[2] Characterization of carbohydrate units of glycoproteins

26

ANALYTICAL METHODS

[2]

Color Reaction. The alkaline hydroxylamine solution is prepared just before use by mixing equal volumes of 0.35 M hydroxylamine HC1 and 1.5 M NaOH. The ferric perchloric acid solution is prepared by dissolving 1.9 g of FeC13.6 H20 in 5 ml of concentrated HC1, into which is further added 5 ml of 70% HC104. It is then evaporated almost to dryness in a vacuum rotator and then diluted to 100 ml with water. The colorimetric determination is performed directly in tube C. Several samples can be distilled and kept stoppered in the refrigerator until they are ready for analysis. Prior to the colorimetric analysis, the tubes are brought to room temperature by standing them in a water bath at 20-25 ° . One milliliter of water is added to each tube, and this is followed by 2 ml of the alkaline hydroxylamine solution. The tube is shaken and allowed to stand 10 minutes; 2 ml of 0.75 M perchloric acid is then added, followed by shaking and the addition of 1.0 ml of the ferric perchloric acid solution. The red color is determined in 5-10 minutes at 520 m~. The quantity of methyl acetate in the sample is determined by comparison with ethyl acetate standards which are analyzed colorimetrically in each run. The standard solution consists of 5 micromoles of ethyl acetate per milliliter in methanol-water, 1 : ! (v/v). Appropriate aliquots of this standard are pipetted into tubes and made up to a volume of 2 ml with methanol-water, 1:1. The recovery should be checked by hydrolyzing and distilling N-acetylglucosamine and Nacetylneuraminic acid standards under the same conditions as the samples. In addition to performing aeetyl analyses, evidence as to occurrence of N-acetylated hexosamines may also be obtained by releasing the hexosamines without prior deacetylation through the use of glucosaminidase or weak acid hydrolysis (this volume ['2] ). The released Nacetylhexosamines can be identified by paper chromatography and measured by a modification of the Morgan-Elson reaction. 43 ~J. L. Reissig, J. L. Strominger, and L. F. Leloir, Biochim. Biophys. Acta 217, 959 (1955); see also this volume [3], [23].

[2] Characterization of Carbohydrate Units of Glycoproteins By ROBERT G. SPmo In a structural investigation of the carbohydrate portion of a glycoprotein, there are several major problems which have to be considered. These are: (1) the number, size, and composition of the carbohydrate units present in the glycoprotein; (2) the structure of these carbohydrate

[2]

CARBOHYDRATE UNITS OF GLYCOPROTEINS

27

units in regard to their monosaccharide sequence, linkages, and branching; and (3) the chemical nature of the glycopeptide bond, the amino acid and sugar involved in this linkage, and the location(s) along the peptide chain(s) at which these attachments occur. It is the aim of this article to indicate some of the methods with which these problems may be approached. It should be appreciated that glycoproteins are a large and varied group of compounds and that only relatively, few of these have so far been the subject of detailed structural investigation.1 Sufficient variation in structural plan has been found from these studies to make it evident that although a general approach to the study of the carbohydrate units of glycoproteins may be given, the details of the experiments will have to be modified from protein to protein to meet the specific problems at hand. Nature of the Carbohydrate Units: Number, Size, and Composition Prior to undertaking an investigation of the sequence of the sugar residues in a glycoprotein, it is essential to determine among what type of carbohydrate units the various monosaccharides are distributed. Considerable variation has been observed in the type of carbohydrate units which may occur in glycoproteins. The carbohydrate may be present in the form of disaccharide units ~ or in fairly large heteropolysaccharides which may contain as many as 17 monosaccharide residues2 The number of units present may vary from several hundred, as in the case of the disaccharides, 2 to only a few3 or even a single one 4 of the larger heteropolysaccharides. A glycoprotein usually has carbohydrate units of only one type, although units with quite different structural features have been found in the same protein2 Digestions with Proteases In order to obtain the carbohydrate portion of a glycoprotein with a minimum number of amino acids attached, extensive digestion with a protease of low specificity can be performed2,5 The most suitable enzyme for achieving such extensive proteolysis appears to be Pronase, a protease from S t r e p t o m y c e s griseus. However, papain and the Bacillus subtilis protease known as Nagarse have also been used effectively for the extensive digestion of the peptide portion of glycoproteins. 1R. G. Spiro, New Engl. J. Med. 269, 566 (1963).

E. R. B. Graham and A. Gottschalk, Biochim. Biophys. Acla 38, 513 (1960). 3R. G. Spiro, Y. Biol. Chem. 237, 382 (1962). 4p. G. Johansen, R. D. Marshall, and A. Neuberger, Biochem. J. 78, 518 (1961). ~R. G. Spiro, J. Biol. Chem. 240, 1603 (1965).

28

ANALYTICAL

METHODS

[2]

Digestions with these nonspecific proteases can usually be carried out on the native molecule. The protein may be present at a concentration of approximately 25 mg/ml dissolved in the appropriate buffer or titrated to the desired pH. In the latter case, it is necessary to maintain the pH of the incubation by the addition of base. The enzyme should be added initially in an amount equal to 0.5-1.0% of the weight of the protein to be digested, and subsequently during the course of the digestion, one or two other additions of approximately 0.5% of the weight of the substrate may be made. The incubation is conducted at 37 ° and small amounts of toluene should be added to prevent bacterial growth. The course of the digestion is followed by analyzing small aliquots with the ninhydrin reagent 8 and leucine as a standard. The incubation is carried out until no further increase in the ninhydrin reaction is observed, and 96-120 hours may be required to achieve maximal digestion. Of the total peptide bonds of a glycoprotein, 40-50% will usually be split during the course of such digestions. The conditions which may be employed for digestions with the various enzymes are as follows3,S: Pronase digestion should be carried out at pH 7.8 in the presence of 0.0015-0.015M CaC12; papain digestions are best conducted at pH 6.5 in the presence of 0.001 M disodium ethylenediamine tetraacetate and 0.005 M cysteine HC1 or 0.005M 2,3-dimercaptopropanol; Nagarse digestions should be carried out at pH 7.5. The use of more specific proteolytic enzymes such as trypsin, chymotrypsin, or pepsin is not advisable, since the peptide fragments obtained wilt be quite large and may contain more than one carbohydrate unit2 Moreover, in such large glycopeptides the identification of the amino acid involved in the glycopeptide bond is made more difficult because of the presence of a large number of amino acids. After digestion with these nonspecific proteases, it may occasionally be useful to further reduce the length of the peptide chain in the vicinity of the carbohydrate-peptide linkage by digesting with leucine aminopeptidase and/or carboxypeptidase. 3,5 The use of two or more proteolytic enzymes in succession may also prove useful to achieve maximal shortening of the peptide chain in the vicinity of the carbohydratepeptide linkage. Separation of Glycopeptides from Peptides and Amino Acids After extensive proteolytic digestion, the carbohydrate units of a glycoprotein are present in the form of glycopeptides, together with other peptides and amino acids. As a rule, these glycopeptides are of subeS. Moore and W. H. Stein, J. Biol. Chem. 211, 907 (1954).

