PIG-V involved in transferring the second mannose in glycosylphosphatidylinositol.

Glycosylphosphatidylinositol (GPI) is a glycolipid that anchors many proteins to the eukaryotic cell surface. The biosynthetic pathway of GPI is mediated by sequential additions of sugars and other components to phosphatidylinositol. Four mannoses in the GPI are transferred from dolichol-phosphate-mannose (Dol-P-Man) and are linked through different glycosidic linkages. Therefore, four Dol-P-Man-dependent mannosyltransferases, GPI-MT-I, -MT-II, -MT-III, and -MT-IV for the first, second, third, and fourth mannoses, respectively, are required for generation of GPI. GPI-MT-I (PIG-M), GPI-MT-III (PIG-B), and GPI-MT-IV (SMP3) were previously reported, but GPI-MT-II remains to be identified. Here we report the cloning of PIG-V involved in transferring the second mannose in the GPI anchor. Human PIG-V encodes a 493-amino acid, endoplasmic reticulum (ER) resident protein with eight putative transmembrane regions. Saccharomyces cerevisiae protein encoded in open reading frame YBR004c, which we termed GPI18, has 25% amino acid identity to human PIG-V. Viability of the yeast gpi18 deletion mutant was restored by human PIG-V cDNA. PIG-V has two functionally important conserved regions facing the ER lumen. Taken together, we suggest that PIG-V is the second mannosyltransferase in GPI anchor biosynthesis.

Glycosylphosphatidylinositol (GPI) 1 is a complex glycolipid that acts as a membrane anchor for many proteins in eukaryotic cells (1). A GPI anchor is important for protein sorting in epithelial cells (2), signal transduction (3), immune responses (4), and the pathobiology of infectious diseases (5,6). Further-more, GPI anchor is essential for the viability of Saccharomyces cerevisiae (7,8) and the bloodstream form of Trypanosoma brucei (9). In mammals, a lack of GPI causes embryonic lethality (10); however, GPI is not essential at a single cell level (11). Paroxysmal nocturnal hemoglobinuria is an acquired hematopoietic stem cell disorder due to a lack of GPI-anchored proteins, in subpopulations of blood cells (12).
The biosynthetic pathway of GPI is mediated by the sequential additions of sugars and other components to phosphatidylinositol (PI). The core structure of a GPI anchor, EtNP-6Man␣132Man␣136Man␣134GlcN␣136-inositolphospholipid (where EtNP, Man, and GlcN are ethanolamine phosphate, mannose, and glucosamine, respectively), is conserved in all eukaryotic cells (1). GPI pre-assembled in the ER is transferred en bloc to proteins bearing the GPI attachment signal sequence at the C terminus. The biosynthetic pathway for GPI is generally conserved among eukaryotic organisms, but a number of significant differences have been found (13).
Three mannoses in the common core and the fourth mannose found in yeast and some mammalian GPIs are transferred from dolichol-phosphate-mannose (Dol-P-Man) and are linked through different glycosidic linkages. The transfer of the first, ␣1,4-mannose to GlcN-acyl-PI is mediated by GPI-MT-I known as PIG-M in mammals or Gpi14p in S. cerevisiae. The mammalian PIG-M has multiple transmembrane domains and a functionally important DXD motif, a characteristic of many glycosyltransferases, within the first ER luminal domain (14). The second mannose, which is the topic of this report, is added by GPI-MT-II, an ␣1,6-mannosyltransferase. Mammalian PIG-B and its S. cerevisiae orthologue Gpi10p transfer the third, ␣1,2-mannose to Man 2 -GlcN-acyl-PI (GPI-MT-III (15)). PIG-B is an ER protein with multiple transmembrane domains. The ␣1,2-linked fourth mannose is transferred by Smp3p in S. cerevisiae and hSMP3 in humans (16,17).
