Glycoprotein Biosynthesis in the aZg3 Saccharomyces cerevisiae Mutant ROLE OF GLUCOSE IN THE INITIAL GLYCOSYLATION OF INVERTASE IN THE ENDOPLASMIC RETICULUM*

on restricted to the that alg3 was slightly leaky, whether the glu- and alkaline phosphatase-conjugated anti-rabbit I g G (Promega Bio-tech) with visualization as described (20) or digoxigenin-labeled ConA and alkaline phosphatase-conjugated anti-digoxigenin (Boehringer Mannheim) with visualization as described (28).

Oligosaccharides on invertase restricted to the endoplasmic reticulum (ER) in alg3,secl8 yeast at 37 "C were found to be 20% wild type ManeGlcNAc  These results suggested that alg3 was slightly leaky, but did not address whether the oligosaccharide-lipid ManeGlcNAca and MansGlcNAcz precursors were glucosylated in alg3 yeast. Therefore, an alg3,secl8,glsl strain was constructed to delete the GLSl-encoded glucosidase I responsible for trimming the terminal al,a-linked glucose from newly transferred GlcSMan,GlcNAcz oligosaccharides. Invertase activity was overexpressed 5-10-fold on transforming this strain with a multicopy plasmid (pRB58) carrying the SUC2 gene, and preparative amounts of the ER form of external invertase, derepressed and accumulated at 37 "C, were purified. The N-linked glycans were released by sequential treatment with endo-8-N-acetylglucosaminidase H (endo H) and peptide-N4-N-acetyl-B-glucosaminyl asparagine amidase. Oligosaccharide pools were sized separately on Bio-Gel P-4, which showed that endo H released about 17% of the carbohydrate as GlcsManeGlcNAc, while peptide-N4-N-acetyl-8-glucosaminyl asparagine amidase released the remainder as HexeGlcNAcz and MansGlcNAcz in a 1:4 ratio. Glycan structures were assigned by 500-MHz two-dimensional DQF-COSY 'H NMR spectroscopy, which revealed that the endo H-resistant HexsGlcNAcp pool contained GlcsMansGlcNAcz and MansGlcNAcz in a 6:4 ratio, the latter a different isomer from that formed by the ER al,a-mannosidase (Byrd, J. C., Tarentino, A. L., Maley  alg3,secl8,GLSl strains. Chromatographic analysis of the products showed that ['HIGlc was removed only in the presence of the GLSl gene product. Thus, the vast majority of the N-linked glycosylation in the ER of alg3 yeast (>75%) occurs by transfer of MansGlcNAcz without prior addition of the 3 glucoses normally found on the lipid-linked precursor.
In both mammalian and yeast cells, GlcsMansGlcNAcn is built up sugar by sugar on dolichol pyrophosphate and then transferred as a unit to specific asparagine residues in nascent proteins in the lumen of the ER' (1). After transfer, the oligosaccharide is modified by specific trimming and elongation reactions, which in yeast lead to the formation of MansGlcNAcz by the stepwise removal of the 3 glucoses and 1 mannose (1). Previous work by Huffaker and Robbins ( 2 ) and by Orlean (3) utilizing the alg3 and dpml strains of yeast, respectively, depict only Man6GlcNAcz in profiles of lipidlinked oligosaccharides released by mild acid hydrolysis. Recent studies from our laboratory have identified the precursor for elongation of endo H-resistant mannan in alg3,secl8 (37 "C) yeast to be Manla+2Manla+2Manla+3(Manla+ 6)Manl/34GlcNAcz (4).
Although GlcaMan6GlcNAcz was isolated from class E Thy-1-murine cells (5), it is not currently known whether a glucosylated MansGlcNAc2 oligosaccharide was initially transferred during protein glycosylation by alg3 yeast. Precursor glucosylation appears necessary in mammalian cells (6, 7), but this is not the case in yeast, where failure to add glucose to oligosaccharide-lipid (alg5-I and alg6-1 mutations) reduces but does not prevent glycoprotein biosynthesis (8). Recent studies with a conditional ts mutation in the yeast Man-P-Dol synthase (DPM1) gene have not resolved this question, as Man5GlcNAcz-PP-Dol accumulated on shift to the nonpermissive temperature (3).

