Lewis X Biosynthesis in Helicobacter pylori

The lipopolysaccharide of certain strains ofHelicobacter pylori was recently shown to contain the Lewis X (Lex) trisaccharide (Galβ-1,4-(Fucα(1,3))-GlcNAc). Lex is an oncofetal antigen which appears on human gastric epithelium, and its mimicry by carbohydrate structures on the surface of H. pylori may play an important part in the interaction of this pathogen with its host. Potential roles for bacterial Lex in mucosal adhesion, immune evasion, and autoantibody induction have been proposed (Moran, A. P., Prendergast, M. M., and Appelmelk, B. J. (1996) FEMS Immunol. Med. Microbiol. 16, 105–115). In mammals, the final step of Lex biosynthesis is the α(1,3)-fucosylation of GlcNAc in a terminal Galβ(1→4)GlcNAc unit, and a corresponding GDP-fucose:N-acetylglucosaminyl α(1,3) fucosyltransferase (α(1,3)-Fuc-T) activity was recently discovered in H. pylori extracts. We used part of a human α(1,3)-Fuc-T amino acid sequence to search an H. pylori genomic data base for related sequences. Using a probe based upon weakly matching data base sequences, we retrieved clones from a plasmid library of H. pylori DNA. DNA sequence analysis of the library clones revealed a gene which we have named fucT, encoding a protein with localized homology to the human α(1,3)-Fuc-Ts. We have demonstrated that fucT encodes an active Fuc-T enzyme by expressing the gene in Escherichia coli. The recombinant enzyme shows a strong preference for type 2 (e.g. LacNAc) over type 1 (e.g. lacto-N-biose) acceptors in vitro. Certain residues in a short segment of the H. pylori protein are completely conserved throughout the α(1,3)-Fuc-T family, defining an α(1,3)-Fuc-T motif which may be of use in identifying new fucosyltransferase genes.

The Gram-negative bacterium Helicobacter pylori is a major cause of chronic gastritis and peptic and duodenal ulcers (1)(2)(3)(4)(5). It has also been implicated in gastric adenocarcinoma (6 -9) and gastric lymphoma (10), leading to its classification as a type I human carcinogen (11). H. pylori is a chronic pathogen, and the means by which this organism is able to persist in the stomach and resist or evade destruction by the immune system is central to its involvement in disease. Some aspects of the host-pathogen interaction have been resolved, including the involvement of the Lewis b (Le b ) 1 epitope on epithelial cells in attachment of H. pylori (12), and characterization of a bacterial cytotoxin responsible for gastric epithelial damage (for a review see Ref. 13), but clearly much remains to be discovered.
Recent structural analysis of H. pylori lipopolysaccharides revealed that the O antigen contains fucosylated carbohydrate structures identical to the mammalian Lewis X (Le x ) and Lewis Y (Le y ) epitopes (14 -17). It was further established that the bacterium contains endogenous galactosyltransferase (Gal-T) and fucosyltransferase (Fuc-T) activities necessary for biosynthesis of these structures (18) suggesting that they are synthesized de novo by H. pylori rather than scavenged from the surface of mammalian cells. Le x is an oncofetal antigen (19,20) also expressed on adult human gastric mucosa (21), and its presence on H. pylori lipopolysaccharides may play a role in survival and pathogenesis. H. pylori infection is known to induce antibodies that cross-react with human gastric mucosa (22). In a recent report, Appelmelk et al. (23) demonstrated that the targets of this autoimmune response include Le x and/or Le y epitopes and provided evidence that anti-Le x/y antibodies may be involved in H. pylori-associated gastritis. Interestingly, molecular mimicry of Le x is also thought to be responsible for autoantibody production by Schistosoma mansoni (24,25). In addition, surface carbohydrate antigens containing Le x structures may play a part in the immunopathology of H. pylori infection by promoting Th-1 to Th-2 switching as has been reported in schistosomal infections (26). Two recent reports (27,28) that over 85% of H. pylori isolates from geographically widespread locations express Le x and/or Le y antigens would also seem to imply selective pressure for maintenance of these structures, given the considerable structural variability often shown by lipopolysaccharides from Gram-negative bacteria.
