Purification and Characterization of UDP-GlcNAc:Galβ1–4GlcNAcβ1–3*Galβ1–4Glc(NAc)-R(GlcNAc to *Gal) β1,6N-Acetylglucosaminyltransferase from Hog Small Intestine*

A β1,6N-acetylglucosaminyltransferase (β1–6GnT) responsible for the formation of the β1,6-branched poly-N-acetyllactosamine structure has been purified 210,000-fold in 2.4% yield from a homogenate of hog small intestine by successive column chromatographies involving CM-Sepharose FF, Ni2+-chelating Sepharose FF, and UDP-hexanolamine-agarose, using an assay wherein pyridylaminated lacto-N-neotetraose (Galβ1–4GlcNAcβ1–3Galβ1–4Glc-PA) was used as an acceptor substrate, and the reaction product was Galβ1–4GlcNAcβ1–3(GlcNAcβ1–6)Galβ1–4Glc-PA. The apparent molecular weight of the purified enzyme was 76,000 under nonreducing conditions. The enzyme has a pH optimum at 7.0 and has no requirement for any divalent metal ions. The K m values for pyridylaminated lacto-N-neotetraose and UDP-GlcNAc were 0.96 and 2.59 mm, respectively. For its activity, this enzyme was shown to have an absolute requirement of at least a complete LacNAc (LacNAc = Galβ1–4GlcNAc) residue bound to position 3 of the acceptor Gal residues, i.e. it is capable of acting only on the Gal residues of internal LacNAc units. The data strongly suggest that this enzyme could be involved in generating branches to central positions of preformed as well as growing polylactosamine chains, but not in synthesizing the distal branches to growing polylactosamine chains.

Linear and branched poly-N-acetyllactosamine chains, which consist of repeating N-acetyllactosamine units (Lac-NAc ϭ Gal␤1-4GlcNAc), 1 occur in glycoproteins, glycolipids, and proteoglycans. In linear poly-N-acetyllactosamine chains, the N-acetyllactosamine units are linked via ␤1,3 linkages. In branched glycans, some of the 3-O-substituted Gal residues in the primary chains are also substituted at position 6 by additional N-acetyllactosamine units. Linear and branched poly-Nacetyllactosaminoglycan backbones represent the blood group i and I antigens, respectively (1-3), the expression of which is known to be regulated in the development of fetal to adult erythrocytes (4 -6), in the course of murine embryonic development (7), and in embryonal carcinoma cells (8).
Until now, the substrate specificity of the purified forms of these enzymes has not been reported. This study reports on the purification of the cIGnT6 from hog small intestine by successive column chromatographies on CM-Sepharose FF, Ni 2ϩ -chelating Sepharose FF, and UDP-hexanolamine-agarose, using pyridylaminated lacto-N-neotetraose (Gal␤1-4GlcNAc␤1-3Gal␤1-4Glc-PA) as an acceptor substrate, the product of which was Gal␤1-4GlcNAc␤1-3(GlcNAc␤1-6)Gal␤1-4Glc-PA. By using several acceptor compounds, this purified enzyme was shown to have an absolute requirement for a complete LacNAc residue bound to position 3 of the acceptor galactose residue for its activity. Consequently, it was devoid of dIGnT6 activity.
Determination of ␤1-6GnT Activity-␤1-6GnT activity was assayed according to the method of Gu et al. (13) with minor modifications. The standard incubation mixture contained the following components in a total volume of 10 l; 200 mM MOPS buffer (pH 7.0), 10 mM UDP-GlcNAc, 0.5% Triton X-100, 5 mM EDTA, 2 mg/ml BSA, 20 M lacto-Nneotetraose-PA, and enzyme fraction (3 l). After incubation at 37°C for 4 h, 40 l of water was added, and the enzyme reaction was quenched by boiling for 1 min. After centrifugation at 10,000 rpm for 5 min, 10 l of the supernatant from the reaction mixture was applied to a TSK-gel ODS-80TM column (4.6 ϫ 250 mm; Tosoh, Tokyo, Japan). The elution was performed at 50°C using an eluent of 20 mM ammonium acetate (pH 4.0) at a flow rate of 1 ml/min. Fluorescence was monitored with excitation and emission wavelengths of 320 and 400 nm, respectively. The specific activity of the enzyme was expressed as picomoles of product/hour/mg of protein. The protein concentration was determined with a BCA kit (Pierce) or a Bio-Rad protein assay reagent (Bio-Rad) using BSA as the standard.