12095
alg3,secl8,gkrl strain was constructed with the belief that any lipid-linked oligosaccharide glucosylated to Glca and transferred to proteins in the ER at 37 "C, the restrictive temperature for ER to Golgi trafficking in secl8 cells, would of necessity retain the glucose residues. The oligosaccharides released from the accumulated ER form of external invertase by sequential digestion with endo H and PNGase F were resolved by gel filtration as Glc3MansGlcNAc, HexsGlcNAc2, and Man6GlcNAca. Structural analysis by high field 'H NMR of the HexsGlcNAcs pool revealed it to be a mixture of GlcsManaGlcNAcz and a novel isomer of MansGlcNAc2. The isolation of nearly 90% of the PNGase F-released oligosaccharide as Man6GlcNAc2 from alg3 yeast invertase supports the conclusion that failure to add glucose to oligosaccharidelipid in yeast may reduce, but does not serve to limit, glycoprotein biosynthesis under normal laboratory growth conditions.

Materials
Saccharomyces cerevlsiae haploid strain X79-1A(a) (alg3-lpecl8l,ade2-IOI,ura3-52) was provided by T. C. Huffaker and P. W. Robbins (2); haploid strain SF732-8A(a) (secl8-l,glsl,pep4) was provided by R. Schekman (9). These two strains were mated, zygotes isolated and sporulated, the tetrads dissected, and the four spores grown out. Temperature-sensitive isolates requiring uracil and adenine were screened for retention of glucose on oligosaccharides as invertase of an increased size by SDS-PAGE. A representative of each phenotype was transformed using the lithium acetate procedure (IO) with a multicopy plasmid (pRB58) obtained from M. Carlson (II), which carries the external invertase (SUC2) gene. Bio-Gel P-4 and SDS-PAGE reagents were from Bio-Rad and Immobilon membranes from Millipore. Sephacryl S-300 Superfine was a product of Pharmacia LKB Biotechnology Inc. and DE52-cellulose was obtained from Whatman. Anti-rabbit IgG alkaline phosphatase immunoblot kit was obtained from Promega Biotech. Sigma was the source for 99. 8

Methods
Cell Growth-S. cerevisiae strains were grown in media consisting of the following in grams/liter of distilled water adjusted to pH 6.5 with NaOH: Bacto-yeast nitrogen base with amino acids, 6.7; Bactoyeast extract, 4.0; Bacto-peptone, 8.0; adenine sulfate, 0.2; succinic acid, 2.4. A 100-ml starter culture grown overnight at 26 "C was used to inoculate a 2.5-liter Bio-Flow I11 benchtop fermentor (New Brunswick Scientific Co., Inc.) containing 2 liters of medium and 600 ml of 50% (w/v) glucose at 26 "C. The pH was automatically controlled at 6.5 by the addition of 4 M NaOH using an Ingold 465 pH electrode, while air saturation was maintained at 80-100% using a polarographic dissolved-oxygen probe (pHoenix Electrode Co.). Growth was monitored by withdrawing 10-ml samples at timed intervals, diluting to an optical density of 0.5 or less, and reading the optical density at 600 nm by comparison with a standard curve prepared by direct cell count with a Helber Chamber (A. H. Thomas Co.).
After overnight growth, cells were harvested at 4 "C by centrifugation at 10,000 X g for 20 min and resuspended in fresh medium (37 "C) containing 10% glucose and incubated at 37 "C for 15-45 min to establish the secl8 ts defect. The cells were then diluted 10-fold with 37 'C, glucose-free medium to derepress invertase synthesis and incubated for a minimum of 3 h (one doubling time) after exhaustion of the glucose. Sodium azide was added to 0.06%, cells were harvested at 4 'C by centrifugation at 10,000 X g for 10 min, resuspended in 20 mM sodium citrate, pH 5.5, containing 5 mM sodium azide, and washed once by centrifugation. The growth cycle was repeated as needed by the sterile addition of 1.9-liter of medium and 600 ml of 50% (w/v) glucose to the 100 ml of culture remaining in the fermentor.
