MNN6, a Member of the KRE2/MNT1 Family, Is the Gene for Mannosylphosphate Transfer in Saccharomyces cerevisiae *

In yeast Saccharomyces cerevisiae theN-linked sugar chain is modified at different positions by the addition of mannosylphosphate. The mnn6 mutant is deficient in the mannosylphosphate transferase activity toward mannotetraose (Karson, E. M., and Ballou, C. E. (1978) J. Biol. Chem. 253, 6484–6492). We have cloned the MNN6gene by complementation. It has encoded a 446-amino acid polypeptide with the characteristics of type II membrane protein. The deduced Mnn6p showed a significant similarity to Kre2p/Mnt1p, a Golgi α-1,2-mannosyltransferase involved in O-glycosylation. The null mutant of MNN6 showed a normal cell growth, less binding to Alcian blue, hypersensitivity to Calcoflour White and hygromycin B, and diminished mannosylphosphate transferase activity toward the endoplasmic reticulum core oligosaccharide acceptors (Man8GlcNAc2-PA and Man5GlcNAc2-PA) in vitro, suggesting the involvement of the MNN6 gene in the endoplasmic reticulum core oligosaccharide phosphorylation. However, no differences were observed in N-linked mannoprotein oligosaccharides between Δoch1 Δmnn1 cells andΔoch1Δmnn1Δmnn6 cells, indicating the existence of redundant genes required for the core oligosaccharide phosphorylation. Based on a dramatic decrease in polymannose outer chain phosphorylation by MNN6 gene disruption and a determination of the mannosylphosphorylation site in the acceptor, it is postulated that theMNN6 gene may be a structural gene encoding a mannosylphosphate transferase, which recognizes any oligosaccharides with at least one α-1,2-linked mannobiose unit.

In yeast Saccharomyces cerevisiae the biosynthesis of Nlinked oligosaccharides has been studied in detail. The core oligosaccharide (Man8GlcNAc2) synthesized in the endoplasmic reticulum (ER) 1 is identical in yeast and mammals. The outer chain attached to the core-like oligosaccharide contains an ␣-1,6-linked polymannose backbone with branches of ␣-1,2linked mannobiose capped with terminal ␣-1,3-linked mannose residues (1)(2)(3)(4)(5). In yeast, N-linked sugar chains are also modified at different positions by the addition of mannosylphosphate (6 -9). Although this oligosaccharide modification significantly contributes to a major negative charge of the cell wall (8 -10), less is known about its biosynthesis and function in yeast.
Ballou and co-workers (11) have isolated mnn mutants that are blocked at various stages of outer chain elongation. The mnn1 mutant lacks ␣-1,3-mannosyltransferase activity and is defective in adding terminal ␣-1,3-linked mannose to both Nlinked and O-linked oligosaccharides (11,12). The ␣-1,3-mannosyltransferase has a property that competes with mannosylphosphate transferase (10,14,15). The mnn4 and mnn6 mutants are known to produce phosphate-deficient mannan relative to wild type cells, presenting a phenotype of less binding to the phthalocyanin dye, Alcian blue (10,15). The MNN4 gene has been cloned and predicted to encode a large protein containing 1,178 amino acids functioning as a positive regulator for mannosylphosphate transferase (16). The mnn6 is a recessive mutation, indicating a lack of mannosylphosphate in the branches of the mannose outer chain in vivo, and is deficient in the mannosylphosphate transferase activity toward mannotetraose in vitro (15). Although mannosylphosphate transferase activity was decreased in the mnn6 mutant, it was still uncertain whether the mnn6 mutation affected the Nlinked core oligosaccharide phosphorylation.
In this work, we have cloned the MNN6 gene by functional complementation and analyzed the effects of the mnn6 null mutation on enzymatic activities toward the core oligosaccharide acceptors. The ⌬mnn6 mutation does not affect the apparent profiles of N-linked core oligosaccharide phosphorylation in vivo, suggesting the presence of redundant functional genes. Characterization of enzymatic reaction product suggested that the MNN6 gene may encode a mannosylphosphate transferase, which recognizes any oligosaccharides with at least one ␣-1,2linked mannobiose unit.