In mammals, the defining step of Le x biosynthesis is fucosylation of a type 2 core structure (Gal␤134GlcNAc). This reaction is catalyzed in humans by one or more members of a family of ␣(1,3)-fucosyltransferases which employ GDP-fucose as an activated sugar donor (29 -38). Fuc-T and Gal-T activities have been detected in H. pylori extracts (18), but although the order of sugar transfer appears to follow the same course as in mammalian systems, with galactosylation preceding fucosylation, little is known about the bacterial Fuc-T and how it is related to the mammalian transferases. If, as evidence is be-* 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  ginning to suggest, cell-surface Le x/y epitopes play an important role in H. pylori persistence and pathogenesis (23,39), the ␣(1,3)-Fuc-T may offer a nonbactericidal therapeutic target for eradication of H. pylori without otherwise disturbing the balance of gut fauna.
Hybridization to the H. pylori Plasmid Library-A 363-bp sequence fragment from the H. pylori genomic data base containing the Fuc-T homology region was amplified from H. pylori NCTC 11637 genomic DNA by PCR using the primers HPFT1 (5Ј-CTT TGA AAA GAG GGT TTG CCA) and HPFT2 (5Ј-CAA GTA TCT CAC GTA ATC AAT). Amplified product was purified using the QiaQuick PCR purification system (Qiagen) following the manufacturer's instructions. Approximately 0.5 g of the purified fragment was used to prepare DIG-labeled probe using the DIG Hi-prime labeling kit (Boehringer Mannheim). Probe was used without further purification.
Library Clone Retrieval and Sequence Analysis-Clones which hybridized strongly to the probe were retrieved from 384-well library storage microtiter plates (stored at Ϫ80°C) and grown overnight in 2 ml of L broth containing 100 g/ml ampicillin. DNA was prepared by the rapid alkaline lysis method (46). Larger quantities of plasmid DNA were obtained from 1-200-ml cultures using the Qiagen plasmid (maxi) system. DNA sequencing reactions performed using AmpliTaq FS with dye terminators (Perkin-Elmer) and run on an Applied Biosystems ABI 373 automated sequencer. Sequence analysis was performed using GCG 8.1 (47) and BLAST (48) software. Sequence alignments were created with Pileup (part of the GCG package) with a gap penalty of 5.0 and gap extension penalty of 0.5.
Subcloning of fucT into an E. coli Expression Vector-A 1.4-kb DNA fragment containing the fucT gene was amplified from H. pylori NCTC11637 genomic DNA by PCR using the primers HPFT3 (5Ј-GAG TGT CTA ATG GGA TCC TTA TTT TTT AAC CCA CCT) and HPFT5 (5Ј-TAG CCC TAA TCA AGC CTT TG). PCR product was purified using the QiaQuick PCR purification system (Qiagen), ligated to the A/T cloning vector pCR TM II (Invitrogen) and introduced into Escherichia coli XL-1 Blue (Stratagene) by electroporation. Recombinant (white) clones were mapped with BssHII to identify plasmids containing the cloned fragment in the desired orientation (direction of transcription of fucT in the same direction as lacZ). A 1.4-kb fragment containing the fucT gene was excised from a suitable clone with BamHI and ligated to pET-11a vector DNA (Novagen) which had been linearized with BamHI and dephosphorylated using shrimp alkaline phosphatase (Amersham, Little Chalfont, UK). Ligated DNA was introduced into E. coli BL21(DE3), and transformants were selected on L agar containing ampicillin (100 g/ml). Recombinant clones were identified by restriction mapping with BssHII.
