Purification and Characterization of UDP-N-Acetylglucosamine: α1,3-d-Mannoside β1,4-N-Acetylglucosaminyltransferase (N-Acetylglucosaminyltransferase-IV) from Bovine Small Intestine*

A new β1,4-N-acetylglucosaminyltransferase (GnT) which involves in branch formation of Asn-linked complex-type sugar chains has been purified 224,000-fold from bovine small intestine. This enzyme requires divalent cations, such as Mn2+, and catalyzes the transfer of GlcNAc from UDP-GlcNAc to biantennary oligosaccharide and produces triantennary oligosaccharide with the β1–4-linked GlcNAc residue on the Manα1–3 arm. The purified enzyme shows a single band ofM r 58,000 and behaves as a monomer. The substrate specificity demonstrated that the β1–2-linked GlcNAc residue on the Manα1–3 arm (GnT-I product) is essential for the enzyme activity. β1–4-Galactosylaion to this essential β1–2-linked GlcNAc residue or N-acetylglucosaminylation to the β-linked Man residue (bisecting GlcNAc, GnT-III product) blocks the enzyme action, while β1–6-N-acetylglucosaminylation to the Manα1–6 arm (GnT-V product) increases the transfer. Based on these findings, we conclude that the purified enzyme is UDP-N-acetylglucosamine:α-3-d-mannoside β-1,4-N-acetylglucosaminyltransferase IV (GnT-IV), that has been a missing link on biosynthesis of complex-type sugar chains.

The complex-type of oligosaccharides are synthesized through elongation by glycosyltransferases after trimming of the precursor oligosaccharides transferred to proteins in the endoplasmic reticulum. N-Acetylglucosaminyltransferases (GnTs) 1 take part in the formation of branches in the biosynthesis of complex-type sugar chains. In vertebrates, six GnTs, designated as GnT-I to -VI, which catalyze the transfer of GlcNAc to the core mannose residues of Asn-linked sugar chains, have been identified (1), GnT-VI-GlcNAc␤1 GnT-II -GlcNAc␤1 GnT-IV-GlcNAc␤1 GnT-I -GlcNAc␤1 The complex-type sugar chains play important roles in many biological phenomena, such as control of glycoprotein hormone activity and cell-cell interactions. Some of the functions of sugar chains are closely related to their branching structure. For example, highly branched complex-type sugar chains are essential for the effective expression of in vivo biological activity of human erythropoietin since the activity correlates to the ratio of tetra-antennary to biantennary oligosaccharides (2). It has been reported that the branching of sugar chains increases with the malignancy of tumor cells (3,4). Recently Yoshimura et al. (5) demonstrated that the sugar chains produced by the action of GnT-V are involved in invasion and cell attachment in the extravation stage of lung metastasis (5).
To elucidate the regulation and the roles of branching structure of complex-type sugar chains, proteins or genes of GnT-I, -II, -III, and -V, have been isolated from mammalian tissues and cells (6 -16). No sequence homology was found among these GnTs (17). Among known mammalian GnTs, only GnT-IV has not been purified, and its gene structure remains unknown.
Determination of Activities of GnT-IV-GnT-IV activity was assayed using fluorescence-labeled substrate according to the method of Nishikawa et al. (19) with some modifications (26). Preparation of the substrate Gn 2 (2Ј,2)core-PA ( Fig. 1) was done as reported (26). Enzyme solution (15 l) was incubated at 37°C for 4 h with 125 mM MOPS buffer, pH 7.3, containing 0.8 mM substrate, 20 mM UDP-GlcNAc, 7.5 mM MnCl 2 , 200 mM GlcNAc, 0.5% (w/v) Triton X-100, 10% glycerol, and 5 mg/ml bovine serum albumin in a total volume of 50 l. After the incubation, 50 l of water was added and the enzyme reaction was stopped by boiling for 2 min. After filtration, 5 l of the reaction mixture was applied to a TSK ODS-80TM column (4.6 ϫ 150 mm, Tosoh, Tokyo, Japan). Reverse phase chromatography was performed at 50°C with a 50 mM ammonium acetate buffer, pH 4.0, containing 0.15% 1-butanol at a flow rate of 1.2 ml/min. Fluorescence was monitored using excitation and emission wavelengths of 320 and 400 nm, respectively. The specific activity of the enzyme was expressed as moles of products per hour of incubation per mg of protein. The Gn 3 (2Ј,4,2)core-PA (Takara, Kyoto, Japan) was used as the standard.
