Biosynthesis of yeast mannan. Properties of a mannosylphosphate transferase in Saccharomyces cerevisiae.

A homogenate of mechanically broken, freshly grown Saccharomyces cerevisiae X2180 cells catalyzes the transfer of mannosylphosphate units from guanosine diphosphate mannose to reduced alpha1 leads to 2-[3H]mannotetraose to yield reduced mannosylphosphoryl [3H]-mannotetraose. The product is analogous in structure to the phosphorylated mannan side chains, which suggests that the enzymic activity is involved in mannoprotein biosynthesis in the intact cell. The mannosylphosphate transferase activity, localized in a membrane fraction obtained by differential centrifugation at 100,000 x g, was solubilized by Triton X-155 and purified 250-fold by ammonium sulfate precipitation and by ion exchange and gell filtration chromatographies. The enzyme requires MN2+ OR Co2+ ions for activity and is stimulated by various detergents. The mnn2 and mnn3 mannan mutants of S. cerevisiae possess normal levels of mannosylphosphate transferase activity, whereas the mnn4 mutant cells contain very low, if any, activity. This is consistent with a previous conclusion that the mnn4 mutation affects the mannosylphosphate transferase activity, whereas the mnn2 and mnn3 strains possess phosphate-deficient mannans because they are unable to synthesize the appropriate side chain precursors. A new mannan mutant class with the mnn4 chemotype was isolated, but the mutation proved to be recessive and nonallelic with the mnn4 locus. This new locus is designated mnn6.

The product is analogous in structure to the phosphorylated mannan side chains, which suggests that the enzymic activity is involved in mannoprotein biosynthesis in the intact cell. The mannosylphosphate transferase activity, localized in a membrane fraction obtained by differential centrifugation at 100,000 x g, was solubilized by Triton X-155 and purified 250-fold by ammonium sulfate precipitation and by ion exchange and gel filtration chromatographies.
The enzyme requires Mn2+ or Co2+ ions for activity and is stimulated by various detergents. The mnn2 and mnn3 mannan mutants of S. cerevisiae possess normal levels of mannosylphosphate transferase activity, whereas the mnn4 mutant cells contain very low, if any, activity. This is consistent with a previous conclusion that the mnn4 mutation affects the mannosylphosphate transferase activity, whereas the mnn2 and mnn3 strains possess phosphate-deficient mannans because they are unable to synthesize the appropriate side chain precursors.
A new mannan mutant class with the mnn4 chemotype was isolated, but the mutation proved to be recessive and nonallelic with the mnn4 locus. This new locus is designated mnn6.
Saccharomyces cerevisiae cell wall mannoprotein contains mannosylphosphate and mannobiosylphosphate groups attached by a phosphodiester linkage to oligosaccharide side chains in the polysaccharide component of the glycoprotein (l-3). The function of these substituents is unknown, but they provide the major negative charge to the cell surface and appear to represent the only phosphate in the wall.
Changes observed in the mannan produced by yeast mutants with altered glycosyltransferase activities suggest that these glycosylphosphate units could be formed in a two-step process consisting of an initial transfer of the mannosylphosphate group from a sugar nucleotide to the acceptor side chain, followed by a transfer of mannose from sugar nucleotide to the mannosylphosphate group to form the ( disaccharide (4). The latter step would presumably be catalyzed by the known enzyme of broad specificity that adds al -+ 3-linked mannose in several different locations in the mannan (5,6).
Some of the S. cereuisiae mannan mutants lack phosphate in the mannoprotein because they are unable to make the appropriate acceptor side chains (the mnn2 and mnn3 classes),' whereas the mnn4 mutant appears to be deficient in the mannosylphosphate transferase activity (5,7). This latter mutation is dominant; that is, the heterozygous diploid shows the mutant phenotype. The present study was undertaken to devise an assay for the mannosylphosphate transferase activity in S. cereuisiae, to investigate the properties of the enzyme, and to elucidate the mechanism of dominance by the mnn4 mutation. and then resuspended in ice cold assay buffer equal in volume to two-thirds of the original digest. The protoplast suspension was sonicated for 30 s in a Branson Sonifier bath, centrifuged at 10,000 x gin a refrigerated centrifuge, and the pellet was resuspended in the same amount of assay buffer.
