Direct carbohydrate analysis of glycoproteins electroblotted onto polyvinylidene difluoride membrane from sodium dodecyl sulfate-polyacrylamide gel

Direct carbohydrate analysis of glycoproteins electroblotted onto polyvinylidene difluoride membrane from sodium dodecyl sulfate-polyacrylamide gel

ANALYTICAL BIOCHEMISTRY 190,165-169 (1990) Direct Carbohydrate Analysis of Glycoproteins Electroblotted onto Polyvinylidene Difluoride Membrane fr...

697KB Sizes 1 Downloads 4 Views

Recommend Documents

Detection of hemoprotein peroxidase activity on polyvinylidene difluoride membrane
The feasibility of detecting hemoproteins after electroblotting was examined. Hemorproteins were subjected to lithium do

Equilibrium of octadecylsilica gel with sodium dodecyl sulphate
The retention times of tyrosine (Tyr), 3,4-dihydroxyphenylalanine (DOPA), 3,4-dihydroxyphenylethylamine (DA), norepineph

Detection of citrulline residues in deiminated proteins on polyvinylidene difluoride membrane
We have developed a new detection method of deiminated proteins on polyvinylidene difluoride membranes. Citrulline resid

Correction factors for membrane protein molecular weight readouts on sodium dodecyl sulfate–polyacrylamide gel electrophoresis
Membrane proteins are known to migrate anomalously on sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAG

Evaluation of urinary proteins by sodium dodecyl sulphate polyacrylamide disc gel electrophoresis and molecular mass analysis
Patterns of urinary proteins were examined on sodium dodecyl sulphate polyacrylamide gels and compared to those obtained

Multimeric analysis of von Willebrand factor by molecular sieving electrophoresis in sodium dodecyl sulphate agarose gel
We report the development and optimisation of an agarose gel electrophoretic method for the separation and detection of

Sodium dodecyl sulfate-protein polypeptide complexes in 8 M urea with special reference to sodium dodecyl sulfate-polyacrylamide gel electrophoresis
The effects of 8 M urea on the complexes formed between sodium dodecyl sulfate and protein polypeptide were found to be

Improved Method for Silver Staining of Glycoproteins in Thin Sodium Dodecyl Sulfate Polyacrylamide Gels
A method for detection of glycoproteins in thin sodium dodecyl sulfate polyacrylamide gels was developed by a combinatio

Electrophoretic transfer of proteins from sodium dodecyl sulfate-polyacrylamide gels to a positively charged membrane filter
Zeta-bind, a positively charged nylon membrane, was tested as an immobilizing matrix for the electrophoretic transfer of

ANALYTICAL

BIOCHEMISTRY

190,165-169

(1990)

Direct Carbohydrate Analysis of Glycoproteins Electroblotted onto Polyvinylidene Difluoride Membrane from Sodium Dodecyl Sulfate-Polyacrylamide Gel Haruko

Ogawa,

Department

Received

Miyoko

of Chemistry,

April

Ueno,

Haruhi Uchibori, Isamu Matsumoto, and Nobuko Seno Faculty of Science,Ochanomizu University, Bunkyo-ku, Tokyo 112, Japan

10, 1990

A procedure for the carbohydrate analysis of glycoproteins electrotransferred to a polyvinylidene difluoride membrane is described. The glycoproteins (plant lectins, transferrin, and vitronectin) were first separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and then electroblotted onto a membrane. Each of the glycoprotein bandsvisualized by staining with Coomassie brilliant blue R-250 was excised from the membrane and subjected to direct hydrolysis either in 2.5 M trifluoroacetic acid at 100°C for 6 h for neutral sugars and hexosamines, or in 0.05 M H,SO, at 8O’C for 1 h for sialic acids. The hydrolysate obtained was analyzed for neutral sugars, hexosamines, and sialic acids independently by three different systems of high-performance liquid chromatography. The analytical values were reproducible with reasonable accuracy and agreed with those expected with recoveries of 57-66%. The method was successfully applied to a mannose-specific lectin of Sophom japonica bark, which is composed of four different subunits that aggregate sugar specifically. Because the four subunits could be separated by SDS-PAGE alone, the method proved useful for determining their carbohydrate compositions. Three of them were shown to contain carbohydrates typical of N-linked oligosaccharides of plant origin, which agreed well with the results of the binding assay carried out on a membrane using various horseradish peroxidase-labeled lectins. o 1990 Academic press, IN.