Gn 3 (6Ј,2Ј,2)core-PA was prepared from Gn 2 (2Ј,2)core-PA by the reaction of GnT-V using CHO-K1 cell extracts as a crude GnT-V preparation. Harvested CHO-K1 cells were sonicated in 2 volumes of 5 mM Tris-HCl buffer, pH 7.5, containing 2 mM MgCl 2 and 1 mM phenylmethylsulfonyl fluoride. After centrifugation at 900 ϫ g for 10 min, the supernatant was used as the crude GnT-V preparation. Sixty microliters of the crude GnT-V preparation, 40 l of 1.6 mM Gn 2 (2Ј,2)core-PA, and 100 l of the reaction buffer (250 mM MES, pH 6.25, 80 mM UDP-GlcNAc, 20 mM EDTA to reduce other GnTs activity, 400 mM N-GlcNAc, and 1% (w/v) Triton X-100) were combined and incubated at 37°C. After 4 h, an additional 60 l of the crude GnT-V preparation was added and further incubated for 20 h. After boiling and filtration, the reaction mixture was applied to a reverse phase column (Vydac 218TP152010, 10 ϫ 250 mm, Hesperia, CA). Gn 3 (6Ј,2Ј,2)core-PA was eluted at 50°C with 50 mM ammonium acetate buffer, pH 4.0, containing 0.15% 1-butanol at a flow rate of 2.5 ml/min. Gn 3 (2Ј,2,bis)core-PA was prepared from the reaction mixture after measurement of GnT-IV activity in the bovine small intestine homogenate.
Analytical Methods-500 MHz 1 H NMR measurement of the enzymatic product was performed with a JEOL JMG-GSX500 at 30 and 60°C using acetone as an internal standard at ␦ 2.225 ppm. Smith's degradation was carried out according to the method described by Kobata and Takasaki (27). Molecular mass of pyridylaminated sugar chains was determined with MALDI-TOF-MS (LASER MAT 2000, Finnigan MAT, Hemel Hempstead, United Kingdom).
Gel Electrophoresis-SDS-polyacrylamide gel electrophoresis (PAGE) was done by the method of Laemmli (28) using 12% gel. The molecular weight marker kit (Bio-Rad) was used for size standards. Native PAGE was carried out at pH 8.8 on a 7.5% gel containing 0.1% Triton X-100 by the method of Davis (29). Proteins in the gels were stained with silver (Silver Stain Plus, Bio-Rad).
Preparation of Microsome Fraction-All procedures were carried out at 4°C. Frozen bovine small intestine (2 kg) was minced and homogenized in 4 volumes of Buffer A with a Polytron homogenizer (Kinematica AG, Luzern, Switzerland) for 30 s four times at 1-min intervals. The homogenate was centrifuged at 900 ϫ g for 10 min. The superna- tant was centrifuged further at 105,000 ϫ g for 60 min and the microsome fraction was obtained as a precipitate.
Solubilization of GnT-IV-The microsome fraction was suspended with 3 volumes of Buffer A containing 1% (w/v) Triton X-100. The suspension was gently stirred for 1 h and then centrifuged at 105,000 ϫ g for 1 h. The supernatant fraction was collected and the residual pellet was suspended again in 1.5 volumes of the same buffer above. The first and second Triton extracts were combined (9 liters).