Among these mannosyltransferases, PIG-B/Gpi10p and hSMP3/Smp3p are related to each other and, together with Alg9p and Alg12p, both of which are Dol-P-Man-dependent ER mannosyltransferases involved in N-glycan synthesis, form a protein family. Further, other Dol-P-monosaccharide-utilizing mannosyltransferases and glucosyltransferases involved in Nglycan synthesis and protein O-mannosylation have a common multi-transmembrane topology (18,19). Many of them have conserved amino acid residues implicated in catalytic sites within the first large and conserved loop on the luminal side (20). It was also reported that a large luminally oriented hydrophilic loop is essential for the function of protein O-mannosyltransferase Pmt1p (18). It appears that these enzymes belong to a superfamily of ER-resident, Dol-Pmannosaccharide-utilizing glycosyltransferases (20).
Aerolysin is a cytolytic toxin secreted by the Gram-negative * This work was supported in part by grants from the Ministry of Education, Science, Sports, Culture and Technology of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AB196341.
Here, using aerolysin, we report the isolation of GPI-MT-II-defective CHO cells termed class V mutant cells, and the cloning of the responsible gene, PIG-V. We also report that yeast S. cerevisiae Gpi18p is a functional homologue of PIG-V. Human PIG-V consists of 493 amino acids, bears multiple transmembrane domains, and has two functionally important highly conserved regions in the ER lumen; notably, the first luminal conserved region being compatible with a characteristic of many members in the superfamily of the ER resident glycosyltransferases.

EXPERIMENTAL PROCEDURES
GPI-anchored Protein-deficient Cells and Other Cells-A parental CHO cell line D9WT for the isolation of GPI-deficient mutants was generated in a similar way as described previously (24). We stably transfected PIG-L, DPM2, SL15, PIG-A, PIG-O, PIG-U, GnTI, and PIG-F cDNAs (to avoid the isolation of known GPI-deficient mutants) into CHO K1 IIIB2A cells that stably expressed CD59 and DAF as marker GPI-anchored proteins (27). D9WT cells were cultured in Ham's F-12 medium containing 10% fetal calf serum, 600 g/ml G418, 200 g/ml hygromycin, 5 g/ml puromycin, and 25 g/ml blasticidin to ensure the maintenance of the cDNAs. For mutagenesis, D9WT cells (1 ϫ 10 7 cells in a 15-cm dish) were treated with 400 g/ml ethylmethyl sulfonate (Sigma) for 48 h and cultured for 4 more days. They were then treated with 3 nM proaerolysin (Protox Biotech, Victoria, Canada) for 2 days, washed, and cultured for 2 days. Surviving cells were retreated with 5 nM proaerolysin for 1 day and cloned by limiting dilution. CHO K1 cells and CHO (U) cells defective in the PIG-U gene (28) were cultured in Ham's F-12 medium containing 10% fetal calf serum. JY25 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum.
Expression Cloning of Rat PIG-V-We used a rat C6 glioma cDNA library made with a mammalian expression vector, pME, bearing the polyoma virus origin for replication (27). We transfected 1.44 ϫ 10 8 CHO D9PA15.6 cells (see "Results" for characterization) with 480 g each of the library plasmid and pcDNA-PyT(ori-) plasmid for the expression of the polyoma large T by electroporation at 360 V and 960 microfarads. Two days later, transfected cells were stained with biotinylated anti-CD59 plus phycoerythrin-streptavidin, and sorted with a FACS-Vantage (BD Biosciences). Plasmids were recovered in Escherichia coli from the sorted CD59-positive cells, amplified, and transfected again with pcDNA-PyT(ori-) into the same mutant cells. CD59positive cells were sorted again. Each colony of E. coli transformed with the recovered plasmids was individually transferred into a well of 96-well plates. Plasmids that could restore CD59 expression on the CHO D9PA15.6 cells were selected and sequenced.
Metabolic Labeling of Cells and Analysis of GPI-Metabolic labeling with D-[2-3 H]mannose and lipid extraction were performed as described previously (29). Glycolipids were separated by thin layer chromatography on Kiesel gel 60 (Merck, Darmstadt, Germany) with a solvent system of chloroform:methanol:water (10:10:3) and detected by a Fuji BAS1500 image analyzer (Fuji Film Co., Tokyo, Japan) or x-ray film. To study the effect of YW3548/BE49385A on GPI synthesis, cells were incubated overnight in a medium containing 10 M BE49385A (a gift from Banyu Pharmaceutical Co., Tokyo, Japan), followed by metabolic labeling with [ 3 H]mannose and lipid extraction, as described above. In some cases, the glycolipids were treated with Jack bean ␣-mannosidase (Sigma) before TLC analysis.