The carbohydrate moieties of glycoproteins have been shown to perform important biological roles such as stabilization, protection from proteases, modulation of activities, recognition sites for other molecules, clearance from blood stream, and processing and targeting of glycoproteins (1,2). The monosaccharide composition of purified glycoproteins must be analyzed in order to $3.00

0003-2697/90

Copyright All rights

0 of

by Academic Press, reproduction in any form 1990

elucidate the structure of their oligosaccharide chains. Minute amounts of glycoproteins can conveniently and effectively be separated by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE),’ and several procedures have been reported for analyzing the protein portions of the bands obtained by SDS-PAGE. The conventional method is to analyze the samples extracted or electroeluted from polyacrylamide gel, but this presents problems of recovery and contamination because of the many steps involved. Direct analysis of the gel after staining the bands is a convenient method, but this gives unidentified peaks from the gel (3). Western blotting is used to detect antigens (4) and the specific carbohydrate chains (5-8) on bands by using labeled antibodies and lectins, respectively. Recently, the direct analysis of bands blotted onto a polyvinylidene difluoride (PVDF) membrane has been developed, by which amino acid sequencing (9) and direct amino acid analysis (10,ll) have been achieved. In this paper, we applied the method to the carbohydrate analysis of glycoprotein bands. The results obtained provided chemical evidence for the oligosaccharide structures of the glycoprotein bands estimated by lectin staining on the same membrane. MATERIALS

AND

METHODS

Clerodendron trichotomum lectin (CTA) was prepared from the fruit of C. trichotomum as described previously (7). Sophoru juponica bark lectin (B-SJA-II), which is specific for mannose and glucose, was purified by two steps of affinity chromatography on lactamyl- and mal’ Abbreviations used: SDS, sodium dodecyl su1fat.e; PAGE, polyacrylamide gel electrophoresis; PVDF, polyvinilydene difluoride; CTA, Ckrodendrm trichotomum agglutinin; B-SJA-II, mannose-specific Sophora japoniea bark Iectin; HRP, horseradish peroxidase; HPLC, high-performance liquid chromatography. 165

Inc. reserved.

166

OGAWA

tamyl-Sepharose columns prepared as described previously (12). Human serum apo- and holotransferrin was purchased from Wako Pure Chemicals (Osaka, Japan). Bovine vitronectin was purified according to Yatohgo et al. (8,13). Horseradish peroxidase (HRP)-conjugated concanavalin A, Pisum sativum lectin, Lens culinaris lectin, and Vicia faba lectin were prepared as described previously (7). Acrylamide, N,N’-methylenebisacrylamide, ammonium persulfate, and N,N,N’,N’tetramethylethylenediamine were electrophoretical grade and 2-mercaptoethanol was biochemical grade of Wako Pure Chemicals. Coomassie brilliant blue R-250 was electrophoretical grade of Nacalai Tesque Inc. (Kyoto, Japan). Molecular weight markers (Marker L; 12,300-78,000) were obtained from BDH (Poole, England). Trifluoroacetic acid was amino acid sequence grade and sulfuric acid was super special grade of Wako pure Chemicals. Other reagents were of the special grade of Wako Pure Chemicals. PVDF membrane (Immobilon; pore size, 0.45 pm) was purchased from Milipore Corp. (Bedford, MA). Nitrocellulose membrane (pore size, 0.45 pm) was purchased from Advantec (Tokyo, Japan). Electrophoresis

and Electroblotting

SDS-PAGE was carried out by the method of Laemmli (14) using minigels (7 X 9 cm, 1 mm thick) of 8.0 or 13.0% polyacrylamide. A blank gel or blank lane was run simultaneously. Electroblotting was performed by the method of Hirano (15). Blotting buffers contained: (A) 0.3 M Tris, 20% (v/v) methanol; (B) 25 mM Tris, 20% (v/v) methanol; (C) 25 mM Tris-boric acid, 20% methanol (pH 9.5). Three sheets of PVDF membrane cut to the size of the gel were wetted with 100% methanol and then soaked in transfer buffer C. After electrophoresis, the gels were washed with transfer buffer C for 15 min with shaking. Two sheets of filter paper wetted in buffer A, another two sheets wetted in buffer B, and three sheets of PVDF membrane, gel, and two sheets of filter paper wetted in buffer C were piled in this order on the anode plate of an Atto Horizblot AE6670 apparatus. Electroblotting was carried out at 7080 mA (1.0 mA/cm2) constant current for 120 min at room temperature. Since vitronectin was hardly transferred to the membrane under these conditions, it was electroblotted in 3-(cyclohexylamino)-l-propanesulfuric acid at 160 mA for 150 min using a Sartoblot II apparatus. The electroblotted membranes were washed with 10 mM sodium borate buffer (pH 8.0)-25 mM NaCl for 5 min and then twice with redistilled water for 10 min. Thev were then stained with 0.1% Coomassie brilliant blue R-250 in water/methanol/acetic acid (5/5/2) for 5 min, destained in 90% methanol with several changes, and washed with and stored in redistilled water at 4°C overnight.

