Porcine submaxillary gland GDP-L-fucose: beta-D-galactoside alpha-2-L-fucosyltransferase is likely a counterpart of the human Secretor gene-encoded blood group transferase.

Partial amino acid sequence of GDP-L-fucose:β-D-galactoside α-2-L-fucosyltransferase purified from porcine submaxillary glands was determined. Amino acid sequence analysis yielded 100, 93.3, and 84.2%, and 75, 46.6, and 84.2% sequence identity between 12-, 15-, and 19-amino acid tryptic peptides generated from porcine enzyme and amino acid residues 61-72, 111-125, and 308-326 and 89-100, 139-153, and 338-356 of the human Secretor and H type α-2-fucosyltransferases, respectively. Higher amino acid sequence homology of the porcine enzyme with the predicted sequence for the human Secretor locus as compared with H gene-encoded blood group β-D-galactoside α-2-L-fucosyltransferase suggests that porcine α-2-fucosyltransferase highly corresponds to the human Secretor gene-encoded enzyme.

The minimum structure of the H antigenic determinant is a terminal Fuc␣132Gal. 1 Two distinct GDP-L-fucose:␤-D-galactoside ␣-2-L-fucosyltransferases, encoded by the H and Secretor (Se) 2 genes, are known to control the biosynthesis of the monoand difucosylated lactoseries structures in normal tissues and in epithelial carcinomas in humans (1), although evidence from normal and tumor tissues suggests that the human genome contains additional structural genes encoding ␣-2-fucosyltransferases (2)(3)(4)(5). A two-loci model for expression of ␣-2-fucosyltransferase was proposed based on the presence of ␣-2-fucosyltransferase enzymatic activity in the serum of Bombay, Sepositive individuals, who lack H gene expression (6). In this model, both Se and H represent structural genes encoding ␣-2-fucosyltransferases, and the enzymes under the control of the Se and H genes could preferentially use type 1, 2, 3, and 4 precursors or type 2 structures and are expressed in cells of endodermal and mesodermal origin, respectively (7). The twoloci model for expression of human ␣-2-fucosyltransferases was firmly established by cloning H (8) and a candidate for human Secretor locus-encoded ␣-2-fucosyltransferases (9). The Se gene-encoded enzyme shares 68% identity with the human H ␣-2-fucosyltransferase protein sequence, which confirms the hypothesis that H and Secretor loci represent two distinct but closely linked ␣-2-fucosyltransferase genes.
Evolutionary evidence has suggested that the Se gene is ancestral in mammals, with the evolutionarily newer H gene present only in humans and anthropoid apes. Indeed, ABO blood group antigens are present in red blood cells, vascular endothelium, and secretions in man and anthropoid apes but only in secretions in lower mammals (10).
The evidence that the ␤-D-galactoside ␣-2-fucosyltransferase from porcine submaxillary glands is equivalent to the human serum Se-type and not H-type enzyme is based on striking similarities with the Se enzyme with respect to the kinetic pattern, specificity toward various oligosacchride acceptors and physicochemical properties. Both, the porcine and human Se ␣-2-fucosyltransferase display the preference for type 1 and 3 lactoseries oligosaccharide acceptors, are represented by 55-kDa glycoproteins, and demonstrate binding properties to S-Sepharose (11)(12)(13)(14)(15)(16). We postulated that porcine ␣-2-fucosyltransferase is equivalent to the human Se gene-encoded enzyme and different from the H blood group ␣-2fucosyltransferase.
Tryptic peptides generated from the porcine submaxillary gland ␤-D-galactoside ␣-2-fucosyltransferase, share higher amino acid sequence homology with human Se blood group enzyme as compared with human H gene-encoded enzyme and results confirm the hypothesis that porcine submaxillary gland enzyme is equivalent to the human Se ␣-2-fucosyltransferase.

Protein Determination and SDS-PAGE
Protein was determined, and electrophoresis was performed as described previously (13,14).