Purification of the Glycosyluted 37 "C ER Form of Invertase-Purification of the nonglycosylated internal and glycosylated external invertases has been described in detail in recent publications (14, 15) and the procedures are essentially the same as those employed here. All steps were performed at 0-4 "C. Harvested cells were broken with a Bead Beater (Bio-Spec) using 0.5-mm acid-washed glass beads; beads were removed by filtration and the crude extract centrifuged at 30,000 X g for 30 min. Solid ammonium sulfate was added to the supernatant fraction to 40% saturation, stirred for 30 min, and the precipitate removed by centrifugation. The invertase-containing supernatant was concentrated and dialyzed against 10 mM sodium phosphate buffer, pH 6.5, using a hollow-fiber system (Amicon; Model DC2, HIP10 cartridge). The pH of the retentate was brought to 4.0 by the dropwise addition of 2 N CH3COOH with stirring, and the precipitate removed by centrifugation. The invertase-containing supernatant was brought to pH 6.5 by the addition of NaOH and dialyzed against 10 mM sodium phosphate buffer, pH 6.5, using the hollow-fiber system.
The sample was applied to a 2 X 22-cm column of DE52-cellulose equilibrated in 10 mM sodium phosphate buffer, pH 6.5, and eluted with a linear NaCl gradient (0-0.2 M) in 10 mM sodium phosphate buffer, pH 6.5. The glycosylated form of invertase eluted at about 0.12 M NaC1, while the internal nonglycosylated invertase remained adsorbed to the column and could be eluted at 2 M NaCI. Invertasecontaining fractions were pooled, concentrated by ultrafiltration (Amicon; Diaflo YMlO), and chromatographed on a 2.2 X 82-cm column of Sephacryl S-300 equilibrated in 10 mM sodium phosphate buffer, pH 6.5, containing 50 mM NaCl. The invertase-containing fractions were pooled and concentrated by ultrafiltration. A typical purification from 595 g of cells (wet weight), summarized in Table I, yielded 125 mg of the glycosylated ER form of external invertase, or 1 mg of external invertase from 5 g of cells. The major loss in activity seen in steps 3 and 4 of Table I is the removal of the nonglycosylated internal form of invertase, which represents about 50% of the total activity and as shown in Figs. I and 2 migrates at 59 kDa on SDS-PAGE (15).
Oligosaccharide Release and Purification-N-Linked oligosaccharides were hydrolyzed from invertase sequentially by treatment with endo H (purified from transformed Esche~ichia coli) (16) followed by PNGase F according to standard deglycosylation protocols (17); endo H will cleave N-linked chains only if they are substituted with a mannose residue linked a1,3 to the upper arm al,g-linked mannose of the core (18). Invertase (100 mg) was denatured in a boiling water bath for 5 min in 0.05 M sodium citrate buffer, pH 5.5, containing 0.10 M 8-mercaptoethanol and 120 mg of SDS (1.2-fold weight excess). After cooling, endo H was added to -100 milliunits/ml and the sample was incubated for 17 h at 30 "C, after which the presence of endo H activity was confirmed by assay with [3H]dansyl-AsnGlcNAc4Mang as substrate (17).
Protein and oligosaccharides were precipitated at 4 "C by adding 4 volumes of ice-cold acetone. The 80% salt-and SDS-containing acetone supernatant was discarded and the pellet extracted twice with ice-cold 60% methanol. Pooled methanol supernatants contained the released oligosaccharides and were concentrated by rotary evaporation. The glycoprotein pellet was dissolved with 0.05 M NaOH, and immediately adjusted to pH 8.0 by the addition of 0.05 M H3P04, supplemented with P-mercaptoethanol and SDS, and heated for 5 min at 100 'C. PNGase F was added to 50 milliunits/ml and the mixture incubated for 16 h at 30 "C, after which the presence of PNGase F activity was confirmed by assay with [3H]dansyl-fetuinpentaglycopeptide as substrate (17). After hydrolysis, N-linked oligosaccharides again were isolated by solvent precipitation/extraction as described above. Residual salt and peptides were removed from oligosaccharide pools by passage through a Bio-Gel P-2 column in water (2 x 20 cm) and a Sep-Pak Cls cartridge (Waters), respectively, and oligosaccharides were chromatographed on a calibrated Bio-Gel P-4 column (<25 pm, lot 44671A) with a Man3GlcNA~[~H]ol internal marker (13).