Plasmids-The full-length human PIG-V cDNA was identified in the GenBank TM data base (accession number AK000484) and was provided from the NEDO Human cDNA Sequencing Project (a gift of Dr. S. Sugano, University of Tokyo, Japan). To add an epitope tag at the N terminus of human PIG-V, we amplified human PIG-V cDNA with forward and reverse primers having SalI and NotI sites, respectively; 5Ј-CCTGGTGGTGTCGACTGGCCCCAGGACCCATCCCGGAAG-3Ј and 5Ј-GTCCCTGGAGGCGGCCGCTCATGTCCAAGGCAGGAAGTT-GCAATGTAG-3Ј. The amplified PCR fragment was cut with SalI and NotI, subcloned into SalI-and NotI-digested pME-3HSV to generate pME-3HSV-hPIG-V, in which the triple HSV tag was linked to the N terminus of PIG-V. The same insert sequence was ligated into pME-GST and pMEEB-GST-FLAG to generate pME-GST-hPIG-V and pMEEB-GST-FLAG-hPIG-V, respectively. Site-directed mutants of PIG-V were generated using an oligonucleotide-directed mutagenesis method.
Subcellular Localization of PIG-V-JY25 cells were stably transfected with a cDNA of GST-tagged PIG-V. Transfected cells (3 ϫ 10 8 ) were lysed by a Teflon homogenizer in a solution consisting of 0.25 M sucrose, 10 mM HEPES/NaOH buffer (pH 7.4), 1 mM dithiothreitol, 0.1 mM 1-chloro-3-tosylamido-7-amino-2-heptanone, and 1 g/ml leupeptin. After treatment with 1 g/ml DNase I for 20 min on ice, unbroken cells and nuclei were removed by centrifugation at 1000 ϫ g for 10 min. The supernatant was further separated into pellet (P10) and supernatant by centrifugation at 10,000 ϫ g for 10 min. Supernatant after 10,000 ϫ g centrifugation was centrifuged at 100,000 ϫ g for 1 h and separated into pellet (P100) and supernatant (S100). The pellet was dissolved in 1% Nonidet P-40 and 150 mM NaCl. Nonidet P-40 was also added to S100 to 1%. GST-tagged PIG-V was precipitated with glutathione-Sepharose beads and Western-blotted with anti-GST antibody. In some experiments, the P100 fraction was further separated with sucrose density gradient centrifugation as described (30). Membranes were collected from subfractions by centrifugation at 100,000 ϫ g for 2 h and dissolved in 1% Nonidet P-40 and 150 mM NaCl. GST-tagged PIG-V was assessed as above. Subfractions were characterized by assaying membrane marker enzymes, alkaline phosphodiesterase I for the plasma membrane, ␣-mannosidase II for the Golgi, and Dol-P-Man synthase for the ER, as described (30).
Membrane Topology of PIG-V Protein-We transfected pMEEB-GST-FLAG-hPIG-V into CHO K1 cells, selected, and maintained in a medium containing 200 g/ml hygromycin. Cells were cultured on 14-mm diameter glass coverslips for 1 day, permeabilized, and stained as described previously (31). In brief, they were washed with PBS containing Ca 2ϩ and Mg 2ϩ (PBS(ϩ)), fixed in 4% paraformaldehyde in PBS for 15 min at room temperature, again washed with PBS(ϩ), and incubated for 20 min in 50 mM NH 4 Cl to quench residual paraformaldehyde. To permeabilize the plasma membrane selectively, the cells were incubated in a buffer consisting of 0.0005% digitonin, 0.3 M sucrose, 0.1 M KCl, 2.5 mM MgCl 2 , 1 mM EDTA, and 10 mM HEPES/NaOH buffer (pH 6.9), for 15 min at 4°C. Thereafter, cells were washed three times with PBS(ϩ) and blocked with 1% bovine serum albumin in PBS for 30 min at room temperature. For complete permeabilization, 0.1% Triton X-100 was added to the blocking solution. Cells were then incubated with 2.5 g/ml goat anti-GST antibody (Amersham Biosciences) or 2.5 g/ml rabbit anti-BiP antibody (ABR, Inc.) in 0.1% bovine serum albumin in PBS for 2 h. After washing four times in PBS, cells were incubated with fluorescein isothiocyanate-conjugated donkey anti-goat IgG or rhodamine-conjugated donkey anti-rabbit IgG antibodies (Chemicon International) in 0.1% bovine serum albumin in PBS for 1 h. Slides were mounted in Mowiol and studied under a confocal laser scanning microscope (Bio-Rad).