Q-Sepharose FF Column Chromatography-The combined Triton X-100 extracts were applied to the Q-Sepharose FF column (5 ϫ 30 cm) equilibrated with Buffer B. The column was washed with 3 liters of Buffer B until the protein concentration in the eluate was reduced to below 0.1 mg/ml. Bound protein was eluted with a linear gradient established between 2.5 liters of Buffer B and 2.5 liters of the same buffer containing 0.5 M KCl at a flow rate of 2.5 ml/min. The fractions containing GnT-IV activity (1 liter) were pooled.
Cu 2ϩ -chelating Sepharose FF Column Chromatography-The pooled fraction from Q-Sepharose FF column chromatography was applied to the Cu 2ϩ -chelating Sepharose FF column (5 ϫ 10 cm) equilibrated with Buffer C. The GnT-IV activity was bound completely to the column. After washing with 5 bed volumes of Buffer C, proteins were eluted with a linear gradient established between 1 liter of Buffer C and 1 liter of the same buffer containing 0.1 M glycine at a flow rate of 1 ml/min. The fractions containing GnT-IV activity (540 ml) were pooled and concentrated to 100 ml on a YM 30 membrane with a Diaflow Ultrafiltrater (Amicon, Beverly, MA).
UDP-Hexanolamine-Agarose Affinity Column Chromatography I: Elution with KCl-The concentrate was dialyzed extensively against Buffer D. Half of the dialyzed sample (50 ml) was applied to a UDPhexanolamine-agarose column (1.2 ϫ 4.5 cm) equilibrated with Buffer D. The column was washed successively with 30 ml of Buffer D and 30 ml of Buffer E. GnT-IV was eluted with Buffer F. The fractions containing GnT-IV activity (4.5 ml) were pooled. The remaining dialyzed sample was fractionated using the identical column, and the fractions containing GnT-IV activity were combined.
UDP-Hexanolamine-agarose Affinity Column Chromatography II: Elution with Buffer Lacking MnCl 2 -After dialysis against Buffer G, the active fractions (10 ml) were applied to a UDP-hexanolamineagarose column (1.0 ϫ 6.5 cm) equilibrated with Buffer G. After washing the column with 20 ml of the same buffer, GnT-IV was eluted with 20 ml of Buffer H at a flow rate of 0.2 ml/min. The fractions containing GnT-IV activity (6 ml) were pooled.
Superdex 200 Gel Filtration Column Chromatography-The pooled fractions were concentrated to 0.5 ml with a small column of Q-Sepharose FF. The concentrate was applied to a column of Superdex 200 HR10/30 (1 ϫ 30 cm) equilibrated with Buffer I. GnT-IV was eluted at a flow rate of 0.25 ml/min. After measurement of the protein concentration of the fractions, MnCl 2 was added to give a final concentration of 10 mM. An low molecular weight kit (Pharmacia, Uppsala, Sweden) was used for molecular weight calibration.
Protein Assay-The protein in the eluate of Q-Sepharose, Cu 2ϩchelating Sepharose, and UDP-hexanolamine column chromatographies were monitored by a Bio-Rad protein assay kit. The protein assays shown in Table I and the protein of the eluate of Superdex 200 column chromatography were measured by the more sensitive BCA protein assay kit (Pierce), using bovine serum albumin as a standard.

RESULTS
Purification of GnT-IV-Activity of GnT-IV was assayed using the fluorescence-labeled biantennary sugar chain, Gn 2 (2Ј,2)core-PA, as an acceptor substrate (19). The assay method was originally developed to measure GnT-III, -IV, and -V activity at one time, which enabled the purification of GnT-III and -V (12,14). However, the sensitivity of this method for GnT-IV activity did not appear sufficient for purification. The substrate concentration used in the previous assay protocol (0.08 mM, Ref. 19) was not high enough to detect GnT-IV activity; moreover, GnT-III and -V, which co-existed in the crude enzyme preparation, competed with GnT-IV against the same substrate. To compensate for these problems, the substrate concentration was raised to 0.8 mM. The addition of 10% glycerol and 1% bovine serum albumin in the assay solution were also effective to stabilize GnT-IV activity. As a result of these improvements, the sensitivity for detection of GnT-IV activity in the crude enzyme preparation was eight times higher than that under the previous assay condition.
