Studies on the Chemical and Enzymatic Modification of Glycoproteins A GENERAL METHOD FOR THE TRITIATION OF SIALIC ACID-CONTAINING GLYCOPROTEINS

SUMMARY A general method is described for the radioactive labeling of proteins containing terminal sialic acid in the carbohydrate moieties. Quantitative conversion of these residues to a radioactive 7-carbon analogue of sialic acid was achieved by sequential periodate oxidation and tritiated borohydride reduction. Evidence is presented to establish the identity of the radioactive derivative as 5-acetamido-3,5-dideoxy-r,-arabino-2-heptulosonic acid and to indicate it as the sole site of tritium incorporation in the carbohydrate chain. Upon injection into rats, labeled preparations of ceruloplasmin and orosomucoid exhibited a normal half-life in the plasma and proved suitable for metabolic studies. The data indicate that the 7-carbon sialic acid analogue mimics sialic acid with respect to hydrolysis by neuraminidase and in regulating survival of glycoproteins in the circulation. In an earlier study (I), a method for the preparation of radioactive glycoproteins was described in which the penultimate galactose residues, exposed by the enzymatic removal of sialic acid, were sequentially oxidized with gala&se oxidase and reduced by tritiated borohydride.

Quantitative conversion of these residues to a radioactive 7-carbon analogue of sialic acid was achieved by sequential periodate oxidation and tritiated borohydride reduction.
Evidence is presented to establish the identity of the radioactive derivative as 5-acetamido-3,5-dideoxy-r,arabino-2-heptulosonic acid and to indicate it as the sole site of tritium incorporation in the carbohydrate chain. Upon injection into rats, labeled preparations of ceruloplasmin and orosomucoid exhibited a normal half-life in the plasma and proved suitable for metabolic studies.
The data indicate that the 7-carbon sialic acid analogue mimics sialic acid with respect to hydrolysis by neuraminidase and in regulating survival of glycoproteins in the circulation.
In an earlier study (I), a method for the preparation of radioactive glycoproteins was described in which the penultimate galactose residues, exposed by the enzymatic removal of sialic acid, were sequentially oxidized with gala&se oxidase and reduced by tritiated borohydride. Subsequent studies in viva revealed that the resulting tritium-labeled asialoproteins were rapidly cleared from the circulation and deposited within the parenchymal cells of the liver (2,3). As a consequence, these preparations were not suitable for metabolic studies requiring a normal serum survival time, and a procedure was sought in which the sialic acid derivative remained covalently bound to the intact glycoprotein.
With the realization that the two terminal exocyclic carbon atoms of sialic acid are selectively cleaved upon exposure to periodic acid without destruction of the ring structure or cleavage of the glycosidic bond (4, 5), it was deemed likely that borohydride reduction of the resulting aldehyde would yield a stable radioactive derivative of the original glycoprotein.
was prepared by the method of Whitehead and Sammons (7) modified for larger quantities.
To 13 liters of human plasma, obtained from the National Institutes of Health Blood Bank, were added 130,000 NIH units of topical thrombin (Parke-Davis) and the mixture incubated for 1 to 2 hours at 37" prior to storage overnight at 4". The resulting clot was removed by centrifugation and the serum dialyzed at 4" against 5 volumes of 0.05 M sodium acetate buffer, pH 4.3. After 1 day, the dialysate fluid was changed and the dialysis continued for an additional 24 hours. A flocculent precipitate was removed by centrifugation and the supernatant fluid was charged onto a column of diethylaminoethyl-cellulose, 10 X 27 cm (Whatman microgranular DE 52), that had been equilibrated with the above buffer. The column was then washed with the equilibrating buffer until the optical density at 280 nm fell to 0.075.
A linear gradient was then arranged with a mixing chamber containing 4 liters of the above buffer connected to a reservoir containing an equal volume of 0.10 M sodium acetate buffer, pH 4.0. The single large peak l'ritiation of Xialoproteins Vol. 246, No. 6 emerging toward the end of the gradient was pooled and 50 g of ammonium sulfate per 100 ml of eluate were added. The precipitate was removed by centrifugation and an additional 18 g of ammonium sulfate was added for each 100 ml of the supernatant fluid.