Transfection of Human PIG-V into S. cerevisiae-To construct a plasmid for expressing human PIG-V in yeast, we amplified two fragments of the 310-bp upstream region and the 120-bp downstream region of the ORF YBR004c/GPI18 from a S. cerevisiae genomic library by PCR, and digested PCR products with KpnI plus XhoI and with XbaI plus SacI, respectively. Human PIG-V cDNA fragments were obtained from pME-HSV-hPIG-V by digestion with XhoI and XbaI. These three fragments were ligated and subcloned into pRS316 between KpnI and SacI to generate pPIG-V. The heterozygous gpi18 (YBR004c) deletion strain BY4743 MATa/␣ his3⌬1/his3⌬1 leu2⌬0/lue2⌬0 ura3⌬0/ura3⌬0 lys2⌬0/met15⌬0ybr004c::KAN R /YBR004c generated by the Saccharomyces Gene Deletion Project was obtained from ResGen TM . We transformed the heterozygous GPI18/gpi18::KAN R diploid with pPIG-V containing URA3 and selected with SD medium without uracil. Transformed diploid cells were sporulated, and complementation was assessed by tetrad analysis. Four viable haploids from a tetrad were grown on a YPD (1% yeast extract, 2% peptone, 2% glucose) agar plate, a YPD plate containing G418 (200 g/ml), and a YPD plate containing 5-fluoroorotic acid.

Isolation of Novel Class Mutants Defective in GPI Biosynthe
sis-To screen mutant CHO cells defective in GPI biosynthesis, we used aerolysin, a GPI-recognizing pore-forming toxin produced by Aeromonas hydrophila (24,25). We chemically mutagenized the parental CHO D9WT cells, which were stably transfected with human CD59 and DAF as markers of GPIanchored proteins (Fig. 1A), and with cDNAs of known genes involved in GPI biosynthesis, to avoid generation of known complementation classes of GPI-deficient mutants. After treatment with 5 M proaerolysin, three aerolysin-resistant cells termed D9PA15.6, -25.2, and -72.1 were obtained (D9PA GPI(Ϫ) cells). As expected, these three mutant cell lines were deficient in the surface expression of CD59 and DAF. D9PA15.6 and D9PA72.1 cells were almost completely deficient in CD59 expression and had only 1% of the normal level of DAF (Fig. 1, C and E). D9PA25.2 cells showed heterogeneous expression of CD59 and DAF (Fig. 1D). Because the defective phenotypes of these mutants were not restored by transfection of any known cDNAs involved in GPI biosynthesis, we considered that they belong to a new class of mutant.

D9PA GPI(Ϫ) Cells Are Defective in Transfer of the Second
Mannose-To determine a step in GPI biosynthesis at which D9PAGPI(Ϫ) cells are defective, we analyzed the biosynthesis of GPI by metabolically labeling the cells with [ 3 H]mannose ( Fig. 2A). As a control, we used GPI transamidase-defective PIG-U mutant cells, C311PA16 (25), which accumulated various GPI intermediates (lane 1 in Fig. 2, A-C). All D9PAGPI (Ϫ) cells accumulated two GPI intermediates H2 and H5 ( Fig. 2A,  lanes 2-4), suggesting that they belong to the same class of mutants.
