Structural Analysis of N-Linked Oligosaccharides from Glycoproteins Secreted by Dictyostelium discoideum IDENTIFICATION OF MANNOSE &SULFATE*

The N-linked oligosaccharides found on the lysoso- mal enzymes from Dictyostelium discoideum are highly sulfated and contain methylphosphomannosyl residues (Gabel, C. A., Costello, C. E., Reinhold, V. N., Kurtz, L., and Kornfeld, S. (1984) J. Biol. Chern. 269, 13762-13769). Here we report studies done on the structure of N-linked oligosaccharides found on proteins secreted during growth, a major portion of which are lysosomal enzymes. Cells were metabolically labeled with [2-sH]Man and s6S04 and a portion of the oligosaccharides were released by a sequential digestion with endoglycosidase H followed by endoglycosi- dase/peptide N-glycosidase F preparations. The oligosaccharides were separated by anion exchange high performance liquid chromatography into fractions containing from one up to six negative charges. Some of the oligosaccharides contained only sulfate esters or phosphodiesters, but most contained both. Less than 2% of the oligosaccharides contained a phosphomon-oester or an acid-sensitive phosphodiester typical of the mammalian lysosomal enzymes. A combination of acid and base hydrolysis suggest$ that most of the sulfate esters were linked to primary hydroxyl groups. The presence of Man-6-SO4 was demonstrated by the appearance of in acid hydrolysates of base-treated,

Structural Analysis of N-Linked Oligosaccharides from Glycoproteins Secreted by Dictyostelium discoideum IDENTIFICATION OF MANNOSE &SULFATE* (Received for publication, July 1, 1985) Hudson H. Freeze and David Wolgast From the University of California Sun Diego, Cancer Cent The N-linked oligosaccharides found on the lysosomal enzymes from Dictyostelium discoideum are highly sulfated and contain methylphosphomannosyl residues (Gabel, C. A., Costello, C. E., Reinhold, V. N., Kurtz, L.,  J. Biol. Chern. 269, 13762-13769). Here we report studies done on the structure of N-linked oligosaccharides found on proteins secreted during growth, a major portion of which are lysosomal enzymes. Cells were metabolically labeled with [2-sH]Man and s6S04 and a portion of the oligosaccharides were released by a sequential digestion with endoglycosidase H followed by endoglycosidase/peptide N-glycosidase F preparations. The oligosaccharides were separated by anion exchange high performance liquid chromatography into fractions containing from one up to six negative charges. Some of the oligosaccharides contained only sulfate esters or phosphodiesters, but most contained both. Less than 2% of the oligosaccharides contained a phosphomonoester or an acid-sensitive phosphodiester typical of the mammalian lysosomal enzymes. A combination of acid and base hydrolysis suggest$ that most of the sulfate esters were linked to primary hydroxyl groups. The presence of Man-6-SO4 was demonstrated by the appearance of 3,6-anhydromannose in acid hydrolysates of base-treated, reduced oligosaccharides. These residues were not detected in acid hydrolysates without prior base treatment or in oligosaccharides first treated by solvolysis to remove sulfate esters. Based on high performance liquid chromatography quantitation of percentage of sH label found in 3,6-anhydromannose, it is likely that Man-6-SO4 accounts for the majority of the sulfated sugars in the oligosaccharides released from the secreted glycoproteins.
The lysosomal enzymes of Dictyostelium discoideum display a diverse series of N-linked oligosaccharides (l), which appear to be derived from the usual lipid-linked oligosaccharide precursor (2,3). Many of the oligosaccharides found on the slime mold enzymes contain a variable number of sulfate esters and Man-6-P in a methylphosphodiester linkage (Man-6-P-OCH3) (4). Certain sulfate esters block the release of the majority of the oligosaccharides by Endo H' (1); however, *This research was supported by National Institutes of Health Grant GM32485. 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.