ET

AL.

Hydrolysis

and Carbohydrate

Analysis

Each band was cut from the membrane, sliced, and placed into a glass hydrolysis tube which had been washed with boiling 50% nitric acid and water. An equal-sized piece of the electroblotted membrane corresponding to the same migration position was cut from the blank lane and used as a blank for hydrolysis. For neutral sugar and amino sugar analyses, hydrolysis was carried out in vacua with 500 ~1 of 2.5 M trifluoroacetic acid at 100°C for 6 h (16). After hydrolysis, ribose (1 pg) was added to the tube as an internal standard and mixed well. The supernatant was transferred to another tube and evaporated to dryness. The residues were dissolved in 30 ~1 of redistilled water and filtered through Ultrafree C3GV (Millipore). The resulting monosaccharide mixtures were analyzed by high-performance liquid chromatography (HPLC) as described by Mikami and Ishida (17). The following equipment was used: a Tosoh HPLC 803D for pumping eluant, a Rheodyne injector carrying a loo-p1 loop, a Jusco 860-CO column oven, Jusco BIP-1 twinpiston pumps for postcolumn labeling, ASB-200 reaction bath, FP-210 spectrofluorometer, a SIC Labchart 180 integrator, and a Jusco Finepak GEL SA-121 column (0.6 X 10 cm). Elution was performed with 0.3 M boric acid (pH 8.35) at 60°C at a flow rate of 0.60 ml/ min. The eluate was mixed with 2% arginine-3% boric acid at the same flow rate of 0.60 ml/min, and the resultant effluent was passed through a reaction coil maintained at 150°C. After the reaction mixture had passed through a cooling coil, the fluorescence was recorded by using excitation and emmision wavelengths of 320 and 430 nm, respectively. Since D-glucosamine and D-galactosamine cannot be distinguished by the HPLC system, amino sugar composition was determined with a Hitachi 835 amino acid analyzer using the same hydrolysate. The ninhydrin method was used for detection. For sialic acid analysis, hydrolysis was carried out in 25 ~1 of 0.05 M sulfuric acid per band of membrane at 80°C for 60 min in vacua (18). After filtration of the hydrolysate through Ultrafree C3GV, 20-~1 portions were analyzed using an ion-exclusion HPLC system comprising a Tosoh HPLC 803D, a Shodex AO-50 column oven, and a Tosoh uv 8000 spectrophotometer with a column of Shodex Ionpak KC-811 (8 X 300 mm), as previously described (8). Elution was performed with 0.3% phosphoric acid at a flow rate of 1.0 ml/min at 60°C and sialic acids were monitored by their absorbance at 205 nm. Interaction

of B-SJA-II

with HRP Lectins

HRP-concanavalin A, L. culinuris lectin, P. sativum lectin, and V. faba lectin were used to detect carbohydrates of B-SJA-II subunits. B-SJA-II (7 pg) was separated into subunits by SDS-PAGE and transferred to

CARBOHYDRATE TABLE Neutral Human

Sugar and Transferrin

ANALYSIS

Compositions onto PVDF

CTA

Recoveryd

Human

Direct hydrolysis’ (mol %)”

32 42 11 0 9.0 0

Glucosamine Mannose Fucose Galactose Xylose Glucose (%I

66

GLYCOPROTEINS

of CTA Membrane

membrane*

(mol%)”

