A Homologue of Saccharomyces cerevisiae Dpm1p Is Not Sufficient for Synthesis of Dolichol-Phosphate-Mannose in Mammalian Cells*

Dolichol-phosphate-mannose (Dol-P-Man) serves as a donor of mannosyl residues in major eukaryotic glycoconjugates. It donates four mannosyl residues in the N-linked oligosaccharide precursor and all three mannosyl residues in the core of the glycosylphosphatidylinositol anchor. In yeasts it also donates one mannose to the O-linked oligosaccharide. The yeastDPM1 gene encodes a Dol-P-Man synthase that is a transmembrane protein expressed in the endoplasmic reticulum. We cloned human and mouse homologues of DPM1, termedhDPM1 and mDPM1, respectively, both of which encode proteins of 260 amino acids, having 30% amino acid identity with yeast Dpm1 protein but lacking a hydrophobic transmembrane domain, which exists in the yeast synthase. Human and mouse DPM1cDNA restored Dol-P-Man synthesis in mouse Thy-1-deficient mutant class E cells. Mouse class E mutant cells had an inactivating mutation in the mDPM1 gene, indicating that mDPM1 is the gene for class E mutant. In contrast, hDPM1 andmDPM1 cDNA did not complement another Dol-P-Man synthesis mutant, hamster Lec15 cells, whereas yeast DPM1restored both mutants. Therefore, in contrast to yeast, mammalian cells require hDPM1/mDPM1 protein and a product of another gene that is defective in Lec15 mutant cells for synthesis of Dol-P-Man.

Dolichol-phosphate-mannose (Dol-P-Man) 1 serves as one of the two mannosyl donors for glycoconjugates. It donates four mannosyl residues used in the N-linked oligosaccharides (reviewed in Refs. 1 and 2) and all three mannosyl residues used in a core of glycosylphosphatidylinositol (GPI) protein anchors (3)(4)(5). It also donates one mannosyl residue used in the Olinked oligosaccharide in yeasts (4).
Dol-P-Man is synthesized from GDP-mannose and dolicholphosphate by the enzyme Dol-P-Man synthase. The enzyme is expressed in the endoplasmic reticulum. The Saccharomyces cerevisiae dpm1 mutant is defective in synthesis of Dol-P-Man. Its gene, DPM1, has been cloned (6). Proteins encoded by DPM1 (Dpm1p) expressed in Escherichia coli showed synthase activity in vitro, indicating that Dpm1p is yeast Dol-P-Man synthase (6,7). Dpm1p, a 267-amino acid protein, has a hydrophobic segment near the carboxyl terminus that may be a transmembrane domain (6). Dpm1p expressed in CHO cells showed reticular immunofluorescence staining, suggesting expression in the endoplasmic reticulum (8).
There are at least two mammalian mutant cells that are defective in synthesis of Dol-P-Man. Mouse Thy-1 negative thymoma mutant cells of complementation class E (9) do not synthesize Dol-P-Man (10) and consequently do not synthesize the GPI core, resulting in the defective surface expression of GPI-anchored proteins, such as Thy-1 (11). The Lec15 mutant of CHO cells is also defective in Dol-P-Man synthesis (12). Because they lack Dol-P-Man, the donor of the last four mannosyl residues in the N-linked oligosaccharide precursor, neither Lec15 nor class E cells assemble a mature oligosaccharyl precursor but rather, accumulate an intermediate bearing five mannosyl residues (10,12). Since this intermediate transfers incomplete oligosaccharides to proteins and are modified to complex type oligosaccharides without trimming of mannosyl residues by swainsonine-sensitive ␣-mannosidase II, these mutants are not sensitive to swainsonine treatment (8,13). Somatic cell hybridization between class E and Lec15 cells indicated that they are different mutants (5).