EXPERIMENTAL PROCEDURES
Yeast Strains and Media-Strain TO3-6D, used for the cloning of MNN6, was a meiotic segregant from a cross of LB1425-1B, kindly provided by C. E. Ballou (University of California, Berkeley) and a strain of LB1-10B, purchased from the American Type Culture Collection (ATCC). KK4 and its isogenic mnn6 disruptant strain XW44 were used for the Calcoflour White (CFW; Sigma) and hygromycin B (Sigma) sensitivity test. Strains YS125-15B, XW27, YS131-30A, and YS131-30D were used for microsomal membrane preparation. YS126-47D and its isogenic mnn6 disruptant strain XW43 were used for oligosaccharide analysis. Yeast strains used in this study are summarized in Table I. * This work was supported in part by a grant-in-aid from the Research and Development Project of Basic Technologies for Future Industries from the Ministry of International Trade and Industry, 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) U43922.
¶ To whom correspondence should be addressed: National Institute of Bioscience and Human Technology, 1-1 Higashi, Tsukuba, Ibaraki 305, Japan. Tel.: 81-298-54-6224; Fax: 81-298-54-6220. 1 The abbreviations used are: ER, endoplasmic reticulum; HPLC, high performance liquid chromatography; MALDI-TOF, matrix-as- Yeast strains with multigene disruptions were constructed by standard genetic methods (17,18). Yeast strains were grown either in YPD (2% Bacto-peptone, 1% yeast extract, and 2% glucose) or in a complete minimal medium containing 4% glucose, 0.67% Bacto-yeast nitrogen base without amino acids (Difco), 0.3 M sorbitol and supplemented with the appropriate auxotrophic requirements (18). CFW and hygromycin B were separately added to autoclaved YPD/agar to a final concentration of 50 g/ml just prior to pouring plates. Cloning of MNN6 by Modified Alcian Blue Staining-Yeast genomic DNA library in YCp50, "CEN BANK" A and B, were purchased from ATCC. Transformation of yeast cells was carried out by the lithium acetate procedure (19). Transformants were selected and maintained on an SD-ura plate. The colonies from master plates were transferred to nitrocellulose filter and incubated for one more day at 30°C. Those colonies on the filter were fixed by autoclaving at 120°C for 1 h and stained by immersion into 0.1% Alcian blue solution until the blue color was developed on the wild type (MNN6) colonies at room temperature (20 -30 min). Positive clones were screened as those providing a blue stain (wild type phenotype). The putative clones were further reassayed and confirmed individually by conventional Alcian blue assay (11).
DNA Sequencing-Bacterial strain JM109 was used for the preparation of plasmids. Restriction fragments containing a portion of the MNN6 gene were subcloned into pRS316 vector (20). Sequencing was performed by the dideoxy chain termination method with dye primers (21) and done with the SequiTherm™ Long-Read Cycle Sequencing Kit-LC by the LI-COR model 4000L Automated Sequencer. Sequence comparisons against the GenBank or GenPept sequence data bases were performed using the FASTA (22) and BLAST (23) programs. The hydrophobicity plot was generated by the method of Kyte and Doolittle (24).
Gene Disruption-Disruption of MNN6 gene was made by the singlestep gene replacement procedure (25). The 4.3-kilobase pair HpaI-SacI fragment containing MNN6 from pSA9-7 was digested with BglII and BclI restriction endonucleases. The BglII site is located 333 base pairs upstream from the ATG, and the BclI site is found 558 base pairs upstream from the stop codon (see Fig. 4A). This digestion removed a 1116-base pair fragment encompassing 261 amino acids of the MNN6 sequence and further replaced it with a BglII fragment containing the complete ADE2 gene from pASZ11 (26). Haploid yeast strains were used to transform with the linearized mnn6::ADE2 DNA fragments (Fig. 4A). The disruption of MNN6 was confirmed by Southern hybridization (data not shown). Selection of mnn6 disruptants (⌬mnn6) was done as follows: for strains carrying the ade2 mutation, ⌬mnn6 was selected on an SD-ade plate; for strains carrying no ade2 mutation (like KK4), ⌬mnn6 was selected by QAE-Sepharose adsorption according to the method described by Ballou (11).
