Purification and Characterization of UDP-N-Acetylgalactosamine: Polypeptide N-Acetylgalactosaminyltransferase from Bovine Colostrum and Murine Lymphoma BW5147 Cells*

UDP-Ga1NAc:polypeptide N-acetylgalactosaminyl- transferase has been purified from two sources. A soluble form, purified 517,000-fold to homogeneity from bovine colostrum, has a molecular mass of 70,000 daltons by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and 69,000 daltons by gel filtration. A membrane-bound form, partially purified 2,500-fold from BW5147 mouse lymphoma cells, has a molecular mass of 70,000 daltons by sodium dodecyl sulfate- polyacrylamide gel electrophoresis and 71,500 daltons by gel filtration. The purified colostrum enzyme ex-hibits specificity for UDP-GalNAc, has itspH optimum between pH 7.2 and 8.6, and requires Mn2+ for activity. The K , is 8 PM for UDP-GalNAc and 2.5 mg/ml for deglycosylated bovine submaxillary mucin. gly-coprotein raised against cross-react

the assembly of these structures. Although a number of kinetic studies have been performed with partially purified enzyme preparations from bovine, ovine, and porcine submaxillary glands, the enzyme has not been fully characterized (4)(5)(6)(7)(8). More recently, Sugiura et al. (9), using affinity chromatography on apomucin-Sepharose, reported a purification of the transferase to near homogeneity from ascites hepatoma AH66 cells. A number of the properties of this enzyme were studied.
In this paper we have used a slight modification of the procedure of Sugiura et al. to purify the soluble Nacetylgalactosaminyltransferase of bovine colostrum to electrophoretic homogeneity. Our preparation differs from the ascites hepatoma enzyme in several aspects. We also report a 2500-fold purification of an intracellular membrane-bound form of the enzyme from murine lymphoma BW5149 cells and describe an antibody raised against the purified colostrum transferase that cross-reacts with the intracellular murine lymphoma BW5149 enzyme.

RESULTS'
The procedures used to purify the N-acetylgalactosaminyltransferase from bovine colostrum and BW5147 lymphoma cells are described in the Miniprint Section. The following sections deal with some of the properties of the enzymes.
Enzymatic Properties of the Bovine Colostrum Transferase Donor Specificity-The purified N-acetylgalactosaminyltransferase is highly specific for UDP-GalNAc, as shown in Fig. 6. Of the nucleotide sugars tested, only UDP-GalNAc competed effectively with the radiolabeled sugar nucleotide under the conditions used. A 50% decrease in counts incorporated was achieved at an added nucleotide concentration of -18 p~ similar to the theoretically expected value of 15 p~, the donor concentration used in the assay.
Effect of Substrate Concentration-The N-acetylgalactosaminyltransferase has an apparent Km for UDP-GalNAc of 8 pM. The concentration of apomucin required for half-maximal reaction velocity was -2.5 mg/ml. The acceptor concentration is expressed as mg of protein/ml rather than M due to the heterogeneity in the apomucin preparation.
pH Dependence-The assays to determine the pH optimum were performed in the absence of added Mn2+, since this cation caused precipitates at pH >8. Three different buffers were used covering a pH range from 5.0 to 10.6. Under these conditions the purified enzyme was found to have a broad pH optimum between 7.2 and 8.6 (Fig. 7). Ion Dependence-The purified transferase is dependent on divalent cations (Table 11). The presence of a 0.5 mM EDTA completely inhibited the enzyme. This effect was suppressed by performing the assay in the presence of excess Mn2+. In the absence of added Mn2+, the enzyme retained -60% of its activity, indicating that it contains some endogenous divalent cations (Fig. 8). Therefore, when the effect of different cations was measured, the enzyme was preincubated in 0.5 mM EDTA to chelate any endogenously bound cations. Table I1 shows that of the 9 cations tested, MnZ+ was the most effective in restoring the activity. Co2+ was almost as effective as Mn2+, and Cd" and Ni2+ were able to restore the activity partially. Fig. 8 shows that maximum enzyme activity is reached at -2.5 mM Mn2+.

Physical Properties of the Bovine Colostrum Transferase
Amino Acid Composition-The amino acid composition of the bovine colostrum enzyme is shown in Table 111. Since the analytical method used deaminates asparagine and glutamine, the values for aspartic acid and glutamic acid include also asparagine and glutamine, respectively. The overall composition agrees fairly well with the average composition of proteins as defined by Dayhoff (20). The major differences are in glycine, glutamic acid, and cysteine contents which are 208, 65, and 62%, respectively, of the average values.