Synthesis of GDP-Hexanolamine-Sepharose
GDP-hexanolamine and GDP-fucose were synthesized as described previously (17). GDP-hexanolamine-Sepharose (250 ml) was synthesized as described previously (18) with the following modifications. CNBr-activated Sepharose 4B was freshly prepared using a solution of CNBr in dioxane (1 g/ml) instead of powdered CNBr, and the pH of the reaction was brought to 11.0 instead of 11.5-12.0 with 1.5 M LiOH. The temperature of the reaction was kept at 2-4°C. The ligand concentra-* This work was supported by Grants AI28679 (to J. T.) and CA45363 (to M. B.-T.) from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Purification of Porcine Submaxillary Gland ␤-D-Galactoside ␣-2-L-Fucosyltransferase
Purification of porcine submaxillary gland ␣-2-fucosyltransferase was based on a previously published purification scheme (15). Therefore, only steps representing alterations to the original method are described in detail. Full details of the purification procedure could be obtained directly from the authors.
Step 8: High Pressure Size-Exclusion Liquid Chromatography (HPLC)-HPLC separation was performed on a HPLC LKB 2150 apparatus with a TSKG3000 SW size-exclusion column (Pharmacia) and with an injection loop (100 l) (13,14). The HPLC column was preequilibrated with 25 mM sodium cacodylate buffer, pH 7.0, containing 0.035 M NaCl. Pooled fractions eluted from the GDP-hexanolamine column III were concentrated (10-fold) using Speed-Vac, and a sample (100 l) was subjected to HPLC size-exclusion chromatography at a flow rate 0.5 ml/min. Fractions (0.3 ml) were collected every 36 s. Enzymatic activity was monitored by standard assay using phenyl-␤-D-galactoside as an acceptor; protein and GDP were monitored with a variable wavelength monitor LKB 2141 at 223 nm.

Amino Acid Sequence Determination
The 55-kDa enzymatic protein (180 pmol), purified by HPLC sizeexclusion chromatography, was submitted to the Wistar Institute Mass Spectrometry/Protein Microchemistry Laboratory for amino acid sequence analysis. Protein sample (50 g) eluted from the GDP-hexanolamine-Sepharose III affinity chromatography column containing 60-, 55-, and 18-kDa proteins was submitted to determine the amino acid sequence of the 60-kDa protein. Proteins were separated by SDS-PAGE on 5-15% gradient gel and electrotransferred to a polyvinylidine diflu-oride membrane. The 55-and 60-kDa individual proteins were excised from the membrane following SDS-PAGE of fractions obtained from HPLC and GDP-hexanolamine-Sepharose III columns, respectively. Tryptic peptides generated from the both isolated proteins were separated by reverse-phase chromatography. The molecular mass of the selected fractions was established using mass spectrometry, and peptides were subjected to automated amino acid sequence analysis using an Applied Biosystems gas/liquid phase sequencer equipped with a Nelson analytical data analysis system.

RESULTS
The partial amino acid sequence of the ␤-D-galactoside ␣-2fucosyltransferase from porcine submaxillary glands, which was purified to homogeneity, was determined.
Purification ␤-D-Galactoside ␣-2-L-Fucosyltransferase-A purification procedure was established based on a previously described method, with effective adsorption of porcine ␣-2fucosyltransferase on a S-Sepharose cation-exchange matrix at pH 6.0 (15). Table I summarizes the purification results of ␤-D-galactoside ␣-2-fucosyltransferase from porcine submaxillary glands.
Pooled active fractions eluted from S-Sepharose were subjected to three consecutive steps of affinity chromatography on GDP-hexanolamine-Sepharose columns. GDP-hexanolamine-Sepharose was synthesized, and a matrix with the same ligand concentration of 5.4 -6.6 mol/ml was used for all steps of affinity purification.
A sharp peak of enzymatic activity was eluted from the GDP-hexanolamine-Sepharose affinity column I at 0.6 -0.8 M NaCl using a 0.4 -2.0 M linear NaCl gradient. The active, pooled fractions from the GDP-hexanolamine-Sepharose-I column were desalted on a Sephadex G-50 (fine). The enzyme was recovered quantitatively (97%), and some degree of purification (1.4-fold) in addition to the desalting effect of the Sephadex G-50 gel filtration column was achieved.
