Aglycosylantibody EFFECTS OF EXOGLYCOSIDASE TREATMENTS ON AUTOCHTHONOUS ANTIBODY SURVIVAL TIME IN THE CIRCULATION*

Rabbit anti-hapten antibodies were purified by affinity chromatography and characterized immuno-chemically for in uiuo studies of their blood clearance rate and organ distribution after treatment with various glycosidases. Following sequential removal of sialic acid, galactose, and N-acetyYglucosamine with the appropriate cellulose-immobilized exoglycosidases, the antibody populations were recharacter-ized, radiolabeled, and introduced intravenously into the original animals. Using double radioiodine labels it was possible to demonstrate alterations in purified antibody survival times in the circulation and altered organ distribution after glycolytic cleavage. Removal of terminal sialic acid resulted in rapid blood clearance and enhanced localization of asialoantibody in the liver. Subsequent removal of penultimate galactose residues returned both antibody survival time in the circulation and organ distribution to near normal. Removal of subpenultimate N-acetylglucosamine moieties resulted in aglycosylantibody survival values which were intermediate between asialo-and asialoagalactoantibodies. Removal of the three saccharides also increased kidney localization. The results are evaluated based on current concepts of the biological roles of protein-linked carbohydrate


Rabbit
anti-hapten antibodies were purified by affinity chromatography and characterized immunochemically for in uiuo studies of their blood clearance rate and organ distribution after treatment with various glycosidases. Following sequential removal of sialic acid, galactose, and N-acetyYglucosamine with the appropriate cellulose-immobilized exoglycosidases, the antibody populations were recharacterized, radiolabeled, and introduced intravenously into the original animals. The concept that the terminal sialic acid moiety is crucial to the continued circulation of many, if not most, core-type and mucin-type (1) glycoproteins has been developed by . This hypothesis is based on the fact that neuraminidase treatment of ceruloplasmin (5), orosomucoid, fetuin, haptoglobin, cu,-macroglobulin, chorionic gonadotropin, follicle-stimulating hormone (6), and plasminogen (7) has resulted in rapid clearance of these asialoglycoproteins from the blood when compared with untreated controls. The loss of aeialoglycoproteins from the circulation was subseyuently shown to be due to their binding by hepatic (parenchymal) cell surface receptors which "recognize" exposed, penultimate galactosyl moieties (8,9). The hepatic cell surface receptor was initially proposed to be an enzyme-UDP-Gal-glycoprotein glycosyltransferase (10). but the involvement of transferase activity in the purified receptor (11) has been questioned by Hudgin and Ashwell (12) (Table I). Activities were reexamined after immobilization on BAc-cellulose confirming th: t the enzymes were coupled essentially without loss of activity except for N-acetylglucosaminidase which decreased nearly 5-fold in activity after immobilization.
This loss in Imglucosaminidase activity might have been due to enzyme sensitivity to the pH conditions used for optimal coupling to BAccellulose or to reactive groups near the enzyme acticie site.
Purified native antibodies (25-or 50-mg aliquots) were treated sequentially with Im-enzymes and the released monosaccharides determined (Fig. 2). Im-neuraminidase liberated 50 to 60% of the total amount of sialic acid on the antibody ( Fig. 2A). From these data it was possible to estimate that 0.45 mol of AcNeu/mol of IgG was released by the immobilized enzyme. Subsequent treatment of Im-neuraminidase treated (aSia-antibodies) and untreated antibodies with Im-fi-galactosidase resulted in release of measurable amounts of galactose from aSia-antibodies but not from untreated antibodies (Fig. 2B). Assuming that there were 3 to 4 mol of galactose/IgG "Units/mg of enzyme protein as described under "Materials and Methods." * -, not detectable. c Im-enzyme concentration expressed in milligrams of enzyme/ml of packed volume was: Im-P-galactosidase, 4.9; Im-N-acetylglucosaminidase, 3.9; Im-neuraminidase, 4.6; Im-lactoperoxidase, 4.9. d BAc-cellulose inactivated by bicarbonate-ethanolamine treatment (see "Materials and Methods").
molecule (40) this corresponded to cleavage of about 20% of the total galactose. Untreated antibodies, aSia-antibodies and aSia-aGal-antibodies were next incubated with Im-N-acetylglucosaminidase.
The results (Fig. 2C) indicated that Im-Nacetyl-glucosaminidase failed to release detectable amounts of GlcNAc from untreated antibody, but could release about 5% of the total GlcNAc from aSia-aGal-antibodies assuming that there are 10 to 11 mol of GlcNAc/mol IgG (41, 42). There was a small, measurable amount of GlcNAc released from aSiaantibodies suggesting that (a) Im-N-acetylglucosaminidase had some endoglycolytic activity, (b) antibody molecules possessed some penultimate GlcNAc, or (c) the Im-neuraminidase contained small amounts of contaminating P-galactosidase activity which was not detected by our assays (Table I).
Average intrinsic association constants (K,) and heterogeneity indices (a) of the purified anti-hapten antibodies and aglycosylantibodies indicated that glycosidase treatment(s) did not affect the antigen-binding properties (Table II). Differences between untreated and aglycosylantibodies are not significant and fall within experimental error. These results suggest that antibody molecules were not "denatured" by treatment with the glycosidases or by exposure to pH 5.5 to 6.3 for 80 min.
Plasma Survival Times-The in vivo fates of untreated and aglycosylantibodies was followed simultaneously by injection of 5.0 mg of '9-untreated antibody with 5.0 mg of Yaglycosylantibody (12sI:1311 = 1). Rabbits were bled periodically and the blood analyzed for percentage of recoverable radioactivity/ml.
Figs. 3 to 6 show the survival times of autochthonous aglycosylantibodies compared with their untreated counterparts. The average clearance rate for untreated homologous antibodies was O.OCi%/hour and the recovery of total label was 0.3 to 0.5%/ml of blood (Fig. 3). aSia-antibodies were cleared at a rate of O.P%/hour (Fig. 4) while aSia-aGal-antibodies were cleared at near normal rates (Fig. 5). The aSia-aGa1 aGlcNAc-antibodies showed a significant increase (O.l%/hour) in blood clearance rates (Fig. 6). Organ Distribution of Aglycosylantibodies-In order to determine the organ distribution of the various antibody populations after measuring clearance from the circulation, rabbits were sacrificed and their organs removed, weighed, minced into approximately 5 mm2 pieces, and monitored for total radiolabel content. The data summarized in Table III shows that aSia-antibodies were localized primarily in the liver while aSia-aGal-antibodies possessed essentially normal organ distribution compared with untreated antibodies injected simultaneously. The aSia-aGal-aGlcNAc-antibodies were found to a large degree in the kidney with some localization in the liver (Table III). The specificity of organ localization was tested in blocking experiments by injecting two additional rabbits with 5 mg of autochthonous Y-labeled aglycosylantibodies plus 5 mg of unlabeled native aglycosylantibodies.
After 120 min, animals were sacrificed and their liver and kidneys evaluated for percentage of total radiolabel injected. Table IV shows that in comparison with lSII-labeled aglycosylantibodies alone, both liver and kidney radiolabel uptake could be specifically inhibited with unlabeled aglycosylantibodies.

