T H E JOURNAL OF BIOLOGICAL CHEMlSTRY Vol. 255, No. 24, Issue of December 25, pp. 11782-11793, 1980 Prrnted in U.S.A.

Synthesis of the N-linked Oligosaccharides of Glycoproteins ASSEMBLY OF THE LIPID-LINKED PRECURSOR OLIGOSACCHARIDE AND ITS RELATION TO PROTEIN SYNTHESIS IN VIVO*

(Received for publication, June 24, 1980)

S. Catherine Hubbard$ and Phillips W. Robbins From the Center for Cancer Research and Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139

The asparagine-linked oligosaccharides of chick em- bryo fibroblast glycoproteins were previously shown to derive from a common lipid-linked precursor, Glc3MansGlcNAcz. The formation of this precursor oli- gosaccharide was examined in intact chick embryo fibroblasts, NIL-8 cells, and Chinese hamster ovary cells. The labeling kinetics and compositions of the lipid-linked oligosaccharides were examined, and the results indicate that lipid-linked Man5GlcNAc2 is rap- idly assembled (~1.5 min) and then extended (t2.5 min) to Glc3Man~GlcNAcz via the intermediate MansGlcNAc2. Chain elongation from Man5GlcNAcz- to MansGlcNAcz- lipid probably occurs by addition of single mannose residues. The pool of lipid-linked GlcsMan9GlcNAcz turns over with a half-time of 3.5 to 6 min; since there is little if any degradation (the mannose residues do not turn over), this reflects the rate at which completed chains are transferred to acceptor proteins. The same intermediates and similar kinetics were observed in all three cell types.

Oligosaccharide-lipid assembly was also examined in cells in which protein synthesis was decreased (using actinomycin D to depress levels of mRNA) or abolished (using cycloheximide). The results indicate that the rate of oligosaccharide-lipid synthesis is proportional to the rate of protein synthesis. The regulated step is prior to the Man5GlcNAcz stage, and we suggest that the most likely control mechanism is limitation of available oli- gosaccharide carrier lipid.

The high mannose and complex asparagine-linked oligosac- charides of vertebrate glycoproteins have widely divergent structures, but they generally have in common the sequence

Manz -% Man - GlcNAc - GlcNAc’ at the site of protein attachment (1, 2). During the past three years it has become clear that this invariant pentasaccharide core is no coincidence but rather reflects the fact that N-linked oligosac- charides are synthesized via a series of specific glycosidases

P P

This work was supported by Grants CAI4142 and CA14051 (to P. W. R. and S. E. L.) from the National Cancer Institute, Department of Health, Education, and Welfare. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accord- ance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ Supported by National Institutes of Health Postdoctoral Fellow- ship I-F32-CA06114-01.

’ The abbreviations used are: GlcNAc, N-acetylglucosamine; CHO, Chinese hamster ovary; MeOH, methyl alcohol; DME medium, Dul- becco’s modified Eagle’s minimum essential medium; ME medium, Eagle’s minimum essential medium; Endo H, endo-P-N-acetylglucos- aminidase H.

and glycosyltransferases from a common precursor (3-5; for review, see Ref. 6). This precursor oligosaccharide, which is assembled on a lipid carrier (dolichol) (7-9) and transferred en bloc (10) to a nascent polypeptide chain (11, 12), has the composition GlcaMansGlcNAcz in cultured CHO (13), NIL-8 (14), and chick embryo fibroblast cells (10). Its structure in CHO cells has been almost completely elucidated (13) (Scheme l), and the proposed structure is consistent with results obtained with GlcjMan9GlcNAc2 from NIL-8 (14) and chick cells (10). An examination of the lipid-linked oligosac- charide from calf thyroid indicates that the glucose residues are in the (Y configuration (15).

Recently, a series of lipid-linked oligosaccharides, Manl- through MaQGlcNAcs, has been isolated from CHO cells. The branching structures of these species are consistent with their being step by step intermediates in the synthesis of the completed oligosaccharide (16). However, with the exception of Man5GlcNAcz (17, 18), none of these oligosaccharides has been shown to be a precursor of lipid-linked GlcaMan9GlcNAc2 rather than a product of its degradation. Detailed kinetic studies are required to resolve this problem and answer a number of other questions: Which of the steps in lipid-linked oligosaccharide assembly is (are) rate-limiting? How is the pathway regulated? How much time is required for the various stages of assembly and for turnover of the pool of completed oligosaccharide? Is lipid-linked GlcaMansGlcNAcZ stable until transfer to protein or is it subject to degradation? How does lipid-linked oligosaccharide formation vary from cell type to cell type? The following kinetic study was‘performed to obtain answers for these questions.

MATERIALS AND METHODS”

RESULTS

Incorporation of [3H]Mannose into the Lipid-linked Oli- gosaccharide Fraction-When subconfluent cultures of chick embryo fibroblasts, NIL-8 cells, or CHO cells were incubated in medium containing [3H]mannose, their lipid-linked oligo- saccharides were rapidly labeled. Typically, incorporation into the washed CHC13/MeOH/H20 (10:10:3) fraction (“oligosac- charide-lipid fraction”) was biphasic, with an exponential increase during the f i s t few minutes and then a slower in-

’ “Materials and Methods,” portions of “Results,” and Figs. 11 to 16 are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, Md. 20014. Request Document No. 80M-1297, cite authors, and include a check or money order for $2.10 per set of photocopies. Full size photocopies are also included with the microfilm edition of the Journal that is available from Waverly Press.