Interestingly, yeast DPM1 DNA complemented both class E and Lec15 mutants (8,14). Transfection of DPM1 into class E cells restored Dol-P-Man synthesis and the surface expression of Thy-1 (14). Its transfection into Lec15 restored the assembly of mature oligosaccharyl precursors and swainsonine sensitivity (8). These results indicated that synthesis of Dol-P-Man in mammalian cells is mediated by a similar enzyme. However, the question as to why there are two different mutants with a similar phenotype remains (5). Here we report the cloning of human and mouse homologues of DPM1 and show that the DPM1 homologue is defective in class E mutant and that another protein, presumably that is defective in Lec15 mutant, is also required for the expression of the synthase activity.

EXPERIMENTAL PROCEDURES
Cloning of hDPM1 cDNA-We searched the human cDNA data base (15) of The Institute for Genomic Research (TIGR, Gaithersburg, MD) with a sequence of yeast DPM1 (6) and identified a cluster of ESTs, THC9298. To obtain the upstream and downstream sequences by RACE methods we synthesized the oligonucleotide primers 5Ј-TGCTTTTAC-CTACCTACAACGAGC and 5Ј-TCCCAACTTTTTCTCTCGTGGTCT. With 5Ј-RACE we obtained a 263-base pair upstream sequence, and with 3Ј-RACE we obtained a 938-base pair downstream sequence. The composite sequence contained 1043 base pairs excluding poly(A). This entire sequence was amplified by polymerase chain reaction from a placental cDNA library using the most 5Ј primer and the most 3Ј * This work was supported by grants from the Ministry of Education, Science, Sports and Culture 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) D86198 for hDPM1 cDNA, D86199, D86200, D86201, and D86202 for hDPM1 genomic sequences, and AB004789 for mDPM1 cDNA.
Cell Lines and Culture-The mouse thymoma BW5147 cell line and its Thy-1-negative mutant of complementation class E were gifts from Dr. R. Hyman (Salk Institute, San Diego, CA). Cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Lec15 mutant (Lec15.2) of CHO cells (a gift from Dr. M. A. Lehrman, University of Texas Southwestern Medical Center, Dallas, TX) (17) and CHO-K1 cells were cultured in Ham's F-12 medium containing 10% fetal bovine serum. To inhibit ␣-mannosidase II, cells were cultured in the presence of 5 g/ml swainsonine (Wako Chemicals, Osaka, Japan) for 4 days. For transfection, cells (5 ϫ 10 6 ) were suspended in 0.8 ml of HEPES-buffered saline (18), mixed with 20 g of plasmids, and electroporated with a Gene Pulser (Bio-Rad) at 350 V/250 F for class E cells and 350 V/960 F for Lec15 and CHO-K1 cells.
Assay of Dol-P-Man Synthase Activity-Cells were destroyed by a Teflon homogenizer. After removal of cell debris and nuclei by centrifugation at 1500 ϫ g for 10 min, membranes were collected by centrifugation at 100,000 ϫ g for 60 min and suspended in a buffer consisting of 50 mM Hepes/NaOH (pH 7.4), 25 mM KCl, 5 mM MgCl 2 , and 5 mM MnCl 2 . Dolichol phosphate (4 g, Sigma) dried under a stream of N 2 was suspended in 10 l of the same buffer containing 0.2% Triton X-100 by vigorous agitation and sonication, and then GDP-[ 3 H]mannose (0.4 Ci, American Radiolabeled Chemicals, St. Louis, MO) and the membranes (60 g of protein) were added to a final volume of 100 l. The mixture was incubated for 15 min at 37°C. The Dol-P-[ 3 H]Man generated was assessed by measuring radioactivity extracted into the organic phase (19).
Cloning and Analysis of Genomic hDPM1 Clone and Ribonuclease Protection Assay-A human genomic library (Stratagene) was screened with an hDPM1 cDNA probe (1-kilobase pair XhoI-NotI fragment). The ribonuclease protection assay was done with a HybSpeed RPA kit (Ambion, Austin, TX) and total RNA of HeLa cells.