DISCUSSION
The carbohydrate moiety of IgG constitutes 3 to 4% by weight of the total molecule and is known to consist of galactose, GlcNAc, AcNeu, GalNAc, mannose, and fucose (40). The exact sequence of these sugars on the antibody molecule is not known. However, the terminal oligosaccharide sequence, AcNeu*+GalBGlcNAc-, appears on an IgG myeloma protein (43) at a frequency equivalent to that found on many other glycoproteins (1) suggesting that this may be an important trisaccharide. The site of attachment of carbohydrate to the IgG of both rabbit and human origin is an asparagine residue in the Fc portion of the heavy chain. In rabbit IgG isolated from pooled sera, about 15% of the heavy chains Aglycosylantibody Survival in the Circulation contain an additional serine-or threonine-linked carbohydrate (44). These latter oligosaccharide residues generally contain GlcNAc, mannose, and galactose but no N-acetylneuraminic acid or fucose.
There are several roles which can be proposed for antibody The carbohydrate portion of IgG is known to be involved in complement binding (24,45). In addition, in a manner perhaps similar to parenchymal cell surface receptors which "recognize" exposed galactose moieties, macrophage and lymphocytes have receptors which may specifically recognize the carbohydrate sequences (perhaps, respectively, the Fd and Fc attached sugars) in the constant region of the IgG molecule (46, 47). Predictions have also been made for a role of carbohydrate in antibody secretion (48). Since serum albumin contains little or no carbohydrate, it appears that sugar moieties are not essential for secretion per se. It seems more likely that protein-linked oligosaccharides function in molecular recognition schemes and may confer important physicochemical characteristics to glycoproteins. For example, the carbohydrate portion of an arctic fish "antifreeze" glycoprotein has been shown to be responsible for lowering the freezing temperature of serum (49-52). The presence of large amounts of carbohydrate on glycoproteins like fetuin could enhance solubility (53) or decrease partial specific volume (54).
The concept that plasma glycoprotein carbohydrate, specifically terminal AcNeu and penultimate galactose, plays a role in cell surface recognition has been under experimental investigation for several years (Z-6, 8, 9, 13). The basic approach has been to remove terminal sialic acid residues with neuraminidase and to study the circulatory survival times of asialoglycoproteins. From the results of these studies, Ashwell and Morel1 concluded that the parenchymal cell of the liver has a surface receptor which recognizes exposed, AcNeu-penultimate, galactose moieties and binds the asialoglycoprotein for subsequent endocytosis and degradation (2-4). This mechanism is postulated to occur naturally as a means for maintenance of glycoprotein homeostasis.   (Fig. 6). This enhanced clearance corresponded to increased localization of radiolabel in the kidneys (Table  III). Elevated liver uptake of radiolabeled aSia-antibodies and kidney uptake of aSia-aGal-aGlcNAc-antibodies were specific processes which could be inhibited with the appropriate, unlabeled aglycosylantibody population (Table  IV).