11782

Regulation of Lipid-linked Oligosaccharide Assembly

Man a12 Man ,=1,6

SCHEME 1

‘“i“-1

FIG. 1 (left). Incorporation of [3H]mannose into oligosaccha- ride-lipids of CHO cells. Subconfluent cultures were incubated for various periods with [’Hlmannose as follows: 30 or 45 s, 2.1 mCi/ml (three plates labeled sequentially and pooled); 1 or 1.5 min, 900 pCi/ ml (three plates labeled sequentially and pooled); 2.5 min, 1.5 mCi/ ml; 5, 10, 15, or 30 min, 700 pCi/ml. In some cases, cells were labeled for 2.5 or 5 min and then chased in nonradioactive medium. Cells were harvested in CHC1:JMeOH (2:l) and washed extensively, and the lipid-linked oligosaccharides were extracted with CHCly/MeOH/ Hz0 (10103). Additional details of the labeling and extraction pro- cedures are given under “Materials and Methods.” The radioactivities of aliquots of the oligosaccharide-lipid fractions were determined and their total calculated radioactivities were adjusted for different con- centrations of [:‘H]mannose in the incubation medium (standardized to 1 mCi/ml). e, pulse-labeled cells; 0, cells chased in nonradioactive medium after a 2.5- or 5-min pulse. The arrows indicate the starts of chases.

FIG. 2 (rcght). Effect of exogenous mannose and glucose on the chase of [3H]mannose-oligosaccharide-lipids from prela- beled NIL4 cells. Cultures were incubated for 5 min in medium containing [’Hlmannose (260 pCi/ml), then washed and chased in media containing various concentrations of mannose and glucose. After harvest, the total radioactivities of the washed CHCI:I/MeOH/ Hz0 (10103) fractions were determined. The glucose concentration of DME medium is 4.5 mg/ml (25 mM). 0, chase in DME medium + 1 mM mannose (standard condition); V, glucose-free ME medium; 0, DME medium; 0, DME medium + 5 mM mannose; A, DME medium + 20 mM mannose.

crease until a maximum was reached after 10 to 20 min of incubation. The results from an experiment with CHO cells are shown in Fig. 1; chick cells and NIL-8 cells were found to behave similarly.

If prelabeled cells were washed and chased in nonradioac- tive medium, [”Hlmannose-oligosaccharide-lipid disappeared, again in a biphasic fashion. For the fist 1 to 2 min, the radioactivity dropped slowly or might actually rise slightly (Fig. l),,presumably due to residual incorporation from pre- labeled pools of mannosylphosphate, GDP-mannose, and other precursors. Thereafter, label was lost exponentially with a half-time of 3.5 to 6 min; in the example shown, the chase half-time was 5 min. This rate of oligosaccharide-lipid turn- over is consistent with the fact that 15 to 20 min were required to attain maximum [“Hlmannose incorporation into this frac- tion, since it would take 15 min for % of the pre-existing nonradioactive material to be replaced with labeled molecules.

Similar chase kinetic data were observed with NIL-8 and chick cells; the results of an experiment with NIL-8 cells are shown in Fig. 2. It can be seen that the half-time for chase of prelabeled [3H]mannose-oligosaccharide-lipids (3.5 min in this experiment) was not significantly affected by variations of the

I I 1 I

2.5 min pulse

6

J :U B 2 3 4 5 CHICK - .<#

2 rnin pulse -

1

3 6

I

N I L I \ 2 minputse 1 ,I

80 90 1 0 0 110 120 130

11 783

FRACTION NUMBER FIG. 3. Chromatography of oligosaccharides from CHO,

chick, and NIL-8 lipid-linked oligosaccharides. Cells were la- beled as indicated with [‘Hlmannose (CHO cells, 1.5 mCi/ml; chick cells, 1.3 mCi/ml; NIL-8 cells, 1 mCi/ml). After mild acid hydrolysis of the washed lipid-linked oligosaccharide fractions, samples of the free oligosaccharides were resolved on a column (115 cm) of Bio-Gel P-4 in sodium phosphate buffer. The fraction size was 0.5 ml, and the inclusion and exclusion volumes for C were at fractions 162.5 and 55.8, respectively. A, CHO oligosaccharide-lipids after a 2.5-min pulse; B, chick, 2-min pulse; C, NIL-8, 2-min pulse. For identification of the numbered peaks, see the text.

concentration of mannose or glucose in the chase medium. Thus, expansion (dilution) of intracellular pools of mannose- containing precursors in the presence of exogenous mannose either does not occur or else has little effect on the rate at which label disappears from the oligosaccharide-lipid fraction.

Identification of Individual Lipid-linked Oligosaccha- rides-Further understanding of the kinetics of lipid-linked oligosaccharide assembly required analysis of the individual oligosaccharides present in the washed CHCls/MeOH/H20 (10:10:3) extracts. When lipid-linked oligosaccharides are sub- jected to mild acid hydrolysis, the resultant free oligosaccha- rides can be resolved by chromatography on columns of Bio- Gel P-4 (-400 mesh) (10). Profiles obtained from [”]man- nose-labeled CHO, chick, and NIL-8 cells are shown in Fig. 3. Although there were reproducible differences in the propor- tions of individual oligosaccharides in the three cell types examined, the same array of peaks was always obtained. Studies detailed in the miniprint supplement to this article,