[2]

CARBOHYDRATE UNITS OF GLYCOPROTEINS

29

stantially higher molecular weight than the noncarbohydrate-containing peptides in the digest. This is due to the additional bulk of the carbohydrate unit of these glycopeptides and to the fact that their peptide portion often has not been degraded as extensively by the proteolytic enzyme as peptides from other segments of the chain, owing to steric hindrance from the carbohydrate. The noncarbohydrate-containing material, on the other hand, is present primarily in the form of amino acids and di- or tripeptides. Several techniques have proved useful in separating the glycopeptides from other peptides as well as amino acids. The use of two or more of the methods to be described will usually result in a high degree of purification. The resulting glycopeptide fraction may then be used for the resolution of the individual glycopeptides present. In all the techniques, high recovery of the original carbohydrate material should be sought, as any unexplained loss may indicate the existence of carbohydrate in more than one type of unit. Moreover, a change in the individual sugar ratios of the recovered material compared to those in the original glycoprotein would strongly suggest the presence of more than one type of carbohydrate unit. 1. Dialysis. This technique has proved of great value in separating peptides and amino acids from glycopeptides. The pore size of dialysis tubing varies widely. It has been found that 18/32 Visking cellophane tubing has a pore size which will retain glycopeptides with an average molecular weight of 4400 during many days of dialysis and yet permit the passage of essentially all the noncarbohydrate-containing material obtained from a papain or Pronase digestion, s, ~ After dialysis of a papain digest of fetuin for 4 days in such a tubing, 94% of the carbohydrate material stayed nondialyzable, while only 8% of the peptide material was retained2 Dialysis should be carried out at 4 °, first against 0.1 M NaC1 for 6-8 hours to remove adsorbed charged molecules, and then against distilled water with frequent changes for as long as is necessary to remove noncarbohydrate-containing material. If carbohydrate units of differing size are present in the digest, this technique may result in the separation not only of glycopeptides from peptide material, but also of dialyzable from nondialyzable glycopeptides. An indication of the presence of such dialyzable glycopeptides may be obtained by performing dialysis on aliquots of the digest for varying periods of time up to 8-9 days and analyzing the sugar and peptide content of the nondialyzable material at various times. For example, when a Pronase digest of calf thyroglobulin was treated in this manner, it was possible to obtain after only 6.5 hours of dialysis

30

ANALYTICAL METHODS

[2]

a nondialyzable fraction containing less than 4% of the peptide portion but more than 90% of the carbohydrate. With more prolonged dialysis, there was a preferential loss of the mannose and glucosamine, which was particularly striking in the case of the mannose component, which dialyzed out to the extent of more than 50% in a period of 8-9 days. In contrast, even after such prolonged dialysis, there were only small decreases in the amounts of sialic acid, galactose, and fucose present. These results were a function of the presence of two types of carbohydrate units in thyroglobulin, the smaller, dialyzable one consisting only of mannose and glucosamine, and the larger having as its components sialic acid, galactose, and fucose, in addition to mannose and glucosamine2 2. Gel Filtration. Because of the differences in molecular weight, separation of glycopeptides and from other peptide material can also be achieved by gel filtration of the proteolytic digest on Sephadex G-25 or G-50. 7 In this type of fractionation, the glycopeptides emerge from the column prior to the peptides and amino acids because of their larger size. In such separations, columns varying from 80 to 130 cm in height have been used effectively, and the recovery of the carbohydrate material is more than 90%. The sample can be dissolved in buffer~ or, if no sialic acid is present, in 0.1 N acetic acid. 7 Elution of the column is then performed with the same solvent, and the eluate is collected in small fractions, which can be analyzed for carbohydrate by the anthrone reaction (see this volume [1]) and for peptide or amino acids by the ninhydrin reaction. 6 Fractions containing the carbohydrate material can be pooled and the buffer removed by brief dialysis. When acetic acid is used, this can be volatilized. If carbohydrate units differing in size are present, the carbohydrate-containing material may emerge from the column as more than one component ~ (see below). 3. Paper and Cellulose Column Chromatography. Most glycopeptides will not move from the origin when chromatographed in many of the systems used for amino acids or sugarsY ~ On the other hand, most of the non-carbohydrate-containing peptides and amino acids will migrate in such systems. Using this property of glycopeptides, it is possible to obtain further purification after dialysis or gel filtration by applying the glycopeptide fraction as a streak on Whatman No. 1 or 3 M M paper2 The paper should be previously washed with the solvent mixture to be used, followed by distilled water, and then dried. Chromatography may be carried out in butanol-acetic acid-water (4:1:5) for 24-30 hours. G. S. Marks, R. D. Marshall, A. Neuberger, and H. Papkoff, Biochim. Biophys. Acta 63, 340 (1962). s R. G. Spiro, unpublished observations (1963).

[2]

CARBOHYDRATE UNITS OF GLYCOPROTEINS

31

From 15 to 20 mg of glycopeptide material can easily be handled on a single sheet of Whatman No. 1 paper. After chromatography, the paper is dried and the origin area is cut out and eluted with water; a substantial purification of the carbohydrate-containing material results. Glycopeptides can also be separated from amino acids and other peptides by chromatography on cellulose powder columns. This procedure has been used for purification of the glyeopeptides of the al-acid glycoprotein2 Whatman cellulose powder (ashless, standard grade) is packed into an appropriate column. For the fractionation of 10.5 g of digest, a column 5 cm in height and 6 cm in diameter was used. The sample was mixed with 15 g of the cellulose powder and placed on top of a previously prepared column. This was followed by a 1.5 cm layer of additional cellulose powder. The amino acids and carbohydrate-free peptides were eluted with 2 liters of n-propanol-ethanol-water-acetie acid (5: 15:5:0.75). The glycopeptides were then eluted with 500 ml of water at a rate of 2-3 ml per minute and collected in fractions in order to obtain a sharp separation between any slow-moving peptides and the glycopeptides.

~. Passage of Glycopeptide

Fraction through Dowex 50-X16.

Further purification of glycopeptides from small peptides and amino acids may be achieved by passage of the mixture through a column of Dowex 50-X16, 20-50 mesh (H ÷ form)2 ,5 The glycopeptides, because of their large size are excluded from the beads of this resin, while the amino acids and small peptides are adsorbed. In this procedure, an aqueous solution of the salt-free glycopeptides is passed through a column of Dowex 50 which has been well washed with water before the sample is applied. An amount of resin containing several times the number of equivalents to be adsorbed should be used. After the sample has passed through the column, the resin is washed with several column volumes of distilled water and the effluent and wash are combined and titrated to pH 7 with dilute :NH4OH. Recovery of the carbohydrate is complete on such a column. Dowex 50-X8, 20-50 mesh has been used, but with this resin complete recovery of the carbohydrate material may not occur. 1° 5. Ethanol Precipitation. Purification of glycopeptides from non-carbohydrate-containing peptides has been achieved by ethanol precipitation. In this procedure, an aqueous solution of the sample is added to 9 volumes of absolute alcohol and the precipitate is recovered by centrifugation and redissolved in water; the procedure repeated 2 or 3 times. When applied to glycopeptides from ~,-globulin, 84% of the hexose 9 S. Kamiyama and K. Schmid, Biochim. Biophys. Acta 58, 80 (1962). ~oj. W. Rosevear and E. L. Smith, J. Biol. Chem. 236, 425 (1961).

32

material but recoveredY °

ANALYTICAL

only

18~

of the

METHODS

ninhydrin-reacting

[2]

material

was

Resolution of Glycopeptides The procedures outlined above yield preparations reasonably free of amino acid and peptide contaminants. Although ultracentrifugation studies may indicate that the components present in such a fraction are of similar molecular weights, electrophoresis will usually indicate the presence of many glycopeptides. Because of the low specificity of the proteolytic enzymes used, numerous glycopeptides, varying in the number of amino acids still attached to the carbohydrate units, are to be expected. Moreover, when several carbohydrate units are present in a glycoprotein, the number of these glycopeptides will increase proportionately, with a "family" of glycopeptides resulting from each carbohydrate-peptide attachment. For the purpose of characterizing the carbohydrate units, it is essential to resolve the glycopeptides into as many components as possible and to determine whether the differences between these glycopeptides are solely a function of the peptide portion or are due also to differences in the carbohydrate. The following methods may be used for such fractionation. 1. Ion Exchange Chromatography. Many glycopeptides are anionic because of their sialic acid content. This property makes possible their chromatography on diethylaminoethyl(DEAE)-cellulose columns. It is advisable to carry out this type of chromatography at a pit fairly close to neutrality in order to avoid removing or degrading the sialic acid. The sample should be placed on DEAE-cellulose columns at a very low ionic strength, as it has been found that otherwise complete adsorption of all of the negatively charged material is not obtained. 3,~ Elution can be achieved with a linear gradient. Glycopeptides obtained from proteolytic digests of fetuin or thyroglobulin have been chromatographed on DEAE-cellulose in the following manner2 ,5 The DEAE-cellulose is washed with 0.1 M buffer at the pH to be used, and then with the dilute buffer in which the sample is to be applied. The suspension is poured into a column approximately 2 cm in diameter, fitted with a sintered-glass plate, and is allowed to settle to a height of 65 cm. Such a column has been used to chromatograph the salt-free glycopeptide fraction from approximately 300 mg of fetuin 3 or 1500 mg of thyroglobulin2 The sample is titrated to pH 7.6 with dilute NaOH, placed on the column, and washed in with the starting buffer. For the fractionation of the fetuin glycopeptides, the initial con-