'H NMR Spectroscopy-Oligosaccharides were exchanged twice by rotary evaporation from 99.8% 'H2O and once by lyophilization from 99.96% 2Hz0. Samples were dissolved in 1.0 ml of 99.96% 'Hz0 and an aliquot containing 1.0-1.5 pmol of oligosaccharide was lyophilized.
After drying in vacuo over P206 for several days, the samples were taken up in 0.72 ml of 99.996% 'Hz0 and 5 pl of 1% acetone was added as an internal chemical shift marker (6 = 2.225 ppm relative to 4,4-dimethyl-4-silapentane sulfonate). Samples (0.71 ml) were transferred to 5-mm tubes (Wilmad Co., NO. 535pp), previously washed and exchanged with 99.8% 2 H~0 , and flame sealed. Compounds were examined at 296 K by 1D and 2D DQF-COSY 'H NMR spectroscopy at 500 MHz at the Albert Einstein College of Medicine NMR facility. The ID spectra were recorded using 16,000 data blocks, a cycle delay time of 3.1 s, a sweep width of 4,000 Hz, and 90" pulses. The 2D DQF-COSY spectra were obtained by the method of States et al. (19). The sweep width in the 11.7 tesla field was 900-1OOO Hz using 512 increments in tl. There were 1024 data points in t2 and an acquisition time of 0.4 s with a recycle time of 1.1 s. Both t , and t 2 were zero-filled to 2048 points. A total of 64 transients were run for each free-induction decay. Data were analyzed on a Varian VXR4OOO work station and on a Silicon Graphics Iris using Hare Research, Inc., software. Resonance intensities were integrated by cutting out and weighing peaks from expansions of the anomeric and C2-H regions of one-dimensional spectra. The anomeric proton of the core @1,4-linked mannose (residue 3) was obscured by the residual H02H peak. However, residue 3's C2-H resonance is found at 4.233 ppm in Man6GlcNAc2 (4) and upfield at 4.158 ppm when the core al,3-linked lower-arm mannose is 6-0-substituted (20). General Methods-Neutral hexose was determined by a scaleddown version (21) of the phenol-sulfuric acid assay (22) with mannose as standard. Protein was determined either according to Bearden (23) a t 595 nm, using bovine serum albumin as a standard, or by absorbance at 280 nm (14). Invertase activity was assayed colorimetrically by a modification (24) of the two-step method described by Goldstein and Lampen (25). Radioactivity was measured in "Ready Protein" (Beckman) with a Beckman LS-3801 scintillation spectrometer. Invertase antibodies were raised in rabbita (26) using carbohydrate-free internal invertase as the immunogen (15). SDS-PAGE was performed on 8% acrylamide slab gels (0.5 mm) (27), which were either stained for protein with Coomassie Blue R-250 or transferred to Immobilon membranes and Western blotted using either rabbit anti-invertase and alkaline phosphatase-conjugated anti-rabbit I g G (Promega Biotech) with visualization as described (20) or digoxigenin-labeled ConA and alkaline phosphatase-conjugated anti-digoxigenin (Boehringer Mannheim) with visualization as described (28).

Cell Growth and Invertase Derepression
To determine the role of glucose in protein glycosylation in the alg3 mutant background, it was necessary to restrict the secreted glycoprotein form of invertase to the ER under conditions where any glucose present on oligosaccharides initially transferred from oligosaccharide-lipid to the protein would not be removed. Thus, a strain was constructed, alg3,secl8,glsl, which included the secl8 mutation to prevent trafficking of invertase from the ER at 37 "C (29) and the glsl mutation to prevent normal glucose trimming in the ER (9).
When grown in the 2.5-liter benchtop fermentor this strain typically yielded 0.5 g of (wet weight) cells/g of glucose, or 40 g of (wet weight) cellsfliter of media containing 10% (w/v) glucose. This result agrees with data previously obtained by Wang et al. (30) who demonstrated a similar ratio of glucose consumed to wet weight of cell yield.