11784 Regulation of Lipid-linked Oligosaccharide Assembly

together with previous work from this laboratory (10, 14), have established the following assignments: Peak 1, Glc3MangGlcNAcz-X'; Peak 2, Glc3MangGlcNAcz; Peak 3, Man8GlcNAcz, with traces of species tentatively identified as GlcsMansGlcNAcz and ManSGlcNAcz-X; Peak 4, Man7GlcNAcz; Peak 5, Man6GlcNAc2; and Peak 6, Man5GlcNAcz. The results of treatment with glycosidases are consistent with the oligosaccharide branching patterns pro- posed by other workers for CHO lipid-linked oligosaccharides (13, 16). In addition to the major peaks listed above, there were also minor peaks migrating between GlcaMangGlcNAc2 and MansGlcNAcz. These included a species behaving as Man8GkNAcz-X and barely detectable traces of MangGlcNAcz and what appeared to be GlclMangGlcNAc2. Synthesis of all these oligosaccharides was abolished in cells treated with the antibiotic tunicamycin, which has been shown to block the initial transfer of GlcNAc to the carrier lipid (19) (data not shown).

I t should be noted that although the ['Hlmannose-labeling medium contained no glucose, it is unlikely that glucose starvation significantly affected the oligosaccharide profiies observed. Most (>95%) of the [:?H]mannose remained in the medium after a 30-min incubation. Furthermore, neither ad- dition of 100 pg/ml glucose to the labeling medium nor prein- cubation for 6 h in glucose-free medium significantly affected the ratio of the individual lipid-linked oligosacharides in chick cells labeled for 20 min with ['Hlmannose (data not shown).

Kinetics of [3HJMannose Incorporation into Individual Oligosaccharides-In order to ascertain precursor-product relations among the individual lipid-linked oligosaccharides, the distribution of radioactivity among the major species GlcaMangGlcNAcz, MansGlcNAc2, and Man5GlcNAca was de- termined for CHO cells incubated for various amounts of time with ['Hlmannose. Representative Bio-Gel P-4 column pro- files from this experiment are shown in Fig. 4; note that equal amounts of radioactivity rather than equal portions of the samples were chromatographed. Clearly, the smaller species MansGlcNAc2 and MansGlcNAcn were much more rapidly labeled than Glc3MangGlcNAc2. Quantitative determination of the results in terms of total radioactivity of individual oligosaccharides (Fig. 5) revealed that incorporation into MansGlcNAcn and MansGlcNAc2 reached what appeared to be steady state in only about 1.5 min of labeling, whereas incorporation into Glc3MangGlcNAcn required much longer, at least 15 min, to do so.

In the experiment just described, ['Hlmannose labeling of the three lipid-linked oligosaccharides MansGlcNAcz, Mans- GlcNAcz, and GlcaMangGlcNAc2 did reach plateaus, but this in itself is not proof that the plateaus corresponded to steady state conditions. Additional evidence was necessary. Before steady state was reached, the outer mannose residues of an oligosaccharide would have a higher specific radioactivity than those near the reducing terminus, while at steady state, all the residues would be equally labeled. The enzyme a- mannosidase removes accessible a-linked mannose residues but cannot hydrolyze mannose residues when they are in the p configuration or when they are blocked by distal glucose residues. Therefore, this enzyme was utilized as a probe for uniformity of labeling in MansGlcNAc2 and Glc3MangGlcNAcz

As described in the miniprint supplement, X is an as yet uniden- tified neutral (at pH 6) moiety attached to the terminal GlcNAc residue of the molecule. It does not interfere with the action of Endo H and is not Man or GlcNAc. It is probable that X is an artifact of sample preparation. The ratio of Glc3Man9GlcNAcs-X to GlcsMan9GlcNAcn was invariant, regardless of length of pulse or chase.

FRACTION FIG. 4. Chromatography of mild acid-hydrolyzed, lipid-

linked oligosaccharides from CHO cells labeled for various periods with [3H]mannose. For incubation conditions and infor- mation on the amount of label incorporated, see the legend to Fig. 1. Mild acid-hydrolyzed oligosaccharide-lipid samples (15,000 cpm of each) were chromatographed as described in the legend to Fig. 3. The inclusion and exclusion volumes were at fractions 156.5 and 53.8, respectively, for B. A, [3H]mannose-oligosaccharides after 1-min pulse (17% of total sample); B, after a 2.5-min pulse (8%); C, after a 30-min pulse (2%). G, Glc; M , Man; N , GlcNAc.

isolated from CHO cells labeled with [3H]mannose for 1, 1.5, 2.5, 5, 10, and 15 min (Table I). For the smaller oligosaccha- ride, the predicted ratio of 4:l for label recovered as free mannose to that recovered as the resistant fragment Man

-GlcNAc;l was found for samples labeled for 2.5 min or longer. Mannosidase treatment of the Glc3MangGlcNAc~ sam- ples, which would be expected to remove an average of 4 of the 9 mannose residues under the conditions used in this laboratory (lo), gave a different result. The expected ratio for free [3H]mannose to ['H]Gl~3Man-~GlcNAc2 (4:5, or 0.8) was approached only after 15 min of labeling. These results sup- port the interpretation that the labeling plateaus of Fig. 5 corresponded to steady state conditions.

