their C termini. Furthermore, these expression plasmids were under control of the CWH43 promoter. These two plasmids were referred to as pRS316T-CWH43-N-HA and pRS316T-CWH43-C-HA, respectively. The expression plasmids were introduced into cwh43 cells, and the phenotypes of the cwh43 cells harboring them were characterized (Figure 3). The cwh43 cells harboring the empty vector were CFW and temperature sensitive, and these sensitivities were rescued by introduction of CWH43 gene. The cwh43 cells harboring the CWH43-N expression plasmid were also CFW and temperature sensitive. Although the CFW sensitivity of cwh43 cells was rescued by the CWH43-C expression plasmid alone or both the CWH43-C and CWH43-N expression plasmids, they did not rescue the temperature sensitivity of cwh43 cells (Figure 3). All cells were able to grow on YPAD plates containing 0.3 M KCl, indicating that cells with high temperature sensitivity also have a defect in the cell wall. These results suggest that CWH43-C partially suppresses the phenotype of cwh43 cells.

View larger version (42K): [in this window] [in a new window]   Figure 3. CWH43-C but not CWH43-N rescues the CFW sensitivity of cwh43 cells. CWH43-C partially suppresses the phenotype of cwh43 cells. Serial dilutions (4-fold) of cells were spotted on YPAD, YPAD containing 10 µg/ml CFW, or YPAD containing 0.3 M KCl and then incubated for 2 d at 30 or 39°C.

View larger version (42K): [in this window] [in a new window]   Figure 4. The C-terminal domain of Cwh43p is essential for lipid remodeling to ceramides. (A) Expression of CWH43-HA, CWH43-C-HA, and CWH43-N-HA was confirmed by immunoblotting by using an anti-HA antibody. The cwh43 cells harboring the CWH43-HA (CWH43), CWH43-N-HA (N), and CWH43-C-HA (C) expression plasmids, respectively, and cwh43 cells harboring both CWH43-N-HA and CWH43-C-HA expression plasmids (N+C) were cultured at 30°C. The lysates were extracted, and ER-rich fractions were concentrated by centrifugation at 13,000 x g and separated by SDS-PAGE. (B) Analysis of PI moieties obtained from GPI anchor in wild-type (WT), gpi7, and cwh43 cells harboring CWH43, CWH43-N (N), CWH43-C (C), and both CWH43-N and CWH43-C (N+C) expression plasmids. Cells were labeled with [3H]inositol. PI moieties obtained from delipidated glycoproteins by deamination were analyzed by TLC as described in Figure 2A. (C) Cells were labeled with [3H]DHS in the presence of myriocin. The incorporation of [3H]DHS derivatives into proteins in the same cells as shown in Figure 4B was analyzed as described in Figure 2B.

To confirm that CWH43-C is important for the remodeling of GPI anchors to ceramides, we further examined the incorporation of [³H]DHS derivatives into proteins in CWH43-C–expressing cells. The cells were labeled with [³H]DHS, and cell lysates were prepared. Glycoproteins were concentrated with Con A-Sepharose, and they were separated by SDS-PAGE. In cwh43 cells harboring the CWH43 expression plasmid, [³H]labeled DHS derivatives were incorporated into proteins, whereas the incorporation of ³H-labeled DHS derivatives was scarcely detected in cwh43 cells harboring the empty vector (Figure 4C). In contrast, in CWH43-C cells but not CWH43-N cells, ³H-labeled DHS derivatives were incorporated into proteins. Moreover, in CWH43-N/C cells, there was a higher level of incorporation than in CWH43-C cells. These results suggest that the C-terminal region of Cwh43p is important for GPI lipid remodeling from the diacylglycerol type to the ceramide type and that the N-terminal region of Cwh43p enhances ceramide formation by the C-terminal region. Therefore, we designated these N- and C-terminal regions of Cwh43p as Cwh43-N and Cwh43-C domains, respectively.