To confirm that the accumulated GPI intermediates are indeed H2 and H5, we used YW3548/BE49385A and Jack bean ␣-mannosidase. YW3548/BE49385A (a terpenoid lactone from fungi) inhibits PIG-N that transfers the EtNP-side chain to the first mannose (32)(33)(34). By treatment with YW3548/BE49385A, the spot with the lower R F value disappeared, whereas that with the higher R F value remained, confirming that the lower R F spot was H5, which has one mannose with EtNP modification (Fig. 2B, lanes 3 and 4). When these intermediates were treated with Jack bean ␣-mannosidase, only the one with the higher R F was sensitive, indicating that it was H2 having one mannose without modification (Fig. 2C, lanes 3 and 4). Therefore, these mutants are defective in the second mannosyltransferase (GPI-MT-II). We termed these mutant cells as class V.

PIG-V cDNA Restores the Surface Expression of GPI-anchored Proteins on D9PAGPI(Ϫ)
Cells-To obtain the gene responsible for class V mutant cells, termed PIG-V for phosphatidyl inositol glycan class V, we transfected a rat cDNA expression library into D9PA15.6 cells, collected CD59-positive cells using a cell sorter, and rescued the plasmids. One plasmid containing the 2429-bp cDNA fragment restored the surface expression of CD59 on D9PA15.6 cells as well as D9PA25.2 and D9PA72.1 after transfection (data not shown) (accession number AB196341). Based on the sequence homology, we cloned human PIG-V cDNA (AK000484), which consists of 2080 bp, and stably transfected it into D9PA15.6 cells. The human PIG-V cDNA recovered the surface expression of CD59 on D9PA15.6 cells (Fig. 3A) and normalized the profile of GPI mannolipids (Fig. 3B). Fig.  2A. H2 and H5 have one mannose without and with an EtNP side chain, respectively. One of the unsolved issues in mammalian GPI biosynthesis pathway is whether H2 or H5 is the substrate of GPI-MT-II. To address this, we transiently trans-fected PIG-V cDNA or mock vector into PIG-U-defective C311PA16 cells, which accumulate various GPI intermediates, including H2 and H5. Two days after transfection, cells were metabolically labeled with [ 3 H]mannose, and GPI intermediates were analyzed by TLC.

Overexpression of PIG-V in the PIG-U-defective Cells Increased the Amount of H3, a GPI with Two Mannoses-PIG-Vdefective CHO cells accumulated H2 and H5 as shown in
If H2 is the main substrate, it would be decreased and H3 would be increased after PIG-V transfection. Alternatively, if H5 is the substrate, it would be decreased, and B and H6 would be increased. The mock transfected cells accumulated various GPIs (H2-H7) (Fig. 3C, lanes 1 and 3) (25). When cells were transfected with PIG-V, the amount of H2 was decreased from 44% of total GPIs to 29% with a concomitant increase of H3 from 13% to 24% (lanes 2 and 4). Because this is the biggest change after PIG-V transfection, it is suggested that PIG-V mainly mediates the H2 to H3 conversion, i.e. H2 is the major substrate of GPI-MT-II (Fig. 3D). H5 was also decreased in the PIG-V overexpressed cells, but there was no increase of B, and H6 increased only slightly (12.7% to 13.4%). A decrease of H5 was, therefore, possibly due to the decreased level of H2. We propose that the H2 to H3 conversion is the major path, that H2 to H5 is a minor path, and that the H5 to B conversion mediated by PIG-V is inefficient (Fig. 3D).
PIG-V Is a Highly Hydrophobic Protein Expressed in the ER-Rat and human PIG-V cDNAs encoded 492 and 493 amino acid residues, respectively (Fig. 4A), and they have 85% iden- tity in their amino acid sequences. The human PIG-V gene consisted of four exons and was located in chromosome 1p36.11. PIG-V homologues of Drosophila melanogaster (accession number CG6657) and S. cerevisiae (P38211 and YBR004c) encoded 449 and 433 amino acids, respectively, having 25 and 31% amino acid identity to human PIG-V, respectively (Fig.  4A). Human PIG-V is a highly hydrophobic protein (Fig. 4B) with eight putative transmembrane domains (35) (Fig. 4A,  underlining). This transmembrane topology of PIG-V is similar to that of a family of ER-resident glycosyltransferases that use dolichol-phosphate-mannose/glucose (Dol-P-Man/Glu) as donor substrates (20).