Construction of High Copy Plasmid Carrying MNN6 -A complementing fragment (HpaI/NruI fragment) containing the entire MNN6 gene was excised by digestion with KpnI/SacI, whose sites are located in multicloning sites of pRS316-based plasmid pRSMNN6 and then inserted into the multicopy vector of pET351 and named pETMNN6. The pET351 high copy vector was constructed based on YEp351 (27), in which a BamHI/HpaI fragment containing the LEU2 gene was replaced with a BamHI/PvuII fragment carrying TRP1 gene from pJJ246 vector (28).
Mannosylphosphate Transferase Assay-The microsomal membrane proteins containing mannosylphosphate transferase activity were prepared according to the previous method (29), except that the cell pellets were frozen at Ϫ20°C for 1 h before the cells were destroyed by glass beads using a B. Braun homogenizer. The enzyme assay was carried out by using 400 g of protein of the high speed pellet (centrifugation at 100,000 ϫ g for 60 min) in 50 l of 50 mM Tris-HCl (pH 6.0), 10 mM MnCl 2 , 25 or 50 pmol of acceptor (depending on the acceptor used), 0.6% Triton X-100, 1 mM GDP-mannose, and 0.5 mM 1-deoxy-mannojirimycin as an inhibitor of ␣-mannosidase in yeast (30) at 30°C for 60 min. The enzyme reaction was terminated by boiling for 5 min, and the reaction mixture was ultrafiltrated with ultrafree C3LGC (Millipore). The filtrated solution was lyophilized and used for high performance liquid chromatography (HPLC) analysis. Man8GlcNAc2-PA, purchased from Takara Shuzo Co. (Kyoto, Japan), was used as the acceptor (50 pmol) in the enzyme assay. Man5GlcNAc2-PA acceptor (25 pmol) prepared from strain YS133-1D mannoproteins was used for the enzyme assay.
Isolation of N-Linked Oligosaccharides from Yeast Mannoproteins-Preparation of oligosaccharides was the same as described previously (31). In brief, yeast cells were grown in YPAD medium at 25°C and harvested at stationary phase. Mannoproteins were hot citrate bufferextracted, ethanol-precipitated, and further purified by concanavalin A-Sepharose. The N-linked oligosaccharides were liberated from the bulk yeast mannoproteins using glycopeptidase A (Seikagaku Kogyo Co., Tokyo, Japan), an enzyme specific to release N-linked oligosaccharides from glycoprotein or glycopeptide. Pyridylamination of the oligosaccharides was performed using a commercial reagent kit (Takara Shuzo Co., Kyoto, Japan). The PA-oligosaccharides were obtained by gel filtration of pyridylaminated products on a Toyopearl HW-40F column (1.0 ϫ 40 cm) and used for detection by fluorescence (excitation ϭ 310 nm; emission ϭ 380 nm).
HPLC Analysis-The separation of PA-oligosaccharides was carried out by HPLC using a Tosoh CCPM-II pump, a Tosoh PX-8020 controller, and a Shimadzu spectrofluorometric detector, RF-550. Phosphorylated oligosaccharides were fractionated by their size and polarity with Asahipak NH 2 P-50 (0.46 ϫ 25 cm) (Asahi Chemical Co., Tokyo, Japan) at a flow rate of 1 ml/min. The retention time of oligosaccharide largely depends on the number of sugar residues in the amine-modified column chromatography. Samples were resuspended in buffer A and injected in up to 20-l aliquots. For analysis of mannosylphosphorylated Man8GlcNAc2-PA, the ratio of 200 mM acetic acid adjusted with triethylamine (pH 7.3) to acetonitrile was 30:70 (v/v) for buffer A and 70:30 (v/v) for buffer B. Initial solvent composition was 80% buffer A with 20% buffer B, and a linear gradient was run over 40 min, in which the percentage of buffer B increased from 20 to 100% with a flow rate of 1 ml/min. For analysis of mannosylphosphorylated Man5GlcNAc2-PA, buffer A contained a 10:90 (v/v) ratio of 200 mM acetic acid adjusted with triethylamine (pH 7.3) to acetonitrile, and buffer B was 100% 200 mM acetic acid adjusted with triethylamine (pH 7.3). The initial solvent of 100% buffer A (0% buffer B) was run for 5 min, and then the percentage of buffer B was linearly increased from 0 to 50% within 30 min; finally, the percentage of buffer B was linearly increased from 50 to 100% for another 15 min with a flow rate of 1 ml/min. ␣-1,2-Mannosidase Digestion-Samples (enzymatic reaction product or in vivo acidic oligosaccharide product prepared from ⌬och1⌬mnn1 cells) were dissolved in 8 l of 0.1 M sodium acetate buffer (pH 5.0). One microunit/2 l of ␣-1,2-mannosidase solution (from Aspergillus saitoi; Oxford Glycosystems, Inc.) was added and then incubated for 15 h at 37°C.
