Biosynthesis of Methylphosphomannosyl Residues in the Oligosaccharides of Dictyostelium discoideum Glycoproteins EVIDENCE THAT THE METHYL GROUP IS DERIVED FROM METHIONINE*

The phosphorylated oligosaccharides of Dictyoste- lium diecoideum contain methylphosphomannosyl residues which are stable to mild-acid and base hydrolysis (Gabel, C. A., Costello, C. E., Reinhold, V. N., Kurtz, L., and Kornfeld, S. (1984) J. Biol. Chem. 259,13762- 13769). Here we present evidence that these methyl groups are derived from [methyl-SH]methionine, in vivo and [methyl-’HIS-adenosylmethionine in vitro. About 18% of the macromolecules secreted from vege-tative cells labeled with [methyl-SH]methionine are released by digestion with preparations of endoglycos-idaselpeptide N-glycosidase F. The majority of the re- leased molecules are sulfated, anionic high mannose-type oligosaccharides. Strong acid hydrolysis of the [SH]methyl-labeled molecules yields [‘Hlmethanol with kinetics of release similar to those found for the gen-eration of Man-6-P from chemically synthesized methylphosphomannose methylglycoside. Treatment of the [‘Hlmethyl-labeled molecules with a phosphodiesterase from Aepergillua niger which is known to cleave this phosphodiester also releases [‘Hlmethanol from a portion of the oligosaccharides.

The N-linked oligosaccharides of mammalian lysosomal enzymes contain residues of Man-6-P which are synthesized in a two-step process. First, GlcNAc-1-P is transferred from UDP-GlcNAc to selected Man residues to form a mild acidlabile phosphodiester (Man-6-P-GlcNAc), and second the GlcNAc residue is rapidly removed by a specific glycosidase to generate Man-6-P in monoester linkage. Both of these reactions occur in the Golgi and are essential for the transport of these newly synthesized enzymes to the lysosome (reviewed in Ref. 1).
The lysosomal enzymes of the slime mold Dictyostelium discoideum contain anionic N-linked oligosaccharides many of which are sulfated and also carry Man-6-P (2). These residues occur almost exclusively in an acid-stable phosphodiester different from the mammalian type (2). This phosphodiester has recently been identified as Man-6-P methyl ester (Man-6-P-OCH3) (3). The physiological function and the route of synthesis of this molecule are unknown, but it is likely that the phosphate and methyl groups are added after cotranslational glycosylation of the proteins (4). Therefore, these modifications probably occur during the transit of the proteins from the endoplasmic reticulum to the Golgi or to the lysosome (4). The sulfation of the oligosaccharides of amannosidase is known to occur in the Golgi (4).
Most of the known methyltransferase reactions use Sadenosylmethionine as a methyl group donor (reviewed in Ref. 5). Since methionine is converted to S-adenosylmethionine in a single step, [methyL3H]Met is often used as a convenient precursor for the in vivo labeling of a variety of methylated molecules (5).
In this report we have examined the in vivo incorporation of [methyL3H]Met into phosphorylated oligosaccharides derived from secreted glycoproteins and of [met/~yl-~H]AdoMet' into endogenous acceptors. The results suggest that AdoMet can serve as a direct or indirect source of this methyl group.

RESULTS
Labeling and Endo/PNGase F Digestion of Secreted Macromolecules-Cells of D. discoideum were grown in the presence of [methyL3H]Met for 3 days and allowed to secrete lysosomal enzymes and other proteins into the culture medium (6). The secreted macromolecules were treated with Endo/PNGase F to release the susceptible N-linked oligosaccharides (7), and chromatographed on Sephadex G-50. As shown in Fig. 1, nearly 18% of the total 3H label was released by this digestion. The released material occurred in two regions of the column. The major peak (R-1) coincided with the distribution of a similar enzymatic digestion [3H]Manlabeled secreted macromolecules (8) (not shown), and the second region (R-2) had an apparent size between a MansGlcNAc oligosaccharide and a monosaccharide.
The total cell material from this labeling was similarly analyzed. Less than 3% of the total label was released in the R-1 region (data not shown). Thus, the secreted macromole-The abbreviations used are: AdoMet, S-adenosylmethionine; Endo/PNGase F, endoglycosidase preparation from Flavobacterium menigwsepticum which contains endoglycosidase and peptide N-glyeosidase activities; SDS, sodium dodecyl sulfate; TCA, trichloroacetic acid; Man-6-P-OCH3, methylphosphomannosyl residue; HPLC, high performance liquid chromatography; Endo, endoglycosidase.  cules were considerably enriched in the EndoIPNGase F releasable radioactivity.