Whole cell homogenates, from 100 g of cells or more, were prepared in the Manton-Gaulin Laboratory Homogenizer from a slurry in assay buffer, 2 ml/g of wet cells. Smaller samples of material were disrupted in a Braun MSK homogenizer.
For samples of 1 g or less, an aluminum adapter similar to that of Needleman and Tzagoloff (21) was designed to hold four Pyrex ignition tubes (10 x 75 mm), each accommodating 1.25 g of beads and 0.75 ml of cell suspension.
The crude extract, that material remaining after the beads and whole cells were sedimented at 2000 x g for 10 min, was centrifuged in a Beckman preparative ultracentrifuge at 100,ooO x g for 1 h in a type 40 rotor. The pellet was washed in buffer, resuspended at 50 mg of protein/ml and the solution was made 1 mM in sodium EDTA, pH 7. After 30 min, it was either centrifuged and rewashed or dialyzed exhaustively against the buffer. This "EDTA-treated enzyme" was diluted with a 5% detergent solution to a concentration of 0.5 to 1% detergent and mixed gently overnight. The suspension was centrifuged at 100,000 X g for 1 h and the supernatant fraction was collected. For (NH&S04 fractionation, the powdered solid was added protein solution. All fractions were dialyzed exhaustively before assay and were concentrated by vacuum dialysis. A 5-ml sample of (NH4)&0,-fractionated material (30 mg of protein/ml) was fractionated on a Bio-Gel A-1.5m column (2 X 100 cm) equilibrated with 25 mu imidazole acetate, pH 6.5, and the eluted fractions were monitored for protein, carbohydrate, and transferase activity.
Unfractionated solubilized enzyme was treated similarly on a Bio-Gel A-0.5m column containing buffer with 0.5% (v/v) Triton X-155. Enzyme eluted from Bio-Gel A-1.5m was made 10 mu in buffer and applied to a column (2 x 20 cm) of Cellex P in that buffer. Although the enzyme did not bind, some purification was achieved by removal of contaminating proteins that remained on the column. Assays for Mannosylphosphate Transferase Activity-Reaction mixtures routinely contained enzyme extract (0.2 to 1 mg of protein), assay buffer (25 mM imidazole acetate buffer, pH 6.5 or 7.5), 6 mM MnCl*, 6 mu GDP-mannose, 10 mM reduced ["Hlmannotetraose-I (2 to 5 Ci/mol), and 0.2 to 1% Triton X-155, in a volume of 20 to 50 ~1. After incubation for 1 h at 3O"C, the reaction was terminated by the addition of 1 ml of cold water and the sample was applied to a Dowex 1 (Cl-) column (0.5 x 3 cm). The column was washed with 1 ml of water and 2 ml of 0.01 N HCl, after which the labeled product was eluted with 1.5 ml of 0.05 N HCl and a 0.5-ml sample was counted for radioactivity.
In  ) column (4 x 45 cm), which was washed with 500 ml of water and eluted with a 0 to 0.5 M NH,HCOa gradient (1 liter each), and the doubly labeled material was collected.