Homogenization and Preparation of the Microsome Fraction (Step 1)-All purification steps were carried out at 4°C. Frozen hog small intestine (127 g) was homogenized with a Waring blendor in 4 volumes of Buffer A. After centrifugation at 900 ϫ g for 10 min, the supernatant was pooled, and the pellet was subjected to two more extractions, after which all the supernatants were combined. After centrifugation at 78,000 ϫ g for 2 h, the microsomal fraction was obtained as a precipitate.
Solubilization of ␤1-6GnT (Step 2)-The microsomal fraction was suspended in 3 volumes of Buffer B, gently stirred for 2 h, and then centrifuged at 105,000 ϫ g for 1 h. The supernatant fraction was collected, and the residual pellet was subjected to another extraction followed by ultracentrifugation. The first and the second Triton extracts were combined and used for further enzyme purification.
CM-Sepharose FF Column Chromatography (Step 3)-Three volumes of Buffer C was added to the combined Triton X-100 extracts and then applied to the column of CM-Sepharose FF (5.0 ϫ 14.5 cm), which had been equilibrated with Buffer C. The column was washed with Buffer C until the protein concentration was reduced to below 0.05 mg/ml. The elution was carried out with a linear gradient between 850 ml of Buffer C and 850 ml of Buffer D. The fractions containing ␤1-6GnT activity were combined.
Ni 2ϩ -chelating Sepharose FF Column Chromatography (Step 4)-An equivalent volume of Buffer E was added to the pooled enzyme fraction TABLE I Structures of the acceptor and the product oligosaccharides Structures of the acceptor oligosaccharides and products obtained in reactions catalyzed by the ␤1-6GnT of hog small intestine. Denotation of monosaccharide residues is also shown. from the CM-Sepharose FF column chromatography step and then loaded on a column of Ni 2ϩ -chelating Sepharose FF (2.5 ϫ 8.2 cm), which had been equilibrated with Buffer E. Ni 2ϩ -chelating Sepharose resin was layered on the chelating Sepharose FF resin without metal ions (2.5 ϫ 4.1 cm), to avoid any possible leakage of Ni 2ϩ into the enzyme fractions. The ␤1-6GnT activity was completely bound to the column. After washing the column with Buffer E until the protein concentration was reduced to below 0.02 mg/ml, ␤1-6GnT activity was eluted with a linear gradient between 200 ml of Buffer E and 200 ml of Buffer F. The fractions that contained ␤1-6GnT activity were pooled, and the buffer in this fraction was replaced by Buffer G by means of an Amicon Diaflow Ultrafiltrater using a YM 30 membrane (Amicon, Beverly, MA) (the final volume of this concentrated fraction was 42 ml).
UDP-Hexanolamine-Agarose Affinity Column Chromatography (Step 5)-After the above step, the column was siliconized with Sigmacote (Sigma) and siliconized tubes (Assist, Tokyo, Japan) were used for fractionation. The UDP-hexanolamine-agarose column (1.5 ϫ 15 cm) had been equilibrated with Buffer G, and the concentrated enzyme fraction from the Ni 2ϩ -chelating Sepharose FF column chromatography was applied to the column, followed by washing with 90 ml of Buffer G, 90 ml of Buffer H, 90 ml of Buffer I, and 90 ml of Buffer J. The elution was performed using a linear gradient between 90 ml of Buffer J and 90 ml of Buffer K. The fractions containing ␤1-6GnT activity were pooled. This buffer in this fraction was replaced by Buffer L by means of a Centriprep-30 (Amicon) to deplete UDP and NaCl (the final volume of this concentrated fraction was 2 ml). This fraction was used for the enzyme characterization.