Molecular Size-Comparison of the Sephadex G-100 superfine elution position of the purified N-acetylgalactosaminyltransferase with the elution positions of 5 protein standards yielded a molecular size of approximately 69,000 daltons ( Fig. 4A). This is in good agreement with the molecular size of 70,000 daltons calculated from the migration distance of the enzyme on SDS-PAGE (Fig. 5A).
Evidence for N-Linked Glycosylation-Studies using affinity chromatography on various lectin columns indicated that the purified N-acetylgalactosaminyltransferase probably contained N-linked oligosaccharide units. The protein bound to TABLE I1 Effect of cations on the activity of bovine colostrum UDP- Ga1NAc:polypeptide N-acetylgalactosaminyltransferase The standard assay system and 10 milliunits of enzyme was used (see "Experimental Procedures"). Assay time was 10 min. Endogenous MnZ+ was removed from the purified enzyme by preincubation in 0.5 mM EDTA. All ions were added as chlorides to a final concentration of 10 mM.   Aspartic acidb  47  Threonine  27  Serine  55  Glutamic acid"  68  Proline  27  Glycine  83  Alanine  31  Cysteined  8  Valine  29  Methionine"  8  Isoleucine  21  Leucine  35  Tyrosine  14  Phenylalanine  15  Histidine  12  Lysine  24  Arginine  22  Tryptophan ND' Values corrected for 7% carbohydrate content. Includes asparagine. e Includes glutamine.
Determined as cysteic acid after performic acid oxidation. e Determined as methionine sulfone after performic acid oxidation. ND, not determined. and could be partially eluted with the appropriate hapten sugars from concanavalin A-Sepharose and lentil lectin-sepharose (data not shown). When the lZ5I-labeled enzyme was treated with endo-P-N-acetylglucosaminidases H and F and analyzed by SDS-PAGE'/autoradiography, a distinct shift in migration position was observed for the endo F-treated material ( Fig. 9, lanes 3 , 4 , 8 , and 9). The magnitude of the shift corresponds to a M, difference of approximately 5000. By contrast, most of the material was resistant to endo H with only a small fraction shifting to a lower M, (Fig. 9, lanes 2 and 7). This pattern was not altered by increasing the incubation time and/or the amount of added endo H (data not shown). The faint 60,000-dalton band is a contaminant, most likely immunoglobulin heavy chain, present in the preparation used for these experiments.

Immunological Properties of the Transferase from Bovine
Colostrum and Murine Lymphoma BW5147 Cells Characterization of Antiserum-Rabbit antibodies raised against purified bovine colostrum N-acetylgalactosaminyltransferase precipitate transferase activity from a partially purified (Apomucin Sepharose I1 eluate) preparation of the colostrum enzyme (Fig. 1OA). In this experiment, 72 pg of IgG (corresponding to 50 pl of antiserum) added to an incubation mixture containing 69 units of N-acetylgalactosaminyltransferase precipitated 81% of the activity or approximately 30 ng of transferase (based on a specific activity of 1860 units/ mg). A purified IgG fraction was used rather than whole antiserum since rabbit serum contains interfering Nacetylgalactosaminyltransferase activity. When titrating the antiserum with purified lZ5I-labeled bovine colostrum trans-The abbreviations used are: SDS-PAGE, sodium dodecyl sulfatepolyacrylamide gel electrophoresis; PBS, phosphate-buffered saline; PMSF, phenylmethylsulfonyl fluoride; or-MEM, minimal essential medium; Endo F and H, endo-6-N-acetylglucosaminidase F and H, respectively; 1, liter. suspension of protein A-Sepharose in PBS was then added and the incubation continued for an additional 50 min on ice during which time the tube was vortexed every 5 min. Following incubation, the gel was sedimented a t 10,000 X g and the supernatant was assayed for remaining N-acetylgalactosaminyltransferase activity. All values have been corrected for nonspecific adsorption. B, 4 ng (-9000 cpm) of "'I-conjugated bovine colostrum N-acetylgalactosaminyltransferase was incubated with increasing amounts of anti-Nacetylgalactosaminyltransferase antiserum for 18 h on ice. Final volume was adjusted to 50 pl with PBS, and the incubation mixture also contained 0.1% taurodeoxycholate. 100 pl of a 1:l suspension of protein A-Sepharose in PBS was added to precipitate antigen-antibody complexes and the incubation was continued on ice for 30 min with vortexing every 5 min. The gel was then sedimented at 10,000 X g for 5 min, washed 3 times with PBS containing 0.1% taurodeoxycholate, and counted. Radioactivity precipitable with preimmune serum has been subtracted from each value.