Mannosylphosphate
Transferase Assay-The presence in S. cereuisiae mannan of mannosylphosphate units attached to al + 2-linked mannooligosaccharide side chain units (24) suggested the existence of an enzyme capable of transferring the mannosylphosphate group to the mannan. An assay for this activity was designed utilizing unlabeled GDP-mannose and labeled oligosaccharide acceptors. The acidic labeled product was absorbed on a small ion exchange column, which was washed free of excess labeled acceptor before the enzymic product was eluted with acid. Potential labeled acceptors were prepared by reduction of oligosaccharides with ["Hlborohydride, which converted the sugar unit at the reducing end to a tritiated polyol. The standard assay employed reduced [3H]mannotetraose-I (Table I), aMan -+ 'aMan + 'aMan + 2["H]mannitol, but other acceptors were tested and are described below. Although the partially purified enzyme preparation contained other enzyme activities associated with mannoprotein biosynthesis, the al -+ 6-mannosyltransferase could not act because the mannose unit at the reducing end was converted to mannitol in the reduced acceptor, and the of Yeast Mannan ail + 3-mannosyltransferase has very low activity with al + 2-mannotetraose (6). The reaction was dependent on protein concentration and was linear with time up to 2 h. The enzyme required Mn*+ or Co'+ for activity and it was stimulated by nonionic detergents. Purification and Properties of the Enzyme- Table  II summarizes purification of the mannosylphosphate transferase from S. cerevisiae X2180 cells grown to stationary phase. Such cells bind Alcian Blue dye (18) and were presumed to possess high transferase activity. Subsequent experiments revealed that S. cerevisiae mnnl cells isolated during early logarithmic growth were also a good source of the mannosylphosphate transferase, and this was the enzyme used in most of the experiments.
Several nonionic detergents, representing a range of hydrophile-lipophile balance values (25), were tested for their ability to solubilize and stimulate the transferase activity. Brij 58 and Brij 35 solubilized almost as much activity as Triton X-155 and Brij 56, but they also solubilized more protein. Triton X-155 was selected for routine use because it gave good activity and formed a stable suspension in water at 4°C. The ratio of detergent to protein was important in determining whether the detergent stimulated or inhibited the activity. A detergent to protein weight ratio of 0.05 to 0.25 gave the highest activity, 2 to 4 times that without detergent. Although a mixture of 2 M urea and Triton X-100 is suitable for solubilization of the mannosyltransferases from S. cerevisiae (6), urea inactivated the mannosylphosphate transferase, even under conditions in which the detergent alone stimulated enzyme activity. Enzyme, solubilized by Triton X-155, was chromatographed on a Bio-Gel A-0.5m column equilibrated with buffer containing 0.5% detergent. The activity co-migrated with a single broad peak of protein that was eluted about halfway through the resolving space of the column. In the absence of detergent, activity recovered by (NH&SO4 fractionation was excluded from Bio-Gel A-1.5m, indicating that the solubilized enzyme may aggregate when the detergent concentration is low. Both the EDTA-treated and the detergent-solubilized enzyme exhibited maximal mannosylphosphate transferase activity from pH 6.5 to 8.5. Activity was inhibited by salt, and at 0.4 M NaCl or 0.3 M sodium phosphate, pH 7, less than 20% of the control value was observed. Full activity was restored by dialysis to remove the salt. Mn2+ and Co'+ at 6 InM stimulated activity about 25-fold and Fe"+ stimulated slightly, whereas Ca'+, Cup+, Hg", and Zn"+ inhibited, and Mg", Kt, Li+, and Na+ had no effect.
The enzyme was routinely stored in detergent. After 3 weeks, the extract in 1%~ Triton X-155 containing 25 mM imidazole acetate, pH 6.5, retained 85% of the activity when stored at 4°C and 60% if stored at -20°C. When incubated in the presence of substrates, the detergent-solubilized activity was stable at 30°C for 2 h. In the absence of substrates, all activity was lost within 5 min at 50°C.
Evidence for the Phosphodiester

Structure of the Assay
Product-'rhe products formed in an incubation of solubilized enzyme, reduced ["Hlmannotetraose-1, and GDP-["C]mannose were chromatographed on a DEAE-Sephadex A-25 column, which fractionates on the basis both of size and charge. The sole doubly labeled compound was eluted in a position ahead of reduced mannotetraose-II phosphate, suggesting that it was larger and less charged than this monoester reference (data not shown). Mild acid hydrolysis of the compound, under conditions that selectively break glycosylphosphate linkages, yielded ["C]mannose and reduced ["HImannotetraose-I phosphate (Fig. l), showing that both labeled substrates were incorporated.
