THE JOURNAL 0 1988 by The American Society for Biochemistry and Molecular Biology,

OF BIOLOGICAL CHEMISTRY Inc.

Vol. 263, No. 25, Issue of September 5, pp. 1277612782,1988 Printed in U. S. A.

Site-specific Glycosylation in Animal Cells SUBSTITUTION OF GLUTAMINE FOR ASPARAGINE 293 IN CHICKEN OVALBUMIN DOES NOT ALLOW GLYCOSYLATION OF ASPARAGINE 312*

(Received for publication, November 24, 1987)

Bradley T. ShearesS From the Center for Cancer Research, Massachusetts Znstitute of Technology, Cambridge, Massachusetts 02139

It has been shown previously that chicken ovalbumin synthesized and secreted in a heterologous cell system is glycosylated at the correct site and that the oligosac- charides at that site, similar to the protein made in hen oviduct, are predominantly of the hybrid type (Sheares, B. T., and Robbins, P. W. (1986) Proc. Natl. Acad. Sci. U. S. A. 83, 1993-1997). This site-specific glycosylation of AsnZB3, but not Asns”, suggested a prominent role for the nascent protein chain rather than the specific cell type in directing the proper at- tachment of oligosaccharide chains. In the present study, the effect of glycosylation at Asn203 on the gly- cosylation of Asn312 has been investigated. Using a 20- base oligodeoxynucleotide primer containing a 2-base mismatch, the codon for AsnZB3 in the chicken ovalbu- min gene (AAC) was changed to that for Gln (CAA), thereby preventing glycosylation at amino acid 293. Constructions containing this mutation were trans- fected into mouse L (tk-) cells which were subsequently labeled with [3sS]methionine. Ovalbumin secreted by these cells was recovered by immunoaffinity chroma- tography and analyzed for the presence of an oligosac- charide attached at AsnSl2. Treatment of the material with peptide:N-glycosidase F demonstrated that oval- bumin molecules containing Gln substituted for AsnzBs were not glycosylated. This further supports our ear- lier hypothesis that the nascent protein chain is re- sponsible for directing site-specific glycosylation of ovalbumin, and that the presence of an oligosaccharide chain at the first site has no influence on glycosylation at the second site.

Despite extensive studies on the biosynthesis of proteins containing N-linked glycans, relatively little is known about the factors determining the site choice and the type of the oligosaccharide chain. To determine why only certain aspar- agine residues are glycosylated and only selected oligosaccha- rides are extensively processed, I have examined glycosylation of chicken ovalbumin synthesized in a heterologous cell. Oval- bumin has proven useful for such studies for two reasons. First, ovalbumin has two potential glycosylation sites (Asn- X..-Ser/Thr), AsnZg3 and Asn3lZ, of which only the former is glycosylated in uiuo. Second, ovalbumin is unique in that it has large amounts of hybrid oligosaccharides. In an earlier study ( l ) , we showed that ovalbumin secreted by mouse L

* This work was supported in part by a grant from the Lucille P. Markey Charitable Trust and by National Institutes of Health Grant CA98765 to P. W. Robbins. 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 accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ Lucille P. Markey Scholar.

cells is glycosylated at the correct site with large amounts of hybrid oligosaccharides, and that this type of sugar chain was unusual and not detectable on other glycoproteins secreted by mouse L cells. This result suggested that the polypeptide chain has greater influence on proper site glycosylation and oligosaccharide processing of ovalbumin than the specific cell type.

Studies by Lennarz and his co-workers (2) have shown that Asn312 can be glycosylated in vitro when the protein is in a denatured state. Hence, it seemed possible that the presence of an oligosaccharide at Amzg3 interferes with glycosylation at Asn312 in uiuo. Using oligonucleotide site-directed mutagen- esis, a glutamine residue was substituted for asparagine at position 293 in ovalbumin. Mouse L cells transfected with the ovalbumin gene carrying this mutation synthesized and se- creted ovalbumin into the tissue culture media. Analysis showed that this material was not glycosylated at Asn312, and, therefore, the nascent polypeptide was responsible for direct- ing the proper site glycosylation of ovalbumin.