To determine the intracellular expression site of PIG-V, JY25 cells stably transfected with GST-FLAG-tagged PIG-V were hypotonically lysed and postnuclear supernatant was fractionated by sucrose density gradient centrifugation. Each fraction was characterized by measuring protein content and the activities of organelle marker enzymes (Fig. 5A). GST-FLAG-PIG-V was detected mainly in the fractions containing the ER by Western blotting (Fig. 5B), indicating that PIG-V is an ER membrane protein. This result is compatible with roles in GPI biosynthesis in the ER.
Two Regions in PIG-V Are Conserved, Luminally Oriented, and Functionally Important-There is a report that members of the ER-resident, Dol-P-Man-utilizing glycosyltransferase family have the first loop region on the luminal side of the ER membrane and that those loop regions have conserved amino acids implicated in the glycosyltransferase active site (20). We found two highly conserved regions among various PIG-V homologues (Fig. 4A, dotted underlining). The first conserved region was within a hydrophilic domain between the first and the second transmembrane domains. Because such regions are usually important for function and because we wanted to know membrane orientation of the first loop, we determined the membrane orientation of the N terminus, which should be on the opposite side of the first loop, to see whether the conserved region is on the luminal side of the ER. Human PIG-V tagged with GST at the N terminus that restored the mutant phenotype was expressed in CHO K1 cells. After treatment with 0.0005% digitonin for selective permeabilization of the plasma membrane, or with 0.1% Triton X-100 for permeabilization of both the plasma and the ER membranes, cells were stained for GST (green) and for endogenous BiP (red), an ER luminal chaperone. The GST-tagged N terminus of PIG-V was stained after permeabilization of only the plasma membrane (Fig. 6B). Further permeabilization of the ER membrane with Triton X-100 did not significantly increase their staining intensity (Fig. 6E), whereas BiP was stained only after permeabilization of both the plasma and ER membranes (Fig. 6, A versus D) as expected. Because these results indicated that the N terminus of PIG-V faces the cytoplasmic side of the ER, we conclude that the first conserved region faces the luminal side of the ER.
To determine the membrane orientation of the second conserved region between the fifth and sixth transmembrane domains (Fig. 7A), we introduced an artificial N-glycosylation site within the loop having the second conserved region. We gener-

FIG. 5. PIG-V is localized in the ER.
JY25 cells stably transfected with GST-FLAG-tagged hPIG-V were hypotonically lysed, and the postnuclear supernatant was fractionated by sucrose density gradient centrifugation. A, each fraction was characterized by activities of organelle marker enzymes. Squares, plasma membrane marker (alkaline phosphodiesterase I); triangles, Golgi marker (␣-mannosidase II); crosses, ER marker (dolichol-phosphate-mannose synthase) and hatched bars, total proteins. B, GST-FLAG tagged PIG-V was detected by Western blotting using anti-GST antibody. ated a site-directed mutant, in which two proline residues (Pro-293 and Pro-294) were changed to Thr-293 and Ala-294 to create an acceptor site for N-glycosylation (Fig. 7A). We transiently transfected CHO K1 cells with FLAG-tagged wild type and mutant PIG-V cDNAs. PIG-V activity was not affected by two amino acid substitutions (Fig. 7B). The P293T/P294A mutant was N-glycosylated as shown by peptide N-glycosidase F-dependent mobility shift in the SDS-PAGE gel (Fig. 7C, lanes  3 and 4). Wild type PIG-V has one N-glycosylation site between the sixth and seventh transmembrane domains that is not N-glycosylated as shown by insensitivity to peptide Nglycosidase F (lanes 1 and 2). The molecular size of the mutant PIG-V was larger than that of the wild type PIG-V, but became similar to wild type PIG-V after treatment with peptide N-glycosidase F (Fig. 7C), consistent with the Nglycosylated state of mutant PIG-V. This result clearly demonstrated that the region between the fifth and sixth transmembrane domains is located within the ER lumen, as is the first conserved region.
