The Biosynthesis of the Major Lipid-linked Oligosaccharide of Chinese Hamster Ovary Cells Occurs by the Ordered Addition of Mannose Residues*

Intact Chinese hamster ovary cells were incubated with [3H]mannose or [3H]galactose to label the mannose and glucose residues, respectively, of the lipid-linked oligosaccharides. The lipid-linked ollgosaccharides were then extracted and the oligosaccharide units released by mild acid hydrolysis and separated by descending paper chromatography. A family of oligosaccharides were isolated which contained from 1 to 8 mannose residues linked to a di-N,N’-acetylchitobiose unit. Each oligosaccharide was subjected to a-mannosidase digestion and methylation analysis. The larger oligosaccharides were also subjected to acetolysis. With this information the arrangement of the mannose residues in each of the oligosaccharides could be deduced. We conclude that each oligosaccharide of a given size consists of one predominant isomer and that the addition of mannose residues to the growing lipidlinked oligosaccharide is highly ordered. In addition to the predominant (glucose)3-(mannose)s(N-acetylglucosamine)~ lipid-linked oligosaccharide previously characterized (Li et al. (1978) J. Biol. Chem. 253,7762-7770), the Chinese hamster ovary cells were also found to synthesize a second glucose-containing lipid-linked oligosaccharide composed of 3 residues of glucose (Glc), 5 residues of mannose (Man), and 2 residues of iV-acetylglucosamine (GlcNAc). The structure of this oligosaccharide was determined to be: Glcl + 2Glcl-+ 3Glcl ---f 3Mancul-+ 2Manal + aManal+ 3(Manal+ 6)Man -+ GlcNAc + GlcNAc.

Intact Chinese hamster ovary cells were incubated with [3H]mannose or [3H]galactose to label the mannose and glucose residues, respectively, of the lipid-linked oligosaccharides.
The lipid-linked ollgosaccharides were then extracted and the oligosaccharide units released by mild acid hydrolysis and separated by descending paper chromatography.
A family of oligosaccharides were isolated which contained from 1 to 8 mannose residues linked to a di-N,N'-acetylchitobiose unit. Each oligosaccharide was subjected to a-mannosidase digestion and methylation analysis.
The larger oligosaccharides were also subjected to acetolysis. With this information the arrangement of the mannose residues in each of the oligosaccharides could be deduced. We conclude that each oligosaccharide of a given size consists of one predominant isomer and that the addition of mannose residues to the growing lipidlinked oligosaccharide is highly ordered. In addition to the predominant (glucose)3-(mannose)s(N-acetylglucosamine)~ lipid-linked oligosaccharide previously characterized  J. Biol. Chem. 253,7762-7770), the Chinese hamster ovary cells were also found to synthesize a second glucose-containing lipid-linked oligosaccharide composed of 3 residues of glucose (Glc), 5 residues of mannose (Man), and 2 residues of iV-acetylglucosamine (GlcNAc). The structure of this oligosaccharide was determined to be: Glcl + 2Glcl-+ 3Glcl ---f 3Mancul-+ 2Manal + aManal+ 3(Manal+ 6)Man -+ GlcNAc + GlcNAc.
The biosynthesis of the two complex-type oligosaccharide units of the vesicular stomatitis virus G protein is initiated by the en bloc transfer of a high molecular weight oligosaccharide from a lipid carrier to the nascent polypeptide.
Following transfer to the G protein, the oligosaccharide is processed extensively to give rise to the completed complex-type oligosaccharide (l-3). The complete structure of the carbohydrate unit of the lipid-linked oligosaccharide is shown in Structure 1 (4).