Cloning of mDPM1 cDNA from Wild-type and Class E Mutant BW5147 Thymoma Cells-We identified a number of ESTs of mDPM1 in a data base of EST (National Center of Biotechnology Information) and synthesized polymerase chain reaction primers from those sequences. Using cDNA prepared from RNA of wild-type BW5147 and its class E mutant cells as templates, we amplified mDPM1 cDNA in two overlapping fragments and cloned and sequenced them. The primers used were 5Ј-GTCCATGGCTTCCACGGGGGCGAG and 5Ј-TGGTGGG-AGAGGTCAGCATCCA to amplify the region spanning nucleotides 1-371, and 5Ј-GCAGAGATCTATGGGCCCGACAG and 5Ј-TGAACAC-GGGAAGCATAAGTGAATCA to amplify the region spanning nucleotides 235-815.
Analysis of the Expression Site of Transfected hDPM1-An expression plasmid for hDPM1 tagged with glutathione S-transferase (GST-hDPM1) was prepared as described previously (16). Flag-tagged microsomal aldehyde dehydrogenase (16) was used as a control membranebound protein. Wild-type CHO and Lec15 cells expressing GST-hDPM1, free GST, and Flag-aldehyde dehydrogenase were hypotonically lysed and separated into soluble (cytoplasm) and insoluble (membranes) fractions by centrifugation at 100,000 ϫ g for 1 h. These fractions were analyzed by SDS-polyacrylamide gel electrophoresis/Western blotting against goat anti-GST (Pharmacia Biotech Inc.) visualized by horseradish peroxidase-conjugated anti-goat IgG antibody (Organon Teknika) plus chemiluminescence reactions (Renaissance, DuPont). The same samples were also Western blotted against anti-Flag antibody M2 (Kodak) (16).

Cloning of a Human Homologue of S. cerevisiae DPM1
A search of the TIGR EST data base for a homologue of a nucleotide sequence of yeast DPM1 identified a cluster of ESTs. This cluster contained a contiguous sequence of 451 nucleotides (Fig. 1, nucleotides 5-455) that has 41% identity to the corresponding sequence of yeast DPM1. We amplified both the 5Ј and 3Ј ends of this cDNA sequence by RACE methods from a cDNA library derived from human placenta and ligated and cloned them in a mammalian expression vector, then transfected the vector into mouse class E mutant cells. The cloned cDNA, 1043 base pairs (Fig. 1, nucleotides 5-1047), partially restored the surface expression of Thy-1 (data not shown), indicating that Dol-P-Man synthesis was partially restored leading to a synthesis of the GPI anchor and the surface expression of GPI-anchored proteins. Since there was no methionine codon near the 5Ј end and a reading frame was open to the 5Ј end, we tried to amplify the 5Ј end by RACE methods again, but only three more nucleotides (Fig. 1, nucleotides 2-4) were obtained. It is possible that some structural constraint hampered reverse-transcription of the 5Ј end.
To obtain the 5Ј end of the coding region, we isolated a genomic clone of human homologue of DPM1. We identified four exons in the genomic clone of 17 kilobase pairs by comparing its sequence with the cDNA sequence. The exon sequences were exactly the same as the corresponding cDNA sequences, indicating that this clone represented the human gene homologous to DPM1. The four exon boundaries corresponded to cDNA nucleotide numbers 161, 261, and 295 (nucleotide number 1 at A of the initiation codon ATG; see Fig. 1).
The genomic clone contained an exon that included a methionine codon at two residues amino-terminal to the end of the partial cDNA (Fig. 1). This methionine codon was within a Kozak consensus sequence (21). We prepared an expression construct including these two residues (spanning nucleotides Ϫ54 to 1047) and transfected it into a class E mutant. The surface expression of Thy-1 was completely restored (Fig. 2), indicating that we cloned the full-length coding region. Based on this functional activity, we termed the cloned gene hDPM1.