Preparation of Cell Extracts-Recombinant clones were grown overnight at 37°C from single colony inocula in 5 ml of L broth containing 100 g/ml ampicillin. 20 ml of fresh L broth was inoculated with 200 l of the overnight culture and incubated at 37°C. E. coli XL-1 Blue containing library plasmids were grown for 5-8 h prior to harvesting. E. coli BL21(DE3) containing the expression plasmid pHPFT was incubated until an A 600 of 0.4 -0.6 was attained. Isopropyl-1-thio-␤-D-galactopyranoside was added to a final concentration of 0.5 mM, and incubation continued for an additional 3 h. Bacteria were harvested by centrifugation (4000 ϫ g, 15 min, 4°C) and washed in phosphatebuffered saline. Pelleted bacteria were resuspended in 0.5 ml of chilled solubilization buffer (0.1% (w/v) Triton X-100, 0.1 M NaCl, 25% (w/v) glycerol, 0.1 M NaCl, 2 mM dithiothreitol, 50 mM Tris-HCl, pH 7.0) and sonicated on ice for 4 ϫ 15 s bursts (MSE Soniprep 150), with a 60-s cooling period on ice between bursts. Sonicate was cleared by centrifugation (20,000 ϫ g, 30 min, 4°C) and stored at Ϫ80°C.
Fucosyltransferase Assays-Fucosyltransferase activity was measured by a modification (49) of the method of Prieels et al. (50). Briefly, 12.5 l of cell extract was incubated with 20 M GDP-fucose, 100,000 cpm of GDP-[ 3 H]fucose, 5 mM acceptor, 5 mM MnCl 2 , 1 mM ATP, buffered to pH 7.2 with 50 mM HEPES-NaOH in a total volume of 50 l for 1 h at 37°C. Sensitivity of the H. pylori enzyme to the inhibitor Nethylmaleimide (NEM) was assessed by including NEM at final concentrations of up to 15 mM in Fuc-T assay reactions. Reactions were stopped by addition of 1 ml of mixed bed resin slurry (AG 1-X8 (Cl Ϫ form) 1:4 (w/v) in water), vortexed briefly, and centrifuged for 5 min at 20,000 ϫ g at room temperature. Radioactivity in 600 l of supernatant was measured by scintillation counting. Allowance was made for nonspecific breakdown of GDP-fucose and fucose transfer to endogenous acceptors by performing control reactions in the absence of acceptor. Assay reactions were performed in duplicate. K m for the acceptor Nacetyllactosamine (LacNAc) was determined by measuring reaction rates with 0 -100 mM LacNAc and 200 M GDP-fucose, while the donor K m was obtained using 0 -100 M GDP-fucose and 5 mM LacNAc.
Purification of Fucosylated Oligosaccharides Generated using H. pylori Fuc-T-0.5 ml of H. pylori Fuc-T (approximately 0.2 milliunit) was incubated with 5 mM acceptor (LacNAc or LNT), 3 mM GDP-fucose, 800,000 cpm of GDP-[ 3 H] fucose, 0.1% (w/v) BSA, 2 l (2 units) of shrimp alkaline phosphatase, 5 mM MnCl 2 , buffered to pH 7.2 with 50 mM HEPES-NaOH in 1 ml total volume for 40 h at 37°C. Incubation mixtures were passed through a 2.5-ml Dowex AG 1-X8 ion exchange column, washed through with 7.5 ml of water. Column eluant and washings were pooled, evaporated to dryness, and redissolved in 0.5 ml of water. Samples from the Dowex column were applied to a Bio-Gel P2 gel filtration column (20 ϫ 1 cm) eluted with water at 20 ml/h. 1-ml fractions were collected, and 5-l aliquots were removed for liquid scintillation counting. Radioactive fractions were pooled and lyophilized, keeping discrete eluant peaks separate. To remove residual Triton X-100, products were dissolved in 500 l of water and applied to disposable 100-mg Amprep C18 reverse phase columns (Amersham), washed through with 3 ml of water. Pooled eluant and washings were lyophilized and redissolved in 100 l of water. Finally, fucosylated products were purified by HPLC on a Neutropak NH 2 column in water/ acetonitrile (25:75 by volume for products derived from LacNAc, 30:70 by volume for those produced from LNT). 0.5-ml fractions were collected, and radioactive fractions were pooled, lyophilized, and redissolved in 100 l of water.