EXPERIMENTAL PROCEDURES

Materials-Mouse L353 cells, herpes simplex virus type 1 ts502A305, and pRB353 were provided generously by Dr. B. Roizman, University of Chicago. Mouse L (tk-) cells (CCL 1.3) were purchased from American Type Culture Collection. Rabbit anti-chicken egg albumin IgG and nonimmune rabbit serum were purchased from Cooper Biomedical. EN3HANCE, [y3’P]ATP (6000 Ci/mmol), and [36S]methionine (800 Ci/mmol) were obtained from Du Pont-New England Nuclear. Reverse sequencing primer, pTZ18, and helper phage M13K07 were supplied by Pharmacia LKB Biotechnology, Inc., and peptide:N-glycosidase F was obtained from Genzyme Corp. Other materials and reagents were purchased from commercial sources at the highest purity available.

Recombinant DNA and Site-directed Mutagenesis”pRB353 (3) contains the genomic ovalbumin gene under the control of the a- regulated promoter of the a-protein 4 gene from herpes simplex virus type 1. For selection in animal cells, the plasmid also contains a functional TK gene. After Asp,,,/SalI digestion of the plasmid, a 5.1- kilobase pair fragment including the DNA sequence coding for Amrn3 was ligated into pTZ18, a multifunctional phagemid vector (Fig. 1). Escherichia coli NM522 were transformed, and the plasmid DNA (pTZOV1) was isolated and mapped by restriction endonuclease digestions. Single-stranded template DNA was prepared as described by the manufacturer. Bacteria from single colonies were inoculated into 2.5 ml of LB media containing 100 rg/ml ampicillin and allowed to grow at 37 “C to 0.5 Asso. Two ml of the suspension was removed, and the cells were infected for 1 h at 37 “C with 10 plaque-forming units/cell of the helper phage M13K07. Four hundred pl of the cell suspension was inoculated into 10 ml of 2~ LB media containing 70 pg/ml kanamycin sulfate and allowed to shake overnight at 37 “C. After removal of bacteria by centrifugation, phage particles were precipitated from the supernatant by the addition of 1 ml of 25% polyethylene glycol, 3 M NaCl at room temperature for 15 min and collected by centrifugation. The pellet was resuspended in 400 r l of 20 m M Tris, pH 7.5, containing 20 mM NaCl and 1 mM EDTA, the phage particles were extracted with phenol, and the template DNA was isolated by ethanol precipitation. The sequence of single-stranded

12778

Protein Glycosylation in Animal Cells 12779

DNA was checked (4) prior to use as a template for in vitro mutagen- esis.

The procedure used for double primer mutagenesis was according to Zoller and Smith (5). A mutagenic oligodeoxynucleotide (20-mer) containing a %base mismatch to induce the desired mutation was synthesized by manual phosphoramidite chemistry similar to that described by Adams et al. (6). The mutagenic primer was phospho- rylated at the 5' end with T4 polynucleotide kinase and ATP. During the annealing reaction, 40 ng of mutagenic oligonucleotide were incubated with 4 ng of reverse sequencing primer and 0.5 pmol of template DNA. The annealed DNA was elongated and ligated over- night in the presence of DNA polymerase I (Klenow fragment) and T4 DNA ligase. This DNA was used to transform E. coli NM522 and the resulting colonies were screened for mutants as described below.