Using the method to survey the distribution of GnT-IV activity in rat tissues (19), we surveyed GnT-IV activity in commercially available bovine tissues and cultured cells. Since the relative activity of GnT-IV against that of GnT-III and -V in the bovine small intestine was the highest among samples examined, we chose bovine small intestine as an enzyme source.
Like other GnTs, GnT-IV was enriched in microsomes. However, about 20% of the GnT-IV activity in the homogenate was found in the cytosol fraction, suggesting that limited hydrolysis might occur. GnT-IV activity was successfully solubilized from the microsome fraction by Triton X-100 extraction. CHAPS and n-octyl-␤-D-thioglucoside were less effective than Triton X-100 (data not shown). The addition of benzamidine hydrochloride, a trypsin inhibitor, was effective for maintenance of GnT-IV activity during the whole procedure. The majority of proteins in the Triton extract were removed from the GnT-IV activity by the Q-Sepharose FF column ( Fig. 2A) and Cu 2ϩ -chelating Sepharose FF column chromatography (Fig. 2B) steps. After the two chromatographies, the GnT-IV active fractions still contained GnT-III and -V activities as well (data not shown). Separation of GnT-IV activity from other GnTs was carried out by affinity column chromatography (Fig. 2C), using an analogue of the common donor substrate for GnTs (UDP-hexanolamine) as ligands. Activities of GnT-III and -V passed through the column under the chromatographic conditions as described under "Experimental Procedures." GnT-IV activity was bound completely to the column and eluted with 1 M KCl at pH 8.0. Since the eluted fractions still contained several proteins of various molecular weights (Fig. 3, lane 4), another chromatography was performed. The GnT-IV activity was retarded on an affinity column (12) to which Gn 2 (2Ј,2)core-Asn was covalently bound, however, the GnT-IV activity could not be isolated from other proteins (data not shown). Neither blue, red, Q-, nor Cu 2ϩ -chelating Sepharose columns were effective for purification in this step. The separation of GnT-IV from other proteins was accomplished by the second UDP-hexanolamine column chromatography with modified elution conditions. The bound GnT-IV was eluted immediately from the column by the buffer lacking MnCl 2 (Fig. 2D). The fraction containing GnT-IV was further purified by gel filtration (Fig. 2E). The purified GnT-IV showed a single band with a molecular weight of 58,000 on SDS-PAGE (Fig. 3, lane 6). It also showed a single band on native PAGE (Fig. 4A), and GnT-IV activity was accompanied by a protein band (Fig. 4B). No activity of GnT-I, -II, -III, or -V was detected in the purified GnT-IV preparation (data not shown). Table I summarizes the purification of GnT-IV. GnT-IV was purified to homogeneity and over 224,000-fold by these procedures. A protein with a molecular weight of 45,000 in lane 4 (Fig. 3) was identified as bovine ␣1,3-galactosyltransferase (30) by sequence analyses of its peptide fragments.
Protein Chemistry of GnT-IV-SDS-PAGE analysis of purified GnT-IV before (lane 1) and after (lane 2) digestion with peptide N-glycosidase F is shown in Fig. 5. Apparent molecular weight was reduced by 3,000 as a result of glycosidase treatment indicating that GnT-IV is a glycoprotein with Asn-linked sugar chains. As shown Fig. 2E, the apparent molecular weight of GnT-IV was estimated to be 73,000 by gel filtration in the presence of Triton X-100. This estimated molecular weight may be larger than the actual size, since membrane proteins migrate as protein-detergent complexes on gel filtration columns. These observations together suggest that GnT-IV is a monomer with a molecular weight of 58,000.
Optimum pH-GnT-IV was the most active between pH 7.0 and 8.0 with an optimum of pH 7.3 (Fig. 6).