Manal + 2Manal
Fiecently a second lipid-linked oligosaccharide has been isolated and its structure determined to be that shown in Structure 2 (5). Manal 6 ManPI + 4(3)GlcNAcPl+ 3 4GlcNAc Manal + 2Manal ---f 2Manal STRUCTURES This lipid-linked oligosaccharide, which is present in both vesicular stomatitis virus-infected and uninfected Chinese hamster ovary cells, appears to be a biosynthetic precursor of the major lipid-linked oligosaccharide shown in Structure 1. The finding of only one isomer with the composition Man5GlcNAcs indicated that the biosynthesis of the major lipid-linked oligosaccharide might occur by the ordered addition of mannose and glucose residues. To test this hypothesis, the various lipid-linked oligosaccharide intermediates of Chinese hamster ovary cells were labeled with ["Hlmannose, isolated, and then subjected to structural analysis. In each instance the predominant species was a single isomer. These data demonstrate that the addition of mannose residues during the assembly of the lipid-linked oligosaccharide is highly ordered. Based on these findings a scheme for the sequence of mannose addition to the lipid-linked oligosaccharide is proposed.  provided  the methylglycoside  derivatives  of the following  partially  Omethylated  mannose standards:  2,3 di-; 2,3,4-tri-;  3,4,6-tri-;  2,3,6-tri-;  2,3,4,6-tetra-O-methylmannose.  2,4-di-; 3,6-di-; and 2,4,6-tri-o-methylmannose,  as well as other partially  methylated  mannose  species,  were obtained from permethylated  ovalbumin  glycopeptides  and their  acetolysis fragments (4). Enzymes-Jack bean a-mannosidase was purified by the method of Li and Li (6). P-Mannosidase was purified from hen oviduct by the method of Sukeno et al. (7). Endo-B-N-acetylglucosaminidase Ci, was isolated by the method of Ito et al. (8). Endo-P-N-acetylglucosaminidase D was a kind gift from Dr. Jacques Baenziger, Washington University.
Isolation of Radiolabeled Lipid-linked Oligosaccharides-The radioactive labeling and the isolation of lipid-linked oligosaccharides from Chinese hamster ovary cells has been previously described (2,4 Analysis-Methylation analysis, borohydride redt ition, and acetolysis were performed as previously described (4).

Isolation
of Oligosacchar ides-Chinese hamster ovary cells were labeled for 10 to 15 min with either [2-"Hlmannose or vH]galactose as described under "Experimental Procedures." The lipid-linked oligosaccharides were then extracted, subjected to mild acid hydrolysis, and the released oligosaccharides were separated by descending paper chromatography in Solvent A (Fig. 1). A series of ten [3H]mannose-labeled oligosaccharide fractions, in addition to the major lipid-linked oligosaccharide were detected as shown in Fig. 1. Each fraction was eluted as indicated, and purified by repeated descending paper chromatography in Solvent A, or Solvent B, or both. Oligosaccharide Fraction V has been characterized previously as MansGlcNAcp (5). Based on their relative mobility on paper chromatography oligosaccharide Fractions I through IV, isolated from the chloroform/methanol (2:l) extract, were estimated to contain 3 through 6 glycose units, respectively (Fig.  1A). Similarly, oligosaccharide Fractions IV through IX, isolated from the chloroform/methanol/water (1:1:0.3) extract, were estimated to contain 6 through 11 glycose units, respectively ( Fig. 1, B and C). In the experiment shown in Fig. 1 there was too little radioactivity in the Peak VII region (Man7GlcNAcz) to analyze further. However in a second similar experiment we were able to isolate enough of the Peak VII species to characterize. The overlapping oligosaccharides in Fractions VIIIa and VIIIb were clearly separated after chromatography for 6 days in Solvent B. The faster migrating material (VIIIa) proved to contain glucose in addition to mannose (see below). Region IX turned out to be a mixture of oligosaccharides.
We were not able to obtain enough Man9GlcNAcz for complete characterization. There was also 3 j' 4  to Glcl-3MangGlcNAcz plus a second oligosaccharide which migrated in the region of MaaGl.cNAc2 (Fig. 1D). The latter oligosaccharide (A in Fig. 1D) migrated as a single peak in Solvent B. The small amount of oligosaccharide material which migrated slower than Glcl-3Mans-GlcNAct (C in Fig. 1D) was not further characterized.