Screen for Mutants-Transformants from the mutagenesis were selected with toothpicks and placed into 96-well microtiter plates containing 100 p1 of 30% glycerol and then transferred onto nitrocel- lulose filters on LB/ampicillin plates. The colonies were allowed to grow overnight at 37 "C. Screening of colonies was performed as described by Woods (7). Colonies were lysed with NaOH prior to baking the filters for 1 h at 80 "C under vacuum. The filters were washed for 20 h at 65 "C with several changes of 3X SSC/O.l% SDS' and colonial matter was removed by rubbing with a gloved finger. The filters were prehybridized for 2 h at 37 "C in 6X SSC containing IX Denhardt's solution, 0.5% SDS, 250 Fg/ml salmon sperm DNA, and 0.05% sodium pyrophosphate prior to overnight hybridization at 37 "C in 6X SSC containing l x Denhardt's solution, 0.5% SDS, 100 pg/ml salmon sperm DNA, and 320 ng of 3*P-labeled mutagenic oligonucleotide. Filters were washed for 3 min in 6X SSC at increasing temperatures followed by autoradiography. Plasmid DNA was iso- lated from candidates for mutant colonies and used to transform bacteria. Four transformants were picked from each plate and re- screened as described above. Appropriate base pair substitutions were confirmed mutants by sequencing of single-stranded DNA and by dot-blot analysis of plasmid DNA. After Asp,,$SalI digestion of plasmid DNA from mutant colonies, the fragment containing the mutation was ligated into the large Asp,,,/SalI fragment of pRB353 to generate pRB353m. This plasmid was identical with pRB353 except for the 2-base pair substitution expected for the Asn to Gln change.

DNA Transfection-Mouse L (tk-) cells were transfected with 10 pg of plasmid DNA by the calcium phosphate method (8). This resulted in foci growing in hypoxanthine-, aminopterin-, and thymi- dine (HAT)-containing medium that were passaged without cloning.

Preparation of Radiolabeled Ouahumin-Methods for maintenance and metabolic labeling of cells have been described previously (1).

Isolation of Radiolabeled Oualbumin-Hemoglobin (10 mg) was added to 10 ml of tissue culture medium, and the solution was dialyzed against 1OX diluted phosphate-buffered saline (0.015 M NaCl and 0.001 M phosphate). Radiolabeled ovalbumin was isolated by immu- noaffinity chromatography on 1-ml columns of anti-chicken egg al- bumin IgG bound to Sepharose 4B (5 mg/ml) as previously reported (1). Fractions containing labeled ovalbumin were lyophilized for subsequent analyses.

imately 4000 cpm) were resuspended in 40 pl of 50 mM Tris-HC1, pH Peptide:N-glycosidase F Digestions-Samples (containing approx-

8.5, 50 mM EDTA, 1.25% Nonidet P-40, 0.1% SDS, 250 mM 2- mercaptoethanol, 0.05% sodium azide, heated to 100 "C for 5 min, and cooled on ice. After addition of 0.75 unit of peptide:N-glycosidase

terminated by heating at 100 "C for 5 min followed by SDS-10% F, the samples were incubated overnight a t 37 "C. Reactions were

polyacrylamide gel electrophoresis (9). Radioactivity was detected by fluorography with EN3HANCE.

RESULTS

Chicken ovalbumin has two potential N-glycosylation sites of which only one, AsnZQ3, is glycosylated in uiuo. Using this model glycoprotein, the possibility was explored that glyco- sylation at the first site prevents glycosylation at the second site in uiuo. Hence, the oligosaccharide chain at amino acid 293 was eliminated by introducing as innocuous a change as possible in the primary structure of ovalbumin, and glycosyl- ation at Asn3" was examined. A mutation was introduced into

The abbreviations used are: SDS, sodium dodecyl sulfate; HAT medium, hypoxanthine/aminopterin/thymidine medium.

the ovalbumin gene which allowed substitution of a glutamine residue for asparagine at amino acid 293 (see "Experimental Procedures" and Fig. 1). The expression of this mutant protein in a characterized cell line allowed the gene product to be isolated and examined for glycosylation at Asn312.