Extension of MansGlcNAc2 and ManaGlcNAc2 to Glc&fangGlcNAcz-Several lines of evidence suggest that Man5GlcNAcz is a precursor of Man~GlcNAc2, which is in turn a precursor of Glc3Man9GlcNAcs. First, as discussed above,

P

Regulation of Lipid-linked Oligosaccharide Assembly 1 1785

FIG. 5. Quantitative determination of radioactivity incor- porated into individual lipid-linked oligosaccharides as a func- tion of length of incubation with [311]mannose. Cultures of CHO cells were incubated as described in the legend to Fig. I , the washed oligosaccharide-lipid fractions were mild acid-hydrolyzed, and the free oligosaccharides were resolved by column chromatography (Fig. 4). The total radioactivity recovered as Man5GlcNAcz, Man&lcNAcn, and GlcaMansGlcNAc2 was determined and adjusted for different concentrations of r3H]mannose present during the incubations (stand- ardized to 1 mCi/ml). G, Glc; M , Man; N , GlcNAc.

TABLE I a-Mannosidase digestion of MansGlcNAcz and GlcsMunrGlcNAc2

from CHO cells Lipid-linked oligosaccharides extracted from CHO cells labeled for

various periods with ["Hlmannose (see the legend to Fig. 1 for details of the incubations) were mild acid-hydrolyzed and chro- matographed on a Bio-Gel P-4 column (115 cm) in pyridinium ac- etate buffer. Fractions corresponding to Man5GlcNAcp and Glc,TMan9GlcNAc2 were pooled and lyophilized, and a variable portion was digested with jack bean a-mannosidase. The reaction products were resolved by rechromatography and their radioactivity was de- termined.

Man5GlcNAcz products GlcsMansGlcNAc? products "

of incu- Length Ratio of bation M I N ~ " Man free Man G3M-:,N2 Free free Man

CLM. :N.,

Free Ratio of

to MIN2 Man to

-~ min

. cpm cpm

1.0 2,140 14,400 (6.74) 320 3,080 (9.66) 1.5 4,840 23,600 (4.87) 1,720 7,870 (4.57) 2.5 3,180 13,300 (4.17) 9,560 18,000 (1.88) 5.0 1,090 4,670 (4.29) 10,200 17,600 (1.73)

10.0 750 3,240 (4.32) 24,700 33,500 (1.36) 15.0 1,060 4,300 (4.05) 5,150 6,010 (1.17) a The abbreviations used are: M. Man; N, GlcNAc; G, Glc. -

Mans- and MansGlcNAcp are labeled more rapidly than GlcsMangGlcNAc2 during incubation of CHO cells with [3H]mannose. Similar results were obtained with chick cells and NIL-8 cells, although in these cases, Mans- and

Man8GkNAcz reached steady state somewhat more slowly than in CHO cells (data not shown). Second, in chick cells, in which the ratio of Ma%GlcNAcn to MansGlcNAc2 is >1 a t steady state, it is possible to demonstrate that MansGlcNAcz is more rapidly labeled than MaGGlcNAc2 (Fig. 6). Third, incubating [3H]mannose-labeled chick, NIL-8, or CHO cells in nonradioactive medium results in the rapid disappearance of Man5- and MamGlcNAc2, so that within 2.5 min of chase, only Glc3MangGlcNAcz is detectable (Fig. 7 ) . This is an over- estimate of the time necessary for elongation of MansGlcNAcz t o Glc3MangGlcNAcz, since the 2.5 min includes the time necessary to exhaust the intracellular pools of prelabeled mannosylphosphate, GDP-mannose, and mannosylphospho- ryldolichol. Finally, such a precursor -product relationship is consistent with the proposed branching patterns of the lipid - linked oligosaccharides Man5GlcNAcz through GlcnMangGlcNAcl from CHO cells (13, 16).

It should be added that although only the major interme- diates Man5- and Man8GlcNAcz are discussed, Man6- and Man7GlcNAcl were also present, and their behavior during pulses and chases always paralleled that of Mans- and Man8G1cNAcz. Thus, elongation from Mans- t o ManRGlcNAcz

80 90 I00 110 120

FRACTION NUMBER FIG. 6. Incorporation of [3H]mannose into individual chick

oligosaccharide-lipids. Chick cells were incubated for various in- tervals with [3H]mannose (see the legend to Fig. 8 for incubation conditions and total label incorporation). The washed CHCL3/MeOH/ H20 (1010:3) fractions were mild acid-hydrolyzed and aliquots were subjected to column chromatography as described in the legend to Fig. 3. The inclusion and exclusion volumes for D were at fractions 163.3 and 56.0, respectively. A, [?H]mannose oligosaccharides after 30-s pulse (entire sample); B, 1-min pulse (entire sample); C, 2-min pulse (32% of the sample); D, 5-min pulse (10% of the sample). G, Glc; M , Man; N , GlcNAc.

11786 Regulation of Lipid-linked

I I I 1 I

2.5 min pulse -1

80 90 loo 110 120

FRACTION NUMBER FIG. 7. Effect of a brief chase on prelabeled oligosaccharide-

lipids of CHO cells. Cells were incubated for 2.5 min with [3H]- mannose (1.5 mCi/ml) and harvested before or after a 2.5-min chase in nonradioactive medium. For information on the total radioactivity of the lipid-linked oligosaccharide fractions, see the legend to Fig. 1. After mild acid hydrolysis, the oligosaccharides were chromato- graphed as described in the legend to Fig. 3. A, ["Hlmannose-oligo- saccharides after a 2.5-min pulse; B, after a 2.5-min chase following a 2.5-min pulse. G, Glc; M, Man; N , GlcNAc.

probably involves addition of single mannose residues. Regulation of Lipid-linked Oligosaccharide Synthesis-

Using the situation in CHO cells as an example, the labeling kinetics may be summarized as follows. Upon incubation of cells in ['Hlmannose-containing medium, lipid-linked Man5GlcNAc2 is rapidly assembled ( 4 . 5 min), then extended (c2.5 min) to Glc3MangGlcNAcz via Man8GlcNAc2. A longer period of time (about 20 min) is necessary for replacement of the pre-existing nonradioactive Glc3Man9GlcNAcn pool with uniformly labeled molecules, and pulse-chase data suggest that this replacement occurs with a half-time of about 5 min. Since lipid-linked oligosaccharide is stable in the absence of protein synthesis (see below), this appears to reflect the rate of transfer of completed oligosaccharide chains to protein.