Mass Spectrometry-Matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry was performed in the negative ion mode by using ␣-cyano-4-hydroxy cinnamic acid as a matrix. The mass spectrometer used in this work was a Finnigan Lasermat (Finnigan MAT Ltd., Hempstead, United Kingdom). Samples (100 -1000 pmol) were desalted by HPLC using a Tosoh TSK-GEL Carbon-500 column (0.46 ϫ 10 cm). Two solvents, A and B, were used. Solvent A was water containing 0.1% trifluoroacetic acid. The column was equilibrated with solvent A. After the sample injection, the proportion of solvent B was increased linearly up to 100% over 60 min. PAoligosaccharides were detected by fluorescence (excitation ϭ 320 nm; emission ϭ 400 nm).
1 H NMR-1 H NMR spectra of PA-oligosaccharide were measured on a JNM-A500 (JEOL Co.) at 50°C. Samples (ϳ10 nmol) were dissolved in 99.96% D 2 O and lyophilized. After three repetitions of the above procedure, the samples were finally dissolved in 700 l of 99.996% D 2 O. The chemical shifts (␦) are expressed in ppm downfield from internal sodium 4,4-dimethyl-4-silapentane-1-sulfonate, but they were actually measured by reference to internal acetone (␦ ϭ 2.217 ppm).

RESULTS
Cloning of the MNN6 Gene-The original mnn6 mutant exhibited a reduced amount of phosphomannan on the cell wall and showed less binding to Alcian blue, a dye that binds to a phosphate moiety of cell surface mannoproteins (11). The wild type cells exhibited a blue color after staining with Alcian blue, while the mnn6 mutant showed as white. This characteristic was used to clone the wild type MNN6 gene by complementation. Strain TO3-6D (mnn1 mnn6) was transformed with a yeast genomic DNA library constructed in the centromeric vector YCp50, which carries the URA3 as a selective marker. Transformants were selected and maintained on an SD-ura plate. About 1.2 ϫ 10 4 transformants were screened, and two positive clones (blue) were selected from among a majority of colorless white transformants.
Restriction maps of the insert DNA on the plasmids from two positive clones were identical. A ϳ10-kb DNA fragment that can complement the mnn6 mutation was isolated. To identify the smallest complementing region, further subcloning into pRS316 vector was carried out. Plasmid pA9 (Fig. 1) contained a ϳ10-kb Sau3AI fragment insert at the BamHI site on YCp50 vector. pSA series plasmids ranging from pSA9-1 to pSA9-7 and pRSMNN6 ( Fig. 1) were the subclones derived from plasmid pA9 and inserted into single-copy yeast vector pRS316. These plasmids were introduced into strain TO3-6D, and the complementation analysis was carried out by Alcian blue staining. The complementing region was assigned into a 2.7-kb HpaI-NruI fragment (pRSMNN6) (Fig. 1). To exclude the possibility of the suppressor gene cloning of mnn6 mutation, strain XW13 (⌬mnn6::ADE2) (Table I and Fig. 4A) was crossed with strain TO3-6D (mnn6), and after sporulation, tetrads were analyzed by Alcian blue staining. All of the segregants derived from 20 tetrads showed the mutant phenotype, which was not able to bind to Alcian blue, demonstrating the gene disruption at the original mnn6 locus and confirming the cloning of the MNN6 gene.