In our previous studies on the structure of the [3H]Manlabeled oligosaccharides we were not able to identify the sulfated sugar (1,2). There were several reasons for this, but the most difficult problem was the lack of quantitative methods which could be readily applied to metabolically labeled molecules.
In this report we present evidence for the release of a large portion of the N-linked oligosaccharides from an enriched source of lysosomal enzymes. We also identify the presence of a novel sulfated sugar, Man-6-SOI in these oligosaccharides using approaches which can be applied to the analysis of other biosynthetically labeled sulfated oligosaccharides.

RESULTS
Labeling and Release of N-Linked Oligosaccharides from Secreted Macromolecules-The array of N-linked oligosaccharides found on each of three different purified lysosomal enzymes from the growth medium of Dictyostelium were very similar to each other, in terms of size, degree of sulfation, the amount and state of esterification of the Man-6-P (1). Furthermore, all of the lysosomal enzymes and a large portion of the secreted proteins are known to contain an antigenic determinant which requires a sulfated oligosaccharide for its recognition (5,6). Therefore, it seemed likely that the secreted proteins would be a rich source of material for characterization of the lysosomal enzyme oligosaccharides. To successfully analyze the oligosaccharides, they must first be cleaved from the proteins. To accomplish this we used a combination of digestions of Endo H (7) and the newly described Endo/ PNGase F to release the oligosaccharides (8). The notation is used because it has recently been reported that this latter preparation contains at least two enzymatic activities (9). One of these is a true endoglycosidase and cleaves between the 2 GlcNAc residues in the di-N-acetyl chitobiosyl core, and the other activity is a peptide N-glycosidase which cleaves the amide linkage to Asn. Thus, there is uncertainty as to which activity is responsible for the cleavage seen. This, however, does not change the overall interpretation of the results.

N-Linked Oligosaccharides
In preliminary experiments (data not shown) at least 80% of the [3H]Man-labeled oligosaccharides were released from purified a-mannosidase or from a mixture of three purified lysosomal enzymes by digestion with this preparation. These results suggested that the great majority of the lysosomal enzyme oligosaccharides could be successfully cleaved by the enzyme. Due to the limited amount of labeled oligosaccharides available from the purified enzymes, we decided to first analyze the oligosaccharides released from total cellular secretions.
Macromolecules secreted by cells labeled with 35s04 and [3H]Man were denatured in sodium dodecyl sulfate/2-mercaptoethanol and chromatographed on Sephadex G-50. The fractions from the void volume were pooled, digested with Endo H, and rechromatographed on the Sephadex G-50 column ( Fig. 1, Panel A). The released oligosaccharides ran within the column while the noncleaved glycoconjugates remained in the Vo region of the column. About 9% of the 3H and 7% of the 35S04 were released by this treatment. &-digestion of the material in the Vo region of the column with Endo H did not release any more oligosaccharides (not shown). The material which still remained in the Vo was re-digested with Endo/ PNGase F (Endo H + Endo/PNGase F) and again rerun on the same column ( in 0.5 M sodium borohydride) releases about 35% of the 3H label. Strong acid hydrolysis showed about 20% of the 3H is [3H]mannitol, and suggests that some Man residues could be directly linked through an 0-glycosidic linkage to serine or threonine. About 40% of the unreleased 3H was bound to ConA-Sepharose both before and after solvolysis treatment, indicating that some of the N-linked high mannose-type oligosaccharides were not released by either Endo H or Endo/ PNGase F digestions.
Cells were also labeled with [3H)GlcN and treated in the same way as the [3H]Man-labeled molecules. Only 0.4% of the label was released by an Endo H digestion, while subsequent Endo/PNGase F digestion released about 5%. Nearly all of the label was found in GlcNAc in both the released and total macromolecules. The characterization of this material was not pursued further. Approximately 77% of the total released 3H and 60% of the 35S04 were bound to ConA-Sepharose. Following solvolysis to remove the sulfate esters (13), about 90% of the 3H label was bound to the lectin column. This indicates that the presence of the sulfate esters block the binding of some of these oligosaccharides to the column.