An important implication of this scheme is that the rate at which lipid-linked oiigosaccharides are synthesized is regu- lated by the rate at which the pre-existing pool of completed GlcsMangGlcNAcn chains are transferred to newly synthesized protein. I t would be logical to presume that this might in turn reflect the rate of protein synthesis itself.

To test this hypothesis, lipid-linked oligosaccharide forma- tion was examined under conditions in which protein synthesis was completely or partially blocked by inhibitors. I t was found that a 60-min preincubation of CHO, chick, or NIL-8 cells in medium containing 100 pg/ml of cycloheximide, which com- pletely (>95%) blocked [%]methionine incorporation into trichloroacetic acid-insoluble cellular material, also abolished incorporation of ['Hlmannose into lipid-linked oligosaccha- rides. The data for NIL-8 cells are presented in Table 11. The inhibition by cycloheximide of [3H]mannose-oligosaccharide assembly was very nearly complete; in one experiment, after correction for the amount of residual free ['Hlmannose found

Oligosaccharide Assembly

TABLE I1 Effect of cycloheximide on synthesis of lipid-linked

oligosaccharides In two experiments, cultures of NIL-8 cells were incubated for 10

min with either [3H]mannose (700 pCi/ml) or ["5S]methionine (56 pCi/ml in Experiment 1; 33 pCi/ml in Experiment 2). For some cultures, cycloheximide (100 pg/ml) was present during a preincuba- tion and the labeling period. For ['Hlmannose-labeled cells, the total radioactivity in the washed CHC13/MeOH/H20 (10103) ("CMW- soluble") fractions was determined. [35S]Methionine-labeled cells were lysed, duplicate aliquots were subjected to trichloroacetic acid precipitation, and the total radioactivity was determined.

I"H1Mannose r%lMethionine Experi- Cyclohexi-

~1 L .

merit mide preincu- CMW- Percent- Trichloro- Percent- bation age of acetic acid- age of

control insoluble control Total counts/mm/l@ cells

1 none (con- 1.10 X IO5 (100) 1.77 X lo6 (100)

2 none (con- 1.34 X IO5 (100) 1.34 X 106 (100)

2 15 min 1.50 X lo4 (11.2) 3.05 X lo4 (2.3) 2 30 min 9.86 X 107 (7.4) 7.12 X lo4 (5.3) 1 60 min 1.86 X loJ (1.7)" 7.31 X lo4 (4.1) 2 60 min 9.62 X IOJ (7.2) 1.66 X IO4 (1.2)

trol 1)

trol2)

Only 8.3% of this material was oligosaccharide; it migrated as Glc3ManyGlcNAcl on a Bio-Gel P-4 column. The remainder of the label migrated as free mannose.

in the washed CHC13/MeOH/HzO (10:10:3) samples, it was found to be 99.9% after a 60-min preincubation.

If the rate of protein synthesis determines the rate of lipid- linked oligosaccharide formation, and if the size of the lipid- linked oligosaccharide pool is relatively constant, it can be predicted that decreasing the rate of protein synthesis by a factor N would have at least three effects on the kinetics of [3H]mannose labeling of the oligosaccharides. First, the half- time for chase of label from radioactive lipid-linked Glc3MansGlcNAcz would be increased by the factor N . Sec- ond, the Glc3MangGlcNAcz pool would take N times longer to reach constant specific radioactivity. Third, assuming that the rate at which Man5GlcNAcz is elongated to Glc3MangGlcNAcz is affected much less than is the rate of oligosaccharide chain initiation, the steady state levels of MansGlcNAcn and MallsGlcNAco would be reduced relative to that of Glc3MangGlcNAcz by a factor of N . This last effect would occur because the rate at which the ManBGlcNAc2 and MansGlcNAcz pools were depleted (by elongation to larger oligosaccharides) would remain constant, but the rate at which they were replenished would decrease.

In order to test these predictions, chick cells were incubated in medium containing 1 pg of actinomycin D/ml for intervals sufficient for the inhibition of RNA transcription to result in decreased protein synthesis (4 to 8 h). Again, the ability to exclude trypan blue was essentially unaffected for the duration of the experiment. As a control for possible nonspecific effects of the drug, chick cells infected with Sindbis virus, in which protein synthesis is directed by the viral RNA genome, were also treated with the inhibitor and examined in parallel with uninfected cultures.

In the uninfected cultures treated for 6 to 8 h with actino- mycin D, the incorporation of [35S]methionine into trichloro- acetic acid-insoluble material during a 30-min incubation was reduced by a factor of 3.1 relative to untreated cells. In the Sindbis-infected cells treated with actinomycin D for 4 to 6 h, [%]methionine incorporation was reduced by approximately 20% relative to the untreated, uninfected cells (data not shown). Sodium dodecyl sulfate-polyacrylamide gel electro-

Regulation of Lipid-linked Oligosaccharide Assembly

phoresis of equal aliquots of [35S]methionine-labeled material showed that the actinomycin-treated, uninfected cells pro- duced essentially the same array of radioactive proteins as did the untreated cells, but with a fainter overall pattern. In contrast, the Sindbis-infected cultures incorporated label only into the viral proteins (data not shown).