Multiple Cloning Site

pRB353

\DNA Ligase ,x' , I ,

4 Transform E. Coli NM522

k helper phage

4 I

ssDNA Rescue

DNA Sequencing

Site-specific Mutagenesis

+

Asn,,, IAAC' +

CTC-CTT - TTT-ATG, T ,GAG-TGT-AGA - T G

pTZOVI ssDNA

DNA DNA Ligase Polymerase I I

Transform E. Coli NM522 + I

Screen for Mutants

FIG. 1. Scheme for site-directed mutagenesis. pRB353 con- tains ovalbumin genomic DNA located downstream from and in the same transcriptional orientation as the a-regulated promoter of a- protein 4 from herpes simplex virus. A 5.1-kilobase pair (kb) fragment containing the sequence coding for Amm3 was removed from pRB353 and cloned into the multiple cloning site of pTZ18. The resulting plasmid, pTZOV1, was used to transform bacteria from which single- stranded template DNA was isolated for dideoxy sequence analysis and site-specific mutagenesis. A 20-base oligodeoxynucleotide con- taining a 2-base mismatch was used to induce the single amino acid change from asparagine (AAC) to glutamine (CAA) at position 293. After elongation and ligation of the template, E. coli were transformed with the DNA and screened for mutants. The asterisk shows the approximate location of the triplet coding for Amm3.

12780 Protein Glycosylation in Animal Cells

Isolation of Mutants-Transformants from the mutagenesis were replica-plated onto nitrocellulose filters and screened by colony hybridization with the 32P-labeled mutagenic oligode- oxynucleotide as described under "Experimental Procedures." In the primary screen for mutants, shown in the upper portion of Fig. 2, colonies with strong autoradiographic signals re- maining after the 50 "C wash were picked for having a high probability of containing mutant plasmid DNA. Due to the possibility of mixed plasmid populations in the isolated bac- terial colonies, plasmid DNA was prepared from these poten- tial positive colonies and used to transform E. coli NM522 a t low concentrations of DNA. Four colonies were picked from each plate, and they were transferred onto nitrocellulose filters in groups of four in horizontal rows. The lower portion of Fig. 2 shows the secondary screen of these transformants, and it is apparent from this figure that the potential positive colonies from the primary screen were heterogeneous. For instance, the eight colonies across the top row of the plate (four colonies from two different transformations) are all positive, whereas the first four colonies in the third row from the top are a mixture of one negative and three positive colonies. The last four colonies in the fourth row from the top are a mixture of one negative and three positive colonies. The

last four colonies in the fourth row from the top, similar to controls containing plasmid DNA without the 2-base change (the eight colonies in the sixth row from the top), are all negative. Plasmid and single-stranded DNA were isolated from single colonies selected from groups where all four members tested positive in the secondary screen. Dot-blot hybridization and dideoxy sequence analysis were performed, and these tests confirmed the identity of the positive colonies as mutants containing plasmid DNA with the desired 2-base change (pTZOVlm). The Asp,,~/SalI fragment containing the sequence coding for AsnZg3 in pRB353 was replaced with the analogous fragment from pTZOVlm which contained the 2- base change. The resulting vector, pRB353m, was identical with pRB353 except for the AAC to CAA substitution.

Glycosylation of Ovalbumin in Mouse L Cells-Mouse L (tk-) cells were transfected with pRB353 or pRB353m, and the resulting transfectants (L353 and L353m, respectively) were selected in HAT-containing medium. Cells were cultured overnight in the presence of [3sS]methionine after superinfec- tion with herpes simplex virus type 1 ts502A305. This virus carries a temperature-sensitive mutation such that at the nonpermissive temperature (39 "C), infected cells produce significant quantities of a-gene products, and, therefore, oval-

Primary Screen For Mutants

,

40" 50" 60" 70"

Secondary Screen For Mutants

40' 50" 60" 70" FIG. 2. Selection of mutants. Single colonies from the mutagenesis were transferred onto nitrocellulose filters,

probed with the "P-labeled mutagenic oligodeoxynucleotide, and washed with 6X SSC at the temperatures indicated prior to autoradiography. In the lower panel (secondary screen), DNA was isolated from potential mutant colonies in the primary screen and used to transform E. coli. Four transformants from each plate were transferred onto nitrocellulose filters and rescreened. The eight colonies along the bottom row of the filters in this secondary screen represent control colonies containing the vector without the 2-base change.