UDP-Hexanolamine-Sepharose 4B Affinity Column Chromatography (Step 6)-The concentrated enzyme fraction was applied to a UDPhexanolamine-Sepharose 4B column (1.0 ϫ 14.5 cm), which had been equilibrated with Buffer L, followed by washing with 10 ml of Buffer L and 30 ml of Buffer J. The elution was performed with a gradient of up to 20 mM UDP-GlcNAc in Buffer J. In total, 70 fractions (fraction volume 1.2 ml) were collected and analyzed by SDS-PAGE.
Gel Electrophoresis-SDS-PAGE was performed by the method of Laemmli (20) using 10% gels. Molecular weight markers (Amersham, Little Chalfont, United Kingdom) were used for size standards. Proteins in the gels were stained by a silver staining kit (2D silver stain II "DAIICHI"; Daiichi Pure Chemicals, Tokyo, Japan).
Acceptor Substrate Specificity of the Purified ␤1-6GnT-The activity of ␤1-6GnT was assayed in the standard assay mixture when pyridylaminated sugar substrates were used. When glycans 1, 2, and 3 (Table I) were used as the acceptor substrates, the enzyme reactions were carried out at 37°C for 48 h in a total volume of 25 l containing 100 nmol of acceptor oligosaccharide, 3.6 mol of UDP-GlcNAc, 200 mM MOPS (pH 7.0), 20 mM EDTA, 0.5 mM ATP, 2 mg/ml BSA, 0.5% Triton X-100, 8 mM NaN 3 , and the enzyme fraction (12.5 l). The reactions were terminated by heating at 100°C for 5 min, followed by passage through a mixed bed of Dowex AG1 (AcO Ϫ ) and Dowex AG50 (H ϩ ). The products were further purified by chromatography on a column of Superdex Peptide HR 10/30 column (10 ϫ 300 mm, equilibrated with water or 50 mM NH 4 HCO 3 ) with a flow rate of 0.5 or 1 ml/min or on a column of Bio-Gel P-4 (1 ϫ 145 cm, equilibrated with water, fraction volume 1.5 ml). The progress of the elution was monitored by UV absorbance at 205 nm.
NMR Spectroscopy-One-dimensional proton nuclear magnetic resonance ( 1 H NMR) spectra of the oligosaccharides were recorded at 23°C in D 2 O at 500 MHz on a Varian Unity 500 spectrometer as described by Leppä nen et al. (15).
MALDI-TOF MS-MALDI-TOF MS was performed in the positiveion delayed extraction mode with a BIFLEX TM mass spectrometer (Bruken-Franzen Analytik, Bremen, Germany), using a 337 nm nitrogen laser. 1 l of aqueous sample solution (10 pmol) and 1.5 l of 2,5-dihydroxybenzoic acid matrix (10 mg/ml in water) were mixed on the target plate and dried under a gentle stream of air. Maltohexaose and N-acetylgalactosaminodecamer were used for external calibration.

RESULTS
Purification of ␤1-6GnT-The activity of ␤1-6GnT was assayed using lacto-N-neotetraose-PA, basically according to the method of Gu et al. (13). BSA (2 mg/ml) was found to be effective in preserving enzyme activity at 37°C, especially for the highly purified enzyme fraction (Step 5 and Step 6). The activity of this fraction was 3.8-fold higher than that under the assay conditions in the absence of BSA, and as a result, BSA was routinely included in the standard assay mixture.
Hog small intestine was chosen as an enzyme source, based on the following two criteria. First, a survey of various hog and rat tissues revealed that intestine showed the highest enzyme activity per tissue protein (13). Second, hog intestine, which has levels of enzyme activity comparable with rat intestine, was more convenient in terms of cost and quantity.