QAE-Sephadex Analysis of the Oligosaccharides-The anionic oligosaccharides were separated by QAE-Sephadex chromatography ( Fig. 2) (1). Molecules with zero, one, two, and three charges are eluted at 0, 20, 70, 125 mM NaC1, respectively, and 400 mM elutes molecules with 4-6 negative charges. Solvolysis treatment of both Endo H and Endo H + Endo/ PNGase F released molecules (Panels B and D, respectively) generated species with only up to three charges, indicating that sulfate must account for the remainder of the charge.
To determine whether any of the charged oligosaccharides contained phosphomonoesters, each of the solvolysis treated, anionic oligosaccharides were pooled separately, desalted and digested with alkaline phosphatase, and reanalyzed again on QAE-Sephadex. None of the molecules with two charges contained any phosphomonoester, while about 20% of the molecules with three charges contain a phosphomonoester. This is only about 1.3% of all released oligosaccharides. Anion Exchange HPLC Separation of the Released Oligosaccharides-Anion exchange HPLC analysis was used to more effectively separate the most highly charged oligosaccharides (14). The results of the fractionation are shown in Fig. 3. Each of the pools marked by the brackets on Fig. 3 was pooled, desalted, and analyzed as described below.
Characterization of the Various Species-Each fraction was analyzed for amount and state of esterification of Man-6-P, sulfate ester content, and relative percentages of molecules with various combinations of phosphate and sulfate esters. The results of these analyses are shown in Table I. It is important to give the rationale for these analyses here. The relative amounts of molecules with multiple sulfate esters and mono-or di-esterified Man-6-P residues are based on the fact that all of the sulfate esters are sensitive to solvolysis which does not destroy any glycosidic linkages (13). Although the phosphodiester found on mammalian lysosomal enzymes (GlcNAc) is cleaved by solvolysis (4), the -0CH3 group of the phosphodiester is totally stable to the procedure. Cleavage of the mammalian-type phosphodiester yields a phosphomonoester which is sensitive to alkaline phosphatase. Thus, phosphatase sensitivity of an oligosaccharide following solvolysis, but not before solvolysis, could be due either to the loss of an acid-labile phosphodiester or to the failure of the phosphatase to cleave a phosphomonoester. Since we cannot distinguish between these possibilities we have grouped these two possibilities together in Table I. In addition, each fraction was treated with alkaline phosphatase and rechromatographed on QAE-Sephadex, both before and after solvolysis. This was important because the presence of the sulfate esters could inhibit the action of the phosphatase on the few molecules  Fig. 1, Panels A and B were desalted and applied to a 300 X 4-mm column of Ax-5 in a total of 50 pl. The column was eluted at a flow rate of 1 ml/min for 2 min with water, followed by a gradient of 15-30 mM NaPOI for 20 min, followed by a gradient of 30-375 mM NaP0, for 22 min. Fractions of 0.5 min were taken and 10 pl of each sample were counted for 3H (0) and %304 (0). The fractions were pooled as indicated in each panel. Panel A, Endo H-released oligosaccharides; Panel B, Endo H + Endo/PNGase F-related oligosaccharides. The standards were: A, Glc-6-SOI; B, C, and D were derived from a partial hydrolysate of dermatan sulfate (IdUA-GalNAc-4-SO4), and are di-, tetra-, and hexasaccharides with two, four, and six negative charges, respectively. which are sensitive (15). Several fractions which had no sulfate esters were also treated with phosphatase without solvolysis as shown in Table I. Each of the fractions shown in Table I was chromatographed on ConA-Sepharose (16). Most of the species were bound, but the proportion elution at 10 mM methylglucoside and 100 mM methylmannoside varied with each fraction.