Aliquots of alg3,secl8,glsl cells growing at 26 "C taken at 3 h before and at selected times after the derepression of invertase synthesis on temperature shift to 37 "C were analyzed for both residual glucose and invertase activity with the results shown in Fig. 1, panel 23. The invertase activity measured represents both the internal nonglycosylated form that migrates on SDS-PAGE as a sharp band a t 59 kDa and an increasing proportion with time at 37 "C of the larger glycosylated form restricted to the ER by the secl8 mutation (Fig.  1, panel A ) . Although external invertase was synthesized maximally when the concentration of glucose in the media dropped below 0.5%, even at high glucose concentrations (>5%), a small background of external invertase routinely was present in cells as observed by Western blot analysis (Fig.  1, -3 h). This pre-existing glycosylated invertase appears to be the source for the small fraction of MaaGlcNAq oligosaccharide which, as will be shown in the accompanying study (311, has the same structure as the major MaaGlcNAc2 isomer found on invertase secreted by alg3 yeast at 26 'C.

Purification of the Glycosylated ER Form of Invertaqe and
Oligosaccharide Isolation Glycosylated invertase of high specific activity (4163 units/ mg protein) was purified with an overall 44% recovery from a crude cell extract by precipitation with ammonium sulfate and acetic acid, and chromatography on DE52-cellulose and Sephacryl S-300 columns (Table I). However, approximately 50% of the initial invertase activity present was the intracellular (internal) nonglycosylated form, which was selectively removed in steps 3 and 4 of the purification scheme. Thus, from the recovery of phenol sulfuric acid-positive material and the low level of residual internal form in the purified preparation used as a source of alg3 oligosaccharides (Fig. 2), recovery of nearly 90% of the secreted (external) glycosylated form is estimated. Maximum recovery of the glycosylated external invertase was achieved only when ammonium sulfate fractionation preceeded the pH 4 precipitation with glacial acetic acid. Fig. 2 shows that the oligosaccharides on purified alg3,seclS,glsl invertase were largely endo H-resistant, although some increase in mobility was observed on digestion of the 37 "C ER form (compare lane 1 with lanes 2, 5, with 6, and 8 with 9 in Fig. 2). PNGase F-treatment markedly increased invertase's mobility, but not to the extent of the internal form (59 kDa) because of charge effects due to the conversion of asparagine to aspartic acid on deglycosylation (17) (Fig. 2, lanes 4, 7, and 10). Nevertheless, the "Carboblot" with ConA (lanes 9 and 10) verifies that the invertase was rendered essentially carbohydrate-free by PNGase F. Fig. 3 shows the Bio-Gel P-4 profiles, which reveal on the basis of phenol sulfuric acid assay, that endo H released 17% of the oligosaccharides as Glc3MansGlcNAc  A and B, respectively, and the expansions of the corresponding Jlg cross-peak regions from their two-dimensional spectra are shown in Fig. 5, panels A and B. The Man6GlcNAc2 pool was saved but not analyzed further, as this compound has been thoroughly characterized previously by methylation and both 1D and 2D 'H NMR analyses (4). Fig. 4 includes the structures deduced in this study with resonance identification numbers h o e each residue for crossreferencing in Figs. 4 and 5, Table 11, and the text. Numbers next to the residue identification numbers are relative molar values of signal intensities integrated under the resonance peaks.