The labeled product, isolated from an enzymatic reaction with reduced ["Hlmannotetraose-1 and unlabeled GDP-mannose, migrated as a single radioactive peak between the reduced mannotetraose acceptor and reduced mannotetraose phosphate monoester ( Fig. 2A). Although this product was unaltered by treatment with phosphomonoesterase ( Fig. 2R), mild acid hydrolysis yielded a new labeled substance with the electrophoretic property of reduced mannotetraose phosphate (Fig. 2C) that was susceptible to phosphomonoesterase digestion which converted it to a neutral labeled compound with the property of the starting oligosaccharide acceptor (Fig. 2D). The position of phosphate attachment to the oligosaccharide acceptor was established by exhaustive digestion with jack bean a-mannosidase (26). Reduced mannotriose phosphate was the smallest radioactive product obtained from the digestion of reduced [:'H]mannotetraose-1 phosphate, whereas similar treatment of the reference compound reduced ["HImannotetraose-II phosphate (Fig. 3) yielded reduced mannobiose phosphate, as expected from the published structure (3). These results indicate that the mannosylphosphate unit in the product formed in the enzyme assay is attached to the mannose next to the nonreducing terminus, analogous to its location on the trisaccharide side chains in the mnnl mutant mannan (24). The reactions employed in this characterization are summarized in Fig. 4.
Acceptor Specificity-Oligosaccharides (Table I) By the labeled components produced by mild acid hydrolysis of the initial product shown in A, which correspond to ['"Cjmannose and reduced l"H]marmotetraose phosphate. The solid line is "11 and the dotted line is Inc.
molecule and the linkage were important for activity (Table  III). The reduced ~yl --f 2-mannotetraose was the best acceptor, although the K,, is apparently so high that it did not saturate the enzyme even at 50 nl M concentration.
Larger oligosaccharides, with an Al-+ 6 linkage connecting two "side chain" units (a), were assayed with GDP["C]mannose as the donor (Fig. 5). None showed significant mannosylphosphate acceptor activity. Instead, the principal labeled product was the neutral oligosaccharide 1 mannose residue larger than the acceptor (established by gel filtration but data not shown), probably resulting from the action of an ~1 + 6mannosyltransferase in the preparation which is known to favor such branched acceptors (6). The large amount of GDP-['"C]mannose remaining at t,he end of the incubation, and the small amount of ["C]mannosylphosphate produced, show that the low activity of the mannosylphosphate transferase is not due to nonproductive degradation of the donor by contaminating enzymes. In fact, as much as 50% of the labeled mannose is incorporated into neutral products in the incubation with the mannohexaose, and less than 5% appears as mannosylphosphate.
Exogenous mannan from the mnnl,mnn4 mutant, which should contain many potential acceptor sites, was not an acceptor and did not compete with the oligosaccharide acceptor.

Specificity of Glycosylphosphate
Donors-GDP-mann3se was the best glycosylphosphate donor although the glucose derivative did show some activity (Table IV). The apparent K,,, for GDP-mannose was about 1 mM, but the enzyme was inhibited at higher concentrations.
Dolichyl [ "'CJmannosylphosphate, from yeast, gave no acidic radioactive product when incubated with enzyme and mannotetraose-I (Fig. 6). Both GDP and GMP, but not guanosine, inhibited the reaction significantly, as did mannose G-phosphate and mannobiose G-phosphate, presumably because of their similarity to the reaction product. The mnnl mannan, which should be analogous to the reaction product, was a poor inhibitor compared to wild type and H. polymorpha mannans.

Mannosylphosphate
Transferase in Yeast Mannan Mutants-The mannosylphosphate transferase activities in various S. cerevisiae strains, except those containing a mnn4 marker rff,,P DISTANCE MIGRATED mutation, were similar to the wild type activity (Table V).