Like other GnTs, ␤1-6GnT was concentrated in the microsome fraction. However, 20% of the ␤1-6GnT activity in the homogenate was found in the cytosolic fraction after ultracentrifugation, suggesting the occurrence of some proteolysis. ␤1-6GnT from the microsome fraction was solubilized by Triton X-100 more effectively at pH 6.5 than at pH 9.0. More than half of the ␤1-6GnT activity failed to bind to CM-Sepharose FF (Fig. 1A), and this flow-through fraction also failed to bind to CM-Sepharose FF (pH 6.0), Ni 2ϩ -chelating Sepharose FF, and Q-Sepharose FF (pH 9.0), none of which was found to be effective for purification. As a result, we used the NaCl eluted fraction for further experiments. The majority of proteins were separated from the ␤1-6GnT activity by the CM-Sepharose FF and Ni 2ϩ -chelating Sepharose FF column chromatography (Fig. 1B). After the Ni 2ϩ -chelating Sepharose FF column chromatography (Step 4), ␤1-6GnT activity was more stable than in Step 2 and Step 3. Different from GnT III and GnT IV (21,22), no ␤1-6GnT activity was detected after Cu 2ϩ -chelating column chromatography. The most effective step for purification was affinity column chromatography (Step 5, Fig. 1C). The majority of the enzyme activity bound to the UDP-hexanolamine-agarose column and was eluted very sharply by UDP (about 1 mM) as a competitive ligand. The addition of DTT (0.5 mM) was about 3.3 times more effective on the activity of the eluted enzyme fraction in Step 5 compared with that without DTT, while it was without effect on the enzyme activity in Step 4. The eluted enzyme fraction in Step 5 showed two bands with molecular weights of 76,000 and 60,000 on SDS-PAGE under nonreducing conditions, followed by silver staining (Fig. 2A). No activity of GnT III, IV, V, and iGnT (␤1-3GnT) was detected in this fraction. All experiments for enzyme characterization described below were performed using this enzyme fraction. Table II summarizes the purification of ␤1-6GnT, which was purified by 210,000-fold by Steps 1-5. The second affinity column chromatography (Step 6) was used to determine which band was ␤1-6GnT, although a complete separation of these two bands was not achieved. Two fractions (fraction 57 and 63), which contained almost the same enzyme activity (24 pmol/h/ ml), were analyzed by SDS-PAGE and stained with silver (Fig.  2B). Only the upper band (76 kDa) showed the same intensity between the two fractions, indicating that the 76 kDa band was ␤1-6GnT.
pH Optimum-The activity of ␤1-6GnT was highest between pH 6.5 and 7.5 and displayed an optimum at pH 7.0.
Effect of UDP-GlcNAc Analogues-To investigate which portion of UDP-GlcNAc was important for enzyme activity, several UDP-GlcNAc analogues (2 mM) were added to the assay mixture containing 0.5 mM UDP-GlcNAc. These results are shown in Table III. UDP and UTP were the most potent inhibitors. The uracil moiety appeared to be essential for the enzyme activity, as evidenced by a comparison of UMP and AMP and of UDP and GDP. The comparison of UMP, UDP, and UTP suggests that the number of phosphodiesters might be important for inhibition. Sugar nucleotides having a UDP moiety also inhibited the enzyme activity, including UDP-hexanolamine which was used as an affinity ligand in Steps 5 and 6.
Acceptor Substrate and Site Specificity-Acceptor substrate and site specificity of the ␤1-6GnT was examined by using a set of oligo-N-acetyllactosamines as shown in Table I.