The neutral oligosaccharides released by both of the enzyme treatments were analyzed by HPLC to determine their size following reduction with sodium borohydride (17). The oligosaccharides released by Endo H have a size equivalent to a standard of Mana-9GlcNAc while the EndoH + Endo/ PNGase F released species are somewhat larger, which is probably the result of the PNGase F activity cleaving the oligosaccharide at the linkage to asparagine (Fig. 4). It should be noted again that the neutral oligosaccharides released by Endo/PNGase F are resistant to Endo H since they could not be released by re-digestion with Endo H. Mild acid hydrolysis followed by paper chromatography showed the presence of about 10% fucose in both. This may account for some of the size heterogeneity of these samples, and variable locations of Fuc on the oligosaccharide may account for the resistance of a portion of these molecules to Endo H digestion (18).
The results presented in Table I show that the oligosaccharides can be quite complex. Thus, none were suitable for complete structural characterization. Nature of the Sulfate Esters-The identification and quantitation of radiolabeled sulfated sugars of glycoproteins has been seriously hampered by the lack of suitable methods. The variable stability of different types of sulfate esters makes the use of any single analytical method unsatisfactory. We have used several methods including differential sensitivity to acid and base hydrolysis to characterize the sulfate residues. Sulfate esters differ in their sensitivity to acid hydrolysis (19-23). The tlh of sulfate esters in 0.25 N HCl at 100 "C is: equatorial, 6-25 min; axial, 60-84 min; and primary, 90-120 min. Base hydrolysis has only been rarely used, in part, because it will occur in only two instances: 1) if a sugar has primary (6-OH) linked ester together with a free 3-OH group, or 2) if there is a free adjacent -OH group located trans to the sulfate ester (24, 25). Table I1 shows the behavior expected for each of the sugars found in a complete (unprocessed) and unsubstituted high mannose-high oligosaccharide (3). It is evident that a combination of acid and base hydrolysis treatments can be used as a preliminary characterization of the type of sulfate esters found on an oligosaccharide. If more than one type is present, the relative amount of each can be estimated graphically from the kinetics of 35s04 loss (19).

Each of the unfractionated pools released by sequential Endo H and Endo
H + Endo/PNGase F dgestion was subjected to acid hydrolysis (Fig. 5). All Endo H-released sulfate appears to exist as primary esters. While about 80% of 35s04 released from the Endo H + Endo/PNGase F is found in primary linkage and the remainder is in an equatorial location. When the same pools were subjected to base hydrolysis after reduction of the oligosaccharides with NaBH,, about 70% 3sS04 from the Endo H (Fig. 6, Panel A), and 90% of the 35S04 from the Endo H + Endo/PNGase F pool were sensitive to base (Fig. 6, Panel B). These results are also consistent with the presence of the majority of the 35s04 being in primary linked ester (Man-6-SO4 or GlcNAc-6-SO,). We can eliminate mannose residues A and C (Table 11) from further consideration since the 6-OH is already in glycosidic linkage. Those oligosaccharides which contain Man-6-P residues, also do not have an available 6-OH position.
To determine whether the acid-labile component found in the Endo H + Endo/PNGase F pool was also base stable, an aliquot was hydrolyzed in 0.25 N HC1 for 1 h at 100 "C. These conditions would be sufficient to hydrolyze 75% of the acidlabile component, but only reduce the amount of the more acid-stable type by 25%. If the acid-labile component is also base-stable, then there should be a large (>75%) decrease in the amount of the base-stable component. The results shown in Fig. 6, Panel B, indicate that the base-stable component is reduced by about 80% by the short acid treatment, and suggest that the acid-labile component is also base-stable. The basestability could be due to its location in the standard oligosaccharide shown in the diagram in Table I1 (Man-4-S04 residue A only or GlcNAc-3-SO4 residues A or B) or to the presence of another acid-labile substituent, which blocks base hydrolysis. Identification of Man-&SO4 in the Released Oligosaccharides-Base treatment of Man-6-S04 in glycosidic linkage leads to the formation of 3,6-anhydromannose, provided that the 3 position is free (in the standard oligosaccharide shown in Table 11, residues A and C do not have a free 6 position). Since this derivative is stable to strong acid hydrolysis, it can be recovered and quantitated by HPLC. The [3H]Man serves as an internal standard to allow the calculation of the percentage of the total label which is present as [3H]3,6-anhydromannose. Fig. 7 shows that analysis of the neutral oligosaccharides or those treated by solvolysis prior to base treatment show no evidence of any 3,6-anhydromannose. In contrast, when the Endo H or Endo H + Endo/PNGase F pools are first treated with base, the amount of the 3,6-anhydromannose is proportional to the amount of sulfate ester. Total acid hydrolysates of non-base-treated pooled fractions also showed no evidence of 3,6-anhydromannose.