The chromatogram of the chloroform/methanol (2:l) fraction of these cells contained only one peak which migrated with free glucose and was presumably derived from glucose-P-dolichol (data not shown active against several of the oligosaccharides provides further evidence to support this assumption. Structure of I (Man,GlcNA&-When the [3H]mannoselabeled oligosaccharide I was methylated followed by hydrolysis and separation of the methylated mannose derivatives by thin layer chromatography only 2,3,4,6-tetra-Me-Man, was detected, indicating that there is only 1 residue of mannose which is located at the nonreducing terminus ( Fig. 2A).' The mannose residue was resistant to jack bean a-mannosidase but was released by incubation with hen oviduct p-mannosidase (Fig. 3, A and B). Based on these data and the fact that the oligosaccharide migrated with authentic Man/31 + 4GlcNAcPl-+ 4GlcNAc when subjected to paper chromatography (Fig. 3A), its probable structure is: ManP + GlcNAc + GlcNAc.

Structure of V (Man5GlcNAcz)-The
complete structural analysis of oligosaccharide V is described in our previous publication (5) and is shown in Fig. 10. Structure of VI (ManfiGlcNAc2)-Methylation of oligosaccharide VI gave rise to 2,3,4,6-tetra-Me-Man, 3,4,6-tri-Me-Man, 2,4,6-tri-Me-Man, and 2,4-di-Me-Man in the ratio of 2.3: 1.7:l.O:l.O (Fig. 6A). a-Mannosidase released 84% of the radioactivity as free mannose and the residual oligosaccharide migrated with Ma@1 -+ 4GlcNAcPl -+ 4GlcNAc (Fig. 7A). Acetolysis was performed on a sample of oligosaccharide VI which had been converted to ManeGlcitolNAc by endo+Nacetylglucosaminidase Cii digestion and subsequent reduction with NaBH,. This was done since our previous studies had shown that the di-N,N'-acetylchitobiose moiety is relatively sensitive to cleavage during the acetolysis procedure (5). The borohydride reduction was performed so that the fragments could be compared with standard ManA-4GlcitolNAc compounds. As shown in Fig. 5B and the other which migrated as a disaccharide. The minor peaks are probably the result of a small degree of overdegradation. Methylation of the disaccharide gave rise to 2,3,4,6tetra-Me-Man and 2,4,6-tri-Me-Man in the ratio of 1.0:0.7. No 3,4,6-tri-Me-Man was detected. Based on these data, the most likely structure for oligosaccharide VI is: Mancul -+ 2Manal -+ 2Manal --f 3(Manul + 3Manal + 6)Manpl --+ GlcNAc + GlcNAc (see Fig. 10).

DISCUSSION
The purpose of this study was to determine whether the synthesis of the major lipid-linked oligosaccharide of intact Chinese hamster ovary cells proceeds by the ordered or random addition of mannose residues. If ordered addition of the residues occurred, then each precursor oligosaccharide of a given size should consist of a single isomer whereas random mannose addition would lead to the formation of multiple isomers of the oligosaccharides.
By incubating the cells with large amounts of [3H]mannose (1 mCi/4 ml) we were able to isolate a series of lipid-linked oligosaccharides ranging in size from ManlGlcNAcz to MamGlcNAcz. The oligosaccharides were then subjected to structural analysis which demonstrated that each species consists of one predominant isomer. In our previous study on the MamGlcNAcz species, we performed a pulse-chase experiment with ["Hlglucosamine which demonstrated that the lipid-linked Man5GlcNAcz, MansGlcNAcs, and MansGlcNAc2 species could be chased into the large GlcaMansGlcNAcn species, indicating that these species are in fact precursors of the major lipid-linked oligosaccharide. Unfortunately there was not enough label in the other species to follow their fate. If we make the reasonable assumption that these oligosaccharides are also biosynthetic ' The arrangement of the 3 glucose residues into the sequence Glcl + 2Glcl+ 3Glc+ rather than Glcl + 3Glcl+ 2Glc+ was chosen because the former is the sequence present in the major lipid-linked oligosacchsride (4). The data in the present paper do not allow us to distinguish between these two possible sequences.