General Characterization of the R-1 and R-2 Fractions (Not Shown)-Greater than 98% of pool R-1 was anionic while pool R-2 contained about 60% anionic molecules, most of which required at least 200 mM NaCl for elution from QAE-Sephadex. About 38% of the R-2 molecules were cationic as shown by binding to Dowex 50.
About 56% of 3H in R-1 was bound to ConA-Sepharose. This was increased to >90% by solvolysis, a procedure which is known to remove sulfate esters without breaking glycosidic linkages (2,9). The 3H label derived from [rnetl~yl-~H]Met is totally stable to this treatment. The removal of sulfate esters also increased ConA-Sepharose binding of 3H-labeled oligosaccharides. R-2 did not interact with ConA-Sepharose either before or after solvolysis. We conclude that the R-1 molecules are high mannose-type, anionic oligosaccharides some of which are sulfated. Pool R-2 appears not to be typical of high mannose-type oligosaccharides and was not extensively characterized beyond this point.
Strong acid hydrolysis of the R-1 pool followed by HPLC analysis showed that none of the 3H chromatographed with Man, fucose, or GlcNAc. Therefore, the 3H was not incorporated into the ring structure of the sugars themselves. Anion exchange HPLC resolves the R-1 oligosaccharides into molecules with 1-6 negative charges. No neutral oligosaccharides were found.
Kinetics of Strong Acid Hydrolysis and Identification of the Products-The Man-6-P-OCH3 diester found in secreted glycoproteins is insensitive to phosphatase digestion and requires strong acid hydrolysis to generate a phosphatase-sensitive form (3).
If the [3H]methyl label is incorporated into the Man-6-P-OCH3, strong acid hydrolysis should produce Man-6-P and [3H]methanol as products. Furthermore, it should be stable to base in contrast to more common methyl esters such as those found in carboxylic acids (3, 11). In fact, it is stable to 0.1 N NaOH for 4 h at 37 "C. Only 15% is degraded by 1 N NaOH at 80 "C for 10 h (data not shown).
The kinetics of acid-catalyzed hydrolysis of [3H]Man-6-P-OCH, shows that 4 h of hydrolysis in 1 N HCl at 100°C is required to achieve >95% conversion of the diester to the monoester form, which is then fairly stable. The [3H]methyllabeled oligosaccharides were subjected to similar acid hydrolysis conditions and the loss of 3H quantified by evaporation in the presence of a non-volatile internal standard ([ "C] mannose). The results are shown in Fig. 2. About 90% of the label was hydrolyzed with a half-life of 95 min, and longer time points (not shown) of up to 24 h gave no further loss of 3H label. This half-life was significantly longer than that of the chemically synthesized Man-6-P-OCH3 (tlh= 55 min). To determine whether this difference was due to the presence of a glycosidic linkage in sugars bearing the methyl group, the a-methylglycoside of Man-6-OCH3 was synthesized and the kinetics of its conversion to Man-6-P were measured (Fig. 2).
The T% was calculated to be 66 min. Although the presence of the glycosidic linkage may influence the rate of hydrolysis slightly, it does not appear to totally account for the difference seen. Furthermore, when the oligosaccharides are desulfated by solvolysis prior to strong acid hydrolysis, the tlh of [3H] methyl-labeled oligosaccharides was still 95 min. It seems unlikely that the presence of the sulfate esters affects the rate of hydrolysis of the ['Hlmethyl groups. The difference in the rate of hydrolysis may depend on structure of the oligosaccharide itself compared to the monosaccharide standard, Man-6-P-OCH3. It is unlikely that the difference is due to an isotope effect ('H uersus 3H) since this would only alter the rate by only -3% (12). A similar acid treatment of the molecules from the R-2 region of the column indicated that only about 10% were labile and that the remainder was stable up to 9 h of hydrolysis.
To confirm that the 3H label in the R-1 region was actually present in a methyl ester, the production of [3H]methanol was confirmed via HPLC, as shown in Fig. 3. Note that the kinetics of the appearance of the product were the same as were hydrolyzed under these conditions as shown in Fig. 3. The TI, is also shown in minutes.   Fig. 2).