[2]

CARBOHYDRATE UNITS OF GLYCOPROTEINS

33

centration of the buffer was 0.0025M sodium phosphate, pH 7.6. The column was washed with an additional 180 ml of this buffer, after which elution was carried out with a linear gradient consisting of 1300 ml of 0.0025 M sodium phosphate, pH 7.6 in the mixing chamber and an equal volume of 0.05 M sodium phosphate buffer, pH 7.6, in the reservoir. All the glycopeptides were eluted when a concentration of 0.03M phosphate buffer had been reached. For chromatography of the thyroglobulin glycopeptides, 5 the concentration of the starting buffer had to be lowered to 0.0005M sodium phosphate pH 7.6. After the sample had passed through, 300 ml of this buffer was run through the column. Elution was in this case achieved by a gradient with 1400 ml of the 5 X 10-* M sodium phosphate buffer in the mixing chamber and an equal volume of 0.03M pH 7.6 sodium phosphate in the reservoir. All of the glycopeptides from thyroglobulin were eluted when a buffer concentration of 0.014M had been reached. The lower degree of adsorption of the thyroglobulin glycopeptides compared to those from fetuin is presumably a function of the smaller number of sialic acid residues in former compounds. The columns may be run at 20-25 ml per hour and collected in fractions of 10-15 ml. The total yield of the carbohydrate placed on such columns ranges from 90 to 100%. The fractions can be desalted by brief dialysis against distilled water in 18/32 Visking dialysis tubing. When glycopeptides not containing sialic acid are to be resolved, chromatography on carboxymethylcellulose may be a useful tool. 2. Charcoal-Celite Chromatography. Although charcoal-Celite columns do not appear to have the resolving power of ion exchangers for the separation of glycopeptides having similar carbohydrate portions and differing only in small variations of their peptide components, they have proved useful in separating glycopeptides whose carbohydrate portions differ substantially from each other. Elution is performed with increasing ethanol concentrations. Generally speaking, glycopeptides with smaller carbohydrate units are eluted prior to those containing a larger number of sugar residues. The sialic acid content of glycopeptides influences their elution from these columns by increasing their affinity for the adsorbent. For example, glycopeptides with large amounts of sialic acid, such as those in fetuin, cannot be eluted from charcoal-Celite columns, even with high concentrations of ethanol. However, after selective removal of the sialic acid, elution of these glycopeptides with ethanol may readily be achieved. 11 Charcoal-Celite chromatography has been employed to separate the ,1 R. G. Spiro, u n p u b l i s h e d o b s e r v a t i o n s (1960).

34

ANALYTICAL METHODS

[2]

glycopeptides containing the two types of carbohydrate units present in thyroglobulin? For such columns, activated carbon (Darco G-60) and Celite 535 (Johns-Mansville),12 previously washed with hot 70% ethanol and distilled water, are mixed in 1:1 proportions by weight and poured as a water slurry into a glass chromatographic column. The column volume should be 1 ml per milligram of hexose present in the glycopeptide sample. The sample is placed on the column in water, then a water wash of approximately 8 ml per milliliter of column volume is employed. Elution can be accomplished in a stepwise manner by increasing the concentration of ethanol 5% each time up to a final concentration of 50-70% and employing approximately 20 ml of each eluent per milliliter of column volume. Elution may also be achieved by a linear concentration gradient with water in the mixer and 50-70% ethanol in the reservoir. The eluate is collected in fractions which are taken to dryness in a vacuum rotator at 40 ° and are then analyzed. Upon fraetionation of the thyroglobulin glycopeptides in this manner, the glycopeptides containing the small mannose-glueosamine unit were eluted with 17% ethanol, while the larger glyeopeptides containing sialic acid, fucose, galactose, mannose, and glucosamine, emerged between ethanol concentrations of 25 and 50%. 5 3. Preparative Zone Electrophoresis. Preparative electrophoresis in various media can be employed for the resolution of glycopeptides. This approach is similar to that of ion-exchange chromatography, being based on differences in the charge of the glycopeptides due to variations in either the amino acid or sialic acid residues. Starch columns have been used to separate glycopeptides from ~,globulin digests, using 0.05 M Veronal buffer at pH 8.5 at 450 volts for 48 hours at 5% A column of 3 X 48 was used to fractionate 200 mg of material. 1° Columns of modified cellulose have been used to separate glycopeptides from ovalbumin in 1 M acetic acid at 15 ° and 20 volts/cm for 24 hours. Approximately 250 mg of material was placed on a column 3 X 38 cm. 18 Preparative paper electrophoresis has been employed for separating glycopeptides from the a~=glycoprotein in pyridine-acetate buffer, pH 6.4 at 300-400 volts for 3 hours on acid-washed paper. The components were located by staining guide strips with ninhydrin and eluted with 10% acetic acid. 14 1~R. L. Whistler and D. F. Durso, J. Am. Chem. Soc. 72, 677 (1950). la R. Montgomery and Y. C. Wu, J. Biol. Chem. 238, 3547 (1963). 14R. Bourrillon, R. Got, and D. Meyer, Biochim. Biophys. Acta 83, 178 (1964).

[2]

CARBOHYDRATE UNITS OF GLYCOPROTEINS

35

4. Gel Filtration. In addition to being useful in separating glycopeptides from non-carbohydrate-containing peptides and amino acids present in proteolytic digests of glycoproteins, gel filtration can be used to separate glycopeptides differing substantially in molecular weight. Since the glyeopeptides originating from one type of carbohydrate unit are likely to be of similar molecular weight, they cannot be resolved by this procedure. However, when carbohydrate units of different size are present in the same protein, the glycopeptides from these different units will appear in different portions of the eluate, depending on their size. The presence of two types of carbohydrate units in thyroglobulin could be demonstrated by gel filtration on Sephadex G-50 or G-25 with the glycopeptides originating from the larger unit (average molecular weight, 4100) emerging prior to those from the smaller unit (average molecular weight 1250).5 For this separation, the glycopeptides originating from 270 mg of thyroglobulin were placed on a Sephadex G-50 column, 2.2 X 80 cm in 0.1 M sodium phosphate buffer, pH 7.0, and eluted with the same buffer. Fractions of 5.2 ml were collected at a flow rate of 15 ml per hour. The void volume of the column was determined to be 88 ml and the glycopeptides containing the larger carbohydrate unit had their peak at 161 ml and were partly resolved from those of the smaller unit which had their peak at 192 ml. 5. Timed Dialysis. As described in a previous section of this chapter, if glycopeptides differing substantially in molecular weight because of differences in their carbohydrate units are present in a digest, conditions of dialysis may be chosen that will permit passage of the small glyeopeptides but retain the larger ones. Composition of Carbohydrate Units Analyses of the monosaecharide (sc. this volume [1]) and amino acid components in the glycopeptides resolved by the above methods should be performed. If significant differences are observed in the sugar composition of the glycopeptides, this would indicate the occurrence of more than one type of carbohydrate unit. If all the glycopeptides have similar carbohydrate compositions, and particularly, if the sugar ratios are the same as in the native glycoprotein, it is likely that only one type of carbohydrate unit is present. Small differences in the carbohydrate composition of the various glycopeptides could be a function of the errors inherent in the analytical methods employed or could represent a microheterogeneity resulting from variation in a basic structural pattern of the carbohydrate units. Such variations, particularly in the sialic acid content, have been observed29"13 It is necessary to perform