Protein Glycosylation in Animal Cells 12781

- 92.5

- 69

- 46

- 30

I 2 3 4 FIG. 3. Peptide:N-glycosidase F treatment of ovalbumin.

Mouse L353 cells (lanes 1 and 2) or L353m cells (lanes 3 and 4 ) were cultured for 20 h in the presence of [35S]methionine, and labeled ovalbumin was isolated from the tissue culture medium by immu- noaffinity chromatography. In both cases, the purified ovalbumin was untreated (lanes 1 and 3 ) or treated with peptide:N-glycosidase F (lanes 2 and 4) before SDS-10% polyacrylamide gel electrophoresis and fluorography. "C-methylated molecular weight standards are phosphorylase b (92,500), bovine serum albumin (69,000), ovalbumin (46,000), and carbonic anhydrase (30,000).

bumin (3). Radiolabeled ovalbumin was secreted into the tissue culture medium, and the protein was isolated by im- munoaffinity chromatography on columns of ovalbumin an- tibody immobilized on Sepharose 4B.

Radiolabeled ovalbumin, secreted into the tissue culture medium, was analyzed for the presence of an N-linked oligo- saccharide by incubation with or without peptide:N-glycosi- dase F prior to SDS-10% polyacrylamide gel electrophoresis and fluorography. This glycosidase removes asparagine-linked oligosaccharides from glycoproteins by hydrolyzing the link- age between the asparagine residue and the first N-acetylglu- cosamine moiety of the core oligosaccharide. Fig. 3 shows that ovalbumin isolated from L353 cells (with Asn a t position 293) is, as expected, glycosylated. This is evidenced by the fact that the sample digested by peptide:N-glycosidase F migrates with a slightly faster mobility than the untreated protein indicating the loss of a carbohydrate unit (compare lanes 2 and I, respectively). However, ovalbumin from L353m cells, with Gln a t position 293, is not glycosylated with an aspara- gine-linked sugar chain, as both the glycosidase-treated (lane 4 ) and untreated (lane 3) samples migrate with the same mobility after SDS-polyacrylamide gel electrophoresis. This is confirmed by the fact that the bands in lanes 3 and 4 co- migrate with the totally deglycosylated ovalbumin from L353 cells (lane 2). Therefore, Asn3" is not glycosylated in the absence of an oligosaccharide a t position 293.

DISCUSSION

Glycoproteins are ubiquitous, and they are known to be involved in a wide array of biological processes. Although in

some cases the function of the oligosaccharide moiety of the glycoprotein is not obvious, in many cases the presence of a carbohydrate unit has been reported to be essential for the proper biological activity, transport, secretion, or stability of a protein (see Refs. 10 and 11 for review). For example, the presence of the oligosaccharide chains of fibronectin (12, 13) and the acetylcholine receptor (12, 14) render these proteins less susceptible to proteolysis than their nonglycosylated an- alogs; the secretion of immunoglobulins IgA and IgE from plasma cells is significantly reduced in the absence of their N-linked oligosaccharide moieties (15); proper glycosylation of human chorionic gonadotropin is not essential for binding to gonadotropin receptors, but it is necessary for this hormone to elicit the cascade of events associated with CAMP synthesis (16, 17); the cell surface expression of many receptor mole- cules is greatly diminished when the attachment of oligosac- charides is inhibited (lo), while in cases such as diphtheria toxin receptor, the number of molecules reaching the cell surface is not affected when sugar chain addition is inhibited, but the affinity of the receptor for its ligand is altered (18), and specific carbohydrate sequences are essential on serum glycoproteins and lysosomal hydrolases for their recognition by asialoglycoprotein (19) and mannose 6-phosphate (20) receptors, respectively. In view of this increased awareness of the importance of carbohydrates in cell physiology, it is im- portant to elucidate the factors that govern protein glycosyl- ation. Although work in many laboratories has led to the discovery of the basic enzymatic steps involved in the attach- ment and processing of carbohydrate chains (21), relatively little is known about the mechanisms for accomplishing site- specific glycosylation and oligosaccharide processing.