Effect of Divalent Cations and 2-Mercaptoethanol on GnT-IV
Activity-GnT-IV activity depends on MnCl 2 concentration, and the optimum concentration of MnCl 2 was 10 mM. When MnCl 2 concentration was increased, the activity gradually decreased and was suppressed by 50% at 60 mM (data not shown). The effects of other divalent cations at 10 mM concentration were examined (Table II). Each divalent cation was added to the GnT-IV preparation which had been treated with 1 mM EDTA followed by dialysis against 20 mM Tris-HCl buffer, pH 7.4, containing 20% glycerol and 0.05% Triton X-100. GnT-IV maintained slight activity without the addition of any metal but lost activity completely by the addition of EDTA and CuCl 2 .
The activity was maximal with MnCl 2 , while CoCl 2 and MgCl 2 were less effective. CaCl 2 and FeCl 2 had no effect. The addition of 2-mercaptoethanol (10 mM) to the enzyme had no effect on the enzyme activity (data not shown). Product Identification-To identify the sugar chain structure produced by the enzyme, the product of GnT-IV using Gn 2 -(2Ј,2)core-PA as a substrate was subjected to Smith's degradation. The molecular weight of the degradation fragments were monitored with a MALDI-TOF mass spectrometer (data not shown). Before Smith's degradation, the molecular weight of the enzymatic products was 1598.2. After the first and second cycle of Smith's degradation, the molecular weights of the fragments were 795.3 and 634.7, respectively. This degradation pattern was consistent with that of the GnT-IV product shown in Fig. 7. 1 H NMR analysis of the enzymatic product showed peaks of chemical shifts at 4.53, 4.54, and 4.55 ppm, which correspond to those of anomeric protons of GlcNAc 7, GlcNAc 5, and GlcNAc 5Ј (31) in Fig. 7, respectively. GlcNAc 7 should link to mannose 4 through ␤-linkage, because its spin coupling constant (J 1,2 ) is 7.9 Hz. These data indicate that the enzymatic product has the structure of Gn 3 (2Ј,4,2)core-PA (Fig. 1).
Effect of UDP-GlcNAc Analogues on GnT-IV Activity-To understand the relative effects of nucleotides and sugar nucleotides against UDP-GlcNAc, the activities of GnT-IV were measured in the presence of 2 mM UDP-GlcNAc analogues (Table III). UDP was the most potent inhibitor. The uracil moiety appeared to be essential for the inhibition of nucleotides, since TDP and CDP did not inhibit enzyme activity. A comparison of uridine, UMP, UDP, and UTP suggested that the number of phosphoesters should be important for inhibition. The sugar nucleotides having the UDP moiety also inhibited enzyme activity, including UDP-hexanolamine which was used as an affinity ligand for enzyme purification.
Kinetic Analysis-Lineweaver-Burk plots of kinetic data obtained with the purified GnT-IV preparation at a 40 mM UDP-GlcNAc concentration were linear, and gave the following kinetic constants: K m values for Gn 2 (2Ј,2)core-PA and Gn 3 (6Ј,2Ј,2)core-PA were 0.73 and 0.13 mM, respectively, and V max values for the same substrates were 3.23 and 1.75 M/ min, respectively. The K m value for UDP-GlcNAc was 0.22 mM in 0.8 mM Gn 2 (2Ј,2)core-PA as an acceptor. DISCUSSION A new ␤1,4-GlcNAc transferase was purified from the membrane fraction of bovine intestine. The enzyme transfers Glc-NAc from UDP-GlcNAc to Gn 2 (2Ј,2)core-PA and produces Gn 3 (2Ј,4,2)core-PA. This enzyme also acts on Gn 2 (2Ј,2)core oligosaccharides attached to glycoproteins (data not shown), suggesting that the enzyme belongs to the branch-forming GlcNAc transferases for Asn-linked sugar chains. Schachter et al. (18) classified such GlcNAc transferases from GnT-I to GnT-VI (1). Among them, GnT-III, -IV, and -VI are ␤1,4-GlcNAc