The Branching Reaction of Tetrasaccharide 1-Tetrasaccharide 1 (100 nmol) and UDP-GlcNAc were incubated together with the purified ␤1-6GnT. The MALDI-TOF mass spectrum of the purified reaction mixture showed sodiated molecular ions of the Gal 2 GlcNAc 3 -product (58%) and the Gal 2 GlcNAc 2acceptor (42%) (Fig. 3A). The pentasaccharide product was separated from the acceptor by gel permeation chromatography on a Superdex Peptide HR 10/30 column. The purity of the isolated pentasaccharide fraction was 95% as estimated by MALDI-TOF MS. Endo-␤-galactosidase treatment failed to cleave the product; the reaction mixture contained only the uncleaved pentasaccharide as assessed by MALDI-TOF MS, indicating that the newly linked GlcNAc residue must be attached to the midchain galactosyl residue rather than to the terminal galactose (23). The one-dimensional 1 H NMR spectrum of the pentasaccharide product (Fig. 4A, Table IV) is identical with the spectrum previously reported for the authentic pentasaccharide 4 (24). In particular, a doublet with the typical chemical shift and coupling constant (4.584 ppm, 8.3 Hz) of the ␤1,6-linked GlcNAc H1 is present in the spectrum of the pentasaccharide product, and the chemical shifts of Gal2 H1 and Gal2 H4 are characteristic to a 3,6-disubstituted Gal, while Gal4 H1 has a chemical shift typical for a terminal galactosyl residue. If the branch-forming GlcNAc had been transferred to C-2 or C-4 of Gal2 of tetrasaccharide 1, the H1 resonances of the new GlcNAc unit and the other reporter group signals of the pentasaccharide product would have been distinctly different from those of pentasaccharide 4 (25). Taken together, these data establish that the pentasaccharide formed from tetrasaccharide 1 by ␤1-6GnT is the branched pentasaccharide 4.
The Branching Reaction of Pentasaccharide 2-The linear pentasaccharide 2 (300 nmol in 100-nmol lots) and UDP-Glc-NAc were incubated together with the purified ␤1-6GnT. The MALDI-TOF mass spectrum of the purified reaction mixture revealed that 29% of 2 had been converted to the hexasaccharide Gal 2 GlcNAc 4 (Fig. 3B). The hexasaccharide product was separated from the acceptor by gel permeation chromatography on a Bio-Gel P-4 column. The purity of the isolated hexasaccharide fraction was 94% as assessed by MALDI-TOF MS. Endo-␤-galactosidase treatment completely cleaved the hexasaccharide product, releasing the tetrasaccharide Gal 1 GlcNAc 3 (obs. (M ϩ Na) ϩ m/z 812.4 (calc. 812.3)) (Fig. 5). The peak of GlcNAcGal, the other cleavage product, was also detected among the matrix signals (obs. (M ϩ Na) ϩ m/z 406.0; calc. 406.1). These data showed that pentasaccharide 2 had been partially converted to glycan 5 by the intestinal enzyme. No signal was detected in the MALDI-TOF spectrum of Fig. 5 at m/z 974.3, suggesting that glycan 2 had not been branched at Gal4. The branching at this site would have generated the hexasaccharide GlcNAc␤1-3Ј(GlcNAc␤1-6Ј)LacNAc␤1-3Ј-LacNAc, which, in turn, had given rise to the pentasaccharide GlcNAc␤1-3Ј(GlcNAc␤1-6Ј)LacNAc␤1-3Gal in the endo-␤-galactosidase experiment and a sodiated molecular ion at m/z 974.3 in the ensuing MALDI-TOF experiment. The 1 H NMR spectrum of the hexasaccharide formed from glycan 2 confirms that the hexasaccharide represents glycan 5 (Fig. 4B, Table  IV). The doublet at 4.584 ppm arises from GlcNAc H1, which is ␤1,6-linked to Gal2 and not from a similar residue linked to Gal4 (14,15). The GlcNAc H1 resonance for the isomeric hexasaccharide GlcNAc␤1-3(GlcNAc␤1-6)LacNAc␤1-3LacNAc at 4.592 ppm (14) was not detected in the spectrum of the hexasaccharide product. These results establish that the structure of the hexasaccharide formed from pentasaccharide 2 by ␤1-6GnT treatment is the branched hexasaccharide 5.
In the 1 H NMR spectrum of the heptasaccharide fraction, the partially overlapping doublets at 4.584 ppm and at 4.592 ppm can be seen (Fig. 4C, Table IV). These signals correspond to the ␤1,6-bonded GlcNAc H1 resonances reported for pure heptasaccharides 6 and 7, respectively (15), indicating that both isomers are present also in the heptasaccharide pool in these experiments. The integrals of these peaks suggest that the heptasaccharide mixture contains 40% of 6 and 60% of 7, which is in reasonable agreement with the results obtained from endo-␤galactosidase cleavage of the heptasaccharide fraction (see above).