ferase, approximately 1.9 ng of transferase protein could be precipitated per p1 of antiserum (Fig. 10B). More than 95% of the radioactivity in the '29-labeled enzyme preparation was precipitable with trichloroacetic acid.
Cross-reactivity between Bovine Colostrum Transferase and the Intracellular Enzyme of Murine Lymphoma BW5147 Cells-When the purified colostrum transferase was subjected to SDS-PAGE followed by immunoblotting with antiserum raised against this enzyme, a single band with a M, of approximately 70,000 was observed (Fig. 11, lane 3). Th' IS was expected, based on the previous characterization of the purified 94 mm 5251 1 2 3 FIG. 11. SDS-polyacrylamide gel electrophoresis and immunoblotting of UDP-Ga1NAc:polypeptide N-acetylgalactosaminyltransferase from bovine colostrum and murine lymphoma BW5147 cells. Triton X-114 extracts from microsomes isolated from murine lymphoma BW5147 cells, partially purified Nacetylgalactosaminyltransferase from murine lymphoma BW5147 cells, and purified N-acetylgalactosaminyltransferase from bovine colostrum were subjected to SDS-PAGE in 10% gels (16). The proteins were transferred electrophoretically to nitrocellulose membranes and incubated with 17 pllml anti-N-acetylgalactosaminyltransferase antiserum followed by 100 ng/ml (5.6 X lo6 cpm) "' I-Protein A. The nitrocellulose membranes were washed, dried, and processed for autoradiography as described under "Experimental Procedures." Lane 1, detergent phase from Triton X-114 extraction of 80 pg of mouse lymphoma BW5147 microsomes; lone 2, 0.22 unit of partially purified N-acetylgalactosaminyltransferase from murine lymphoma BW5147 cells; lane 3,l.S units of purified bovine colostrum N-acetylgalactosaminyltransferase. enzyme. A similar analysis was performed on a Triton X-114 microsome extract and a partially purified preparation of the transferase, both from murine lymphoma BW5147 cells. As shown in lanes 1 and 2 of Fig. 11, an immunoreactive species was observed which had a M , of 71,000, slightly greater than that of the soluble bovine enzyme. The band at the top of the gel in lane 2 most likely represents aggregated material although we cannot exclude the possibility that there is a very high molecular weight form of the enzyme.

DISCUSSION
UDP-Ga1NAc:polypeptide N-acetylgalactosaminyltransferase purified to homogeneity from bovine colostrum is a monomer of approximately 70,000 daltons as determined by both SDS-PAGE and gel filtration chromatography. An antibody raised against the purified 70,000-dalton enzyme precipitates the enzyme activity as well as the radiolabeled protein. The difference in the precipitation capacity of the antibody between the two experiments (-0.7 ng of enzyme precipitated/pl when calculated on the basis of activity versus -1.8 ng/pl when calculated on the basis of the iodinated enzyme protein) is probably due to the presence of inactive enzyme. This would be expected since the enzyme loses activity upon storage. Isolation of the intracellular form of this enzyme from mouse lymphoma BW5147 cells yielded a preparation which, although not homogenous, eluted at a position corresponding to a molecular mass of 71,500 on gel filtration chromatography and contained a band of approximately 70,000 daltons on SDS-PAGE (Figs. 4B and 5B). Blotting of this preparation with anti-bovine colostrum N-acetylgalactosaminyltransferase antibody resulted in a band of approx-imately 71,000 daltons. If one assumes that the soluble transferase is derived from a membrane-bound intracellular form by proteolytic cleavage, these results indicate either that the membrane-anchoring portion of the intracellular transferase is relatively small or that this part of the enzyme is very sensitive to proteases and is readily cleaved during the purification procedure. The former possibility seems more likely since even when a crude membrane extract prepared in the presence of protease inhibitors is blotted, no species of higher molecular mass could be detected. The fact that antibody raised against the soluble bovine enzyme cross-reacts with the membrane-derived murine enzyme lends further support to the conclusion that the soluble enzyme is a cleavage product.