4, panels
Chemical shifts (6, ppm) for C1-H protons are summarized in Table 11. The ManaGlcNAc2 chemical shifta assigned in the previous study (4) for residues lalb, 2, 3, 4, 5, 8, and 11 have been included in Table I1 for comparison with the additional oligosaccharides isolated from alg3,sec18&&1 (37 'C) invertase. Linkage aesignmenta for the mannose and N-acetylglucosamine residues ( Fig. 4 and Table 11) were based on the existing library of chemical shifta for model compounds and oligosaccharides of similar configuration (12, 21, [31][32][33][34][35][36]. Those for the glucose residues were based on assignmenta reported previously for Glc3ManloGlcNAc isolated from the yeast mutant mnn2,secl8,glsl (37). for the model tetrasaccha- ride Glcal~2Glca1~3Glcal-+3Man1a4CH2CH2CH3 (38) and for GlcsMan,GlcNAc, the major species isolated from Lec23/VSV, a Chinese hamster ovary cell mutant deficient in CY -glucosidase I f 39). ~~~a~G~N A c -~n~~c t i o n of the spectnun (Fig. 4, panel  A ) of the oligosacch~ide released from a~3 , s e~l 8 , g~~ by endo H and its integration reveals three anomeric proton signals in addition to those present for the yeast form of MansGlcNAc (21); 1) 1 mol of resonance at 5.535 ppm, which is the reporter for an internal al,3-linked glucose ((22) substituted by a terminal a1,S-linked glucose (Ga); 2) 1 mol of signal at 5.273 ppm, the reporter resonance for the al,3-linked glucose (GI) added to the terminal a1,S-linked mannose residue 11 on the lower arm; and 3) 1 mol of resonance at 5.185 ppm, which is the new terminal al,a-linked glucose (G3). Since the sum of the resonance intensities for residues 5,8,6, 7,11,9, and 4 is 7 mol, this indicates that the only species present is Glc3M~GlcNAc. The resonance at 4.228 ppm in the C2-H region of the 2 0 projection (Fig. 5, panel A ) indicates the 3-0-substi~tion of residue 11. Subtraction of 1 mol of residue ll's C2-H from the 2 mol total at 4,228-4.235 ppm leaves 1 mol of intensity, which is the C2-H of residue 3, whose anomeric proton is obscured by HO'H in the 1D spectrum, but whose J1,2 cross-peak is apparent in the 2D expansion (Fig, 5, panel A) at 4.770(Cl-H)/4.228(C2-H) ppm. This confirms the presence of residue 3, which as a core residue must be present in all species. The three X . 2 cross-peaks for glucose residues are apparent in Fig. 5, panel A, and are found in GlcsMansGlcNAc at the same locations as in Glc3 Man~GlcNAc~ (compare Fig. 5, panels A and B). This confirms the conclusions drawn by Tsai et a l . (37) employing lf) lH NMR spin decoupling analysis to establish the C2-H assignments for the glucotriose constituents. The C2-Hs can be seen to reside at chemical shifts expected for the glucose axial protons, not the region expected for mannose equitorial protons.
HexaGlcNAc2-Inspection and integration of the H e 5 GlcNAcz spectrum (Fig. 4, panel B ) revealed the presence of two species which could be assigned as Glc3ManSGlcNAcz and a unique isomer of ManeGlcNAc' (Fig. 4, p a n e l B, and Table   11). The integrated sum of resonance intensities for residues 5, 8, 4, 2 and 3 (see below) was 5 mol indicating that these residues are common to d l structures in the pool (see Fig. 4,  pawl B ) . The glucose resonance intensities, however, were not 1 mol, suggesting that an endo H-resistant m~n o s y l compound of the same size as Glc~Man~GlcNAcz had copurified in the H e 5 pool. Integration of the 1D spectrum (Fig. 4,  panel B ) provided values for al,3-Iinked glucose residues GI and GZ of 0.6 mol each, and since Gz is only found at 5.535 ppm when substituted by an al,2-linked glucose (G3) (37), this provides evidence that G1a are all present at the same amount. Note that the anomeric proton for G3 overlaps the a-anomer of the reducing-end GlcNAc ( l a ) at 5.190 ppm. The 0.6 mol value for G3 deduced above can be calculated directly by subtracting from the 1.1 mol of signal integrated at 5.190 ppm the value for f a , 0.5 mol, which itself is derived by subtracting the 0.5-mol value for the @-anomer integrated at 4.694 ppm ( l b ) from the 1-mol value for the reducing-end. The presence of a unique r e s o n~c e cross-peak at 5.032(C1-H)/~.224(C2-~) ppm (Fig. 5, panel B ) verses the 3-0-substitution of residue 11 (32). Thus, based on the glucose content, 60% of the HexsGlcNAcz species are GlcaMan5GlcNAcz, in which lower arm residue 11 of the core MansGlcNAc' (4) bears the glucotriose unit. Note that the J1,2 cross-peaks for the monosaccharide constituents of the G1c3 extension are identical to those seen in Glc3Man~GlcNAc (compare Fig. 5,  panels A and 3 ) .