Glycosidase reaction products were analyzed by HPLC on the Neutropak NH 2 column in water/acetonitrile (25:75 by volume for products derived from LacNAc, 30:70 by volume for those generated from LNT). Elution profiles were generated by collecting 0.5-ml fractions for scintillation counting. Retention times for unlabeled oligosaccharide standards (LacNAc and for products derived from LacNAc using H. pylori Fuc-T, LNT, LacNAc, and Le x for those generated using LNT as acceptor) under corresponding chromatographic conditions were determined by monitoring absorbance at 205 nm.

RESULTS
Identification of a Fucosyltransferase Gene in H. pylori NCTC 11637-Human ␣(1,3)-fucosyltransferases (Fuc-TIII-VII) show a high degree of sequence similarity at the amino acid level. To identify H. pylori sequences with homology to the human fucosyltransferase enzymes, we performed a TBLASTN 2 search of a GlaxoWellcome H. pylori genomic data set with part of the catalytic domain (residues 152-303) of human Fuc-TVI, a strongly conserved region among the human ␣(1,3)-Fuc-T family. A number of H. pylori sequence fragments showed weak similarity to the query (maximum BLAST score 0.0025), with matches localized to a short region in each case (17 identities in 30 amino acids). Codon usage plots indicated that this reading frame was likely to be protein coding (data not shown). Closer examination of the sequence alignments revealed that several of the matching residues from the H. pylori sequence are conserved across all five human ␣(1,3)-Fuc-Ts, suggesting that the data base sequence fragments may be part of a related H. pylori gene. Since both ␣(1,3)-Fuc-T and ␤(1,4)-Gal-T Le x forming activities have been reported in H. pylori (18), we carried out a similar search with part of the catalytic domain of human ␤(1,4-Gal-T), but found no matching sequences.
Using primers derived from one of the matching data base sequences, we amplified a short (approximately 400 bp) DNA fragment from H. pylori NCTC 11637 genomic DNA which was subsequently labeled with digoxigenin and used to identify hybridizing clones in a plasmid library of DNA from the same organism. Seven strongly hybridizing clones were retrieved from the library for DNA sequence analysis, which revealed considerable overlap between the cloned sequences in all seven plasmids. DNA sequencing of all seven clones in both strands yielded a total of approximately 2.7 kb of contiguous sequence (Fig. 1A). The probe sequence occurs within the only complete open reading frame in the sequence (designated fucT), spanning 1002 bp and coding for a predicted 333-amino acid polypeptide with localized sequence homology to the human ␣(1,3)-Fuc-Ts. A partial open reading frame occurs approximately 500 bp upstream of the fucT gene, running in the same direction. The predicted translation of this part of the H. pylori sequence shows homology to phosphoserine phosphatase (serB) genes from Gram-negative bacteria, with greatest similarity to the Haemophilus influenzae sequence (37% identity, 65 matching residues over 173 amino acids).
Primary Structure of H. pylori fucT-The nucleotide sequence and predicted translation of H. pylori fucT are shown in Fig. 1. The GC content of the gene (36%) is typical for H. pylori coding sequences (3). The predicted amino acid sequence contains no recognizable signal peptide or transmembrane domain, a Kyte-Doolittle plot revealing that hydrophobic regions of the sequence are small and infrequent (Fig. 1B). A repetitive element occupies 49 amino acids of the C-terminal part of the protein. The repeat unit is imperfect, but leucine appears consistently at 7-amino acid intervals in a pattern reminiscent of the eukaryotic leucine zipper motif.