precursors of the final lipid-linked GlcJMansGlcNAcz molecule which is transferred to nascent proteins, then we can propose a scheme for the synthesis of this molecule, as shown in Fig.  10. This scheme indicates that the addition of mannose residues to the growing lipid-linked oligosaccharide is highly ordered. Several aspects of the scheme deserve special comment. The ManyGlcNAcg species has a structure identical with the core portion of many complex-type oligosaccharides (11). Yet complex-type oligosaccharides are not formed by the direct transfer of the MansGlcNAcz species to protein, but rather by the transfer of the GlcaMangGlcNAct species which is then processed extensively (l-3). The Man-i and Mans species which are involved in the biosynthesis of the major lipid-linked oligosaccharide represent different isomers than the Man7 and Mans species which are generated by processing of the GlcsMangGlcNAcz molecule after it has been transferred to the protein acceptor (12). This provides further evidence that these molecules are not transferred directly to protein but rather represent true intermediates in the biosynthesis of the final lipid-linked oligosaccharide. The Man5GlcNAcz species is of particular interest, since it serves as a branch point, being converted primarily to GlcaMangGlcNAcz but also to GlcaMansGlcNAcz. This latter species has been shown to be transferred to nascent protein in uninfected Chinese hamster ovary cells under certain experimental conditions.3 These results are consistent with the proposal of Turco et al. that glucose residues are required as a signal for oligosaccharide transfer to protein (13). An identical lipid-linked GlcnManr,GlcNAcz species has recently been found in a mutant clone of mouse lymphoma cells which has a block in the conversion of lipid-linked MansGlcNAcs to MansGlcNAcz (14,15). The GlcaManBGlcNAcz species was also shown to be transferred to protein in the mutant lymphoma cell line and to be processed to form typical complextype oligosaccharides ( 14) :' Several limitations inherent in the techniques used in these studies should be pointed out. The acetolysis procedure results in some degree of under-and over-degradation, especially in oligosaccharides that contain a di-N,N'-acetylchitobiose unit (5). This makes it difficult to evaluate the significance of the minor fragments that are detected.
The methylation procedure may be complicated by the presence of variable amounts of radioactivity at or near the origin and the possibility of some variation in the recoveries of the different methylated species. To control for this potential problem, compounds of known structure were included in each methylation run and these gave the expected products in the correct ratios. In addition, the assumption is made that all of the mannose residues are uniformly labeled during the period of incubation with the ['Hlmannose. The larger the oligosaccharide, the greater the possibility for some degree of nonuniform labeling. In spite of these limitations, the data clearly demonstrate that each of the oligosaccharides which was examined consists of one predominant isomer. For the reasons just mentioned, we cannot, however, exclude the possibility that small amounts of other isomers of the various oligosaccharides also exist. Herscovics et al. (16)  linked oligosaccharides in vitro using particulate enzyme preparations from pig aorta (17). Using either GDP-mannose or mannosylphosphoryldolichol as the mannose donor to endogenous lipid-linked oligosaccharides, they obtained a heptasaccharide and larger oligosaccharides as the predominant products. Acetolysis of the heptasaccharide released most of the radioactivity as free mannose, suggesting that the residue transferred in vitro was linked al + 6. This is difficult to understand in view of our findings suggesting that the hexasaccharide MamGlcNAcz is converted to Man5GlcNAcz by the formation of an al + 2 linkage. However, their finding that acetolysis of the octasaccharide released most of the radioactivity as a mannobiose is consistent with our structures for the MansGlcNAcz and ManeGlcNAcn species. Although GlcaMansGlcNAcp and MansGlcNAcz are consistently the two most abundant lipid-linked species found in Chinese hamster ovary cells, significant variations in the relative proportions of the various species have been observed in different experiments.
In addition, other cell types may have different proportions of the various species. Consequently, it is difficult to draw any conclusions concerning regulation of the biosynthetic pathway.
In conclusion, we have proposed a sequence of mannose addition to form the lipid-linked oligosaccharide as it occurs in uivo. This information should be helpful in dissecting out the enzymes involved in this pathway.