Since the reading frame was still open to the 5Ј end, we analyzed a start site of transcription with the ribonuclease protection assay using RNA of HeLa cells. Only one protected fragment of the hDPM1 RNA probe was seen and from its size the start site of transcription was determined to be at around nucleotide Ϫ10 (data not shown). This is consistent with the idea that the methionine codon at position 1-3 is the initiation codon.

Structural Properties of hDPM1
Human DPM1 encodes a protein of 260 amino acids. Fig. 3 compares amino acid sequences of yeast Dpm1p (267 amino acids) and hDPM1 protein. The amino acid identity between homologous regions is about 30%. There are two significant differences. Human DPM1 has 22 extra residues at the amino terminus. This sequence is hydrophilic. Human DPM1 is shorter than Dpm1p at the carboxyl terminus by 29 residues. The hydropathy profile of hDPM1 protein (not shown) indicated that it lacks a transmembrane domain and an aminoterminal hydrophobic signal sequence. So, hDPM1 has characteristics of a cytoplasmic protein or a peripheral membrane protein. In contrast, yeast Dpm1p has a carboxyl-terminal hydrophobic segment that may be a transmembrane domain. Human DPM1 had a sequence RKIIS (Fig. 3, amino acids 161-165) that is similar to a consensus phosphorylation site sequence for cAMP-dependent protein kinase found in Dpm1p (RRVIS corresponding to amino acids 137-141) (6).

Complementation of Class E Mutant by hDPM1
Dol-P-Man Synthase-Restoration of Dol-P-Man synthase activity in class E mutants was tested after stable transfection of an expression construct of hDPM1 (Fig. 4). Class E cells had almost no activity. hDPM1 and yeast DPM1 restored Dol-P-Man synthase in class E cells to levels about 3 times and 2.5 times as high as the wild-type level, respectively.
GPI Anchor Synthesis-To confirm that restoration of Dol-P-Man synthase activity in class E cells by hDPM1 led to restoration of synthesis of GPI anchor precursors, we labeled the class E cells and hDPM1 and DPM1 transfectants with [ 3 H]mannose, and analyzed mannolipids. As shown in Fig. 5, the synthesis of Dol-P-Man and mannose-containing GPI anchor precursors was restored. These results indicate that hDPM1 restored the defective synthesis of Dol-P-Man in class E mutant cells.

A Lack of Complementation of Lec15 Mutant by hDPM1
We next tested whether hDPM1 complements Lec15 mutant. It was reported previously that yeast DPM1 fully complemented defective Dol-P-Man synthesis of Lec15 (8) as well as that of class E. On the other hand, there is a report that Lec15 and class E are different mutants (5). We confirmed the latter by a cell hybridization experiment; heterokaryons expressed Thy-1 on the cell surface 2 days after fusion, indicating complementation (data not shown).
We stably transfected Lec15 mutant cells with hDPM1 and DPM1, and assessed restoration of Dol-P-Man synthase activity. Consistent with its leaky phenotype (17), the lysate of Lec 15 mutant cells synthesized about 20% as much Dol-P-Man as the lysate of wild-type CHO cells (Fig. 6). Yeast DPM1 restored Dol-P-Man synthesis of Lec 15 mutant as reported (8), whereas hDPM1 had no enhancement effect (Fig. 6). Transfection of the same expression vector of hDPM1 cDNA into wild-type CHO cells doubled the Dol-P-Man synthase activity (data not shown), indicating that the expression vector was functional in CHO cells.
To confirm the inability of hDPM1 to complement Lec15, we cultured wild-type CHO cells, Lec15, and Lec15 hDPM1 and DPM1 transfectants in swainsonine for 4 days, before staining them with fluorescein isothiocyanate-conjugated phytohemagglutinin-E4 for complex oligosaccharides. The staining intensity of wild-type CHO decreased by 75% after incubation in swainsonine, whereas that of Lec15 did not change as expected (Fig. 7, panels A and B). In about a half of the DPM1-transfected Lec15 cells, the staining intensity was reduced by 80% after incubation in swainsonine, indicating complementation (panel C). On the other hand, hDPM1-transfected Lec15 cells maintained a mutant phenotype (panel D).