The similarity between the fucT gene product and other ␣(1,3)-Fuc-T is weak outside the short region originally identified by the data base search, spanning residues 101 to 129 of the H. pylori protein. Within this part of the sequence, however, 10 residues are completely conserved in all five human ␣(1,3)-Fuc-Ts and also appear unchanged in bovine, murine, and avian ␣(1,3)-Fuc-T enzymes (Fig. 2). In addition, there are a number of partially conserved positions (occupied by one of two amino acids). Outside this region, similarity to other members of the ␣(1,3)-Fuc-T family diminishes very quickly, although a number of isolated conserved residues can be identified. No significant similarity to any enzyme class other than the ␣(1,3)-Fuc-T family could be found for the H. pylori sequence.
Given the reported occurrence of Le x structures on the H. pylori O antigen and detection by others (18) and ourselves 3 of corresponding ␣(1,3)-Fuc-T activity in cell extracts from this bacterium, we took the exclusive, albeit localized, similarity between the deduced amino acid sequence of H. pylori fucT and ␣(1,3)-Fuc-T enzymes as an indication that it may encode an H. pylori fucosyltransferase enzyme.
Fucosyltransferase Activity-We assayed cell lysates from the clones retrieved from the H. pylori plasmid library for ␣(1,3)-Fuc-T activity using N-acetyllactosamine (Gal␤134GlcNAc) as an acceptor. All seven of the library clones tested showed measurable Fuc-T activity, while neither control clones containing pUC18 nor untransformed E. coli possessed any activity (Table  I), demonstrating that cloned H. pylori sequences contained in the library plasmids encode an active Fuc-T.
Cloning H. pylori fucT into an E. coli Expression Vector-The H. pylori library plasmids contain stretches of flanking sequence on either side of fucT. To exclude the possibility that coding sequences outside the identified fucT gene were responsible for the observed Fuc-T activity, and in an effort to increase levels of recombinant fucosyltransferase activity, we subcloned H. pylori fucT into the E. coli expression vector pET-11a. The resulting plasmid, pHPFT, contains fucT as the sole H. pylori-derived coding sequence under control of the T7lac promoter. E. coli BL21(DE3) containing pHPFT produced Fuc-T activity when induced with isopropyl-1-thio-␤-Dgalactopyranoside, extracts typically showing a specific activity of 100 -200 pmol/min/mg with N-acetyllactosamine as acceptor. Some activity could also be detected in uninduced samples (10 -20% of induced levels), presumably as a result of "leaky" promoter repression. Maximal activity levels produced from pHPFT were not, as we had hoped, substantially higher than those in the library clones, nor was a highly expressed protein of the expected molecular mass (approximately 40 kDa) apparent from SDS-polyacrylamide gel electrophoresis analysis of uncleared cell extracts (data not shown). The limited Fuc-T activity produced by pHPFT thus appears to result from limited expression rather than accumulation of highly expressed but insoluble and inactive recombinant protein.
Acceptor Specificity of H. pylori Fuc-T-We measured the activity of recombinant H. pylori FucT with a panel of oligosaccharide acceptors, as shown in Table II. The enzyme strongly preferred type 2 (Gal␤134GlcNAc) structures over type 1 (Gal␤133GlcNAc) acceptors. The type 2 tetrasaccharide (9) was a better acceptor than LacNAc (1) suggesting that H. pylori FucT may prefer to fucosylate ␤-configured GlcNAc. Similar preferences have been reported for human Fuc-TIV and to a lesser extent for Fuc-TV with these two acceptors (51). With sialylated LacNAc acceptors the H. pylori FucT most closely resembled human Fuc-Ts V and VII in that 3Ј-sialyl-LacNAc (6) was a substrate, while 6Ј-sialyl-LacNAc (7) was not (51, 52). 2 TBLASTN compares a peptide query sequence with the translation in all six frames of a nucleotide data base. 3  inhibitor NEM was measured by performing assays in the presence of NEM at concentrations up to 15 mM. The enzyme showed very limited NEM sensitivity, with Fuc-T activity reduced by only 34% in the presence of 15 mM NEM. For comparison, FucT-III (expressed in COS cells) is inhibited approximately 85% at the same NEM concentration, while extracts from Schistosoma mansoni and COS cells expressing Fuc-TIV retain about 50% of their Fuc-T activity (53).