34 39 16 0 8.2 1.6 100

and

transferrin

Hydrolysis on PVDF

Direct hydrolysis’ (In01 %)O

43 32 0 25 0 0

44 30 0 27 0 0

57

100

’ Values are expressed in mol % as the average of three analyses obtained from an individual blotting. * Samples (100 rg) were applied to SDS-PAGE and electroblotted onto PVDF membrane. The stained bands were cut out, hydrolyzed, and analyzed by HPLC as described under Materials and Methods. A blank membrane corresponding to the same migration position as the band was hydrolyzed and analyzed for each experiment, and used for correction. ’ Samples (100 pg) were hydrolyzed without SDS-PAGE and electroblotting. d Recovery was calculated from total of neutral sugars and hexosamines by taking the value of direct hydrolysis as 100%.

nitrocellulose sheets and stained with each HRP lectin as described previously (7). Lectin binding was detected by using 3,3’-diaminobenzidine/H,O, as the substrate for HRP. RESULTS

AND

Western

Blotting

DISCUSSION

Membrane

A nitrocellulose membrane ysis, since it has been widely

was first used for the analused in the lectin staining

Neutral

of Human Transferrin Human Hydrolysis on PVDF membraneb (mol %)O

N-Acetylneuraminic N-Glycorylneuraminic Recoveryd (%)

acid acid

100 0 62

and

167

ELECTROBLOTTING

Carbohydrate

and Hexosamine

2

Bovine

Vitronectin

Electroblotted

onto

transferrin

Bovine

Direct

Analysis

Human serum transferrin and CTA (100 fig) were each electrophoresed on SDS-PAGE, electroblotted onto PVDF membranes, and stained with Coomassie blue R-250. Both were observed as a single band on the PVDF membrane. The carbohydrate compositions of the hydrolysates of the stained band pooled from three membranes agreed with the analytical results for solutions of each glycoprotein, as shown in Table 1. The carbohydrate composition of transferrin was close to that reported previously (19). D-Glucose was often detected in the blank hydrolysate, especially in that of the membrane corresponding to the dye front of SDSPAGE. This may be because of contamination in reagents used for electrophoresis. To cancel the background contamination, an equal-sized piece of membrane from the same migration position as the corresponding glycoprotein band was taken as a blank membrane for each experiment.

TABLE

Sialic Acid Compositions

BY

assay because of its high efficiency of blotting. The membrane gave a turbid solution on hydrolysis with 2.5 M trifluoroacetic acid, and a clear solution with 0.05 M sulfuric acid. The turbid solution was clarified by filtration with Ultrafree C3GV. When the solutions were analyzed on HPLC systems, however, a large peak of glucose and an ascending baseline were observed in the neutral carbohydrate analysis, as well as many unidentified peaks in the sialic acid analysis. On the other hand, the blank tests with PVDF membrane gave no detectable peaks on HPLC in neutral carbohydrate analysis and very small peaks in sialic acid analysis. These peaks corresponded to neither N-acetylneuraminic acid nor N-glycorylneuraminic acid peaks. Therefore, PVDF membrane was hereafter used for the carbohydrate analyses.

1

Hexosamine Electroblotted

Hydrolysis on PVDF membraneb (In01 %)”

OF

hydrolysis’ (mol %)” 100 0 100

Hydrolysis on PVDF membrane* (mol %)a 48 52 57

PVDF

Membrane

vitronectin

Direct

hydrolysis’ (mol %)” 46 54 100

’ Values are expressed in mol % as the average of three analyses for transferrin and two analyses for bovine vitronectin from an individual blotting. b Samples (12-16 pg per well) were applied to SDS-PAGE and electroblotted onto PVDF membrane. The stained bands were cut out, hydrolyzed and analyzed by HPLC as described under Materials and Methods. Values were corrected by the data of blank hydrolysis as described under Table 1. ’ Samples (4-10 pg) were hydrolyzed without SDS-PAGE and electroblotting. ’ Recovery was calculated from total of sialic acids by taking the value of direct hydrolysis as 100%.

168

OGAWA ET AL.

28,600 (19,400) (18,200) (15,000) (13,200)

FIG.

1. SDS-polyacrylamide gel electrophoresis of B-SJA-II. B(about 8 pg) was applied to SDS-PAGE (14) using 13% polyacrylamide gel for separation and 3% gel for stacking. The gel was stained with Coomassie blue R-250 and destained with 7% acetic acid. The migration positions of molecular weight markers are indicated at left. The molecular weights of the four subunits of B-SJA-II, termed a-l, a-2, b-l, and b-2, were estimated by Ferguson plot (26).