MNN6 Is a Member of the KRE2/MNT1 Mannosyltransferase Gene Family-Sequence analysis of a 2.7-kb fragment revealed one open reading frame with 1338 base pairs, which was translated to a protein of 446 amino acids (MNN6 accession number U43922). From the GenBank TM data base, the MNN6 gene was identical with the KTR6 gene, which was reported by the genome sequencing as a family of killer toxin related genes (accession number U39205). Two potential Nglycosylation sites were found in the Mnn6p sequence ( Fig. 2A). Kyte-Doolittle hydrophobicity analysis showed a potential membrane-spanning region near the N-terminus suggesting a type II membrane protein (Fig. 2B).
In addition, a homology search of the MNN6 sequence revealed that Mnn6p shares 38% identity and 79% similarity with Kre2p/Mnt1p, an ␣-1,2-mannosyltransferase responsible for O-linked glycosylation in yeast (32,33). Sequence alignment of the Mnn6p and Kre2 protein families (Ktr1p, Ktr2p, Ktr3p, and Yur1p) (34 -36) is shown in Fig. 3. Six cysteine residue positions in the latter half of Mnn6p were identical to those of the other proteins, suggesting a similarity of threedimensional structures. Interestingly, Mnn6p has an addi-tional cysteine at the 120th residue, which shares an identical position with Kre2p, but lacks one cysteine at the 229th residue, which is commonly located in the other five proteins.
Disruption of the MNN6 Gene Results in Calcoflour White and Hygromycin B Sensitivities-To study the function of Mnn6p, the MNN6 gene was disrupted by inserting the ADE2 gene (Fig. 4). MNN6 gene disruption did not affect the cell morphology and the rate of cell growth, indicating a nonessential gene for normal cell growth. The mnn6 null (⌬mnn6) mutant displayed the same phenotype as the original mnn6 mutant, which provides a prominent loss of Alcian blue binding ability, while the isogenic wild type cells strongly bound the dye. A single copy plasmid containing the MNN6 gene restored the Alcian blue binding, which was lost in the ⌬mnn6 mutant (data not shown), suggesting the involvement of MNN6 in oligosaccharide phosphorylation. A sensitivity to a negatively charged fluorescent dye CFW and an aminoglycoside antibiotic hygromycin B was examined. As shown in Fig. 4B, the ⌬mnn6 mutant was sensitive to CFW and hygromycin B, while the isogenic wild type was not affected. A single copy of the MNN6 gene recovered the growth defect of the ⌬mnn6 mutant by CFW and hygromycin B, respectively. Since CFW binds to nascent chains of chitin and prevents both microfibril formation and cell wall assembly (37), the result may suggest a lesser charge repulsion between CFW and the cell surface in ⌬mnn6 mutant.
Alcian Blue Staining of Various Strains-The original mnn6 mutant was deficient in the mannosylphosphate addition to the mannose outer chain (15). This is supported by the Alcian blue staining of the isogenic pairs with or without the MNN6 gene (⌬mnn1 cells and ⌬mnn1⌬mnn6 cells in Table II, G-1). To further examine whether the MNN6 gene may affect phosphorylation of the oligosaccharide lacking a mannose outer chain, other isogenic pairs of double and triple mutant cells (⌬och1⌬mnn1 and ⌬och1⌬mnn1⌬mnn6) were constructed. The ⌬och1⌬mnn1 cells showed a significant dye binding, while the ⌬och1⌬mnn1⌬mnn6 cells failed to bind the dye (Table II,  dye binding ability, but the effect of multicopy gene dosage was not observed on Alcian blue staining (Table II, G-2). These results suggest that the MNN6 gene may be involved in the phosphorylation in vivo not only at the outer chain portion but also the N-linked core and/or O-linked oligosaccharides. In contrast, the Alcian blue staining was not changed by the introduction of the ⌬mnn6 mutation into ⌬och1⌬mnn1⌬kre2 cells, which produces the N-linked core oligosaccharide (Man8GlcNAc2) (31) and truncated O-linked chains (Man2) (32), suggesting the possibility of no apparent effect of the MNN6 gene on N-linked core oligosaccharide phosphorylation in vivo.