Analysis of Fucosylated Products Generated by H. pylori Fuc-T-Some of the inferences drawn from the acceptor preferences of H. pylori Fuc-T were investigated further by examining the sensitivity of fucosylated products to glycosidase treatment. Radiolabeled oligosaccharide products were generated by incubating the acceptors LacNAc and LNT with H. pylori Fuc-T in the presence of GDP-[ 3 H]fucose. Following removal of excess sugar nucleotide, free fucose and residual Triton X-100 from the reaction mixture by successive ion exchange, size exclusion, and reverse phase chromatographic steps, oligosaccharide products were purified by HPLC on a Neutropak NH 2 column.
Incubation of LacNAc with H. pylori Fuc-T yielded a single radiolabeled product with a retention time on HPLC corresponding to that of Le x (Fig. 3A). The product was unaffected by treatment with a selective ␣(1,2)-fucosidase, while treatment with ␣(1,3/4)-fucosidase resulted in complete conversion to a new product which eluted from the Neutropak NH 2 column more rapidly than LacNAc. These observations support the conclusion that the major fucosylated product generated by H. pylori Fuc-T with LacNAc as acceptor is the Le x trisaccharide. Accordingly, the product is resistant to ␣(1,2)-fucosidase, but sensitive to ␣(1,3/4)-fucosidase, which liberates fucose as the sole radiolabeled product.
To investigate the fucosylation of the type 1 acceptor LNT (10) by H. pylori Fuc-T, labeled product was treated with endo-␤-galactosidase. The action of this glycosidase is inhibited by fucosylation of residues flanking the ␤-galactoside linkage (54,55) and thus if, as substrate preferences seem to suggest, H. pylori Fuc-T fucosylates the glucose residue of LNT, the product should be resistant to endo-␤-galactosidase cleavage. Since endo-␤-galactosidase activity may also be hampered by fucosylation at more distant sites (for example, on GlcNAc or the distal Gal residue of LNT) (55), the product was also treated with bovine testis ␤-galactosidase, alone or in combination with ␤-N-acetylhexosaminidase, to examine this possibility. Analysis of the labeled oligosaccharides produced by glycosidase digestion of H. pylori Fuc-T-generated fucosyl-LNT is shown in Fig. 3B. As expected, endo-␤-galactosidase had no effect on the The position of Cys 143 in Fuc-TIII, implicated in inhibition by NEM, is arrowed. A repetitive element in the H. pylori sequence is underlined and in bold. Completely conserved residues are indicated below the alignment. A region of strong, localized homology is boxed. B, ␣(1,3)-Fuc-T motif. All known ␣(1,3)-Fuc-Ts contain this short peptide motif. Fully conserved residues are indicated in one-letter code, two letters arranged vertically indicate that either of the corresponding amino acids may occupy this position. As shown here, the motif represents a minimal consensus and those positions which may be occupied by more than two different amino acids, or by two unrelated residues, are indicated as unconserved positions, designated X. The asterisked position is occupied by Ile or Val in all ␣(1,3)-Fuc-Ts except in the recently cloned rat gene, where it is occupied by Asn. fucosylated oligosaccharide, while LNT itself was readily cleaved under similar conditions (data not shown). Digestion with bovine testis exo-␤-galactosidase resulted in substantial (Ͼ50%) conversion to a product with a retention time close to that of LNT. This indicates that the distal Gal residue of the major product is not fucosylated, as this would otherwise block exo-␤-galactosidase action. Combined treatment with ␤-galactosidase and N-acetyl-␤-hexosaminidase yielded a more rapidly eluting product, consistent with removal of distal Gal and GlcNAc residues to leave a fucose-containing trisaccharide. Further ␤-galactosidase action is presumably blocked by fucose branching at the Glc residue of the remaining galactoside. These findings suggest that H. pylori Fuc-T fucosylates LNT predominantly at the glucose residue. Resistance of part of the product material to bovine testis ␤-galactosidase may reflect some degree of ␣(1,2)or ␣(1,4)-fucosylation, although the substrate preferences of this Fuc-T indicate that incomplete galactosidase digestion is perhaps a more likely explanation. DISCUSSION By searching an H. pylori genomic data set with part of the catalytic domain sequence of a human ␣(1,3)-Fuc-T and sequencing corresponding clones from a plasmid library of H. pylori DNA we were able to identify a gene (fucT) with highly localized similarity to known ␣(1,3)-Fuc-T enzymes. Cell ex-tracts from library clones containing the H. pylori gene possessed Fuc-T activity, and by subcloning fucT into an E. coli expression vector we were able to confirm that it encodes an active ␣(1,3)-Fuc-T. H. pylori fucT is the first Fuc-T gene to be cloned from an invertebrate, although enzyme activity has been detected in the freshwater snail Lymnea stagnalis (56) and in the parasite S. mansoni (53).