The Cwh43-N Domain Associates with the Cwh43-C Domain
Because CWH43-N enhanced the function of CWH43-C, we suspected that Cwh43-N protein associates with Cwh43-C protein. To test this possibility, we performed a coimmunoprecipitation assay. Cwh43-N was tagged at its C terminus with three copies of FLAG epitope (encoded by CWH43-N-FLAG gene). CWH43-N-FLAG and CWH43-C-HA were coexpressed from the CWH43 promoter in yeast cells, and the products were immunoprecipitated from total lysates solubilized by mild detergent treatment with digitonin. Cwh43-N-FLAG was immunoprecipitated with FLAG-beads, and Cwh43-C-HA was immunoprecipitated with HA-beads. The immunoprecipitates were separated by SDS-PAGE and examined by immunoblotting. Cwh43-N-FLAG was coprecipitated with Cwh43-C-HA and Cwh43-C-HA with Cwh43-N-FLAG when the two were coexpressed but not in control samples (Figure 5). These results demonstrate that domains Cwh43-N and Cwh43-C associate with each other.

View larger version (21K): [in this window] [in a new window]   Figure 5. Cwh43-N associates with Cwh43-C. The cwh43 cells introduced with both CWH43-N and CWH43-C plasmids (N+C), both CWH43-N plasmid and empty vector (N), both CWH43-C plasmid and empty vector (C), and two empty vectors (V) were cultured at 30°C. Total cell lysates solubilized with digitonin were immunoprecipitated using FLAG-beads or HA-beads. The immunoprecipitated samples were separated by SDS-PAGE and analyzed by immunoblotting using an anti-FLAG antibody or an anti-HA antibody.

View larger version (49K): [in this window] [in a new window]   Figure 6. The highly conserved region containing D862 and R882 is essential for the function of Cwh43-C. (A) Alignment of the amino acid sequences of the highly conserved N-terminal region of Cwh43p containing G57. The amino acid sequences of S. cerevisiae, Aspergillus fumigatus, Schizosaccharomyces pombe, Homo sapiens, and Mus musculus proteins were aligned using ClustalW. Asterisks (*) and colons (:) indicate identical and conserved amino acids, respectively. (B) Alignment of the amino acid sequences of the highly conserved C-terminal region of Cwh43p containing H472A, D693A, H770A/H771A, E807A, D862A, R882A. The amino acid sequences of S. cerevisiae, A. fumigatus, S. pombe, H. sapiens, and M. musculus proteins were aligned using ClustalW. (C) The production of Cwh43-HA, G57R mutant Cwh43-HA, Cwh43-C-HA, and H472A, D693A, H770A/H771A, E807A, D862A, and R882A mutants of Cwh43-C-HA in cwh43 cells was examined by immunoblotting by using anti-HA and anti-Dpm1p antibodies. Cells were cultured at 30°C, and cell lysates were extracted. ER-rich fractions were concentrated by centrifugation at 13,000 x g and separated by SDS-PAGE. Dpm1p was used to confirm equal loading. (D) Analysis of PI moieties obtained from GPI-anchored protein by deamination in the same cell as shown in Figure 6C. Cells were labeled with [3H]inositol, and PI moieties obtained from GPI-anchored proteins by deamination were analyzed as described in Figure 2A. (E) The same cells as in Figure 6C were analyzed for CFW sensitivity. The cells were cultured on YPAD and YPAD containing 10 µg/ml CFW at 30°C for 2 d as described in Figure 3. (F) Analysis of incorporation of [3H]DHS derivatives into proteins in the same cells as in Figure 6C. Cells were labeled with [3H]DHS in the presence of myriocin, and incorporation of [3H]DHS was analyzed as described in Figure 2B.