36

ANALYTICAL METHODS

[2]

analyses on as many of the resolved glycopeptides as possible in order to account for as close to the total carbohydrate of the original molecule as possible. Determination of Molecular Weights of Carbohydrate Units

Different glycoproteins have been shown to have carbohydrate units with substantially different molecular weights, ranging from 5122 to 3400. 3 If the purified glycopeptide fraction from a particular protein appears homogeneous in the ultracentrifuge, it is likely that similar sized molecules are present. Since there are usually only a few amino acid residues left on the glycopeptides obtained after extensive proteolytic digestion, the carbohydrate portion of these glycopeptides accounts for most of their weight. Therefore, ultracentrifugal homogeneity would also indicate that the carbohydrate units are of similar size. If analyses of the various glycopeptides resolved from this fraction have indicated a similarity in their carbohydrate composition, it is likely that the carbohydrate units of the glycoprotein are all of the same type. Average molecular weights of the glycopeptide in such fractions can be performed by the short column sedimentation equilibrium method ~'~ or can be calculated from sedimentation-viscosity or sedimentationdiffusion data. The partial specific volume should be experimentally determined and is usually less than that of proteins because of the large carbohydrate component of the glycopeptides2 From the average molecular weight of the glycopeptides and the percentage of their weight which the carbohydrate forms, the average molecular weight of the carbohydrate units may be calculated2 (The total carbohydrate present may be calculated from the sum of the sugar residues minus the water of glycosidic bond formation.) The number of carbohydrate units present in a particular glycoprotein can be calculated as follows: number of carbohydrate units =

mol. wt. of glycoprotein X % carbohydrate in protein average molecular weight of carbohydrate units

If more than one type of carbohydrate unit is present in a glycoprotein, the glycopeptides containing these units must first be separated from each other prior to performing such molecular weight determination. ~ Molecular weight determinations can also be performed by amino terminal analyses of glycopeptides by the D N P method although these tend to be less accurate than the physical measurements. Most glycopeptides fractions obtained after digestion with nonspecific proteases ~SD. A. Yphantis, Ann. N . Y . Acad. Sci. 88, 586 (1960).

[2]

CARBOHYDRATE UNITS OF GLYCOPROTEINS

37

will contain more than one NH~-terminal amino acid. However, in the absence of contaminating non-carbohydrate-containing peptides, the toolecular weight can be calculated from the sum of the NH2-terminal residues and the carbohydrate composition of the glycopeptides. 3 A rough approximation of the molecular weight of the carbohydrate units can be obtained by gel filtration of the glycopeptides on Sephadex G-25 or G-50 by comparison of their penetration into these gels to that of polymers of known molecular weight. Structure of the Carbohydrate Units

Studies to determine the sequence and linkages of the monsaccharide residues of a glycoprotein can be performed on the entire glycopeptide fraction or even on the native protein if it has been demonstrated by the methods outlined above that only a single or a single type of carbohydrate unit is present in the molecule under investigation. If, however, these studies indicate that there is more than one type of carbohydrate unit in the glycoprotein, the glycopeptides containing each type of unit must be separated from each other before undertaking any studies on the monosaccharide sequence. Several approaches to the study of the sugar sequences and linkages of the monosaccharide residues present in the carbohydrate units of glycoproteins have been employed, and among these the most useful include: (1) graded acid hydrolysis; (2) enzymatic degradation with glycosidases; (3) isolation of oligosaccharides from partial acid hydrolyzates; and (4) periodate oxidation, including (5) the serial periodate oxidation technique. It is important to realize that the information obtained from any one of these approaches used singly is not adequate to establish the structure of the carbohydrate units. It is necessary to obtain concurring data from several of these lines of investigation in order to obtain convincing information on which to base a structural formulation. The methods to be described will have to be adapted to fit the requirements of the individual glycoprotein to be studied, since a large number of variations in structure have been observed so far and more are to be expected. Graded Acid Hydrolysis

Because of the differences in the stability of the glycosidic bonds of the various sugars found in glycoproteins, it is often feasible to obtain information in regard to the sugar sequence by measuring the rate of release of the monosaccharide residues under conditions of mild acid hydrolysis. This approach entails hydrolysis with dilute acid for varying

38

ANALYTICAL METHODS

[2]

periods of time and analysis of the released monosaccharides by appropriate techniques after their separation from the undegraded molecule, as well as from any oligosaecharides. Since acid labile bonds may be located in internal positions of the carbohydrate units, oligosaceharides rather than monosaeeharides may be initially released. These must be separated from the monosaccharides by paper or column chromatography and analyzed to determine the quantity of monosaccharides present in this form. This information is important in interpreting the monosaccharide release data, since if a relatively stable bond is located in a peripheral position, the externally located sugars may be initially released in large amounts in the form of oligosaccharides. Generally speaking, hydrolysis should be performed at a low protein or glycopeptide concentration (2-3 mg/ml)to minimize the possibility of artifactual oligosaccharide formation by acid reversion, as well as the possibility of amino acid-sugar interaction. When sialic acid is present in the glycoprotein, mild conditions usually suffice for its complete release. Hydrolysis with 0.025N to 0.1 N H2SO~ at 80 ° for 1 hour will usually release all this sugar without removing any other monosaccharide, except for a small percentage of the fucose, if present. 16 The released sialic acid may be measured directly by the thiobarbiturie reaction 17 or by the resorcinol-HC1 method TM after separation on Dowex 1 formate columns (see this volume [1] ). It is possible to recover the sialic acid-free protein, for use as starting material for further degradation studies, by dialysis of the hydrolyzate or passage through Dowex l-X8, 50-100 mesh (formate form). In order to release the other sugars present in glycoproteins, 0.05 or 0.1N H~S04 at 100 ° in sealed tubes may conveniently be used. TM Measurements should be made at frequent intervals up to 24 hours and thereafter at longer intervals up to as long as 72-96 hours. For measurement of the monosaccharides released at each time, the hydrolyzate is passed through columns of Dowex 50-X4, 200-400 mesh (H ÷ form) coupled to Dowex 1-XS, 200-400 mesh (formate form), as described in this volume [1]. The Dowex 50 will adsorb glucosamine, in addition to peptides and amino acids, and the Dowex 1 will exchange sulfate for formate ions. The effluent and water wash from these columns contain the neutral sugars. The hexosamines are eluted from the Dowex 50 resin with 2 N HC1 (see this volume [1]). Analysis for free galactose, mannose, glucose, or fucose can be performed on the neutral sugar fraction ~eR. 17L. is L. ~9R.

G. Spiro and M. J. Spiro, J. Biol. Chem. 240, 997 (1965). Warren, J. Biol. Chem. 234, 1971 (1959). Svennerholm, Biochim. Biophys. Acla 24, 604 (1957). G. Spiro, J. Biol. Chem. 237, 646 (1962).