Reduction of Mannosylphosphate Transferase Activity toward N-Linked Core Oligosaccharide Acceptors in a mnn6 Null Mutant-The assay conditions for the mannosylphosphate transferase were established by using 1 mM GDP-mannose as a donor and 50 pmol of pyridylaminated core oligosaccharide Man8GlcNAc2-PA (M8-PA; see structure shown in Fig. 8, A-1) as an acceptor in 50 l of reaction mixture (see "Experimental Procedures"). Under these assay conditions, microsomal membranes from MNN6 wild type cells (⌬och1⌬mnn1, strain YS125-15B) showed two reaction products (peaks 1 and 2) (Fig.  5, A-1), which were already identified as a monomannosylphosphorylated Man8GlcNAc2-PA (ManP-M8-PA) (16). In contrast, microsomal membranes from isogenic ⌬mnn6 cells (⌬och1⌬mnn1⌬mnn6, strain XW27) diminished corresponding peaks (Fig. 5, A-2), and the enzyme activity was restored after the introduction of the MNN6 gene into ⌬mnn6 cells (Fig. 5,  A-3), indicating more directly the involvement of the MNN6 gene in the mannosylphosphate transferase activity toward Man8GlcNAc2-PA in vitro. However, introduction of MNN6 in a multicopy plasmid did not produce any higher enzymatic activities in wild type cells (Fig. 5, A-4), suggesting the presence of some limiting factors for this enzyme reaction.
Taken together, these data strongly suggest that the MNN6 gene may encode a mannosylphosphate transferase, which is involved in the oligosaccharide phosphorylation not only of the mannose outer chain but also of the N-linked core portion.
Characterization of Enzymatic Reaction Products-To determine the structure of reaction products, peaks 1 and 2 in Fig. 5 were analyzed by MALDI-TOF mass spectrometry. The molecular ion peaks were observed at m/z 2048.9 for peak 1 and 2042.5 for peak 2, respectively. These mass values were nearly identical to the molecular mass of ManP-M8-PA (calculated M r 2041.8). The 1 H NMR spectra of peak 2 show the mannosylphosphate signal at ␦ 5.44 (Fig. 6B). The intensity of this signal indicates the presence of one mannosylphosphate group in peak 2. Measurement of 1 H NMR spectra of peak 1 was not successful due to the loss of material during the purification process.
Since the core-like oligosaccharide has two mannosylphosphorylation sites (38), two structures of ManP-M8-PA are possible. One is the structure in which mannosylphosphate attaches to the side of the ␣-1,6-branch of core Man8GlcNAc2 (structure I); the other is the structure in which mannosylphosphate attaches to the side of ␣-1,3-branch of the same Man8GlcNAc2 (structure II). By ␣-1,2-mannosidase digestion, structure I releases three mannoses and yields ManP-M5-PA, but structure II releases two mannoses and yields ManP-M6-PA. The ␣-1,2-mannosidase digestion products of peaks 1 and 2 were analyzed by MALDI-TOF mass spectrometry. The molecular ion peak of each product was detected at m/z 1724.7 (ManP-M6-PA, calculated M r 1717.5) for peak 1 and 1558.9 (ManP-M5-PA, calculated M r 1555.4) for peak 2, respectively. These results revealed the site of monomannosylphosphorylation at Man8GlcNAc2. Peak 1 product was mannosylphosphorylated at the ␣-1,3-branch of core Man8GlcNAc2 (structure II) and peak 2 product was mannosylphosphorylated at the the ␣-1,6-branch of core Man8GlcNAc2 (structure I).