The mnnA mutants showed much less activity. Even the heterozygous mnn4/wild type diploid had less than one-fourth the activity observed for the others, and it was comparable to the haploid mutant itself. If one-fourth of the mnnl pellet was replaced by buffer or mnn4 pellet, 75% of the original mannosylphosphate transferase activity was observed. Thus, the mnn4 extract was inactive and did not inhibit the action of the wild type enzyme. Transferase activity in protoplasm and in detergent-solubilized preparations behaved similarly to that in crude cell extracts. No degradation of [3H]mannotetraose-I mannosylphosphate was observed when it was incubated with the pellet fraction or the crude extract from any of the strains with or without GDP-mannose (data not shown). These results suggest that the extracts did not contain an active phosphodiesterase capable of hydrolyzing the mannosylphosphate diester group in vitro.
Isolation of New Mannosylphosphate Transferase Mutants-cultures of S. cerevisiae XW-452 (mnnl) were mutagenized with ethyl methanesulfonate and negative selection procedures (5) were employed to obtain seven clones that did not bind Alcian Blue dye or agglutinate with anti-K. breuis, anti-mnn2, or anti-X2180 sera. Segregation of the mnn4 trait, determined by Alcian Blue dye binding and immunochemical analysis of the tetrads of matings with parental and wild type strains, showed that each mutation involved a single gene that was not linked to mnnl. The diploids of three of the clones crossed with mnnl did not bind dye; thus, these mutations, like the original mnn4 strain, were dominant. Four of the isolates were recessive, since the heterozygous diploid formed with the mnnl strain exhibited the properties of mnnl cells.

FIG. 2. Electrophoretic
analysis of the product from incubation of the solubilixed enzyme with reduced [3H]mannotetraose and unlabeled GDP-mannose. A, the reaction product eluted from the Dowex 1 column; B, after treatment of the reaction product with phosphomonoesterase; C, after mild acid hydrolysis of the reaction product; D, after sequential acid hydrolysis and phosphomonoestereee treatment of the reaction product.
All of the recessive mutants were found to be allelic by complementation tests in the diploid. Although they gave the acetolysis pattern characteristic of mnn4 mutants (Fig. 7), lacking only the peak for mannotetraose phosphate, they segregated independently of the dominant mnn4 strain (4 parental ditypes, 4 nonparental ditypes and 15 tetratypes in 23 asci dissected). The results suggest that they involve a new genetic locus for mannosylphosphate transferase activity. This new locus is designated mnn6, and its properties are compared to those of other strains in Table V.  (4,27), and selective assays have been developed for studies on biosynthesis of the serine-and threonine-linked sugars (22), the asparagine-linked core sugars (33), and the outer chain sugars (6,34). The focus of this study is the presumed mannosylphosphate transferase that is involved in the incorporation of phosphate into the mannan molecule. Although mannosylphosphate transfer in S. cerevisiae has not been reported  4. Scheme for characterization of the mannosylphosphate transferase assay product. The enzyme-catalyzed reaction (a) yields a product that is stable to phosphomonoesterase, but it is hydrolyzed by mild acid ( b) to give a product that is converted by the phosphatase (cZ) to a neutral oligosaccharide. The same product is converted by a-mannosidsse (c) to a mannotriose phosphate that is degraded by phosphatase (d) to a neutral trisaccharide. * indicates position of 'H.
previously, a number of mutants with phosphate-deficient mannans were known (5,7). Because of the dominance of one of these mutations (35), we felt that a detailed study of the enzyme that catalyzes this reaction could provide new insight into the mechanisms by whichmannoprotein biosynthesis is regulated.