[2]

CARBOHYDRATE

UNITS

OF GLYCOPROTEINS

39

by quantitative paper chromatography 2° (see this volume [1]). Direct analyses of this fraction by reactions such as the anthrone for hexose or the cysteine-sulfuric acid reaction for fucose are misleading as they would determine not only the monosaccharides present, but also the oligosaccharides. During the early stages of graded acid hydrolysis, large peptides may be present in the effluent and wash from the Dowex columns; their presence may make quantitative paper chromatography difficult. These can easily be removed by passage of this fraction, dissolved in water, through small columns of charcoal-Celite, l : l . The effluent and 30% ethanol eluate from these columns are collected and contain the monosaccharides and oligosaccharides. 19 The release of hexosamines can be determined by performing the Elson-Morgan reaction ~1 on the Dowex 50 eluate to measure the unsubstituted hexosamines and the Morgan-Elson reaction 22 on an aliquot of the neutral sugar fraction to determine the N-acetylated hexosamines. The sum of these two determinations will represent the total hexosamines released at any one time. During the course of weak acid hydrolysis, a large portion of the hexosamines are initially released in the N-acetyl form, and only subsequently do they become deacetylated as hydrolysis proceeds further. • Galactose x Hexosamine o Monnose

0 0

o ~::k

12

6

8

16

24

Hours

FIG. 1. Release of monosaccharides from fetuin at various times during hydrolysis with 0.05 N sulfuric acid at 100% From R. G. Spiro, J. Biol. Chem. 237, 646 (1962). ~ R . G. Spiro, J. Biol. Chem. 235, 2860 (1960). =1N. F. Boas, J. Biol. Chem. 204, 553 (1953) ; see also this volume [3]. 2~j. L. Reissig, J. L. Strominger, and L. F. Leloir, J. Biol. Chem. 217, 959 (1955); see also this volume [3], [23].

40

ANALYTICAL METHODS

[2]

The results from such a graded acid hydrolysis of fetuin in 0.05 N H~S04 at 100 ° shown in Fig. 1.1~ Since sialic acid was completely released by even weaker conditions, these results suggested in this glycoprotein the sequence sialie acid--> galactose--~ hexosamine--> mannose. Not only was mannose released slowly as a monosaccharide, but moreover, mannose-containing oligosaccharides did not appear until essentially all the galactose and an almost equimolar amount of the hexosamine had been released. It is desired to study the sugars remaining attached to large peptides at various times during the hydrolysis, the hydrolyzate can be passed through Dowex 50-X12, 50-100 mesh (H ÷ form) to adsorb the glucosamine, amino acids, and small peptides, and through Dowex 50-X2, 200-400 mesh (H +) to adsorb the large peptidcs. 19 These can subsequently be eluted with 5 N N H 4 0 H at 4 ° and then taken to dryness at 25-30 ° in a vacuum rotator. Enzymatic Release of Monosaccharides by Glycosidases Because of the high specificity of many glycosidases and because they generally release sugars only from the terminal, nonreducing positions, their use may give less ambiguous data than those obtained from studies with graded acid hydrolysis, where random cleavage of glycosidic bonds tends to take place. Moreover, glycosidases may give information as to the anomeric configuration of the bond which is split. Although many glycosidases act directly on glycoprotein substrates, they may work more rapidly and completely on glycopeptides, 19 presumably as a result of some interference by the peptide chain of the native glycoprotein molecule with their approach to the carbohydrate units. In order to achieve sequential release of sugars from a glycoprotein, it is usually necessary to recover the substrate after removal of the terminal sugar by the action of one glycosidase in order to submit it to digestion with another glycosidase. If instead of using a highly purified glycosidase, a less pure preparation is employed which contains several glycosidases, a sequential release of sugars may be obtained. This can give valuable information if the order of the monosaccharide release is followed in the same manner as outlined for the graded acid hydrolysis. A glycoprotein may be recovered after incubation with a glycosidase, and separated from the released monosaccharides by dialysis. If glycopeptides are used as substrate, these can readily be recovered by passage of the digest through a small charcoal-Celite (Darco G-60, Celite 535, 1:1) column. In this procedure, 19 1 ml of such a column may be used for 4 mg of substrate and the released monosaccharides obtained in the

[2]

CARBOHYDRATE UNITS OF GLYCOPROTEINS

41

effluent and 4% ethanol eluate (30-40 ml per milliliter of column volume). The glycopeptide can usually be eluted with 50% ethanol (25 ml per milliliter of column volume). This procedure removes the enzyme which remains adsorbed to the charcoal column. Incubations with glycosidases are usually performed in buffer at 37 ° at the appropriate pH for varying periods of time until maximal release of the sugar under study has been achieved. This may require as long as 6-7 days, and small amounts of toluene are added to prevent bacterial growth. Concentrations of substrate of 10-20 mg per milliliter may conveniently be employed. A brief description of some of the glycosidases which have proved to be of use in the study of the sugar sequence of glycoproteins follows. Neuraminidase (Sialidase). This enzyme, which is a ketosidase, has been of great value in demonstrating that sialic acid residues of glycoproteins are usually located in terminal positions. The complete release of the neuraminic acid residues from glycoproteins is often achieved. The enzyme obtained from Vibrio cholerae has been highly purified 2~ and may be incubated with glycoproteins in 0.1 M sodium acetate buffer at pH 5.6 in the presence of 0.001 M CaC12.1~,19 Purified preparations of this enzyme have to be added only as a small fraction of 1% of the weight of the substrate. The sialic acid released may be measured directly by the thiobarbituric acid reaction 1~ (see this volume [1]). The sialic acid-free protein may then be isolated by dialysis or by passage through a Dowex 1 column. fl-Galactosidase. This enzyme has been highly purified from Escherichia coli 2~ and can be employed to release galactose from glycoproteins and glycopeptides. Incubation of this enzyme in 0.05M potassium phosphate pH 7.0, containing 0.01M MgS04 at a concentration of 4% of the weight of the sialic acid-free fetuin or sialic acid-free glycopeptides from fetuin caused a selective release of the galactose 19 (see the table). After incubation the released monosaccharides can be obtained either from the dialyzate or from the effluent and 4% ethanol eluate from the charcoal-Celite column described above. The sugars released may be determined by quantitative paper chromatography or by colorimetric analysis after being desalted by passage through columns of Dowex 50 (H ÷) coupled to Dowex 1 (formate). Controls containing enzyme alone should be run to correct for any releasable carbohydrate present in the enzyme itself. ~3G. L. Ada and E. L. French, Nature 183, 1740 (1959) ; see also this volume [115] and [116]. =4A. S. L. Hu, R. G. Wolfe, and F. J. Reithel, Arch. Biochem. Biophys. 81, 500 (1959).

42

ANALYTICAL METHODS

[2]

RELEASE OF MONOSACCHARIDES FROM FETUIN GLYCOPEPTIDES BY GLYCOSIDASESa'b

Incubation time (hours)

Sialic acid

Galactose

Neuraminidase

24

29.7

0

0

0

fl-Galactosidase

30 43 89

--~ ---

24.6 20.3 21.3

0 0 0

0 0 0

/~-N-Acetylglucosaminidase

80

__d

--~

20.6

0

30 43 89 148 165 190

~ ------

26.4 23.7 22.1 27.9 24.7 24.0

-2.6 8.9 18.1 20.7 23.5

-0 0 0 3.0 6.7

Enzyme

Almond emulsin

N-Acetylhexosamines Mannose

Data partly from R. G. Spiro, J. Biol. Chem. 237, 646 (1962) and partly from R. G. Spiro, unpublished observations (1965). b All values are expressed as micromoles of monosaccharides released from glycopeptides originating from 100 mg of native fetuin. c Sialic acid-free glycopeptides were used in these digestions. a Sialic acid and galactose were removed from these glycopeptides prior to incubation.

fl-N-Acetylglucosaminidase. This e n z y m e has been purified from pig epididymisY 5 W h e n i n c u b a t e d in 0.1 M sodium acetate buffer, p H 4.2, c o n t a i n i n g 0 . 1 M NaC1 at a c o n c e n t r a t i o n of 5 - 1 0 % of the weight of the substrate, this e n z y m e released N - a c e t y l g l u c o s a m i n e f r o m fetuin glycopeptides which h a d been previously t r e a t e d to r e m o v e the sialie acid and galactose 26 (see the table). A l m o n d Emulsin. This is a crude e n z y m e p r e p a r a t i o n which contains galactosidase, N - a c e t y l g l u c o s a m i n i d a s e , and m a n n o s i d a s e activity. W h e n a l m o n d emulsin (fl-glucosidase, M a n n R e s e a r c h Labs.) was i n c u b a t e d with sialic acid-free fetuin a n d fetuin glycopeptides in 0.1 M sodium acetate buffer, p H 5.0, at an e n z y m e c o n c e n t r a t i o n of 2 0 - 3 0 % of the weight of the substrate, there was a sequential release of m o n o s a c charide residues similar to t h a t observed in g r a d e d acid h y d r o l y s i s 19 (see the table). This is p r e s u m a b l y the result of the sequential action b y the several e n z y m e s present on the terminal, n o n r e d u c i n g sugar. Sequential release of sugars f r o m glycoproteins b y induced e n z y m e s ~J. Findlay and G. A. Levvy, Biochem. J. 77, 170 (1960) ; see also this volume [98]. ~*R. G. Spiro, unpublished observations (1965).