The N-Linked Core Oligosaccharide Profiles Were Not Changed by MNN6 Gene Disruption in Vivo-Further to examine the effect of MNN6 gene disruption on the mannosylphosphate addition in the N-linked core portion, we analyzed the oligosaccharides of mannoproteins prepared from ⌬och1⌬mnn1 (strain YS126-47D) and ⌬och1⌬mnn1⌬mnn6 (strain XW43) cells. In HPLC analysis, a peak of neutral M8-PA, two peaks of ManP-M8-PA (peaks 1 and 2), and unknown peak x were observed in ⌬och1⌬mnn1 cells (Fig. 7). Unexpectedly, the same oligosaccharide pattern was observed in ⌬och1⌬mnn1⌬mnn6 cells (data not shown). Especially, the ratio of peak area corresponding to ManP-M8-PA to neutral M8-PA was not changed in both cells. These results suggested the presence of functionally redundant gene(s), which may interfere the appearance of mutant phenotype due to MNN6 gene disruption in vivo.
We also analyzed the structure of peak x, which was eluted at 39.5 min in Fig. 7. The molecular ion peaks were observed at m/z 2294.7, which was nearly identical to the molecular mass of dimannosylphosphorylated Man8GlcNAc-PA (calculated M r 2283.9). 1 H NMR spectra of peak x showed a mannosylphosphate signal at ␦ 5.44, and the intensity of this signal indicated the presence of two mannosylphosphate groups (Fig. 6C). To determine the mannosylphosphorylation sites, the ␣-1,2-mannosidase digestion product of peak x was first subjected to time-of-flight mass spectrometry analysis. However, the molecular ion peak was not observed, presumably due to the increase of negative charge in the molecule. Then the above digestion product was subjected to mild acid treatment followed by alka-line phosphatase to convert to neutral PA-oligosaccharide. This product showed a retention time of 9.5 min in amino column HPLC (Asahipack NH 2 P-50), which was identical to the authentic Man6GlcNAc2-PA (Takara PA-sugar chain 018) (data not shown). Therefore, the sites of two mannosylphosphates in peak x were confirmed as shown by the star in Fig. 8, A-1. DISCUSSION We have reported the cloning and analysis of the MNN6 gene. For the cloning, the original Alcian blue dye binding assay in a test tube was not appropriate for the colony screening due to the laborious work. To solve this problem, a modified procedure for Alcian blue staining was developed on plates. The method established in this work should be applicable to clone other yeast genes, especially genes related to the biosynthesis of cell wall components.
⌬och1 ⌬mnn1 ⌬kre2 ⌬mnn6 ϩϩ a The colors developed on the cell pellet by the Alcian blue assay were classified as follows: Ϫ, white; ϩϩ, light blue; ϩϩϩ, blue; ϩϩϩϩ, dark blue. mannobiose) recognized by all of these enzymes. The functional relations between Mnn6p and Kre2p/Mnt1p will be investigated in future work.
Disruption of MNN6 caused a hypersensitivity to CFW and hygromycin B (Fig. 4). The former phenotype is caused by the loss of charge repulsion between the cell surface and drug, leading to the penetration of drug through the outermost mannoprotein portion in the cell wall. It is noteworthy that hygromycin B-sensitive mutants involve not only the defects in sugar chain length, as reported (39,40), but also in oligosaccharide phosphorylation, although its mechanism is still unclear.
We have shown that MNN6 is involved in core oligosaccharide phosphorylation by demonstrating the loss of mannosylphosphate transferase activity in vitro toward Man8GlcNAc2 and Man5GlcNAc2 in ⌬mnn6 cells. Two reaction products (peaks 1 and 2) corresponding to ManP-M8-PA were identified when M8-PA was used as an acceptor. The mannosylphosphorylation site was determined by time-offlight mass spectrometry after the ␣-1,2-mannosidase treatment. These sites were identical to the phosphorylation sites observed in dimannosylphosphorylated oligosaccharide in vivo (peak x compound in Fig. 7) described in this paper and to those reported for the N-linked core-like Man10GlcNAc2 oligosaccharide from carboxypeptidase Y and mnn1 mnn9 strain mannoproteins (38,41). When M5-PA was used for acceptor substrate, only one peak corresponding to monomannosylphosphorylated product was observed (Fig. 5, peak 3, B-1). Although the phosphorylation site could not be determined by ␣-1,2-mannosidase treatment due to the limited amount of purified material, based on combined results on both the phosphorylation sites determined for ManP-M8-PA in vitro and the structure of dimannosylphosphorylated oligosaccharide determined in vivo, the most reasonable phosphorylation site in ManP-M5-PA is shown in Fig. 8 (A-2).