When these cultures were incubated with r3H]mannose, all the predicted effects of reduced protein synthesis on lipid- linked oligosaccharide metabolism were evident. The results for the Sindbis-infected cells were almost superimposable with those for the uninfected, untreated cells, so for clarity only the latter are shown. The oligosaccharide-lipids of uninfected, actinomycin-treated cells, in which protein synthesis was re- duced 3-fold, required much longer for steady state labeling with [3H]mannose than did the untreated cells (-30 versus -10 min), but they accumulated little excess oligosaccharide- lipid (Fig. 8). As expected, the half-time for chase from pre- labeled oligosaccharides was also increased (-7 uersus 3.5 min). Finally, analysis of individual oligosaccharide species present after the incubations revealed that in all cases the levels of the small precursors MansGlcNAcz and MaaGlcNAq were reduced relative to the level of Glc3MangGlcNAcz (Fig. 9), and the reduction in total radio- activity in the smaller species was approximately 4-fold. This result indicates that the block in lipid-linked oligosaccharide

t Act. D

-Act. D

I

FIG. 8. Effect of reduced rate of protein synthesis on incor- poration of [3H]mannose into chick oligosaccharide-lipids. Chick cells were incubated with [3H]mannose as follows: 30 s and 1 min, 3 mCi/ml (two plates labeled sequentially and pooled); 2 min, 1.8 mCi/ml (two plates); 5, 10, 20, 30, and 60 min, 500 pCi/ml. Some cultures were preincubated for 6 to 8 h with actinomycin D so that protein synthesis was reduced approximately 3-fold. Cells were har- vested before or after a chase period. The calculated total radioactiv- ities of the lipid-linked oligosaccharide fractions were adjusted for differences in the amount of [3H]mannose in the incubation medium (standardized to 1 mCi/ml). U, untreated, pulse-labeled o " 0 , untreated, chased; A- - -A, actinomycin D-treated, pulse- labeled; A- - -A, actinomycin D-treated, chased. The arrows indicate the starts of chases.

11787

'0

FIG. 9. Effect of reduced protein synthesis on the composi- tion of lipid-linked oligosaccharides. Chick cells were incubated for 10 min with ['Hlmannose (500 pCi/ml); some of the cultures had been preincubated for 6 to 8 h with actinomycin D so that protein synthesis was reduced 3-fold. The lipid-linked oligosaccharides were extracted, hydrolyzed, and examined by column chromatography as described in the legend to Fig. 3. The inclusion and exclusion volumes forA were at fractions 163.1 and 56.1, respectively. A, untreated cells; B, actinomycin D-treated cells. G, Glc; M , Man; N , GlcNAc.

formation during protein synthesis inhibition is prior to the MansGlcNAcz stage.

Stability of Lipid-Linked GLc:+MangGLcNAc2--The fact that blockage of protein synthesis by cycloheximide results in rapid and complete inhibition of lipid-linked oligosaccharide assem- bly makes it possible to determine experimentally whether the completed Glc3MangGlcNAc2 moiety is stable until it is transferred to protein or is subject to degradation and resyn- thesis before transfer. Cultures of chick or CHO cells were incubated for 20 min in ["Hlmannose-containing medium without inhibitor, then chased either in the presence or ab- sence of cycloheximide. If the GlcsMangGlcNAca species is stable until transfer to protein, one would expect that a chase in cycloheximide-containing medium would have little or no effect on the level of this oligosaccharide. Both cell types gave the same result; the data for chick cells are presented in Fig. 10. In this experiment, the chase half-time for oligosaccharide- lipid was 3.5 min in untreated cells, but little label was lost during chases in the presence of cycloheximide. In fact, after 20 min, which was shown in an earlier section to suffice for complete turnover of the pre-existing Glc:3MangGlcNAca pool, there was no loss at all. Analysis of the individual oligosac- charides present after 20 or 45 min of chase in cycloheximide- containing medium showed no evidence of degradation; no species smaller than Glc3Man9GlcNAc2 was detected (data not shown). Thus, turnover of lipid-linked oligosaccharides appears to reflect almost exclusively transfer to protein; man- nose residues are not removed until after transfer to protein.

Regulation of Lipid-linked Oligosaccharide Assembly I I 1 I I I

0

h

'.

I

'. '. '. -"""-4 - cx

I I I I I I

IO 20 30 40 50 60 MINUTES

FIG. 10. Inhibition by cycloheximide (CX) of [3H]mannose- labeled lipid-linked oligosaccharide turnover. Chick cells were prelabeled for 20 min with [3H]mannose (700 pCi/ml), then washed and chased for various periods either in standard chase medium or in medium containing cycloheximide (100 pg/ml). The lipid-linked oli- gosaccharides were extracted and their radioactivities were deter- mined. 0- - -0, untreated; -, cycloheximide-treated. The time units on the abscissa refer to length of chase.