Subtraction of the anomeric proton values (0.6 mol) for the monosaccharide constituents of Glc3Man6GlcNAcz leaves 0.4 mol of intensity for residues 5, 8,4, and 2 in the Man6 core structure. Also present were 0.8 mol at 5.138 ppm and 1.4 mol at 5.04 ppm (Fig. 4, puwt B ) . The former chemical shift is where al,6-linked 2-0-substituted (Si') andlor al,3-linked terminal (3t) mannose reside, while the latter is where a1,2linked terminal (2t) and/or 3-0-substituted al,e-linked @i3) mannose reside. The sum of C2-H proton intensities for residues 3 and 11 at 4.229-4.224 ppm plus that at 4.160 pprn was 2 mol. Since core residue 3 is present in all species, by difference 1 mol of intensity at 4.224 ppm belongs to residue 11, signifying that all of this residue is 3-0-substituted. As 60% of this substitution is due to GI in the GlcaMan6GlcNAc~ isomer derived above, 40% must be substituted with mannose. This accounts for 0.4 mol of intensity at 5.138 ppm, the chemical shift of a1,3-linked terminal mannose (32).
The shift in 0.4 mol of residue 3's C2-H from 4.235 to 4.160 ppm is indicative of an al,6-linked substitution of core-linked residue 5 by residue 12 (20). Thus, 40% of the species contain residue 12, which when 2-0-substituted resides at 5.144 pprn and accounts for the remaining 0.4 mol of resonance intensity at this chemical shift. Residue l l ' s anomeric proton is found at about 5.04 ppm and accounts for 1 of the 1.4 mol of signal integrated at this chemical shift. The 0.4 mol difference at 5.04 ppm corresponds to residue 13, the ul,2-linked mannose that terminally substitutes residue 12 (Fig. 4, panel E ) . k Residue names are based on the linkage of a particular residue, how it is substituted, and its relative position in the molecule.
The assignment of cul,2-linked residue 13 is verified by the J1,p cross-peak at 5.048(Cl-H)/4.067(C2-H) ppm (Fig. 5,panel   B). The assignment of the two isomers in the Hexs pool satisfies all integrations in the 1D spectrum (Fig. 4, panel B).
The unique isomer of MansGlcNAcz, representing 40% of the Hexs pool but only 4% of the total oligosaccharides present, has been derived from the Man~GlcNAcz precursor by addition of a terminal cul,3-linked mannose to residue 11 and a1,6linked mannose residue 12 to the lower arm core residue 5, which itself is 2-0-substituted by terminal residue 13. This ManaGlcNAcz isomer, present on external invertase in the cells prior to derepression at 37 "C ( Fig. I), is the major Mans core structure found on invertase secreted by alg3,sec18 yeast at 26 "C in the accompanying paper (31). Of the oligosaccha-rides that were endo H-resistant on alg3 invertase (83% of the total in Table 11), 88% were MansGlcNAcz while only 7% became glucosylated to yield Glc~MansGlcNAc2. The novel ManaGlcNAcz isomer was only 5% of the endo H-resistant oligosaccharides recovered.

DISCUSSION
The current study sought to determine whether glucose was involved in the transfer of the glycan from oligosaccharidelipid to protein in yeast. Structural studies were performed on the oligosaccharides released enzymatically from the alg3,seclB,glsl triple-mutant invertase, which had accumulated in the ER during incubation at 37 "C when ER-Golgi traffic was disrupted by the secl8 ts defect. This mutant provided preparative amounts of oligosaccharide, which were subjected to analyses by one-and two-dimensional lH NMR spectroscopy. Integration of the resonance intensities from the 1D expansions revealed that Glc3MansGlcNAc was the only oligosaccharide released by endo H (Fig. 4, panel A , and Table 11), and that PNGase F-released oligosaccharides corresponded in size to Man6GlcNAcz and HexsGlcNAc2, which were assigned as Glc3Man5GlcNAcz and a novel isomer of MansGlcNAcz (Fig. 4, panel B, and Table 11).