This precipitate was dissolved in a minimal amount of water and dialyzed exhaustively against water prior to recovery by lyophilization; 7.4 g of a white, soluble powder was obtained.
Descending paper chromatographic separations were carried out at room temperature for 15 to 17 hours. The solvent systems employed were : Solvent A, n-butylacetate-acetic acidwater (3:2:1) ; Solvent B, n-butyl alcohol-pyridine-1 N HCl (5:3:2); Solvent C, n-butyl alcohol-n-propyl alcohol-O.1 N HCl (1:2 : 1) ; Solvent D, n-butyl alcohol-acetic acid-water (4 : 1:5 as ceruloplasmin except that the amount of material used was adjusted to maintain a comparable ratio between periodate and sialic acid. Thus, to 18.6 ml of a Buffer A solution containing 41 mg of orosomucoid (15.8 pmoles of sialic acid), 12.3 ml of 0.012 M NaI04 were added and the oxidation carried out as described above.
After reduction and dialysis against Buffer A, and subsequent dialysis against water, the preparation was lyophilized to yield 38.5 mg of product.
These ratios were calculated on a dry weight basis and an assumed molecular weight of 40,000 (13). The specific activity of the final product was 11.5 X 1 O6 dpm per mg.
The specific activity of the control orosomucoid, prepared in the absence of periodate, was 0.55 x lo6 dpm per mg. Preparation of Partially Desialylated [AcNeu7]Ceruloplasmin-An alternate preparation of tritiated ceruloplasmin, 6.1 mg (4.06 X lo6 dpm per mg) was treated with 0.47 unit of C. perfringens neuraminidase in 4.0 ml of sodium acetate-phosphate buffer, pH 5.6, for 50 min at 26". At this point, 45% of the tritium remained in the supernatant fraction after precipitation of an aliquot of the protein with cold 5% trichloracetic acid. The neuraminidase was removed by affinity chromatography employing Sepharose 4B to which had been coupled rabbit anti-C. perfringens neuraminidase y-globulin (14). with rZ51 was carried out as described previously (15).
The final preparation had an estimated specific activity of 2.7 X lo7 cpm per pg. In excess of 69% of the total counts were precipitated by rabbit anti-human orosomucoid antiserum. Preparation of Tritiated r-Carbon Derivative of Sialic Acid-The crystalline P-methylglycoside of AcNeur was synthesized essentially as described by Yu and Ledeen (4) except that the reduction step was carried out with tritiated borohydride. The resulting heptulosonic acid derivative was dissolved in water to give a 1.3% solution and hydrolyzed at 100" for 60 min. The slight discoloration arising during the heating step was removed by filtration through a small (1 to 2 mg) pad of charcoal. Upon lyophilization of the colorless filtrate, the theoretical yield of AcNeu' was obtained with a specific activity of 1.8 x lo8 dpm per mg.

Extensive
periodate oxidation of glycoproteins results in destruction of the terminal, nonreducing residues as well as those pyranosidically linked neutral sugars within the oligosaccharide chain in which carbon 3 is unsubstituted.
This procedure has found application in studies on structure determination and is the basis of the widely used Smith degradation (1'3.
Locus of Tritium Incorporation in Protein-For the purposes of the present study, conditions were sought whereby maximal oxidation of the sialic acid residues could be achieved with little or no destruction elsewhere in the molecule. Under optimal conditions, 2 moles of periodate are consumed and 1 mole of formaldehyde is released per mole of sialic acid present in the intact molecule.
Attainment of these conditions is illustrated by the limited periodate treatment of orosomucoid in which oxidation of the sialic acid residues was essentially completed within 10 min (Fig. 2) L. Van Lente?l and G. Ashwell lS91 judged by maximal formaldehyde formation.
Here, however, the yield of formaldehyde leveled off at 8 to 9 moles per mole of ceruloplasmin.
This value is artificially low and results from interference by ceruloplasmin in the chromotropic acid assay. When a correction is applied, by the use of an internal standard, the release of formaldehyde approaches 12 moles per mole of protein.