Sequence similarity between the mammalian Fuc-Ts and chick fucosyltransferase (CFT1) provided evidence for evolutionary conservation of ␣(1,3)-Fuc-T sequences (45). Conservation between H. pylori FucT and the mammalian enzymes, although limited and highly localized, suggests that aspects of the ␣(1,3)-Fuc-T sequence have survived unchanged through evolution from bacteria to higher mammals and man. The lack of overall sequence similarity to human ␣(1,3)-Fuc-Ts would seem to preclude the idea that H. pylori acquired the Fuc-T gene from a mammalian source. The base composition of the gene (35% GC) is also much closer to the average for H. pylori (36%) than to mammalian and avian ␣(1,3)-Fuc-T genes, which are typically GC-rich (e.g. CFT-1, 69% GC).
Unlike eukaryotic Fuc-Ts which have a hydrophobic transmembrane domain near their N terminus and share a common type II membrane protein topology, the H. pylori enzyme con- FIG. 3. A, glycosidase/HPLC analysis of fucosylated LacNAc generated by H. pylori Fuc-T. Purified labeled product was analyzed by chromatography on a Neutropak NH 2 column under isocratic conditions using water/acetonitrile (25:75 by volume) following no treatment (q), digestion with Xanthomonas manihotis ␣(1,3)-fucosidase (E), or digestion with X. manihotis ␣(1,2)-fucosidase (ϫ). LacNAc and Le x standards eluted at 7.8 and 13.2 min, respectively, under these conditions. B, glycosidase analysis of H. pylori Fuc-T-generated product with LNT as acceptor. Purified labeled product was analyzed by chromatography on the Neutropak NH 2 column using water/acetonitrile (30:70 by volume) following no treatment (q), digestion with B. fragilis endo-␤-galactosidase (E), digestion with bovine testis ␤-galactosidase (ϫ), or digestion with bovine testis ␤-galactosidase and S. plicatus ␤-N-acetylhexosaminidase (f). In this system, LacNAc, Le x , and LNT standards eluted at 4.3, 6.1, and 13.4 min, respectively. Endo-␤-galactosidase cleavage of LNT generated a product with a retention time of 9.3 min (data not shown). tains no recognizable membrane insertion elements. Aligned on the basis of the short, highly conserved region of homology (Fig. 2), the bacterial enzyme appears to lack a region corresponding to the transmembrane and stem domains of other Fuc-Ts. Most of the "hypervariable region" previously implicated in acceptor binding specificity in human Fuc-TIII and -V (residues 34 to 161 in Fuc-TIII) (57) is also absent, suggesting that the architecture of the H. pylori protein is substantially different from the rest of the enzyme family. The alignment also reveals that the C terminus of the bacterial sequence extends for approximately 100 amino acids beyond that of the other Fuc-Ts, half of this C-terminal extension being taken up by a periodic 7-amino acid leucine zipper-like motif. The function of this region, which has no counterpart in mammalian or avian Fuc-T sequences, is unknown. One possibility is that it mediates homo-or heteromultimer formation through coiledcoil type interactions, but at present the subunit structure of the H. pylori protein is unknown and further work will be necessary to establish the role of the zipper-like region.