To determine the influence of the protein structure and cell type on proper site glycosylation and processing, an earlier study was carried out in which the glycosylation of a cloned glycoprotein expressed in a heterologous cell was examined (1). Results from that study showed that chicken ovalbumin synthesized in mouse L cells, similar to the protein made in hen oviduct, was glycosylated exclusively a t Amm3, and that the oligosaccharides at that site were predominantly of the hybrid type. This suggested that the nascent protein chain plays a significant role in directing the site-specific glycosyl- ation and processing of glycoproteins. In a continuation of that study, the present report focuses on factors involved in specific glycosylation of Asnm3, but not Asn312, in chicken ovalbumin. The technique of site-specific mutagenesis was performed to examine whether an oligosaccharide chain at Amm3 is responsible for blocking attachment of an N-linked sugar chain a t Asn31Z, or if the nascent polypeptide itself contains sufficient information to direct proper site glycosyl- ation. By making the conservative substitution of glutamine for asparagine, glycosylation at amino acid 293 is specifically prevented and allows the glycosylation of Asn"' to be exam- ined. The results indicate that the presence of an oligosaccha- ride at AsnZs3 does not affect glycosylation a t Asn"', suggest- ing that there is little communication between the two sites in uiuo. This would suggest that the in vitro glycosylation of Asn312 in the denatured molecule reported by Lennarz and co- workers (2) is likely due to an overall change in the tertiary structure of the protein which makes Asn3" accessible to the glycosylation apparatus and not due to a change in the inter- action of Asn3" with the oligosaccharide a t Amm3.

Oligonucleotide-directed site-specific mutagenesis allows for specific changes in the primary structure of glycoproteins and facilitates studies of factors governing site-specific gly- cosylation and oligosaccharide processing. With this tech- nique, Rose and co-workers have examined the role of sugar

12782 Protein Glycosylation in Animal Cells

chains at specific sites (22, 23). In one such study (23), it was demonstrated that either one of the two N-linked sugar chains of the vesicular stomatitis virus G protein is sufficient for cell surface expression of the G protein in transfected cells, while the nonglycosylated protein does not reach the cell surface. The approach utilized in these reports and in the current study obviate the need for antibiotics which inhibit core glycosylation, such as tunicamycin. For instance, ovalbumin is known to be secreted from cultured cells in the presence of tunicamycin, indicating that an N-linked sugar chain at AsnB3 is not needed for secretion (24, 25). However, the use of this inhibitor is not compatible with examining the glycosylation of the second site since tunicamycin inhibits total cellular glycosylation.

An advantage of using in vitro mutagenesis instead of antibiotics that inhibit protein glycosylation is that specific alterations in the polypeptide can be made without having to be concerned about the effect on other essential cellular processes such as protein synthesis (10). Also, the conse- quence of not glycosylating other proteins in the system under study is not a consideration with site-specific mutagenesis. However, the drawback of the immensely powerful technique of in vitro mutagenesis is that it is difficult to access unin- tended effects of the desired mutation on protein conforma- tion when there is no structural information available for the protein of interest, as is the case with chicken ovalbumin. To minimize the chance that a mutation which would prevent glycosylation of Ams3 would cause changes in the tertiary structure of ovalbumin, it was determined that the most reasonable substitution to make was glutamine for asparagine. This effort notwithstanding, in the absence of precise infor- mation on the effect of the substitution on the conformation of the protein, one cannot be absolutely certain that there is not some ancillary impact of the mutation on protein struc- ture which might influence glycosylation. This is true in the present study where a negative result is found after the amino acid substitution is made, i.e. Asn312 is not glycosylated after the mutation at AsnZg3.