A Mutation in mDPM1 Is Responsible for Class E Mutant
A mDPM1 cDNA encoded a protein of 260 amino acids having 91% identity in amino acids with hDPM1 protein (Fig. 8A). Nucleotide sequence analysis of products of reverse transcriptase-polymerase chain reaction from class E cells demon- strated that all eight clones had a T to C substitution at nucleotide 365 that caused a change of serine 122 to phenylalanine. To test whether this missense mutation causes a loss of function, we transfected the mutant mDPM1 cDNA into class E mutant cells. The cDNA bearing the mutation did not restore the surface expression of Thy-1 while the wild-type cDNA restored it (Fig. 8B), indicating that this base substitution is an inactivating mutation and that the class E cell is a mutant of the mDPM1 gene.

A Possible Role of the Protein Defective in Lec15 Cells
The above results indicate that synthesis of Dol-P-Man in mammalian cells is mediated by a mammalian homologue of Dpm1p and a product of the gene defective in Lec15 cells. hDPM1 protein may be either a cytoplasmic protein or a peripheral membrane protein as described above. If the former is true, the protein defective in Lec15 cells may be required for association of hDPM1 protein with the membrane. In this case, Dol-P-Man synthesis activity would be complemented if the cytoplasm of Lec15 cells that may contain a soluble DPM1 homologue is mixed with the membranes of class E cells that may contain the protein necessary for membrane association of DPM1 homologue. This mixture, however, was inactive (data not shown), indicating that the latter case may be true. To confirm this, we tested whether hDPM1 is membrane-bound even when expressed in Lec15 cells. As shown in Fig. 9, transfected GST-tagged hDPM1 was expressed in the membrane of Lec15 cells (lane 4) but not in the cytoplasm (lane 3), similarly to transfection into wild-type CHO cells (lanes 1 and 2). Therefore, it appeared that the protein defective in Lec15 cells is not required for membrane association of mammalian DPM1. DISCUSSION We have cloned a human homologue of the yeast Dol-P-Man synthase gene DPM1 and termed it hDPM1. An hDPM1 cDNA complemented mouse Thy-1-negative class E mutant cells that are defective in Dol-P-Man synthesis, indicating that the hDPM1 gene product participates in Dol-P-Man synthesis (Figs. 2,4,and 5). However, the hDPM1 cDNA did not complement another Dol-P-Man synthesis mutant, CHO Lec15 (Figs. 6 and 7). To determine whether class E cells are defective in mouse DPM1 homologue, we cloned it, analyzed its sequence in class E cells, and found an inactivating mutation (Fig. 8). This indicates that the defective Dol-P-Man synthesis in class E cells is due to the mutation in mDPM1. Somatic cell hybridization demonstrated that class E and Lec15 cells are of different complementation groups (5). These results together indicate that two genes, a DPM1 homologue and one defective in Lec15 cells, are necessary for synthesis of Dol-P-Man in mammalian cells.