Periodate consumption is also more difficult to determine in the presence of ceruloplasmin because of a larger than normal scatter in the end point of the iodometric titration. However, after 10 to 15 min, approximately 24 moles of periodate were consumed per mole of protein.
Since the sialic acid content of ceruloplasmin was determined to be 12 moles per mole, based upon an assumed molecular weight of 160,000 (17), it is apparent that periodate oxidized all of the sialic acid residues and little else.
Sialic acid is readily cleaved from the native glycoproteins by mild acid hydrolysis or esposure to C. perfringens neuraminidase and it was presumed that the 'i-carbon sialic acid derivative would be similarly sensitive to both acid hydrolysis and to enzymatic cleavage.
Consequently, the radioactive products were hydrolyzed for 1 hour at 80" in 0.1 N HzS04 and precipitated by the addition of 3 volumes of ethanol-chloroform (9:l v/v). The results, summarized in Table I, indicate that more than 80% of the tritium of both proteins was released into the supernatant fraction.
The 4 to 5%) of the original radioactivity remaining in the pellet compares closely with the nonspecific incorporation occurring in the absence of periodate oxidation (see "Experimental Procedures"). Upon treatment with neuraminidase, approximately two-thirds of the radioactivity was susceptible to enzymatic hydrolysis. These yields are minimal since no attempt was made to drive the enzymatic reaction to completion by increasing the amount of neuraminidase or by extending the time of incubation.
The tritiated material released from the protein by both enzymatic and acidic hydrolysis was subjected to paper chromatography in several solvent systems and the papers were developed with AgNO (18) and thiobarbituric acid (19). A characteristic thiobarbituric acid-positive reaction product was seen to cochromatograph with authentic AcNeu?. Upon screening for radioactivity, this area was found to contain the vast majority of tritium (Fig. 3). A minor radioactive component, located between AcNeu7 and N-acetylneuraminic acid, and reactive towards thiobarbituric acid, was presumed to be a trace amount of the 8-carbon analogue of sialic acid and was not further investigated.
No additional thiobarbituric acid reactive area was detectable.
The reaction products isolated after either acidic or enzymatic hydrolysis of both proteins were indistinguishable.
Minimal destruction of carbohydrate constituents elsewhere in the molecule was evidenced by the negligible amounts of radioactivity located in the areas corresponding to typical degradation products such as glycerol, erythritol, and threitol (Fig. 3). Since the terminal, nonreducing fucosyl residues are highly susceptible to destruction by periodate oxidation, the fucose content of ceruloplasmin was determined directly (12) ; more than 90% of the material present in the control was demonstrable in the oxidized and reduced protein.
The migration, in several solvent systems, of the radioactive product relative to that of sialic acid is summarized in Table II After cooling to room temperat,ure, the proteins were precipitated by the addition of 3 volumes of ethanol-chloroform (9:l v/v) and allowed to stand for 30 min.
Following centrifugation, t,he pellets were washed, dissolved iu Buffer A, and the radioactivity determined.
The supernatant fraction was concentrated to a small volume, a slight excess of BaClz was added, and the resulting precipitate of BaS04 was removed by ccntrifugation. Excess barium was removed on a small column of Dowex 50 (l-I+) and the radioactivity of the pooled eluate was determined. U, enzymatic hydrolysis.
Of the above orosomucoid and ceruloplasmin preparations, 3.3 mg and 2.8 mg, respectively, were dissolved in about 1.5 ml of 0. product was recovered, as described in the legend to Table I. An aliquot was chromatographed overnight in nbutylacetate-acetic acid-water (3:2:1) and a radioactive tracing was made on a Vanguard autoscanner.  D The reaction products were isolated from the modified glycoproteius by acid hydrolysis, as described in the legend to Table  I. b The thiobarbituric acid assay was carried out as described by Warren (10) with the exception that the periodate oxidation was conducted at 37' and the chromogen was extracted into 2.5 ml of cyclohexanone.
c The resorcinol assay (11) was scaled to one-half volume.
thiobarbituric acid reagent. In no case was it possible to detect residual, intact sialic acid.