Next, we characterized the PI moieties obtained from the GPI-anchored proteins produced by these mutant cells (Figure 6D). Wild-type CWH43-C–introduced cwh43 cells accumulated a small amount of IPC/B in addition to pG1 and lyso-PI, whereas cwh43 cells harboring the empty vector accumulated pG1 and lyso-PI but not IPC/B (Figure 6D), as described above. Interestingly, in R882A mutant cwh43-C–introduced cwh43 cells, IPC/B lipid moieties were not detected in GPI anchor (Figure 6D, lane 9). Also, the amount of IPC/B was decreased in D862A mutant cwh43-C–introduced cwh43 cells (Figure 6D, lane 8), whereas the amount of IPC/B was not decreased in the other mutant cwh43-C–introduced cwh43 cells. Furthermore, D862A and R882A mutants of CWH43-C did not complement the CFW sensitivity of cwh43 cells, whereas the sensitivity was complemented by the wild type and H472A, D693A, H770A/H771A, and E807A mutants of CWH43-C (Figure 6E). We also checked whether GPI-anchored proteins are remodeled to ceramide, which can be detected by incorporation of ³H-labeled DHS derivatives into proteins in these mutant cells. In the cwh43 cells introduced with the R882A mutant CWH43-C, ³H-labeled signals were not detected in proteins, whereas ³H-labeled signals were observed in the other mutant and wild-type CWH43-C–introduced cwh43 cells (Figure 6F). The ³H-labeled DHS derivatives were incorporated into proteins in D862A mutant CWH43-C–introduced cwh43 cells, implying that these mutant cells had a detectable amount of IPC/B in the PI moiety. These results suggest that R882 is an important amino acid residue for the lipid remodeling function and that the conserved region containing D862 and R882 is important for the function of Cwh43p, although the detailed mechanism is not clear.

The Mouse CWH43 Counterparts FRAG1 and C130090K23 Complement CWH43-deficient Yeast Phenotypes
According to a BLAST search, the mouse homologues of human PGAP2 and FLJ21511 are FRAG1 and C130090K23, respectively. The predicted amino acid sequences of mouse FRAG1 and C130090K23 are 29 and 31% identical to yeast Cwh43p, respectively. To address whether mouse C130090K23 is a functional homologue of CWH43-C, we introduced expression plasmids of FRAG1 and C130090K23 into the cwh43 cells and examined CFW sensitivity and PI moieties obtained from GPI-anchored proteins.

The C130090K23 gene rescued the CFW sensitivity of cwh43, but the FRAG1 gene did not (Figure 7A). This is similar to the results for CWH43-C– and CWH43-N–introduced cwh43 cells, respectively (Figure 3). These results suggest that C130090K23 but not FRAG1 rescues the cell wall defect in cwh43 cells.

View larger version (75K): [in this window] [in a new window]   Figure 7. The mouse C130090K23 gene suppresses the phenotype of yeast cwh43 cells. (A) The mouse C130090K23 gene rescues the CFW sensitivity of cwh43 cells. cwh43 cells harboring CWH43, the empty vector (vector), FRAG1, C130090K23 (C1300), both FRAG1 and C130090K23 (FRAG1+C1300), CWH43-C, and CWH43-N expression plasmids were cultured on YPAD and YPAD containing CFW at 30°C for 2 d as described in Figure 3. (B) The production of Cwh43-HA (CWH43), C130090K23-HA (C1300), FRAG1-HA (FRAG1), and both FRAG1-HA and C130090K23-HA (FRAG1+C1300) in cwh43 cells was confirmed by immunoblotting by using an anti-HA antibody as described in Figure 4A. (C) Mouse C130090K23 recovers the function of GPI lipid remodeling to ceramides. PI moieties obtained from GPI-anchored proteins by deamination were analyzed in cwh43 cells harboring CWH43, the empty vector, FRAG1, C130090K23 (C1300), and both FRAG1 and C130090K23 (FRAG1+C1300) as described in Figure 2A. The results from two independent experiments are shown in left (lanes 1–5) and right (lanes 6–10) panels, respectively.