Apparently, ␣-1,2-linked mannotriose (Man␣1,2Man␣1,2Man) (mannose residue for the phosphorylation is shown in boldface type) is a common structure for the phosphorylation of Man8-GlcNAc2, Man5GlcNAc2, and mannose outer chain branch. Consistent with the previous result (15), we found that ⌬mnn6 mutant diminished the enzyme activity toward the ␣-1,2linked mannotriose (Man␣1,2Man␣1,2Man), which mimics  (10,16,29), are indicated. Peak x is also analyzed and identified as dimannosylphosphorylated Man8GlcNAc2-PA, as described under "Results." The sites of mannosylphosphorylation are shown in Fig. 8, A-1. both the N-linked outer chain branch and the O-linked oligosaccharides (data not shown). As described, ⌬mnn6 also showed a defect in the core oligosaccharide phosphorylation in vitro at the site of the ␣-1,6-branch (Man␣1,2Man␣1,6Man) (Fig. 8, A-1). Based on these data, we have proposed that MNN6 is involved in the phosphorylation of any oligosaccharides containing at least one ␣-1,2-linked mannobiose (Fig. 8B). In mammalian cells, GlcNAc-1-phosphate transferase showed a requirement for the presence of at least one Man␣1,2Man sequence on the glycoprotein acceptor (42). Since the acceptor sites for the core oligosaccharide are identical in yeast and mammals (38), the mannosylphosphate transferase in yeast may share some similar acceptor requirements with the mammalian GlcNAc-1-phosphate transferase.
Oligosaccharide profiles of total mannoproteins were compared between ⌬och1⌬mnn1 and ⌬och1⌬mnn1⌬mnn6 cells. Disruption of MNN6 did not exhibit any significant defect in the oligosaccharide phosphorylation in vivo (data not shown), consistent with the Alcian blue staining of the cells (Table II, G-3), suggesting the presence of redundant enzymes required for N-linked core oligosaccharide phosphorylation. Since the MNN6 is a member of the KRE2/MNT1 gene family, it is most likely that some genes in this family may function as homologues. Disruption of MNN6 in the ⌬mnn1 cells caused a dramatic loss of Alcian blue staining (Table II), indicating the involvement of MNN6 in the phosphorylation of N-linked outer chain in vivo. Since O-linked sugar chains are also phosphorylated (43), MNN6 may also affect the O-linked oligosaccharide phosphorylation.
Phosphorylated oligosaccharides are also found as a component of lipid glycoconjugate, lipophosphoglycan, on the cell surface of the protozoan, Leishmania (13,44). Phosphoglycan polymer contains the repeating disaccharide unit (Gal(␤1,4)Man␣1-PO 4 -6) attached to a conserved core, which in turn is linked to an unusual phosphatidylinositol-lipid anchor. The repeating units are synthesized by the alternating transfer of mannose 1-phosphate and galactose from GDP-mannose and UDP-galactose, respectively (13). Since these reactions involve transfer of mannose 1-phosphate from GDP-mannose, this study may also contribute to the understanding of other mannosylphosphate transferases, such as those in Leishmania. FIG. 8. Mannosylphosphorylation sites at the N-linked core oligosaccharides and its enzymatic reaction. A, structures for ManP-Man8GlcNAc2 and ManP-Man5GlcNAc2. The mannosylphosphorylation sites are indicated by asterisks. B, schematic presentation for the enzymatic reaction by Mnn6p. M and R represent mannose and any type of sugar residues, respectively. At least an ␣-1,2-linked mannobiose structure in mannotriose is required for mannosylphosphate addition by Mnn6p using GDP-mannose as a donor and producing GMP as the other reaction product.