We considered the possibility that amino acid residues in these conserved regions are functionally important. We introduced a number of point mutations into two conserved regions. These mutant cDNAs were transfected into PIG-V-defective D9PA15.6 cells and analyzed by FACS for surface expression of GPI-anchored proteins. W66L, D67A, and W312L mutants almost completely, and Q308A partially, lost PIG-V activity. Asp-67 corresponds to an aspartic acid that is thought to be involved in stabilizing a metal ion during catalysis in several other glycosyltransferases in the same family. These data are consistent with the idea that these conserved regions contribute to a catalytic site.
Human PIG-V Rescued Lethality of Yeast gpi18 Deletion Mutant-The ORF YBR004c, a sequence homologue of PIG-V (Fig. 4A), is essential for growth. We termed the gene GPI18. To obtain evidence that GPI18 is functionally homologous to PIG-V, we tested whether human PIG-V cDNA is able to restore viability of gpi18 null mutants. A heterozygous GPI18/gpi18::KAN R diploid strain was transformed with human PIG-V cDNA. Transformed diploid cells were allowed to sporulate, and complementation was assessed by tetrad analysis. Asci from diploid clones harboring PIG-V gave rise to four viable haploid progenies: two with normal growth (Fig. 8A, growths 2 and 4) and the other two with slower growth (Fig. 8A, growths 1 and 3). The two fast growing haploids must have harbored wild type GPI18 allele, because they were sensitive to G418 (Fig. 8B, growths 2 and 4). The two haploids with slower growth were resistant to G418 (Fig.  8B, growths 1 and 3) and sensitive to 5-fluoroorotic acid (Fig. 8C,  growths 1 and 3), indicating that they harbored the gpi18::Kan R allele and that their viability was dependent upon URA3-containig human PIG-V plasmid. Thus, PIG-V and GPI18 are functionally homologous. DISCUSSION Among the three mannosyltransferases involved in the biosynthesis of the trimannosyl core of GPI, two have been previously cloned, i.e. ␣1,4-mannosyltransferase (GPI-MT-I or PIG-M) (14) and ␣1,2-mannosyltransferase (GPI-MT-III or PIG-B) (15), by using corresponding class M and class B mutant cell lines. However, neither molecular cloning of ␣1,6mannosyltransferase (GPI-MT-II) nor the establishment of mutant cells defective in GPI-MT-II had been reported. This is one of the few remaining enzymes unidentified in GPI biosynthesis. Recently, the isolation and characterization of the GPI-MT-IIdeficient mutant of Leishmania mexicana were reported (36), but the responsible gene remains to be identified.
PIG-V Is Essential for Transferring the Second Mannose to GPI on the Luminal Side of the ER-In this study, we isolated three novel mutant CHO cells, termed class V cells, which were defective in the surface expression of GPI-anchored proteins (Fig. 1). These mutant cells accumulated two GPI intermediates H2 (Man-GlcN-acyl-PI) and H5 (EtNP-Man-GlcN-acyl-PI), both without the second mannose, indicating that the mutants were defective in GPI-MT-II (Fig. 2). By means of expression cloning, we identified a gene previously uncharacterized and termed it PIG-V (Fig. 3). This is the first report of a gene involved in the second mannose transfer. Because PIG-V cDNA fully restored the surface expression of CD59 on, and normalized the profile of GPI-mannolipids in these class V mutants (Fig. 3), we propose that PIG-V is the gene responsible for the mutant's phenotype.
PIG-V Represents a Novel Family within the Glycosyltransferase Superfamily-PIG-V showed some similarity in membrane topology with Dol-P-Man-utilizing glycosyltransferases. Human PIG-V encodes 493 amino acids containing 8 putative transmembrane regions and is targeted to the ER (Figs. 4 and 7A). We found a large number of PIG-V homologues from various eukaryotes in databases, which were also multitransmembrane proteins. These homologues share two highly conserved regions, and we determined that the two regions face the ER lumen ( Figs. 6 and 7). Furthermore, we identified several residues important for enzyme activity in those conserved regions (Fig. 7D). Because the transfer of mannose from the donor substrate Dol-P-Man to the acceptor GPI intermediates should occur on the luminal side of the ER, these results are consistent with the idea that these conserved regions contribute to the catalytic site. In particular, Asp-67 in the first conserved luminal region, which may correspond to the latter Asp residue in the DXD motif of PIG-M, was essential for its activity. The Asp residue in the DXD motif is thought to be involved in stabilizing a divalent metal ion during catalysis in many glycosyltransferases in the Golgi apparatus (37).