In this study, we have purified the cIGnT6 210,000-fold from hog small intestine by successive column chromatographies, using an assay where pyridylaminated lacto-N-neotetraose (Gal␤1-4GlcNAc␤1-3Gal␤1-4Glc-PA) was used as an acceptor substrate and the reaction product was Gal␤1-4GlcNAc␤1-3(GlcNAc␤1-6)Gal␤1-4Glc-PA. The purified enzyme has an apparent molecular weight of 76,000 by SDS-PAGE under nonreducing conditions ( Fig. 2A). The purified cIGnT6 has no metal ion requirement for full activity, which is a property shared by all the other ␤1-6GnTs. The optimum pH is at 7.0, which is the same value as has been reported for rat intestine cIGnT6 (13). The analysis of acceptor specificity indicates that the enzyme has unique properties. The data which shows that GlcNAc␤1-3Gal␤1-4Glc-PA and Gal␤1-3GlcNAc␤1-3Gal␤1-4Glc-PA could not be the acceptor compound revealed that this enzyme was devoid of dIGnT6 activity and that Gal␤1-4 moiety, not Gal␤1-3 linkage at the nonreducing terminal should be necessary for its cIGnT6 activity. A more precise analysis of the acceptor specificity of this enzyme was performed with tetrasaccharide 1, pentasaccharide 2, and hexasaccharide 3. The fact that tetrasaccharide 1 was converted into glycan 4 showed that this IGnT6 was cIGnT6, consistent with the data using a pyridylaminated substrate. The pentasaccharide 2 was converted in a site-specific reaction to glycan 5 by cIGnT6 and not to the isomeric saccharide GlcNAc␤1-3Ј(GlcNAc␤1-6Ј)-LacNAc␤1-3ЈLacNAc. The structure of glycan 5 was established by endo-␤-galactosidase treatment, which cleaved the product into the disaccharide GlcNAc␤1-3Gal and the tetrasaccharide GlcNAc␤1-3(GlcNAc␤1-6)Gal␤1-4GlcNAc. The 1 H NMR spectrum of glycan 5 confirmed its structure by showing a GlcNAc H1 resonance that is characteristic of poly-Nacetyllactosamines that contain a branch at the reducing-end LacNAc unit. The completely site-specific action of hog small intestinal cIGnT6 with the pentasaccharide acceptor 2 is identical to the action of cIGnT6 activities of human serum (17) and rat serum (15) as well as serum of other mammalian species (15). We note that the purified cIGnT6 from hog intestine has no dIGnT6 activity, which should convert pentasaccharide 2 into the isomeric saccharide GlcNAc␤1-3Ј(GlcNAc␤1-6Ј)LacNAc␤1-3ЈLacNAc. This was also confirmed in an experiment where incubation of UDP-GlcNAc donor and trisaccharide acceptor GlcNAc␤1-3Gal␤1-4GlcNAc with this enzyme failed to generate detectable amounts of the expected dIGnT6 product, GlcNAc␤1-3(GlcNAc␤1-6)Gal␤1-4GlcNAc (data not shown). It is interesting that the membrane-bound intestinal cIGnT6 reacted more efficiently with glycan 1 than glycan 2. This is also true for the soluble human and rat serum enzymes (15). The reaction with hexasaccharide 3 led to the production of two isomeric heptasaccharides 6 and 7 in nearly the same quantity and finally also to the doubly branched octasaccharide 8. This clearly shows that this cIGnT6 is capable of reacting with both midchain Gal positions of the linear hexasaccahride 3 without specificity.  A human IGnT6 has been cloned by expression cloning using an antibody against I antigen (32). The cDNA for this enzyme encodes the type II transmembrane protein with 400 amino acids, but it is not known whether it represents dIGnT6 or cIGnT6 (33,13); both have been used to generate multiply branched poly-N-acetyllactosamine structure. Quite recent experiments of ours suggest that this human IGnT6 may represent a cIGnT6 (31). Our current interest is whether the cIGnT6 purified from hog small intestine is the same enzyme as human cIGnT6. Protein sequencing and cloning of this hog cIGnT6 is now currently ongoing in our laboratory.