Each of the fractions shown in Table I was assayed for Man-6-SO4 content. Table I11 shows these values along with the calculated maximum possible Man-6-S04 content assuming that all of the sulfate is present in this sugar. From these calculations, it is clear that many of the fractions could be solely composed of Man-6-SOr while others, notably those released by Endo H + Endo/PNGase F digestions which have 5 and 6 charges must contain other sulfated sugars.
Search for GkNAc-6-S04 Residues-Similar base treatments of [3H]GlcN-labeled released oligosaccharides did not show the appearance of any 3,6-anhydroGlcNAc. Partial acid hydrolysis (0.25 N HCl, 100 'C, 2 h) of lo6 cpm of t3H]GlcNlabeled material failed to show any evidence of GlcNAc-6-S04 by paper chromatography (26). It is possible that this method was not sensitive enough to detect or quantify it.

DISCUSSION
The structure and biosynthesis of sulfated N-linked oligosaccharides has previously received little attention. More recently, a large number of glycoproteins with important physiological functions have been shown to contain sulfated N-linked oligosaccharides. These include a variety of pituitary polypeptide hormones (27-29), cell adhesion, and other developmentally regulated molecules in both mammals and lower eukaryotes (1, 30-32), and several proteins including the low density lipoprotein receptor (33) and the basement membrane proteins (34-36). Although these findings have generated more enthusiasm, a serious limitation in studying these oligosaccharides is that there are few reliable analytical methods of structural characterization. For instance, there is respectively) were hydrolyzed in 0.25 N HC1 at 100 "C for various periods of time and the free was precipitated with barium as described under "Experimental Procedures." The results are calculated as the percentage of ?304 which is not precipitable as normalized to the 3H remaining in solution. Nonspecific precipitation of the 3H is generally 4 0 % . only a single known sulfatase which acts on a oligosaccharide bound sulfate ester (27). This is a chondro-4-sulfatase from Proteus vulgaris which acts on GalNAc-4-S04 in leutropin. This is probably a coincidence due to their similarity to the structure of the natural substrate. Hydrolytic conditions which remove the sulfate are accompanied by destruction of many glycosidic linkages (19). Also, exo-glycosidases which have been invaluable in the structural analysis of N-linked oligosaccharides can be blocked by sulfate esters even when they are present on neighboring sugars (38), and the presence of certain sulfate esters blocks cleavage by various endoglycosidases (1). It would be misleading to say that no methods are available to study the sulfated glycoconjugates, but many of the methods developed in the past were designed to analyze large quantities of polysaccharides (23-25) or proteoglycans (26). For our purposes, it was necessary to develop methods which could be adapted to the analysis of the sulfate esters found on rare proteins, and their biosynthetic intermediates.
Our long-term goal is to understand the roles that the oligosaccharides of lysosomal enzymes play in various developmental and physiological processes in Dictyostelium (39).
To accomplish this we need to understand their structure and biosynthesis. The growth medium is an enriched source of the lysosomal enzymes and many other sulfated anionic proteins. In these initial characterizations, we wanted to examine the broadest possible spectrum of proteins with sulfated N-linked oligosaccharides.