[2]

CARBOHYDRATE UNITS OF GLYCOPROTEINS

43

has been described as another tool for studying the sugar sequence of glycoproteins. ~7 Isolation of Oligosaccharides The isolation and characterization of oligosaccharides obtained from partial acid hydrolyzates of glycoproteins or glycopeptides is an important tool in establishing both sequence and linkages of sugar residues. In order to find optimal hydrolytic conditions for the isolation of oligosaccharides in as high a yield as possible, the neutral sugar fraction obtained at various times during graded acid hydrolysis should be examined by paper chromatography. A preliminary estimation of the number and amounts of oligosaecharides present at each time can be obtained by staining such chromatograms with the silver reagent and examining the region between the origin and the galactose spot; this is the area to which most oligosaceharides will move. Large-scale hydrolyses should then be performed at the times when maximal amounts of various oligosaccharides are observed. These can be isolated from the neutral sugar fraction of such hydrolyses by various methods, including quantitative paper chromatography, charcoal-Celite chromatography, or cellulose column chromatography. If hexosamines are present in the glycoprotein, basic oligosaccharides may be present in the hydrolyzate because of deacetylation of the hexosamines prior to their complete release as monosaccharides. In order to obtain such oligosaccharides, the basic fraction should be studied after adsorption and elution from Dowex 50. Isolated oligosaccharides should be examined in regard to their composition. Reducing end groups can be detected by sodium borohydride reduction, followed by hydrolysis and identification of the sugar alcohols. Information in regard to linkage can be obtained by various techniques, including periodate oxidation and methylation. In addition, color reactions can give valuable information in regard to linkage; for example, the Morgan-Elson reaction can be used to study oligosaccharides containing N-acetylhexosamines on the reducing end. 2s The yield of oligosaccharides obtained from glycoproteins is usually quite low. For this reason it is difficult to generalize from information obtained about their structure alone to that of the entire carbohydrate unit. However, data obtained from their study are valuable in confirming information derived from other approaches. If, for example, isomers of a given oligosaccharide are obtained which vary in linkage, it is likely that more than one type of linkage occurs in the glycoprotein for that particular sequence. ~7S. A. Barker, G. I. Pardoe, M. Stacey, and J. W. Hopton, Nature 197, 231 (1963). 28R. W. ffeanloz and M. Tr6m~ge, Federation Proc. 15, 282 (1956).

44

ANALYTICAL

METHODS

[2]

Periodate Oxidation

Treatment of glycoproteins or glycopeptides with sodium metaperiodate can give valuable information in regard to monosaccharide sequence and linkages if an analysis of the destruction of the different monosaeeharides is made and if the products of the oxidation are identified. If periodate oxidation is performed both on the native carbohydrate units and on those that have been degraded by the action of glycosidases or by partial acid hydrolysis, it is possible to obtain confirmatory information in regard to the sequence of the monosaccharide units. Measurement of periodate consumption itself, although of value in following the course of the reaction, is difficult to correlate quantitatively with monosaccharide destruction, particularly when significant amounts of peptide material are present in the sample. Oxidation of several of the amino acids may take place, 29,~° and this type of oxidation probably accounts for part of the slow phase of periodate oxidation often observed with glyeoproteins. Periodate oxidation should be performed with an excess of periodate over the amount theoretically expected to be consumed by the monosaccharide components so that the periodate consumption by the peptide portion will not limit the periodate present for sugar oxidation. The following procedure may be used for the periodate oxidation of glycopeptides or glycoproteins21 Oxidation may be carried out in 0.05 M sodium acetate buffer at pH 4.5 and 4 ° in the dark. The concentration of the material to be oxidized should be approximately 5 mg/ml and a concentration of sodium metaperiodate varying from approximately 0.01 M to 0.08M may be used. Experiments at two separate concentrations of periodate should be performed to ensure that no significant differences due to the limitation of periodate are encountered. Aliquots may be taken at several times up to 48 or 72 hours for various analyses. These should be taken more frequently during the early period of the oxidation procedure. In order to analyze for the monosaccharides unaffected by the periodate, the oxidation is terminated by adding an excess of ethylene glycol to aliquots taken at several times. The sample may then be extensively dialyzed, first against 0.1 M NaC1 and then against distilled water in the cold. When dealing with dialyzable glycopeptides, the solution may be desalted by passage through Sephadex G-25. Monosaccharide analyses are performed, after suitable hydrolysis, by the P. Desnuelle, S. Antonin, and A. Casal, Bull. Soc. Chim. Biol. 29, 694 (1947). my. C. Lee and R. Montgomery, Arch. Biochem. Biophys. 95, 263 (1961). sl R. G. Spiro, J. Biol. Chem. 239, 56T (1964).

[2]

C A R B O H Y D R A TUNITS E OF GLYCOPROTEINS

45

methods outlined in this volume [1]. It is advisable to measure the neutral hexoses after separation by quantitative paper chromatography in order to avoid the possibility of a contribution by aldehydic oxidation products to the color obtained in the nonspecifie hexose colorimetric reactions. Destruction of the sialic acid must also be followed by paper chromatographic analysis after mild acid hydrolysis and separation on Dowex 1 columns (see this volume [1]). Use of the resorcinol-HC1, thiobarbituric acid assay, or the direct Ehrlich reaction in following the destruction of this sugar will lead to serious errors, as the acidic products of the oxidized sialic acid continue to produce color in these reactionsY It is advisable to analyze controls of the glycoprotein or glycopeptides which have been treated in the same manner except that the sample has been added after reduction of the periodate by ethylene glycol. Aliquots of the oxidized mixture can be analyzed for periodate consumption by the arsenite method of Fleury and Lange2 2 Formaldehyde may be determined by the chromotropic acid reaction after reduction of the periodate by sodium arsenite2 8 This compound is of interest as it will be formed from any sialie acid residues unsubstituted on carbons 8 and 9. From the information about the destruction of the monosaccharides during periodate oxidation, inferences may be made about the structure of the carbohydrate units. These may be based on the following consideration. Sugar residues in terminal nonreducing positions should be completely and fairly rapidly destroyed. If hexoses are present in the pyranose form and in glycosidic linkage, they should be destroyed unless substituted singly on position 3, or if doubly substituted (branched), on positions 2 and 4 or position 3 plus any other. N-Acetylated hexosamines, on the other hand, will be destroyed only if substituted singly on C-6. An example of the application of this type of procedure is the periodate oxidation of the glycoprotein fetuin21 When the native protein was oxidized, there was a rapid destruction of all the sialic acid residues, with negligible destruction of the other sugars in the protein. However, after removal of the sialic acid, destruction of all the galactose residues took place, and the other sugars again remained essentially unaffected. This indicated that the sialic acid residues were terminal in location in the native protein and were linked to carbon 3 of the galactose. In addition it showed that the other sugar residues were linked in such a manner as to be spared oxidation. =J. Dyer, in "Methods of Biochemical Analysis" (D. Glick, ed.), Vol. 3, p. 111. Wiley (Interscience), New York, 1956. ~D. A. MacFayden, J. Biol. Chem. 158, 107 (1945).