In contrast to hDPM1 cDNA, yeast DPM1 DNA complemented both class E and Lec15 mutants (8,14) (Figs. 2 and  4 -7). Since yeast Dpm1p has Dol-P-Man synthase activity by itself, it is not surprising that yeast DPM1 complements both mutants. A major structural difference between yeast and human Dpm1 proteins is that yeast Dpm1p has a hydrophobic carboxyl-terminal segment, whereas human DPM1 protein lacks it (Fig. 3). The hydrophobic segment in Dpm1p may serve as a transmembrane domain (6) and in addition it may act as an acceptor site for Dol-P. hDPM1 protein does not have a hydrophobic segment for a typical transmembrane domain, nevertheless, it associated with the membrane even in Lec15  8. Analysis of mDPM1 cDNA. A, the deduced amino acid sequence of mDPM1. The serine 122 that is changed to phenylalanine in class E mutant is underlined. B, a lack of function of mDPM1 cDNA bearing the T to C missense mutation. The wild-type mDPM1 cDNA (left) and one bearing T to C base substitution (right) were transfected into class E cells. Two days later, the surface expression of Thy-1 was assessed by FACS analysis. The wild-type but not the mutant cDNA restored the surface Thy-1 expression. Thin line is vector; bold line is wild-type mDPM1 (left) and S122F mDPM1 (right). cells ( Fig. 9), suggesting that hDPM1 may be a peripheral membrane protein. The protein defective in Lec15 cells, therefore, does not play a role in association of hDPM1 with the membrane. So, a possible role of that protein may be in a step of Dol-P-Man synthesis, such as binding of Dol-P. It is noticeable, however, that an expression level of GST-tagged hDPM1 protein in Lec15 mutant was much lower than that in wild-type CHO cells (Fig. 9). This suggests another possible role of the protein defective in Lec15 cells, i.e. it may be necessary for stable expression of hDPM1 in the endoplasmic reticulum. This has to be formally shown by a complementation experiment when the gene defective in Lec15 mutant is cloned because it is possible that some nonrelevant mechanism operating in this particular Lec15 cell line caused rapid degradation of GSTtagged hDPM1 and/or its mRNA.
Recently, a hamster SL15 cDNA was cloned (22). SL15 encodes a protein of 248 amino acids bearing at least two putative transmembrane domains and a double-lysine endoplasmic reticulum retention signal near the carboxyl terminus, suggesting that it is an endoplasmic reticulum transmembrane protein (22). SL15 is a candidate that is necessary, together with mammalian DPM1, for Dol-P-Man synthesis, because it has been reported to suppress the Lec15 mutation. It is yet to be determined whether the Lec15 mutation is actually due to a defect in SL15 or some other protein.
The synthesis of Dol-P-Man in yeast and mammalian cells, therefore, differs; Dpm1p alone is sufficient in S. cerevisiae, whereas at least two gene products are necessary in mammalian cells. DPM1 homologues were cloned from a protozoa, Trypanosoma brucei brucei and a fungus Ustilago maydis (23,24). The overall structure of T. b. brucei DPM1 protein is very similar to that of yeast Dpm1p, i.e. 267 amino acids having a hydrophobic putative transmembrane domain near the carboxyl terminus. U. maydis DPM1 protein is of 294 amino acids. It also has a hydrophobic putative transmembrane domain near the carboxyl terminus. These two DPM1 proteins have Dol-P-Man synthase activity in vitro. So, only mammalian DPM1 protein requires an additional protein and lacks the carboxyl-terminal hydrophobic segment. Moreover, S. cerevisiae Dol-P-Man synthase is active in the presence of nonionic detergent (6, 7), whereas mammalian Dol-P-Man synthase is inactive (25). Furthermore, amino acid identities among three microbial Dpm1 proteins are 49 -60%, whereas those between mammalian DPM1 and three microbial ones are only about 30%. Therefore, mammalian DPM1 diverged from three microbial DPM1s.
A consensus sequence for cAMP-dependent protein kinasemediated phosphorylation is conserved in yeast, Ustilago and Trypanosoma DPM1 proteins, i.e. RRVIS (6), RRIIS (24), and RRFIS (23), respectively. The corresponding sequence in hDPM1 was RKIIS (amino acids 161-165). Although the double arginine sequence conserved in three other DPM1 proteins is changed to arginine-lysine in hDPM1, it would still be consistent with a phosphorylation site (26,27). It was reported that Dol-P-Man synthase activities of microsomes prepared from rat, bovine, and hen cells were enhanced by 30 -80% upon phosphorylation with cAMP-dependent kinase (28). Whether this regulation of Dol-P-Man synthesis is mediated through phosphorylation of DPM1 is yet to be determined.