In the purification of N-acetylgalactosaminyltransferase from ascites hepatoma AH66 cells, Sugiura et al. (9) reported an apparent M, of 54,000-56,000. The specific activity of this enzyme was 390 units/mg as compared to 1,860 units/mg for the bovine colostrum enzyme. However, the assays of the ascites hepatoma enzyme were done at UDP-GalNAc concentrations below the K,. The two enzymes differ in several other respects. While both enzymes are specific for UDP-GalNAc, the ascites hepatoma enzyme has a K, of 42 pM for this nucleotide sugar whereas the apparent K, of the bovine colostrum enzyme is 8 p~. Another major difference is in the cation requirement. Both enzymes prefer Mn2+, but the bovine colostrum enzyme utilizes Co2+ almost as well as Mn2+ whereas Ca2+ is poorly utilized. By contrast, the ascites hepatoma enzyme uses Co2+ and Ca2+ equally well, each being 50% as effective as Mn2+. The reasons for these differences in the properties of the two enzymes is not clear. One possiblity is that the two enzymes represent different proteins. Alternatively the ascites hepatoma enzyme may be a proteolytic fragment of the intracellular enzyme resulting in some alterations in its properties. Further studies will be required to resolve this discrepancy.
An intriguing finding is the presence of complex-type Asnlinked oligosaccharides on the bovine enzyme. This indicates that the enzyme has been exposed to the late stage oligosaccharide processing enzymes which have been immunolocalized to the middle and trans cisternae of the Golgi stack (2,21,22). Several lines of evidence indicate that the Nacetylgalactosaminyltransferase resides in the cis Golgi and/ or the endoplasmic reticulum (13,(23)(24)(25)(26)(27)(28)(29)(30)(31)(32). For example, Cummings et al. (33) have shown that the precursor form of the low density lipoprotein receptor of A431 cells has high mannose-type Asn-linked units and 0-linked units containing only N-acetylgalactosamine. By contrast, the mature form of the receptor contains complex-type Asn-linked units plus 0linked oligosaccharides that contain sialic acid, galactose, and N-acetylgalactosamine. These findings demonstrate that the transfer of N-acetylgalactosamine residues occurs before the entry of the low density lipoprotein receptor into the regions of the Golgi that contain the late stage processing enzymes (e.g. middle and trans elements). In support of this, Roth has published an electron microscopy study that has localized GalNAc-containing glycoproteins to the cis Golgi cisternae of intestinal goblet cell (23). If the N-acetylgalactosaminyltransferase acts in the cis Golgi or the endoplasmic reticulum, how did the colostrum form of the enzyme acquire complextype oligosaccharides? The simplest explanation is that the cleavage of the membrane-bound enzyme occurs within the Golgi (or endoplasmic reticulum) and that the formation of complex oligosaccharides procedes as the soluble enzyme is transported through the stack. A less likely explanation is that the enzyme initially passed through the Golgi before returning to reside in the cis cisternae (or the endoplasmic reticulum). In this case it could be initially present in the colostrum in a membrane-associated form, similar to galactosyltransferase (34). A final possibility is that the oligosaccharide processing occurs after the enzyme is released into the colostrum.
The availability of the antiserum described in this report will hopefully enable us to study further the relationship between the intracellular and secreted forms of this enzyme and also to immunolocalize it, thereby providing direct evidence for its subcellular location. The reaction mixture was incubated a t 37' C f o r 5-10 min (see individual experiments) and the reaction product was TCA precipitated and radioactivitymeasured as described (13). SDS-PAGE was done on 10% gels as described i n (16) and autoradiography of the dried gels was performed a t -7O' C using Kodak X-hat AR film. Digestion of native enzyne was done as for denatured e m p e b u t a n i t t i n g t h e b o i l i n g i n 1% SDS. -: Apomucin a f f i n i t y chranatography 11. I n t h i s s t e p t h e same type column was used as i n the previous one. Before loading. the The fractions canprislng t h e a c t i v i t y peak were pooled and concentrated as described above but without any further transferase was carried out using the same procedure but with a smaller column (0.9 x 100 addition of detergent. Analytical gel filtration to detenine the mlecular weight of the cm) and collecting 1.06 ml fractlons.