A second large GDP-hexanolamine-Sepharose column and an NaCl pulse for nonspecific elution was introduced into the procedure to combine and concentrate fractions from four batches of approximately 5 kg of submaxillary gland tissue processed separately on GDP-hexanolamine-Sepharose column I into the total pool from 21.5 kg of tissue. To minimize nonspecific interactions with the gel matrix, cacodylate buffer with higher pH 7.0 than that for S-Sepharose and GDP-hexanolamine-Sepharose I columns (pH 6.0) was chosen for GDPhexanolamine-Sepharose column II. With the 2 M NaCl pulse, 60% of the ␣-2-fucosyltransferase eluted from this column resulted in 10-fold purification, as compared with nonspecific elution with NaCl gradient from GDP-hexanolamine-Sepharose column I, from which enzyme was purified 26-fold with a smaller loss of enzymatic activity (76%). This may be due to the low concentration of protein eluting from GDP-hexanolamine-  Sepharose II as compared with column I. Again, the active fractions eluted with 2 M NaCl were desalted using a Sephadex G-50 (fine) column, which removes approximately 50% of the total protein at this point without loss of enzymatic activity. Desalted active fractions were directly applied to a third affinity purification step, which consisted of small GDP-hexanolamine-Sepharose III column. The enzyme was specifically eluted using 0.5 mM GDP in 25 mM sodium cacodylate buffer (pH 7.0) containing 0.035 M NaCl. A 14-fold purification and considerable loss of enzymatic activity (60%) due to a low levels of protein occurred at this step (Table I). SDS-PAGE analysis of the pooled fractions eluted from the third affinity column upon reduction with ␤-mercaptoethenol showed major protein bands of 60 and 55 kDa and a band at 18 kDa (not shown). This fraction was submitted for tryptic digestion and amino acid sequence analysis of the major 60-kDa protein at the Wistar Institute Mass Spectrometry/Protein Microchemistry Laboratory. Amino acid sequencing of two tryptic peptides generated from this protein band demonstrated 98 and 94% homology with bovine catalase, suggesting that the 60-kDa protein purified by affinity chromatography on GDP-hexanolamine-Sepharose III represents catalase (Table II).
Final purification of the porcine ␣-2-fucosyltransferase was achieved by HPLC size-exclusion chromatography. The pooled enzymatic fractions eluted from GDP-hexanolamine column III were subjected to HPLC size-exclusion chromatography in 25 mM sodium cacodylate buffer (pH 7.0) containing 0.035 M NaCl. Enzymatic activity was monitored by standard assay using phenyl-␤-D-galactoside as an acceptor and absorption at 223 nm (Fig. 1). The HPLC size-exclusion enzymatic activity profile showed a broad peak corresponding to molecular size of 55 kDa. SDS-PAGE analysis of the active fractions in the presence of ␤-mercaptoethanol revealed the 55-kDa protein representing ␣-2-fucosyltransferase (Fig. 1, inset).
Amino Acid Sequence Determination of ␤-D-Galactoside ␣-2-L-Fucosyltransferase-The HPLC-purified protein of 55 kDa representing porcine ␤-D-galactoside ␣-2-fucosyltransferase was submitted for tryptic digestion and amino acid sequencing. Table II lists the results of mass spectrometry and amino acid sequence analysis of tryptic peptides generated from the purified enzyme. 100, 93.3, and 84.2% and 75, 46.6, and 84.2% sequence identity in 12-, 15-, and 19-amino acid overlaps were found between the sequences of the peptides from porcine enzyme with amino acid residues 61-72, 111-125, and 308 -326 and 89 -100, 139 -153 and 338 -356 of the recently published primary structure of human candidate Se (9) and human H type ␣-2-fucosyltransferases, respectively (8). DISCUSSION We have purified a ␤-D-galactoside ␣-2-fucosyltransferase to homogeneity from porcine submaxillary glands, using a modification of a previously published procedure (15). Our procedure included additional GDP-hexanolamine-Sepharose affin-ity chromatography and a 2 M NaCl pulse elution, and HPLC size-exclusion steps instead of Sephadex G-150 gel filtration and chromatography on SP-Sephadex steps of the original procedure.