DISCUSSION

Taken together, the results presented in this study suggest that the following events take place in the assembly of the lipid-linked precursor oligosaccharide GlcsMan9GlcNAcp in intact, subconfluent CHO, chick, and NIL-8 cells. The smaller lipid-linked oligosaccharide MansGlcNAc2 is extended, prob- ably by addition of single mannose residues, to ManxGlcNAcz and then on to GlcsMan9GlcNAca; this elongation is extremely rapid ( ~ 2 . 5 min). Under normal conditions, the entire (active) pool of lipid-linked Glc3Man9GlcNAcz turns over with a half- time of 3.5 to 6 min; thus the pool is completely replaced every 10 to 20 min. This turnover appears to represent pri- marily or solely transfer of completed chains to newly synthe- sized proteins, since our results indicate that degradation without transfer is not a significant reaction. There are intrigu- ing minor differences among the different cell types examined, such as variations in the MamGlcNAc2/MansGlcNAc2 ratio, which is higher in chick cells than in NIL-8 or CHO cells. Nevertheless, in all three cell types, the same intermediates and similar kinetic patterns were observed, and the oligosac- charide donor (lipid-linked Glc3Man9GlcNAca) was present in excess to available acceptor protein.

Our observation that lipid-linked oligosaccharide synthesis can be blocked by protein synthesis inhibitors is consistent with a recent report by Schmitt and Elbein (20) that addition of cycloheximide or puromycin decreased [3H]mannose incor- poration into the oligosaccharide-lipid fraction of canine kid- ney cells. These workers were able to conclude that cyclohex- imide itself does not inhibit the glycosyltransferases, since addition of up to 100 mM cycloheximide (28 times the concen- tration used in the present study) to membrane preparations from porcine aorta did not decrease incorporation of label from GDP-[14C]mannose into the lipid-linked oligosaccharide fraction. Two other potential sources of artifact, direct inhi- bition of protein glycosylation or alteration of [3H]mannose- labeled precursor pools by the drugs, can also be ruled out with a fair degree of certainty. In the present study, two fundamentally different procedures for inhibiting protein syn-

thesis, depressing cellular levels of mRNA with actinomycin D and interfering with ribosome translocation with cyclohex- imide, both blocked lipid-linked oligosaccharide formation. Based on the Schmitt and Elbein study, a third mode, releas- ing puromycin-terminated peptide chains, can be added to the list. Furthermore, preincubation of Sindbis virus-infected cells with actinomycin D, which had little effect on protein synthe- sis, did not significantly change the rate of oligosaccharide assembly or transfer.

The control point is probably at the stage of lipid-linked oligosaccharide chain initiation rather than in elongation, since upon partial inhibition of protein synthesis, the lipid- linked intermediates ManSGlcNAc2 and MansGlcNAcz did not accumulate but were depressed by an amount proportional to the decrease in protein synthesis. At least two possible mech- anisms could account for all the observations. The first, feed- back inhibition of an enzyme early in the assembly pathway by accumulated product in the absence of turnover (transfer to protein), seems unlikely. The level of GlcaMan9GlcNAcz- lipid is the same or only slightly increased in actinomycin D- or cycloheximide-treated cells. The second candidate mecha- nism is simple, direct, and requires no postulation of compli- cated feedback loops: limitation of available oligosaccharide carrier lipid. Degradation of lipid-linked GlcaMan9GlcNAc2 occurs slowly if at all, so the carrier lipid necessary for initia- tion of a new oligosaccharide chain would be made available only upon transfer of a completed chain to protein. Although there is as yet no direct evidence establishing this as the mechanism linking oligosaccharide-lipid assembly with pro- tein synthesis, it is suggestive that lack of phosphoryldolichol has been implicated as the cause of the loss of saccharide-lipid synthesis upon the conversion of reticulocytes, which are capable of protein glycosylation, to erythrocytes, which are not (21, 22). Conversely, the level of phosphoryldolichol in chick oviduct appears to increase upon stimulation of glyco- protein formation by estrogenic hormones (23), and dolichol synthesis in mouse testis is markedly elevated during sper- matogenesis (24). I t should be pointed out, however, that the putative rate-limiting pool of oligosaccharide carrier lipid may involve only a fraction of the total cellular dolichol. Thus, the observation that cycloheximide does not inhibit mannosyl- phosphoryldolichol formation in kidney cells (20) does not necessarily argue against the hypothesis. There is some evi- dence that different pools of phosphoryldolichol are utilized in the syntheses of mannosylphosphoryldolichol and N-ace- tylglucosaminylpyrophosphoryldolichol (25).

Although our results indicate that the mannose residues of completed Glc3Man9GlcNAcz chains are not hydrolyzed until after transfer of the oligosaccharide to protein, they do not exclude the possibility that the glucose residues may turn over. Glucosidases capable of acting on either lipid- or peptide- linked GlcsMansGlcNAcp in vitro (15, 26) are present in the cisternae of the endoplasmic reticulum (27). Since removal of glucose residues from the donor oligosaccharide markedly inhibits its transfer from lipid to protein in vitro (28, 29), it has been suggested that these glucosidases may regulate the availability of fully glucosylated precursor oligosaccharide (15). During the present study, in which cells were harvested in cold CHC13/MeOH (2:l) to prevent oligosaccharide com- position artifacts (lo), no more than trace amounts of labeled lipid-linked Glcz- or GlclMan9GlcNAca-lipid were ever ob- served, even when transfer of lipid-linked oligosaccharides to protein was slowed or stopped by inhibitors. At least in the three cell types examined, therefore, the hypothesis that the glucosidases and/or glucosyltransferases regulate the level of lipid-linked Glc3Man9GlcNAcz available for protein glycosyl-

Regulation of Lipid- linked Oligosaccharide Assembly 11789

ation appears untenable. If glucose residues are removed, they are replaced so quickly that deglucosylated species do not accumulate.