The pathway for the synthesis of N-linked glycans in both mammalian and yeast cells starts with the transfer of Glc3MangGlcNAcz from dolichol pyrophosphate to protein, and continues with the specific trimming to MansGlcNAc, by the stepwise removal of the 3 glucose residues followed by the middle arm al,2-linked terminal mannose residue 10 (21). Evidence suggests that in mammalian cells only glucosylated lipid-linked glycans are transferred to nascent proteins (7), while yeast are capable of transferring truncated lipid-linked oligosaccharides (40). Nonglucosylated oligosaccharides are transferred to protein by the alg5, alg6, and dpgl mutants, which accumulate MangGlcNAcz-PP-Dol (8, 41), and an underglucosylated donor is accumulated by the alga, mutant which transfers GlcMangGlcNAcz to protein (42). By contrast, Glc3Man6GlcNAcz has been isolated from class E Thy-1-cells (5) and Chinese hamster ovary cells (43), which argues for the need for precursor glucosylation in mammalian cells.
'H NMR spectroscopy of the endo H-sensitive oligosaccharide confirmed it to be Glc3MansGlcNAc (Fig. 4, panel A , and Table 11). This agrees with earlier work showing that Nlinked oligosaccharides accumulated with a size corresponding to Glc3MansGlcNAcz in the secl8,glsl mutant at 37 "C (9). The formation of only GlcsMansGlcNAc in the endo Hsensitive oligosaccharide pool (Fig. 3, panel A )  The recovery from invertase accumulated at 37 "C of a small amount of a He% isomer that clearly would require post-ER processing was initially somewhat disturbing. However, under the most restrictive glucose-repression conditions employed, some mature, pre-existing external invertase was always found in cells (Fig. 1, -3 h), which can be identified as the source for the MansGlcNAc2. Again, finding only Glc3ManaGlcNAc in the endo H-released oligosaccharides supports this conclusion, as it is known that GlcsMansGlcNAc is a substrate for elongation not only to Glc3ManloGlcNAc, but also to glucosylated "mannan" (45). Had any of the alg3,secl8,glsl invertase of interest leaked past the secl8 block into elements of the Golgi apparatus, endo H-releasable oligosaccharides larger than those in the column profile in Fig.  3, panel A , would have been found.
Failure to find glucosylated oligosaccharide-lipid in previous studies (2,3) probably reflects their higher efficiency of transfer to protein compared to their nonglucosylated counterparts (46). The results obtained in this study indicate that glucosylation of the Man6 structure is very slow relative to the Mang oligosaccharide-lipid. On the basis of the ratio of Glc3Man6GlcNAcz/Man6GlcNAc2 found on invertase in the ER, the transfer of the Mans structure to protein appears to be about 5-fold more efficient than the glucosylation of Man5 to Glc3Mans in alg3 yeast. We cannot exclude that some Glcl,zManS was synthesized, transferred, and trimmed so quickly to Mans that neither of these were found in the oligosaccharide-lipid pool in uiuo (2, 3), or on invertase oligosaccharides in the current work. However, structural studies on invertase oligosaccharides synthesized and secreted by alg3,secl8 at 26 "C in the accompanying paper (31) reveal a considerable fraction to have the form Glcl,zMan~loGlcNAc2. This suggests that the four additional mannoses added from Man-P-Dol to form the al,6-linked branch of the precursor MangGlcNAc2 significantly enhance both the rate of oligosaccharide-lipid glucosylation and the subsequent ER glucosidase trimming reactions. The latter has been observed in vitro using a Glcl-3Man4,5GlcNAc derived from Glc3MangGlcNAc by jack bean a-mannosidase treatment (47).
In summary, we have shown that in alg3 yeast, the failure to add glucose to lipid-linked oligosaccharides may reduce but does not serve to limit glycoprotein biosynthesis under the laboratory growth conditions employed. Any reduction in the efficiency of invertase glycosylation due to the lack of glucose on oligosaccharide-lipid was minor, as quantitative carbohydrate recoveries in the current work indicate that invertase subunits were associated with an average of two endo Hsensitive and seven endo H-resistant chains, which is very close to the average of 10 occupied sites per subunit on wild type invertase (48). It is not clear whether alg3's vigor is due to leakage of about 20% of the oligosaccharides past the alg3 block to become wild type Glc3MangGlcNAcz, followed by processing to normal mannan, or to the relaxed specificity of yeast oligosaccharyltransferase in glycosylating nascent peptides with truncated nonglucosylated oligosaccharides.