Data base searches by PSI-BLAST revealed that PIG-V shows sequence similarity to PIG-B (E value, 2 ϫ 10 Ϫ21 ) and ALG9 (1 ϫ 10 Ϫ11 ) (38), Dol-P-Man-utilizing ␣1,2-mannosyltransferase involved in N-glycan precursor biosynthesis. PIG-V also shows similarity to members of the protein O-mannosyltransferase family, such as Pmt2p/YAL023c (3 ϫ 10 Ϫ26 ) and Pmt3p/YOR321w (2 ϫ 10 Ϫ24 ) from S. cerevisiae (39) and human POMT1 (AF095136) (3 ϫ 10 Ϫ19 ) (40). Protein O-mannosyltransferases transfer Man from Dol-P-Man to Ser/Thr of proteins and are the integral ER membrane glycoproteins with multiple transmembrane domains (18). Two major hydrophilic domains are located between the first and second transmembrane domains (loop 1) and the fifth and sixth transmembrane domains (loop 5), both of which face the ER lumen and are essential for Pmt1p activity (41). Furthermore, protein O-mannosyltransferases have conserved WD (Trp-Asp) motifs or an Asp residue in the first ER luminal loop. These characteristics of protein O-mannosyltransferases are in common with the characteristics of PIG-V, i.e. two conserved regions on the luminal side of the ER and the essential WD sequence in the first loop. Additionally, PIG-V shows similarity with Stt3p (3 ϫ 10 Ϫ30 ), the catalytic subunit of the oligosaccharyl transferase complex in various eukaryotes (42), and also with the Stt3p homologues found in prokaryotes that transfer oligosaccharide from polyprenyl-PP-oligosaccharide donor to Asn residues in nascent proteins on the periplasmic side of the plasma membrane (43). These similarities may indicate that these Dol (polyprenyl)-PP-saccharide-utilizing glycosyltransferases, including PIG-V, share a distant but common evolutional origin.
There are 75 public families of glycosyltransferases in the CAZy data base (available at afmb.cnrs-mrs.fr/CAZY). Because the sequence of PIG-V is clearly distinct from all previously established glycosyltransferase families even though possessing a slight similarity to the aforementioned glycosyltransferases, it would define a novel glycosyltransferase family (GT family 76) in CAZy. 2 S. cerevisiae Gpi18p/YBR004c Is the Functional Homologue of PIG-V-We found that S. cerevisiae Gpi18p/YBR004c is homologous to mammalian PIG-V, with 25% amino acid identity. We showed that human PIG-V cDNA partially restored the viability of the yeast gpi18 deletion mutant (Fig. 8), suggesting that PIG-V and Gpi18p are functional homologues. We recently found that PIG-M requires an essential subcomponent, PIG-X, for GPI-MT-I activity (26). A combination of Gpi14p and Pbn1p, the yeast homologues of mammalian PIG-M and PIG-X, respectively, worked in mammalian cells when their cDNAs were cotransfected, whereas Gpi14p alone did not work in mammalian cells. Because human PIG-V only partially restored the defective phenotype of yeast gpi18 mutant, PIG-V might require a similar partner subcomponent.
PIG-V Preferably Uses H2 as an Acceptor Substrate-Whether of H2 or H5 is the preferable substrate of GPI-MT-II in the mammalian GPI biosynthesis pathway is an unresolved issue. When PIG-V was overexpressed in the PIG-U-defective cells, we found that H2 was greatly decreased with a concomitant increase in H3. On the other hand, the decrease of H5 was not so remarkable (Fig. 3C). For now, we speculate that PIG-V mainly mediates the conversion of H2 to H3, namely H2 is the main substrate of GPI-MT-II (Fig. 3D). An enzyme assay in vitro using purified enzyme preparation and the substrates is presently in progress. Taken together, the data presented in this report suggest that PIG-V represents a new type of mannosyltransferase family.