46

ANALYTICAL

METHODS

[2]

In order to identify the products formed during periodate oxidation, sodium borohydride reduction of the oxidized material may be performed24 The resulting alcohols are more stable to subsequent acid hydrolysis than the initial aldehydic products. The following procedure may be used for the borohydride reduction of the oxidized glycopeptide or glycoprotein.31 The oxidized material is treated with ethylene glycol and dialyzed or desalted as described above. It is then treated with sodium borohydride added in equimolar sodium borate buffer at pH 8.0 with a final concentration of both borohydride and borate of 0.1M to 0.15M. The reaction is carried out at 0 ° for 12 hours and terminated by lowering the pH to 5 by the addition of acetic acid. Subsequently the reaction mixture is dialyzed in the cold against 0.1 M NaC1, followed by distilled water. When glycopeptides are being studied, the sample may be desalted by passage through Sephadex G-25. A sample not treated with periodate should also be treated with sodium borohydride in a similar manner. The salt-free sample may then be hydrolyzed and the neutral sugar fraction (see this volume [1]) examined by paper chromatography for the presence of various polyols which represent the reduced forms of the lower portions of those sugars cleaved by periodate oxidation. From the sugars present in glycoproteins, glycerol, erythritol, threitol, and 1,2-propylene glycol may result. The polyols can be determined by quantitative paper chromatography on Whatman No. 1 paper in the butanol-ethanol-water (10:1:2) system21 They may be located by staining adjacent guide strips on which standards have been spotted, and after elution with water may be estimated as formaldehyde by oxidation with periodate and use of the chromotropic acid reagent. 3~ In this system the Rgalactose of glycerol is 3.93 and the Rgalactose of erythritol is 2.27. Glycolaldehyde and glyceraldehyde, which represent the upper carbons of the oxidized and then reduced sugars, are not stable to acid hydrolysis and consequently can be identified and measured only indirectly26 Estimation of these components has been attempted by means of the diphenylamine method of Dische and Borenfreund. 37 From this examination of the products obtained after periodate oxidation and borohydride reduction, information in regard to the positions of linkages may be derived. Hexoses substituted only at carbon 4 3, M. Abdel-Akher, J. K. Hamilton, R. Montgomery, and F. Smith, J. A m . Chem Soc. 74, 4970 (1952). M. Lambert and A. C. Neish, Can. J. Res. B28, 83 (1950). ~J. A. Rothfus and E. L. Smith, J. Biol. Chem. 2~38, 1402 (1963). 37Z. Dische and E. Borenfreund, J. Biol. Chem. 18{}, 1297 (1949).

[2]

CARBOHYDRATE UNITS OF GLYCOPROTEINS

47

yield either erythritol (mannose and glucose) or threitol (galactose) as well as glycolaldehyde. Terminal hexoses and those substituted only at position 6 give rise to glycerol and glycolaldehyde, while hexoses linked only at carbon 2 give glycerol and glyceraldehyde. If hexoses are substituted at carbon 3, they will not be oxidized and will yield the undegraded hexose upon hydrolysis. Terminal fucose will yield 1,2-propylene glycol. N-Acetylhexosamines, either terminal or linked at carbon 6, give rise to N-acetylserinal (2-acetamido-2-deoxyglyceraldehyde). Serial Periodate Oxidation It has been shown by Smith that the technique of periodate oxidation followed by sodium borohydride reduction can be used in the degradation of polysaccharides, because of the increased acid lability of the aeetal bonds of the resulting polyalcohols. 38 These acetal bonds can be split by mild acid hydrolysis without breaking the glycosidic bonds of the unoxidized sugars. In polymers, sugars located in terminal nonreducing positions will be oxidized, and the reduced oxidation fragments will be released by this mild acid hydrolysis. If oxidizable sugars are located in positions internal to nonoxidizable sugar residues, the mild hydrolysis will then release the reduced oxidation products of this sugar with the more peripheral, nonoxidized portion of the sugar chain still attached to one of the released fragments of the oxidized sugar. The consecutive application of this technique of periodate oxidation, sodium borohydride reduction, and mild acid hydrolysis to a glycoprotein will bring about the degradation of the carbohydrate units21 The result of such a serial periodate oxidation technique will depend on the structure of the carbohydrate units under study. Under favorable structural circumstances, that is, when only the sugars in nonreducing terminal positions are destroyed by periodate, the sequential destruction and release of one monosaccharide residue after another may be achieved. In such cases, serial periodate oxidation can serve as an alternate tool to graded acid hydrolysis or release by glycosidases as a method for sugar sequential analysis in glycoproteins. When a glyeoprotein has in addition to its terminal residues internal sugars which are susceptible to periodate oxidation, the first step of this degradation technique will result in the release of not only the peripheral sugar but also the entire sugar chain external to this internally located oxidized sugar. Identification of the reduced oxidation products released at each step of the procedure will give further information in regard to linkages. ~*F. Smith and A. M. Unrau, Chem. Ind. (London), p. 881 (1959).

ANALYTICAL METHODS

48

CHzOH

[2]

CH2OH

HH~HH

H

H

OH

H

O~R HNAc

I NaIO (pH 4.5)

CI~OH

OCH +HCOOH

CH~OH H

0 OCH

0

H

H

~R

HNAc

l

(1) Dialysis

(2)NaBI~

CH~OH

CH~OH

.oH,d H H

HNAc

(1) Dialysis (2) 0.05 N H~SO4, 80 °, 1 hour

CH20H

c oH HOH

CH~OH

+

+ R

H

H

HNAc

FIa. 2. One step in the serial periodate oxidation technique. The reactions depicted represent the second step in the degradation of an oligosaccharide chain of fetuin by the application of this procedure; in the previous step the terminal sialic acid was oxidized, reduced, and released. R = inner portion of the heteropolysaccharide unit, which is unaffected by the oxidation in this step. From R. G. Spiro, J. Biol. Chem. 239, 567 (1964).

[2]

CARBOHYDRATE UNITS OF GLYCOPROTEINS

49

From an analysis of the sugars still attached to the peptide portion of the molecule after several applications of this technique, valuable information in regard to the monosaccharide involved in the glycopeptide bond may be obtained. The results obtained when this serial periodate technique was applied to fetuin illustrates its utility for the sequential removal of monosaccharides and identification of the sugar involved in the glycopeptide bond. 31 In this protein, only the terminal sugar residues are susceptible to periodate oxidation, and consequently, the application of 4 consecutive steps resulted in the sequential degradation and release of monosaccharides in the same order as they were released when this protein was treated with glycosidases or graded acid hydrolysis. The first step destroyed all the sialic acid; the second destroyed the galactose; the third, approximately half of the hexosamine, and the fourth, most of the mannose, leaving only hexosamines in significant amounts attached to the peptide moiety. The serial periodate oxidation procedure may be carried out in the following general manner21 The glycoprotein or glycopeptide at a concentration of approximately 5 mg/ml is oxidized with periodate under the conditions already detailed in this chapter, for the period of time necessary to result in complete destruction of susceptible sugars. The periodate is reduced by the addition of ethylene glycol, and removal of salt is accomplished by dialysis or gel filtration. The salt-free material is then reduced with a large excess of sodium borohydride in borate buffer at pH 8 under the conditions already described and then desalted again by dialysis or gel filtration. The acetal bonds are then cleaved by hydrolysis in 0.05 N H2S04 at 80 ° for 1 hour, after which the hydrolyzate is neutralized with NaOH; the process is repeated, beginning with another periodate oxidation. After each sodium borohydride step, the salt-free material may be analyzed for the sugars and polyols present, and after each mild acid hydrolysis step, the components released may be examined by passage of the neutralized mixture through a charcoal-Celite (1:1) column to remove protein or peptide material. The polyols will be recovered in the water wash and can be freed of salt by passage through a mixed-bed ion exchange resin. Any released oligosaccharides may be eluted from the charcoal with increasing concentrations of ethanol. One step of such a serial periodate procedure is shown in Fig. 2. Characterization of the Glycopeptide Bond