Taken together these results suggest that the 13H]methyl group is incorporated into a base-resistant, relatively acidresistant methyl ester which has kinetics of acid hydrolysis similar, but not identical to Man-6-P-OCH3.
Semitivity to Aspergillus niger Phosphodiesterase-The Man-6-P-OCH3 diester can be cleaved by a phosphodiesterase from A. niger (3). This enzyme or a closely related one will cleave glycerol phosphodiesters, glucose 6-phospho-~~-l'-(2'hydroxy-3'-ethoxy)propane, and Man 3,4-cyclic phosphate (3,13). About 85% of synthetic [3H]Man-6-P-OCH3 can be cleaved by this enzyme preparation, as shown in Fig. 4. To determine whether the [3H]methyl-label could be released from an intact oligosaccharide, solvolysis-treated oligosaccharides were separated into molecules containing 1 or 2 phosphodiesters and treated for various times with a partially purified preparation of the phosphodiesterase. As a control, [3H]Man-labeled oligosaccharides were prepared and treated identically. The results in Fig. 4 show that the oligosaccharides with 2 phosphodiesters were completely resistant to the digestion, and that those with only a single phosphodiester were partially sensitive. The reason for the greater sensitivity of the [3H]methyl-labeled oligosaccharides (40%) compared to the same [3H]Man-labeled molecules (20%) is unknown, but it may depend on other structural features of the molecule, such as the number of Man residues in the sensitive compared to the resistant class.
Here again, it appears that the loss of I3H]methano1 and the increase of charge on the [3H]Man-labeled oligosaccharides (converting the diester to a monoester) occurred coordinately for a portion of the oligosaccharides bearing a single phosphodiester.  Table I indicate that nearly 9% of the trichloroacetic acidprecipitable label was bound to ConA-Sepharose. When the homogenate was fractionated into a crude membrane and 100,000 X g supernate fraction, nearly all of the activity was found in the membrane fraction. It was important to gently homogenize the cells to preserve a high activity in the membrane fraction compared to the S-100.
The results of a time course of incorporation of [3H]AdoMet into crude membranes are shown in Fig. 5. Much of the label is incorporated into unknown molecules which loose the label on prolonged incubation. Mild base treatment which is sufficient to destroy carboxymethyl esters (1 1) shows that most of  the [3H]methyl is probably incorporated into such molecules. The label is also stably incorporated into molecules which are resistant to this base treatment. These products could include methylated amino acids and the Man-6-P-OCH3 of the oligosaccharides. Strong acid hydrolysis (1 N HCl, 4 h, 100°C) sufficient to destroy Man-6-P-OCH3, showed that a portion of the label was also stably incorporated into an acid-stable fraction. The incorporation was dependent on time and protein concentration and was totally abolished by 0.1% Triton X-100 and 1 mM non-labeled AdoMet (data not shown).
Subcellular Fractionation-The crude membranes were further fractionated using a step gradient of increasing sucrose concentrations, and the transferase activities measured in each fraction. N-Acetylglucosaminidase and a-glucosidase I1 served as markers of the lysosomal and rough endoplasmic reticulum fractions, respectively (4). The highest specific activity of transfer into ConA-Sepharose bindable material was associated with light density membranes which have been identified by others as Golgi (4) (Fig. 6).
Churaeterizution of the in Vitro Products Released by Endo/ PNGase F Digestion-About 5% of the total 3H could be released by Endo/PNGase F (Fig. 7). Most of this material was found in the R-2 region (see Fig. 1) which did not bind to ConA-Sepharose. The remainder (R-1) appeared to run in a position somewhat smaller than the oligosaccharides released by a similar digestion of labeled in vivo oligosaccharides. The products labeled in vitro from the R-1 region were both neutral and anionic and about 50% bound to ConA-Sepharose (not shown). Acid hydrolysis of the Endo/PNGase F material released volatile 3H (presumably methanol) with kinetics similar to those seen for the oligosaccharides synthesized in vivo (Fig. 8). Therefore, these moleculegappeared to be the same as those labeled with [methyL3H]Met in vivo. However, when the ConA bound oligosaccharides were analyzed by QAE-Sephadex chromatography, only non-sulfated species seemed to have been acceptors. This is shown by the fact that solvolysis treatment of the unfractionated pool changed the QAE-Sephadex profile only slightly.