Next, we analyzed the lipid moieties of GPI-anchored proteins in cwh43 cells harboring C130090K23 and FRAG1. The cells were labeled with [³H]inositol, and the labeled glycoproteins were concentrated by Con A-Sepharose and digested with pronase. The PI moieties were obtained by deamination and were separated by TLC. In cwh43 cells harboring the CWH43 gene, the PI moiety was mainly IPC/B, which consists of PHS and C26 fatty acid (Figure 7C, lanes 1 and 7), and in the cwh43 cells harboring the empty vector, the PI moieties were pG1, a PI with C26 fatty acid in the sn-2 position, and lyso-PI (Figure 7C, lanes 2 and 6), as described above. The cwh43 cells harboring FRAG1 accumulated pG1 and lyso-PI like the cwh43 cells harboring the empty vector (Figure 7C, lanes 3 and 8). In contrast, the cwh43 cells harboring C130090K23 accumulated a small but significant amount of IPC/B in addition to pG1 and lyso-PI (Figure 7C, lanes 4 and 9). Moreover, the cwh43 cells coexpressing FRAG1 and C130090K23 accumulated not only IPC/B but also IPC/C, which consists of PHS and a hydroxylated C26:0 fatty acid (Figure 7C, lanes 5 and 10). The level of IPC in cwh43 cells coexpressing FRAG1 and C130090K23 was higher than that of cwh43 cells expressing C130090K23 alone, as observed in cwh43 cells coexpressing CWH43-N and CWH43-C. These results suggest that the mouse C130090K23 protein can remodel lipids of PI moieties in GPI-anchored proteins from diacylglycerol type to ceramides and that the independent mouse FRAG1 protein is able to enhance lipid remodeling to ceramides by C130090K23.

Because Cwh43-N and Cwh43-C associated with each other, we further investigated whether FRAG1 is coimmunoprecipitated with C130090K23. FRAG1 was tagged with three copies of the HA epitope at its C terminus (FRAG1-HA), and C130090K23 was tagged with three copies of FLAG epitope at its C terminus (C130090K23-FLAG). The tagged mouse genes were coexpressed under the constitutive TDH3 promoter and introduced into cwh43 cells. The digitonin-solubilized total lysates of cwh43 cells coexpressing FRAG1-HA and C130090K23-FLAG were extracted, and FRAG1-HA and C130090K23-FLAG proteins were immunoprecipitated by HA-beads and FLAG-beads, respectively. FRAG1-HA coprecipitated with C130090K23-FLAG in lysates containing C130090K23-FLAG but not in control lysates (Figure 8A). C130090K23-FLAG also coprecipitated with FRAG1-HA in lysates containing FRAG1-HA. Because interaction of FRAG1-HA and C130090K23-FLAG are shown in low efficiency, we further tried the immunoprecipitation by using FRAG1-FLAG and C130090K23-HA. FRAG1-FLAG was coprecipitated with C130090K23-HA and C130090K23-HA with FRAG1-FLAG when the two were coexpressed but not in control sample. These results indicate that, like Cwh43-N and Cwh43-C, FRAG1 and C130090K23 associate with each other.

View larger version (23K): [in this window] [in a new window]   Figure 8. Mouse FRAG1 and C130090K23 proteins associate with each other. (A) The cwh43 cells harboring FRAG1-HA and C130090K23-FLAG expression plasmids or the empty vector were cultured at 30°C. Total cell lysates were extracted with digitonin, immunoprecipitated, and analyzed by immunoblotting as described in Figure 5. (B) cwh43 cells harboring FRAG1-FLAG and C130090K23-HA expression plasmids or the empty vector were used for immunoprecipitation as described in Figure 8A. We think that the upper bands of FRAG1-FLAG (*) may be dimmer form and the lower bands of FRAG1-FLAG (**) may be degradation form. (C) cwh43 cells coexpressed ALG14-HA with FRAG1-FLAG, C130090K23-FLAG and ALG13-FLAG were cultured at 30°C. Total lysates were extracted, immunoprecipitated using FLAG-beads, and analyzed immunoblotting with anti-FLAG, anti-HA, and anti-Dpm1p antibodies.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The lipid remodeling of yeast GPI-anchored protein begins with the removal of an unsaturated fatty acyl chain from diacylglycerol by a GPI-PLA₂ activity. Per1p is involved in this reaction (Fujita et al., 2006a). Next, a saturated long fatty acyl chain is transferred to lyso-GPI by the O-acyltransferase Gup1p (Bosson et al., 2006), generating diacylglycerol containing a C26 fatty acid. Most mature GPI-anchored proteins, however, contain ceramides in their lipid moieties. Our analysis of the lipid moiety obtained from GPI-anchored proteins by deamination showed that cwh43 cells accumulate pG1, which is a PI containing a saturated C26 fatty acid, and lyso-PI but not IPCs. Therefore, it is conceivable that CWH43 is involved in the remodeling of GPI-anchored proteins from the diacylglycerol type to the ceramide type.