An inquiry into the carbohydrate-peptide linkage (s) of a glycoprotein should include the identification of the sugar and amino acid involved,

50

ANALYTICAL METHODS

[2]

as well as a characterization of the chemical nature of the bond. The most definitive manner of obtaining this information is through the isolation of a fragment containing solely the sugar and amino acid directly involved in this linkage. The isolation of such a fragment, particularly in a high enough yield to give conclusive information in regard to all the glycopeptide linkages which may be present in a glycoprotein, is often not feasible. However, it is possible to obtain a large amount of information in regard to the glycopeptide linkage by other less direct means. Identification of the amino acid involved in the glycopeptide linkage can in many cases be made from a study of the amino acid composition of glycopeptides after maximal digestion with proteases and exopeptidases. In the presence of free NH2- and COOH-terminal amino acids, only amino acids with additional functional groups can potentially be involved in the glycopeptide bond. Information in regard to the monosaccharide involved in the glycopeptide linkage can be obtained from a study of the sugars remaining attached to the peptide after removal of the more peripheral sugars by such techniques as serial periodate oxidation, treatment with glycosidases, or partial acid hydrolysis. If no free reducing groups are present in the glycoprotein, as indicated by the absence of polyols after sodium borohydride reduction of the native glycoprotein, it is likely that C-1 of the most internally located sugar is involved in the glycopeptide bond. 31 Up to the present time, evidence for three types of carbohydratepeptide linkages has been obtained: (1) acyl glycosylamines, involving C-1 of glucosamine and the amide nitrogen of asparagine39; (2) an O-glycosidic linkage in which C-1 of a sugar is linked to the hydroxyl group of serine or threonine4°; (3) a glycosidic ester bond between N-acetylhexosamine and the fl-carboxyl group of aspartic acid or the ~/-carboxyl of glutamic acid. 41 These bonds have some distinguishing characteristics which permits their tentative identification in glycoproteins or glycopeptides.

Glycosylamine Linkage In order for the glycosylamine type of bond to be present the glycopeptide must contain stoichiometric amounts of N-acetylhexosamine, aspartic acid, and ammonia. 39 In contrast to the O-glycoside bond and the glycosidic ester, the glycosylamine is fairly stable to alkali. MoreG. S. Marks, R. D. Marshall, and A. Neuberger, Biochem. ,I. 87, 274 (1963). 4oB. Anderson, N. Seno, P. Sampson, J. G. Riley, P. Hoffman, and K. Meyer, J. Biol. Chem. 239, PC 2716 (1964). ~1A. Gottschalk, W. H. Murphy, and E. R. B. Graham, Nalure 194, 1051 (1962).

[2]

CARBOHYDRATE

UNITS OF G L Y C O P R O T E I N S

51

over, it is not split by ester-cleaving reagents, such as hydroxylamine at room temperature. 4~ Amide Analysis. Glyeoproteins or glyeopeptides containing sialie acid and/or hexosamines require special precautions in the performance of amide analysis in order to avoid a substantial contribution to the ammonia from the degradation of these two sugars. Acid causes the decomposition of sialic acid with the conversion of approximately 25% of the amino group to ammonia. 43 To avoid this, the sialic acid should be selectively removed by mild acid hydrolysis prior to performing the stronger hydrolysis for the amide determination. Since volatilization of the released ammonia by strong alkali at room temperature causes deamination of hexosamines, this step should be performed at low temperatures. The following method may be employed. 43 The sialic acid-free material containing 0.25-2.5 micromoles of amide nitrogen is hydrolyzed for 3 hours in 0.4 ml of 2 N HC1 in sealed tubes at 100 °. The sample is then adjusted with NaOH to approximately pH 3.5 using methyl orange as an indicator. The ammonia released during hydrolysis is determined by the Conway microdiffusion technique. 4~ The sample is transferred with deionized water to the outer ring of a diffusion dish containing 1.5 ml of 0.01 N H2SO~ in the center well. The dish is placed in the cold room at 2-4 ° and after ½ hour 1 ml of 5 M K2C03, cooled to that temperature, is pipetted into the outer ring. Diffusion is allowed to proceed for 7-12 hours. The ammonia content of the central well is determined by Nesslerization, 45 ammonium sulfate being used as a standard. The rate of diffusion is temperature dependent, and the minimum time for complete recovery should be determined under the exact conditions to be used. A reagent blank should be carried through the entire procedure. O-Glycosidic Linkage This bond may be split by mild alkaline conditions by a fl-carbonyl elimination of the substituted serine or threonine which converts these amino acids into dehydroalanine and a-aminocrotonic acid, respectively. ~° This O-glycosidic linkage, however, is not split by hydrazine under conditions which split glycosidic esters. Treatment with 0.5N NaOH at 0-4 ° for 19-24 hours under nitrogen has been used to split this bond. In carbohydrate-protein complexes where the ratio of serine or threonine residues to carbohydrate units is small, an indication of the R. H. Nuenke and L. W. Cunningham, J. Biol. Chem. 236, 2452 (1961). M. J. Spiro and R. G. Spiro, J. Biol. Chem. 237, 1507 (1962). E. J. Conway and A. Byrne, Biochem. J. 27, 419 (1933). F. C. Koch and T. L. McKeekin, J. Am. Chem. Soc. 46, 2066 (1924).

52

ANALYTICAL METHODS

[3]

presence of this type of linkage can be obtained if a significant decrease in the serine or threonine is observed in the treated compared to the untreated protein. 4° However, when the ratio of serine or threonine residues to carbohydrate units is large, this comparative technique cannot be employed, since only a small percentage of the total residues of these amino acids could potentially be involved in the bond and the maximal decrease which could be expected would be within the error of the ami~o acid analyses. Many glycoproteins would fall into this category, although in glycopeptides derived from them a much lower ratio of serine or threonine to carbohydrate units would be likely to occur. When the ratio of serine or threonine to carbohydrate units is high, evidence for cleavage of this type of bond by alkali must depend on separation of the peptide and carbohydrate material by techniques such as gel filtration or dialysis. Glycosidic Ester The occurrence of glycosidic esters between the hydroxyl group of C-1 of N-acetylhexosamine and the carboxyl group of aspartic or glutamic acid has been proposed as a form of glycopeptide bond. 41 These linkages are very labile to alkali. In addition, they react like esters by being split by hydroxylamine under mild conditions. Moreover, lithium borohydride also splits this type of linkage with the release of the free carbohydrate and conversion of the dicarboxylic acid to its reduced form, that is, a-amino-v-hydroxy-n-butyric acid from aspartic acid or a-amino-8-hydroxy-n-valerie acid from glutamie acid. A final problem to be considered is the precise location of the carbohydrate-peptide linkage(s) in the glycoprotein molecule. Information in regard to this must be based on a determination of the number of peptides chains present in the molecule and on an analysis of the amino acid sequence around each one of these bonds, as well as of the remainder of the peptide portion.

[ 3 ] A n a l y s i s of S u g a r s F o u n d in M u c o p o l y s a c c h a r i d e s

By E. A. DAVIDSON The definition of mucopolysaccharides is itself a controversial problem, and the sugars found in such can therefore be extended to include almost every sugar found in natural sources. This article will restrict itself to discussion of those sugars found in connective tissue polysac-