trans-  ferases which produce ␤1-4GlcNAc branches on the ␤-linked Man residue, the Man␣1-3 residue, and the Man␣1-6 residue of the core oligosaccharides, respectively (1,32). The ␤1,4-GlcNAc transferase newly purified in this paper was identified as GnT-IV, because the 1 H NMR of the product oligosaccharide clearly indicated that the enzyme produced Gn 3 (2Ј,4,2)core-PA exclusively. The substrate specificity of purified GnT-IV matched almost perfectly to that studied by Gleeson and Schachter (18) using the hen oviduct microsome fraction as an enzyme source: as shown in Table IV, (i) GlcNAc␤1-2Man␣1-3 residue was absolutely required, (ii) galactosylation (GGn(2Ј),Gn(2)core-PA or Gn(2Ј),GGn(2)core-PA) reduced or even prevented the transfer, (iii) bisecting GlcNAc (Gn 3 (2Ј,2,bis)core-PA) prevented the transfer. The only disagreement (to Ref. 18) was the fact that the activity of purified GnT-IV was inhibited up to 50% by fucose residue linked ␣1-6 to non-reducing GlcNAc (Gn 2 (2Ј,2)-coreF-PA versus Gn 2 (2Ј,2)core-PA, Table IV) while the crude enzyme was not. This may due to the structural difference at the reducing termini of the substrate from PA-derivatives to sugar peptides.

TABLE II Effect of divalent cations on GnT-IV activity
GnT-IV inactivated by the addition of 1 mM EDTA was used after dialysis against a buffer containing 20 mM Tris-HCl, pH 7.4, 0.05% Triton X-100, and 20% glycerol. Enzyme activity was assayed in the standard mixture described under "Experimental Procedures" with 10 mM metal chlorides or EDTA. Activity is expressed as percent of ␤1,4-GlcNAc transfer observed in the presence of MnCl 2 .  III Effects of nucleotides on GnT-IV activity As described under "Experimental Procedures," purified enzyme (3 g/ml) was incubated in the reaction mixture with 0.5 mM UDP-GlcNAc and 2 mM inhibitors. Activity is expressed as percent of ␤1,4-GlcNAc transfer observed in the absence of inhibitor. teins with tetra-antennary sugar chains are relatively rare. There must be either an imbalance of those enzymes or some factors in glycoproteins which prevent or promote the access of those enzymes (33,34). It should be quite interesting to investigate the relationship between branch structure of complextype sugar chains and activity valance among GnT-III, -IV, -V, and ␤1,4-GalT.
Our result clearly showed that purified GnT-IV could act on the mono-antennary Gn(2)core-PA (product of GnT-I and ␣-mannosidase II) and produced the Gn 2 (4,2)core sugar chain known as an abnormal biantennary type. Such an abnormal sugar chain appears on hCG from choriocarcinoma (24) and ␥-glutamyltranspeptidase from human hepatocellular carcinoma (35). It would be quite interesting to examine if GnT-IV activity correlates with the advances of malignancy on those cancer cells.
GnT-IV potentially contributes to the production of glycoprotein hormones sharing highly branched sugar chains. The specific activity of GnT-IV in CHO cells used widely for protein production is relatively low (20 pmol/h/mg), which is about 5% of GnT-V specific activity. It is expected that CHO cells with elevated GnT-IV activity by incorporation of the GnT-IV gene will increase the production of tetra-antennae. Cloning of the GnT-IV gene is now in progress in our laboratory.

TABLE IV
Substrate specificity of GnT-IV Purified GnT-IV (4 g/ml) was assayed in the standard mixture described under "Experimental Procedures" except that the concentration of acceptors was 0.15 mM. Relative activity is expressed as percent of ␤1,4-GlcNAc transfer observed in the presence of Gn 2 (2Ј,2)core-PA as an acceptor. The structures of acceptor sugar chains are shown in Fig. 1.