On the basis of our results and previous reports (Sipos et al., 1994; Bosson et al., 2006; Fujita et al., 2006a), we speculated that there are two possibilities for the remodeling pathway of GPI lipids to ceramides (Figure 9). One possibility may be a sequential pathway and the other may be diverged pathway. In former case [Figure 9, (1)] an unsaturated fatty acid is removed from diacylglycerol of conventional PI to generate lyso-PI-type GPI anchor (lyso-PI). Per1p is involved in this process. Next, a C26 saturated fatty acid is transferred to lyso-PI by the O-acyltransferase Gup1p, generating PI with diacylglycerol containing a C26 saturated fatty acid at the sn-2 position (pG1). Diacylglycerol containing a C26 saturated fatty acid may then be converted to a ceramide moiety (IPC). Cwh43p may be involved in this step because the GPI-anchored proteins from cwh43 cells accumulated pG1-type PI. Alternatively, the pathway may diverge. In this case, the lyso-PI generated by Per1p could be used as a substrate for two reactions executed by Gup1p and Cwh43p [Figure 9, (2)]. In one of these reactions, C26 saturated fatty acid is transferred to lyso-PI by Gup1p to generate diacylglycerol containing a C26 fatty acid (pG1). In the other reaction, lyso-PI may be substituted directly for ceramide moiety (IPC), a step that may involve Cwh43p. In this case, the pathway where lyso-PI is transferred with C26 fatty acid to generate diacylglycerol might be enhanced when the conversion to ceramides is defective. A third additional pathway may also exist as an option of both sequential and diverged pathways, as shown by dotted lines in Figure 9 (3). Diacylglycerol with an unsaturated fatty acid in conventional PI is directly exchanged for a ceramide moiety, together with the pathway mediated by Per1p and Gup1p. Cwh43p could be involved in this direct exchange of unsaturated fatty acid for a ceramide moiety because GPI-anchored proteins from per1gup1 double-mutant cells accumulate IPC/C in addition to PI (Fujita et al., 2006a). This third alternative may operate as a bypass or as an emergency route for exchange when there is a defect in the steps mediated by Per1p and Gup1p. Identification of a direct substrate for Cwh43p should help determine the detailed molecular mechanism of lipid remodeling in GPI anchors to ceramides.

View larger version (30K): [in this window] [in a new window]   Figure 9. The hypothetical lipid remodeling pathways of GPI-anchored proteins in yeast. These three pathways [(1), (2), and (3)] of GPI lipid remodeling are proposed based on our results and previous reports (Sipos et al., 1997; Bosson et al., 2006; Fujita et al., 2006a), as described in Discussion. The first possible pathway (1) is sequential, and the second pathway (2) is diverged, and there also might be a bypass (3) for these pathways. Our current studies indicate that Cwh43p is required to form IPC of lipid moiety in GPI-anchored proteins.

Moreover, the cwh43 cells introduced with CWH43-C are able to produce GPI anchor containing IPC/B, suggesting that C-terminal domain is essential for the ability to remodel GPI lipids ceramides. The function of CWH43-C was enhanced when coexpressed with CWH43-N. Furthermore, the G57R mutant of Cwh43p was unstable, and a much smaller amount of GPI-anchor–derived IPC/B was detected in the cwh43 cells expressing the G57R mutant of CWH43 than in cwh43 cells expressing wild-type CWH43, indicating that mutation of the N-terminal domain causes the instability of the Cwh43 protein. These results suggest that the C-terminal region of Cwh43p is an essential functional domain and that the N-terminal region, which is homologous to PGAP2, supports the function and/or stability of the C-terminal domain.

Rat PGAP2 seems to be identical to FRAG1 (Lorenzi et al., 1996), which was isolated from a rat osteosarcoma cell line as a fusion with fibroblast growth factor receptor 2 that has constitutively elevated tyrosine kinase activity due to a chromosomal rearrangement. Recently, rat PGAP2, which is a Golgi/ER-resident membrane protein and conserved in many eukaryotes, was reported to be involved in lipid remodeling from lyso-PI to PI containing saturated fatty acids (Tashima et al., 2006; Maeda et al., 2007). PGAP2 does not have any known motifs (Tashima et al., 2006; Maeda et al., 2007). Therefore, it is possible that, like the N-terminal domain of Cwh43p, PGAP2 interacts with an O-acyltransferase and that it supports the stability of the enzyme or increases its catalytic activity.