SJA-II

The recoveries of total carbohydrates loaded onto the gels ranged from 57 to 66%. The recovery of total carbohydrates from PVDF membrane was higher than that of amino acids: our results, 20% for transferrin, and those by others, 29-47% for soybean trypsin inhibitor (11) and 29% for human serum albumin (10). The difference in recovery between carbohydrate and amino acid would arise from the readsorption efficiency of carbohydrates and amino acids onto PVDF membrane after hydrolysis. The total losses of glycoprotein in this work (3040%) would have occurred during transfer.

Siulic Acid Analysis As little as lo-pm01 amounts of N-glycorylneuraminic acid and N-acetylneuraminic acid were each quantitated in the sialic acid analysis on HPLC system. However, about 150 pmol of glycoprotein (12 pg of transferrin) per single analysis needed to be applied to SDS-PAGE in order to cover the losses during the transfer and hydrolysis process. The sialic acid compositions of the transferrin and bovine vitronectin from the PVDF membrane are summarized in Table 2. Only N-acetylneuraminic acid was detected in human serum transferrin and both N-glycorylneuraminic acid and Nacetylneuraminic acid were detected in bovine vitronectin by the membrane analysis and by the direct analysis of solubilized samples. The sialic acid compositions agreed well with those reported previously (8,19). The recovery of total sialic acid was 57-66%.

Application

to the Multiple

Subunits

of B-SJA-II

B-SJA-II was separated into four subunits, a-l (19.4 kDa), a-2 (18.2 kDa), b-l (15.0 kDa), andb-2 (13.2 kDa), by SDS-PAGE, as shown in Fig. 1. The subunits were electroblotted onto PVDF membrane and stained with HRP lectins. Subunits a-l, a-2, and b-l were stained with all the lectins specific to N-linked oligosaccharide

chains, but subunit b-2 reacted with none of them. The HRP lectins can stain as low as 1 pmol of carbohydrate chains (8), but subunit b-2 failed to react even when sufficient amount of B-SJA-II (7 pg) was subjected to the assay. Because concanavalin A and pea lectin do not contain carbohydrates, the positive staining observed seems to be due to their sugar binding activities toward carbohydrates of B-SJA-II subunits, not to those of BSJA-II. For carbohydrate analysis, B-SJA-II (500 pg) was electrophoresed on SDS-polyacrylamide gel, electroblotted onto PVDF membrane, and stained with Coomassie blue R-250. The carbohydrate compositions of the four subunits stained showed that subunits a-l, a-2, and b-l contained glucosamine, mannose, fucose, and xylose in various molar ratios and that subunit b-2 was free from carbohydrates, as shown in Table 3. The results, together with those of HRP lectin staining, indicate that subunits a-l, a-2, and b-l are glycopeptides containing N-linked oligosaccharides with xylose and fucose, typical of the oligosaccharides of plant origin (7,20-25) and subunit b-2 is a polypeptide containing no carbohydrates. This study showed that the carbohydrates of glycoproteins or glycopeptides can be analyzed after their separation by SDS-PAGE and subsequent electroblotting onto PVDF membrane as efficiently as amino acids can be analyzed on membrane. Furthermore, the results obtained by this method, together with those of lectin staining experiments, give useful indicators of the oligosaccharide structures of glycoproteins and glycopeptides. ACKNOWLEDGMENTS This study was supported in part by a Grant-in-Aid Scientific Research from the Ministry of Education,

TABLE

(01780194) for Science and Cul-

3

Carbohydrate Compositions of B-SJA-II Subunits Electroblotted onto PVDF Membrane Subunit B-SJA-IIb Glucosamine Mannose Fucose Xylose Glucose

3.5 3.0 1.3 1.0 0.7

(molar

ratio)

a-l’

a-2”

b-l’

3.0 3.0 1.3 0.8 0

2.4 3.0 0.9 0.8 0

1.9 3.0 1.4 1.0 0

a b-2’ 0 0 0 0 0

rr Values are expressed as molar ratios, taking the value of mannose as 3.0. b Intact B-SJA-II was directly hydrolyzed without SDS-PAGE and electroblotting. ’ B-SJA-II (500 pg) was applied to SDS-PAGE and electroblotted onto PVDF membrane. The stained bands were separately hydrolyzed and each hydrolysate was analyzed as described under Materials and Methods. Values were corrected by the data of blank hydrolysis as described under Table 1.