Many other questions remain. For example, at what stage are the glucose residues transferred to the growing lipid-linked oligosaccharide chain? The mannosyl branch which will ulti- mately bear them is already complete in Man5GlcNAcz-lipid (18), and under certain conditions, a considerable amount of GlcaMansGlcNAcn may be produced (17). In the present study, however, no more than a trace of Glc3MansGlcNAcz-lipid was ever detected, and MaQGlcNAc2-lipid was shown to be a major intermediate. Since far more MamGlcNAcz- than Man9GlcNAcn-lipid was found in all three cell types examined, it is tempting to speculate that one or more glucose residues may be added before addition of the 9th mannose residue. It should be noted that there has not yet been a direct demon- stration that MansGlcNAcn-lipid, which is the largest lipid- linked oligosaccharide produced with NIL-8 membrane prep- arations in the absence of a glucose donor, can be efficiently converted to the fully glucosylated form.

Acknowledgments-We are grateful to Drs. C. Edson and M. Snider for many stimulating discussions and to Ms. D. Young for typing this manuscript.

REFERENCES

1. Kornfeld, R., and Kornfeld, S. (1976) Annu. Reu. Biochem. 45, 217-237

2. Kornfeld, R., and Kornfeld, S. (1980) in The Biochemistry of Glycoproteins and Proteoglycans (Lennarz, W., ed) pp. 1-34, Plenum Press, New York

3. Robbins, P. W., Hubbard, S. C., Turco, S., and Wirth, D. (1977) Cell 12,893-900

4. Tabas, I., Schlesinger, S., and Kornfeld, S. (1978) J. Biol. Chem.

5. Hunt, L., Etchison, J., and Summers, D. (1978) Proc. Natl. Acad. Sci. USA 75, 754-758

6. Schachter, H., and Roseman, S. (1980) in The Biochemistry of Glycoproteins and Proteoglycans (Lennarz, W., ed) pp. 85-111, Plenum Press, New York

7. Waechter, C., and Lennarz, W. (1976) Annu. Reu. Biochem. 45,

253,716-722

8.

9.

10.

11.

12. 13.

14.

15.

16.

17.

18. 19.

20.

21.

22.

23. 24. 25.

26.

27.

28.

29.

30.

31.

95-112 Parodi, A,, and Leloir, L. (1979) Biochim. Biophys. Acta 559, 1-

37 Struck, D., and Lennarz, W. (1980) in The Biochemistry of Gly-

coproteins and Proteoglycans (Lennarz, w., ed) pp. 35-83, Plenum Press, New York

Hubbard, S. C., and Robbins, P. W. (1979) J . B i d . Chem. 254, 4568-4576

Kiely, M., McKnight, G., and Schimke, R. (1976) J. B i d . Chem.

Rothman, J., and Lodish, H. (1977) Nature 269, 775-780 Li, E., Tabas, I., and Kornfeld, S. (1978) J. Biol. Chem. 253,7762-

Liu, T., Stetson, B., Turco, S., Hubbard, S. C., and Robbins, P.

Spiro, R., Spiro, M., and Bhoyroo, V. (1979) J. Biol. Chem. 254,

Chapman A., Li, E., and Kornfeld, S. (1979) J. Biol. Chem. 254,

Chapman, A., Trowbridge, I., Hyman, R., and Kornfeld, S. (1979)

Li, E., and Kornfeld, S. (1979) J. Biol. Chem. 254, 2754-2758 Tkacz, J., and Lampen, 0. (1975) Biochem. Biophys. Res. Com-

Schmitt, J., and Elbein, A. (1979) J. Biol. Chem. 254, 12291-

Martin-Barrientos, J., and Parodi, A. (1977) Mol. Cell. Biochem.

Lucas, J., and Nevar, C. (1978) Biochim. Biophys. Acta 528,475-

Lucas, J., and Levin, E. (1977) J. Biol. Chem. 252,4330-4336 James, M., and Kandutsch, A. (1980) J. Biol. Chem. 255, 16-19 Godelaine, D., Beaufay, H., and Wibo, M. (1979) Eur. J. Biochem.

Grinna, L., and Robbins, P. W. (1980) J. Biol. Chem. 255, 2255-

Grinna, L., and Robbins, P. W. (1979) J. Biol. Chem. 254, 8814-

Turco, S., Stetson, B., and Robbins, P. W. (1977) Proc. Natl.

Spiro, M., Spiro, R., and Bhoyroo, V. (1979) J . Biol. Chem. 254,

251,5490-5495

7770

W. (1979) J. Biol. Chem. 254,4554-4559

7659-7667

10243-10249

Cell 17,509-515

mun. 65,248-257

12294

16, 111-117

482

96, 27-34

2258

88 18

Acad. Sci. USA 74, 4411-4414

7668-7674 Waechter, C., Lucas, J., and Lennarz, W. (1973) J. Biol. Chem.

248, 7570-7579 Warren, C., Auge, C., Laver, M., Suzuki, S., Power, D., and

Jeanloz, R. (1980) Carbohydr. Res., 82, 71-83

11790 Regulation of Lipid-linked Oligosaccharide Assembly

Regulation of Lipid-linked Oligosaccharide Assembly

11792

a -

7 -

6 -

0 5 - x

h 4 t V I

3 -

2 -

I -

I x

Regulation of Lipid-linked Oligosaccharide Assembly ,

Regulation of Lipid-linked Oligosaccharide Assembly 11 793

A

' 5 1 o n r I \ 1

I800

1600

600

0.2 0.3 0.4 0.5 0.6 0.7

RELATIVE ELUTION CONSTAM