Calorimetric analysis of the isolated reaction product indicated a marked loss of sensitivity in the thiobarbituric assay and a somewhat augmented reaction in the resorcinol assay. In order to utilize these differences for quantitative purposes, a direct comparison of the reactivity of synthetic AcNeu' and intact sialic acid was undertaken. The initial erratic results observed with AcNeu7 in the thiobarbituric acid assay as described by Warren (10) were shown to result from incomplete periodate oxidation. Increasing the temperature of the oxidation step from ambient to 37" resulted in an augmented and more stable color yield for AcNeu7.
This modification diminished the sensitivity of intact sialic acid by approximately 20%. In the resorcinol assay, the absorption maximum of both synthetic AcNeu7 and the isolated reaction product was shifted to a higher wave length and was essentially flat between 590 and 630 nm. All values recorded for this assay, however, were determined at 580 nm where absorption was maximum for sialic acid (11). Table III lists the ratio of extinction coefficients of AcNeu7 and sialic acid in the two color assays. Comparable examination of the radioactive products obtained by acid hydrolysis of the modified glycoproteins yielded a ratio closely paralleling that of authentic AcNeu7 thereby providing additional evidence in support of the identification of the reaction product.
Serum Circulation Time of Labeled Protein-The previously indicated correlation between serum survival time and sialic acid content revealed that desialylated ceruloplasmin was cleared from the circulation within 10 to 15 min after injection, in marked contrast to the survival time of the fully sialylated protein (2). With the subsequent indication that removal of only two of the sialic acid residues of ceruloplasmin was sufficient to effect this rapid disappearance from the circulation (20), it became possible to test this observation on the sialic acid-modified protein.
The characteristic pattern of short time survival curves resulting from the injection of intact and partially desialylated preparations of [AcNeu7]ceruloplasmin, shown in Fig. 4, duplicate closely the results seen earlier with the 64Cu-labeled protein (20). Further confirmation of the similarity between authentic ceruloplasmin and the tritiated [AcNeu7]ceruloplasmin derivative, with respect to their circulation time in the bloodstream, is illustrated in Fig. 5. Upon injection of the latter into a female albino rat, the half-life in the circulation was found to be 25 hours, in close agreement with the reported value of 27 hours for 64Cu-labeled ceruloplasmin (20). In contradistinction to ceruloplasmin, orosomucoid represents a group of proteins for which no specific labeling procedure Moreover, in addition to the serum proteins, this procedure has been successfully extended to include the gonadotrophic hormones; continuing studies2 indicate that essentially full retention of hormonal activity accompanies the incorporation of tritium into the modified sialic acid residues of human chorionic gonadotrophin and follicle-stimulating hormone.
In the design of this method, care was exercised to define conditions such that maximal oxidation of the sialyl residues was achieved with minimal, or no, oxidation elsewhere in the molecule.
The complete destruction of sialic acid, the stoichiometry of periodate consumption with formaldehyde production and the recovery of a single radioactive product, identified as AcNeu', all attest to the essential achievement of this goal. However, it is important to note that exposure to longer periods of periodate oxidation at room temperature, followed by borohydride reduction, may result in marked destruction of the underlying carbohydrate chain of the glycoproteins with possibly untoward effects.
The normal half-life of the modified proteins described here stands in contrast to the striking consequences attendant upon removal of sialic acid (2, 20).
In an earlier study on ceruloplasmin, the rapid clearance from the circulation was shown to be correlated with the presence of intact terminal, galactosyl residues since alteration of the latter with galactose oxidase, or removal with P-galactosidase, diminished the rapid hepatic accumulation of the asialoprotein and resulted in an increased circulation time approaching normality (2). The present results extend these observations with the indication that intact sialic acid is not essential for survival; the heptulosonic acid analogue is equally effective. Significantly, the latter derivative is also cleaved by neuraminidase and the resulting desialylated protein is rapidly cleared from the circulation.
Finally it should be emphasized that the present method of labeling, when combined with the previously described incorporation of tritium into the terminal galactose of the asialoprotein (l), permits a direct examination of the metabolic sequelae specifically related to the removal of sialic acid.