CARBOHYDRATE

ture of Japan, and reported nese Biochemical Society,

ANALYSIS

in part at the 62th Congress Kyoto, 1989.

OF

of The

GLYCOPROTEINS

Japa-

15. Hirano,

1. Rademacher, T. W., Parekh, R. B., and Dwek, R. A. (1988) Annu. Rev. Biochem. 57, 785-838. 2. Sharon, N., and Lis, H. (1981) Chem. Eng. News March 30,21-44.

169

ELECTROBLOTTING

13. Yatohgo, T., Izumi, Cell Struct. Funct. 14. Laemmli,

REFERENCES

BY

M.,

U. K. (1970) H. (1989)

Kashiwagi,

H., and Hayashi,

M.

(1988)

13, 281-292. Nature

J. Protein

fLondon) Chem.

227,680-685.

8,115-130.

16. Arakawa, Y., Imanari, T., and Tamura, Z. (1976) Bull. 24,2032-2037. 17. Mikami, H., and Ishida, Y. (1983) BunsekiKugaku

Chem. 32,

Pharm.

E207-210.

4. Towbin, H., Staehelin, T., and Golden, J. (1979) Proc. Nutl. Sci. USA 76,4350-4354. 5. Kijimoto-Ochiai, S., Katagiri, Y. U., and Ochiai, H. (1985) Biochem. 147,222-229.

Acad.

18. Honda, S., Iwase, S., Suzuki, S., and Kakehi, K. (1987) Anal. Biothem. 160,455-461. 19. Regoeczi, E., Taylor, P., Debanne, M. T., Marz, L., and Hatton, M. W. C. (1979) Biochem. J. 184,399-407.

Anal.

20. Ishihara, H., Takahashi, N., Oguri, Biol. Chem. 254, 10,715-10,719.

6. Rohringer,

144,

21. Takahashi, N., Hotta, T., Ishihara, H., Mori, M., Tejima, S., Bligny, R., Akazawa, T., Endo, S., and Arata, Y. (1986) Biochemistry 25,388-395.

3. Yamagata, 14

T., and Yamagata,

S. (1986)

Connect.

l&146-

Tissue

i.

R., and

Holden,

D. W.

(1985)

Anal.

Biochem.

118-127. 7. Kitagaki-Ogawa, H., Matsumoto, I., Seno, N., Endo, S., and Arata, Y. (1986) Eur. J. Biochem. 8. Kitagaki-Ogawa, H., Yatohgo, wagi, H., Matsumoto, I., and Acta 1033,49-56.

Takahashi,

N.,

161, 779-785.

T., Izumi, M., Hayashi, M., KashiSeno, N. (1990) Biochim. Biophys.

9. Matsudaira, P. (1987) J. Biol. Chem. 262, 10,035-10,038. 10. Tous, G. I., Fausnaugh, J. L., Akinyosoye, O., Lackland, H., Winter-Cash, P., Vitorica, F. J., and Stein, S. (1989) Anal. Bio&em. 179,50-55. 11. Nakagawa, S., and Fukuda, T. (1989) Anal. Biochem. 181,75-78. 12. Matsumoto, I., Kitagaki, H., Akai, Anal. Biochem. 116, 103-110.

Y., Ito, Y., and Seno,

N. (1981)

S., and Tejima,

S. (1979)

J.

22. Hase, S., Koyama, S., Daiyasu, H., Takemoto, H., Hara, S., Kobayashi, Y., Kyogoku, Y., and Ikenaka, T. (1986) J. Biochem. 100, l-10. 23. Kimura, Y., Hase, S., Kobayashi, Y., Kyogoku, Y., Ikenaka, and Funatsu, G. (1988) J. Biochem. 103,944-949.

T.,

24. Ashford, D., Dwek, R. A., Welply, J. K., Amatayakul, S., Homans, S. W., Lis, H., Taylor, G. N., Sharon, N., and Rademacher, T. W. (1987) Eur. J. Biochem. 166, 311-320. 25. Fournet, B., Leroy, R. D., and Goldberg,

Y., Wieruszeski, R. (1987) Eur.

26. Hedrick,

Smith,

J. L., and

126,155-164.

J. M., Montreuil, J. Biochem. 166,

A. J. (1968)

Arch.

Biochem.

J., Poretz,

321-324. Biophys.