Based on these results we conclude that AdoMet can serve as a donor of the methyl group donor of the methyl phosphodiester in an in vitro assay. At this point, however, we do not know whether AdoMet is the immediate donor to a Man-6-P already located on the oligosaccharide at the time of methyl transfer or whether the methyl group is transferred together with phosphate.

DISCUSSION
The results of in vivo labeling presented here indicate that [methyL3H]Met can donate a methyl group to the Man-6-P-OCH3 containing oligosaccharides released by Endo/PNGase F digestion. These conclusions are based on the released by Endo/PNGase F digestion, binding to ConA-Sepharose, HPLC analysis of anionic species, kinetics of acid hydrolysis, base stability, release of [3H]methanol, and sensitivity to A.

niger phosphodiesterase (3). A very large proportion of the 3H
incorporated into secreted macromolecules (90%) appears to be found in the methylphosphomannosyl residues. Although we do not know the absolute amounts synthesized during the labeling period, other data suggests that this phosphodiester may be plentiful on many secreted glycoproteins (8). The remaining 10% of the 3H was not found in the ring structure of the sugars themselves or in neutral oligosaccharides, but it is possible that some could be present as 0-methyl ethers or as N-methyl substitutions of GlcN. Studies are in progress to answer these questions. These characterizations have been done only on the R-1 fractions (Fig. 1).
The nature of the R-2 fraction has not been investigated thoroughly, but its ability to be released by Endo/PNGase F suggests that it is derived from a GlcNAc-Asn linkage typical of N-linked oligosaccharides. The majority appears to be highly anionic and the insensitivity to strong acid hydrolysis suggests that it must be unrelated to the methylphosphomannosyl residues. A reasonable possibility is that the methyl group is present as an 0-methyl ether or as an N-methyl-GlcN. Since the R-2 appears to be the major product of in   The in vitro incorporation of a portion of the 3H from [~nethyl-~HIAdoMet into a fraction which binds to ConA-Sepharose shows that AdoMet can serve as a donor for this reaction. We do not know if it is the immediate donor nor do we know if the synthesis of the methylphosphodiester occurs as a single or multi-step reaction. Of the several possible routes of synthesis, none would appear to be of the type seen for the mammalian lysosomal enzymes (1). Phosphorylation does not occur cotranslationally at the time of initial glycosylation (4). One possibility is that the phosphomethyl group is donated to the appropriate Man residues on an oligosaccharide in a single step. This would require the participation of a high energy donor such as a methylphosphodiester of ATP, although such a compound is not known to exist. In other experiments4 the inclusion of 1 mM ATP or [ T -~~P I A T P did not stimulate incorporation of 3H into ConA bindable in uitro products or incorporation of 32P, respectively.
Alternatively, methylation of a Man-6-P residue already present on the oligosaccharide may occur. The initial phosphorylation might proceed through kinase-type reaction specific for an oligosaccharide. If an oligosaccharide kinase exists, the subsequent methyltransferase reaction must be extremely efficient since <2% of all of the oligosaccharides with Man-6-P are found in phosphomonoester linkage (8).
Yet another possibility is that the methylation of the phosphate displaces another group already in phosphodiester linkage. The use of methylation inhibitors would allow us to distinguish between these possibilities based on the presence of phosphate and its state of esterification; unfortunately, none of the eight inhibitors we tried were effective. No mutant strains of Dictyosteliurn lacking Man-6-P have been isolated, although many post-translational modification mutants have been screened in an enzyme-linked immunosorbent assay using the mammalian Man-6-P receptor and an antibody against it.5 Thus, the route of biosynthesis of this unusual phosphodiester remains unknown.
The cell fractionation studies suggest that oligosaccharide methylation occurs in the Golgi where the sulfation of amannosidase is known to occur (4). The inhibition of this reaction by 0.1% Triton X-100 suggests, but does not prove,  that precursor may be transported into the Golgi analogous to the transport of nucleotide sugars and the activated sulfate donor, adenosine 3"phosphate 5'-phosphosulfate (17)(18)(19). Under the experimental conditions used in uitro, methylation appears to preceed sulfation. The modification of the oligosaccharides occurs very rapidly in Dictyosteliun (4, 20). Like mammalian cells, the processing of oligosaccharides and crucial decisions about the routing and secretion of the proteins bearing these sugars are probably made in the Golgi (1,4, 20).