The assay procedure we have developed is convenient be-  5. Electrophoretic analysis of the products from incubation of solubilized mannosylphosphate transferase with GDP-['%]mannose and various acceptors. A, endogenous acceptor; B, mannotetraose-I; C, mannopentaose-II; D, mannohexaose; and E, mannoheptaose. The mobility of reference compounds is indicated at the top. The labeled product in A is probably polymeric mannoprotein, and the additional peak in B is the mannosylphosphate mannotetraose, whereas that in the other incubations is an oligosaccharide one mannose unit larger than the acceptor. cause the neutral oligosaccharide acceptor is labeled rather than the nucleotide sugar donor. Thus, neither the charged degradation products, which are unlabeled, nor the neutral labeled by-products complicate the measurement of radioactive material converted to an acidic form by enzymic action. The use of an exogenous acceptor has made it possible to demonstrate that the mnn2 mutant, which was presumed to have a phosphate-deficient mannan because it lacked endogenous acceptors (27), does possess a normal level of mannosylphosphate transferase activity.
The solubilized enzyme has properties that are consistent with a role in mannan biosynthesis; it has a preference for acceptor oligosaccharides with an crl + 2 linkage at the nonreducing end of the chain, and it has a much lower activity with acceptors that have been modified by the cul + 3-mannosyltransferase.
Moreover, the structure of the product synthesized in vitro is analogous to that of phosphorylated side chains in mannan from a mnnl mutant.
The acceptor structural specificity for the mannosylphosphate transferase in S. cerevisiae differs from that for the Nacetylglucosamine transferase in Kluyveromyces lactis (36). The latter enzyme shows a specificity for an (~1 + 3 nonreducing terminus, and the mnnl mutant of K. lactis, that is defective in an arl ---, 3-mannosyltransferase, lacks N-acetylglucosamine because the acceptor is missing. This situation is analogous to the phosphate-deficient mannan found in mnn2 and mnn.5 strains of S. cerevisiae, in which the mannosylphosphate acceptor site is absent due to defective al + 2mannosyltransferases.
In each of these mutants, the respective transferase activities are found in extracts when assayed with exogenous acceptors.
Although the relative specificity of the enzyme is apparent, mannosylphosphate transfer to reduced mannotetraose-I is about 200-fold slower than the rate calculated for the corresponding activity in the cell. The K, for the reduced tetrasaccharide acceptor appears to be greater than 1 M, which is far higher than the 0.2 to 7.5 mu values reported for various and mnnl mutants, which would be expected to have acceptor and product sites, did not compete significantly with the reduced mannotetraose.
Thus, we found no evidence that the enzyme can act on large mannan molecules.
No evidence was obtained that a lipid-bound mannosylphosphate derivative was involved in the reaction. Labeled oligosaccharide phosphate was not detected in incubations in which dolichyl mannosylphosphate was present as a potential donor, whereas much of the labeled mannose was incorporated into apparently neutral products. The result is not unexpected because thermodynamic considerations predict the involvement of a mannosylpyrophosphate derivative as the donor for a mannosylphosphate group, and such a derivative of dolichol has not been detected in yeast.
The dominance of the mnn4 mutation, which leads to a reduced level of mannosylphosphate groups even in the mannan of the mnn4/wild type heterozygous diploid (7,35), remains an unexplained phenomenon. Lowered phosphate levels in the mannan of the heterozygote could arise in two ways-either the wild type allele is expressed but cannot compensate for the mutant allele or the mutant allele blocks all wild type expression. Mechanisms consistent with the first explanation could involve the production of an inhibitor of the transferase or an enzyme that degraded the product. We found no support for either of these possibilities; no inhibitor specific to the mnn4 strain was detected by mixing extracts, nor could we demonstrate a phosphodiesterase that cleaved the mannosylphosphate mannotetraose product. An additional possibility is that the transferase is a multimeric enzyme that is subject to negative complementation, i.e. a single defective subunit could lead to complete inactivation of normal enzyme molecules. One example is the "paralysis" mutation in the catalytic subunit of Escherichia coli aspartate transcarbamylase, in which enzyme hybrids containing one mutant allele and one normal allele are almost completely inactive (37).