Future studies must take advantage of emerging technolo- gies that define peptide and protein structure in order to more exactly determine the structure of peptides that act as accep- tors in oligosaccharide transfer reactions. In conjunction with prudent mutagenesis experiments, these types of studies will potentially allow precise determination of the subtleties in- volved in N-linked glycoprotein synthesis. Not only will ex- periments of this type elucidate the specific domains or se-

quences recognized by the glycosylation and processing ap- paratus, they will also more adequately access potential secondary effects of site-specific mutations.

Acknowledgments-I am indebted to Dr. Phillips W. Robbins for his support and encouragement. I thank Dr. M. A. Kukuruzinska for her critical reading of the manuscript.

REFERENCES 1. Sheares, B. T., and Robbins, P. W. (1986) Proc. Nutl. Acud. Sci.

2. Glabe, C. G., Hanover, J. A., and Lennarz, W. J. (1980) J. Biol.

3. Post, L. E., Norrild, B., Simpson, T., and Roizman, B. (1982)

4. Sanger, F., Coulson, A. R., Barrel, B. G., Smith, A. J. H., and

5. Zoller, M. J., and Smith, M. (1984) D N A ( N Y ) 3,479-488 6. Adams, S. P., Kavka, K. S., Wykes, E. J., Holder, S. B., and

U. S. A. 83,1993-1997

Chem. 256,9236-9242

Mol. Cell. Biol. 2, 233-240

Roe, B. A. (1980) J. Mol. Biol. 143, 161-178

Gallupi, G. R. (1983) J. Am. Chem. SOC. 106,661-663 7. Woods, D. (1984) FOCUS 6, 1-3 8. Wigler, M., Silverstein, S., Lee, L., Pellicer, A., Cheng, Y., and

9. Laemmli, U. K. (1970) Nature 227,680-685 Axel, R. (1977) Cell 11, 223-232

10. Albein, A. D. (1987) Annu. Reu. Biochem. 56,497-534 11. Olden, K., Bernard, B. A., Humphries, M. J., Yeo, T.-K., Yeo,

K.-T., White, S. L., Newton, S. A., Baver, H. C., and Parent, J. B. (1985) Trends Biochem. Sei. 78-82

12. Olden, K., Parent, J. B., and White, S. L. (1982) Biochim. Biophys. Acta 650,209-232

13. Olden, K., Prat, R. M., and Yamada, K. M. (1978) Cell 13,461- A72

14. Prives, J. M., and Olden, K. (1980) Proc. Natl. Acud. Sci. U. S. - .- A. 77.5263-5267

15. Hickman, S., Kulczycki, A., Jr., Lynch, R. G., and Kornfeld, S.

16. Rebois, R. V., and Fishman, P. H. (1983) J. Biol. Chern. 258,

17. Kalyan, N. K., and Bahl, 0. P. (1983) J. Biol. Chem. 258,67-74 18. Kranitzky, W., Durham, D. L., Hart, D. A., and Eidels, L. (1985)

19. Ashwell, G., and Harford, J. (1982) Annu. Reu. Biochem. 51,

20. von Figura, K., and Hasilik, A. (1986) Annu. Reu. Biochem. 55,

21. Kornfeld, R., and Kornfeld, S. (1985) Annu. Reu. Biochern. 54,

22. Guan, J. L., Machmer, C. E., and Rose, J. K. (1985) Cell 42,489-

23. Machamer. C. E.. Florkiewicz. R. Z.. and Rose, J. K. (1985) Mol.

(1977) J. Biol. Chem. 262,4402-4408

12775-12778

Infect. Zmmun. 49,336-343

531-554

167-193

631-664

496

Cell. Biol. 5,3074-3083 ’

Commun. 85,762-768

Morser, J. (1981) Eur. J. Biochem. 113,339-348

24. Keller, R. K., and Swank, G. D. (1978) Biochem. Biophys. Res.

25. Coleman, A., Lane, C. D., Craig, R., Boulton, A., Mohun, T., and