The gene products of mouse FRAG1 and C130090K23 are homologues of the N- and C-terminal domains of Cwh43p, respectively. Recombinant mouse C130090K23 protein can remodel the lipid moiety of GPI anchor in yeast cells to ceramides. When both mouse FRAG1 and C130090K23 were coexpressed in yeast cells, the gene products coimmunoprecipitated, suggesting that they physically interact with each other, and the coexpression of mouse FRAG1 enhanced lipid remodeling by C130090K23. This is consistent with the findings in cwh43 cells introduced with both CWH43-N and CWH43-C. Because the amino acid sequence of mouse FRAG1 is 97% identical to that of rat PGAP2, mouse FRAG1 is presumably the functional homologue of rat PGAP2, which is involved in the remodeling of lipid from lyso-PI to diacylglycerol containing saturated fatty acids (Tashima et al., 2006; Maeda et al., 2007). It is possible that mouse FRAG1 protein not only enhances lipid remodeling to ceramides but also the O-acyltransferase activity. Thus, further studies should examine whether mouse FRAG1 and human/rat PGAP2 enhance O-acyltransferase activity.

It is noteworthy that the mouse C130090K23 gene product can remodel lipids to ceramides and that C130090K23 can rescue CFW sensitivity in cwh43 cells. Therefore, C130090K23 protein is a functional homologue of Cwh43-C. Although ceramides are found in yeast, fungi, parasite, Dictyostelium, and certain plants (e.g., pear) (Fankhauser et al., 1993; Haynes et al., 1993; Ferguson et al., 1999; Oxley and Bacic, 1999; Fontaine et al., 2003; Pittet and Conzelmann, 2007), they have not been found in the lipid moieties of mammalian GPI-anchored proteins. Thus, it is possible that in some specific tissues where C130090K23 is produced, FRAG1 and C130090K23 associate with each other and participate in the formation of ceramide-type lipid moieties in GPI-anchored proteins. In contrast, in mammalian cells, alkyl/acyl PI moieties are selectively concentrated during the GPI biosynthesis whereas endogenous PI, which is used for the start of GPI biosynthesis, is generally diacyl species (Houjou et al., 2007). Although it is unknown what genes are involved in these events and how reactions occurred, it is also possible that C130090K23 might be involved in the uncharacterized lipid remodeling reaction as described above in mammalian cells.

According to SGD, Cwh43p has a DNase I-like region at its C terminus. Isc1p, Inp51p, Inp52p, Inp53p, and Inp54p also have this DNase I-like region. Isc1p is an inositol phosphosphingolipid phospholipase C (Sawai et al., 2000), and Inp51/52/53/54 proteins are involved in inositol 5-phosphatase activity (Stolz et al., 1998a,b; Guo et al., 1999; Wiradjaja et al., 2001). Therefore, this motif may be related to the recognition of inositol phosphate, implying that a DNase I-like region in the C-terminal domain of Cwh43p is important for the recognition of inositol phosphate in the GPI anchor. However, it is still unclear whether the conserved D862 and R882 residues, which are likely to be important for its function but are not in the DNase I-like region, are directly involved in the recognition of inositol phosphate.

In this study, we identified how Cwh43p participates in the lipid remodeling of diacylglycerol to ceramides in GPI-anchored proteins; however, how diacylglycerol is replaced with ceramides and what substrate is used for the replacement remains unclear. Future investigations of Cwh43p-mediated lipid remodeling should clarify the molecular mechanism by which the lipid moiety of GPI anchors is replaced with ceramides and its precise biological function.

	ACKNOWLEDGMENTS

 
We are grateful to Ryo Taguchi and Andreas Conzelmann for helpful discussion. We also thank Katsura Hata for providing the anti-Gas1 peptide polyclonal antibody and Roger Tsien for providing the mRFP plasmid. Finally, we thank Michiyo Okamoto and Hiroto Hirayama for stimula