The sensitivity of the sulfate esters to acid hydrolysis ". Kinetics of base hydrolysis of "SO4 from released oligosaccharides. The pools of the oligosaccharides released by Endo H and Endo H -+ Endo/PNGase F (Panels A and B, respectively) were reduced and subjected to base hydrolysis in 1 N NaOH at 80 "C as described under "Experimental Procedures." The per cent of the %SO4 not precipitable by barium was determined and normalized to 3H content. In Panel B, a separate aliquot of the sample was first treated with 0.25 N HC1 at 100 "C for 1 h, (0). The acid was removed by lyophilization and then the base treatment was performed. depends on its orientation on the sugar and is relatively independent of other glycosidic linkages on the same residue (19). The procedure can be applied to monosaccharides, oligosaccharides, and polysaccharides as well (19-22). The lability of a sulfate ester to base hydrolysis, on the other hand, depends upon not only the location of the sulfate, but also on the presence of other glycosidic linkages found on the sugar carrying the sulfate ester (23-25). Thus, if the nature of the underlying structure of the oligosaccharide is known, the general types of sulfate esters can be predicted by acid and base hydrolysis.
The procedure used to identify Man-6-SO4 appears to be tions were analyzed for the presence of Man-6-S04 by quantifying the appearance of 3,6-anhydromannose by HPLC amine adsorption chromatography on Ax-5. Samples were reduced with sodium borohydride and treated with 1 N NaOH for 4 h at 80 'C in the presence of 0.1 M sodium borohydride. Following neutralization and desalting, the oligosaccharides were hydrolyzed in 1 N HCl for 4 h at 100 "C, dried, and the hydrolysate was again reduced with sodium borohydride and desalted as described under "Experimental Procedures." The products were analyzed by HPLC chromatography. The average recovery was about 50% of the initial radioactivity prior to base hydrolysis. Panels are: A: standards a, 3,6-anhydromannitol; b, fucitol; c, mannitol; B: neutral oligosaccharides (Fraction Ha); C, total Endo H-released oligosaccharides; D, solvolysis-treated Endo H-released oligosaccharides; E, Endo H .--f Endo/PNGase F released total oligosaccharides; F, solvolysis-treated Endo H .--f Endo/PNGase F released oligosaccharides. Arrow denotes position of 3,6-anhydromannitol.  Fig. 3.
Calculated as a percentage of the total counts recovered on HPLC analysis of base-treated, acid-hydrolyzed fraction. Corrected for the content of Man-6-P, its degradation to yield manose, and for the completeness of the base hydrolysis (85%).
Determined from the percentage of each fraction containing the variable amounts of sulfate as indicated in Table I and assuming a  Mans structure. quantitative and should be generaliy applicable, but at this time we do not have an independent method to determine quantitative accuracy.
The formation of the anhydrosugar derivative can also be used to assess the location of the sulfated residue in the oligosaccharide, since the resulting ring strain makes the glycosidic linkage extremely acid-labile (40). Mild acid hydrolysis of the base-treated oligosaccharide should generate fragments which are characteristic of the number and location of the sulfate esters in the intact oligosaccharide. Such a procedure could be used for oligosaccharide structural analysis much like acetolysis which also relies on the selective acidlability of certain glycosidic linkages (41).
The presence of sulfate esters does not account for all of the Endo H resistance of newly synthesized a-mannosidase (42). In addition, some of the neutral oligosaccharides in this study are also resistant to Endo H. Ivatt et al. (18) have previously reported that the presence of Fuc may block Endo H digestion. Since we have detected Fuc in the acid hydrolysates of both pools of released neutral oligosaccharides, resistance could be due to the presence of Fuc residues in different positions on the chain, some of which block the cleavage by Endo H. Clearly, the presence of Fuc does not account for all of this resistance, since the highly charged, Endo H -+ Endo/PNGase F releasable oligosaccharides have very little Fuc. Furthermore, the carbohydrate composition of purified secreted a-mannosidase and @-glucosidase also shows the presence of very little Fuc (15), and yet the majority of the oligosaccharides are resistant to the release by Endo H. Thus, other modifications besides sulfation may contribute to the Endo H resistance.