Three proteins, 60, 55, and 18 kDa, were eluted from the third GDP-hexanolamine-Sepharose affinity column by specific elution using 5 mM GDP. Based on a previous analysis suggesting that a 60-and 55-kDa doublet represented ␤-D-galactoside ␣-2-L-fucosyltransferase purified from porcine submaxillary glands (15), we analyzed the amino acid sequence of the 60-kDa protein isolated after the last affinity chromatography step. Amino acid sequence analysis of two tryptic peptides of this protein revealed high sequence homology with bovine catalase, consistent with the previous observation that catalase for undetermined reasons shows high binding capacity to GDP-hexanolamine-Sepharose and copurifies with ␣-2-fucosyltransferase from the affinity column (15). Therefore, to further purify the ␣-2-fucosyltransferase protein, HPLC size-exclusion chromatography was used as a purification step following the affinity chromatography steps. The HPLC step resulted in the isolation of a single major protein of 55 kDa, which coeluted with enzymatic activity, strongly suggesting that the 55-kDa  protein represents porcine submaxillary gland ␤-D-galactoside ␣-2-fucosyltransferase. In this purification scheme, catalase was separated from fucosyltransferase as a tetrameric protein in native conditions by the HPLC size-exclusion step. The isolated porcine enzyme as determined by amino acid sequence analysis is highly homologous to human Se and H type blood group ␤-D-galactoside ␣-2-fucosyltransferases (8,9). As expected, a higher level of homology was observed between the three tryptic peptides generated from the porcine enzyme and a recently cloned human candidate for Se type ␣-2-fucosyltransferase (100, 93.3, and 84.2%) as compared with the human H gene-encoded enzyme (75, 46.6, and 84.2%). The amino acid sequence of the HP 94266 peptide is identical with the Se type ␣-2-fucosyltransferase, whereas 75% of homology was observed with respective sequences of H enzyme. Thus, the 75% homology found between the porcine and H type ␣-2-fucosyltransferases also reflects the level of homology between both human Se and H enzymes. Porcine peptide HP 94278 shares the same degree of homology (84.2%) with both the Se and H type ␣-2-fucosyltransferases. The amino acid substitutions are in exactly the same positions in both human and porcine ␣-2fucosyltransferases, although they are represented by different amino acids in all three enzymes (9). The most striking differences between the protein sequence of porcine and human H enzyme were observed in the amino acid sequence corresponding to HP 94244 peptide, where only 46.6% identity was observed. On the other hand, this peptide shares 93.3% identity with the corresponding amino acid sequence of Se ␣-2-fucosyltransferase. These results support the hypothesis that porcine ␤-D-galactoside ␣-2-fucosyltransferase is equivalent to the human Se type enzyme and different from the human H blood group fucosyltransferase. Very high sequence homology between the peptides from the porcine enzyme and recently cloned rabbit ␣-2-fucosyltransferase, RFT-II (19), which likely corresponds to rat FTB enzyme (20), also suggests that they represent equivalent enzymes.
Porcine peptide HP 94244, which shares the lowest degree of sequence homology with the human H enzyme, is derived from the catalytic domain region where the least sequence identity between corresponding amino acid sequences of Se and H type enzymes was determined (9). These results suggest that this region may determine the oligosaccharide substrate specificity, in particular this region may be involved in ␤-D-galactose binding since H and Se enzymes greatly differ in affinity to ␤-Dgalactose and its derivatives, as determined by K m values (11)(12)(13)(14). Nine amino acid sequences similar to those of HP 94244 were also found in bacterial phospho-␤-and ␤-galactosidases, sharing 55.6 and 62.5% homology with the peptide (21)(22)(23).
The availability of the DNA primary structure of the human H and Se gene-encoded ␤-D-galactoside ␣-2-fucosyltransferases (9, 10) and their animal counterparts (19,20) will enable study of the molecular basis of ␣-2-fucosylated glycoconjugates expression in different species.