Recombinant H. pylori Fuc-T has a strong preference for type 2 acceptors, and analysis of oligosaccharides generated by fucosidase digestion of the product generated by this Fuc-T with LacNAc indicates that H. pylori Fuc-T is indeed capable of synthesizing the Le x epitope. Some type 1 structures are also fucosylated, but our studies suggest that with these acceptors fucose may be transferred predominantly to glucose rather than GlcNAc, implying that the enzyme has little ␣(1,2)or ␣(1,4)-Fuc-T activity, as has been reported for human Fuc-TV (51). Biosynthesis of the Le y epitope found on the surface of many H. pylori isolates is therefore likely to involve a separate ␣(1,2)-Fuc-T activity. Overall, the acceptor specificity of H. pylori Fuc-T does not match that reported for any of the human enzymes or indeed that of S. mansoni ␣(1,3)-Fuc-T (53). Like the schistosome enzyme and human Fuc-Ts IV and VII, however, H. pylori Fuc-T shows only slight sensitivity to NEM inhibition. Interestingly, 3Ј-sialyl-LacNAc is an efficient acceptor (although 6Ј-sialyl-LacNAc is not), implying that H. pylori Fuc-T may be capable of synthesizing the sialyl-Le x (sLe x ) structure which was recently detected in a small number of H. pylori isolates by Wirth et al. (27). The absence of sLe x from the majority of H. pylori isolates may therefore reflect a lack of sialyltransferase activity in these strains.
Mammalian ␣(1,3)-Fuc-Ts are a closely-related family of enzymes, making it difficult to identify residues of potential structural and/or catalytic importance from sequence alignments. The recently cloned avian ␣(1,3)-Fuc-T, CFT-1 (45), also shows a high level of sequence similarity to the corresponding mammalian proteins, with 46.3% sequence identity to human Fuc-TIV. This is not the case with the H. pylori enzyme, which shows significant homology to the other ␣(1,3)-Fuc-Ts only in one short region. A consensus motif derived from this local area of homology (Fig. 2B) is unique to members of the ␣(1,3)-Fuc-T family, including the H. pylori enzyme and an open reading frame from a Caenorhabditis elegans cosmid 4 (58). This highly conserved ␣(1,3)-Fuc-T motif may be useful in identifying novel ␣(1,3)-Fuc-T genes in genomic and expressed sequence tag sequence data, since its appearance seems to be a reliable predictor of membership of this enzyme family. It may also provide a tool for cloning ␣(1,3)-Fuc-Ts in a manner similar to the demonstrated utility of the L-and S-sialyl motifs in cloning novel sialyltransferases (59).
The functional significance of the ␣(1,3)-Fuc-T motif is at present unclear. Marked differences in acceptor preferences among members of the ␣(1,3)-Fuc-T family would seem to ar-gue against a role in acceptor binding. Human Fuc-TIV and -VII for example both contain the ␣(1,3)-Fuc-T motif, but while Fuc-TVII uses 2,3-sialylated acceptors almost exclusively, Fuc-TIV strongly prefers neutral type 2 substrates (41) in in vitro assays. The behavior of Fuc-TIV in vivo is apparently more complex (60). The ␣(1,3)-Fuc-T motif lies outside sequence regions implicated by efforts to define acceptor-discriminating residues in ␣(1,3)-Fuc-Ts (51,57,61). Given that the enzymes transfer fucose from a common sugar nucleotide donor, it seems more likely that the ␣(1,3)-Fuc-T motif is involved in binding GDP-fucose or Mn 2ϩ . The motif lies some considerable distance from a cysteine residue implicated in GDP-fucose protectable inhibition by NEM (62), although it may be much closer in space within the folded protein than the primary sequence suggests. The corresponding position in the H. pylori Fuc-T is occupied by tyrosine ( Fig. 2A), in keeping with observations that enzymes with Cys at this location are inhibited by NEM while those with other amino acids (Fuc-TIV has Ser, Fuc-TVII has Thr) are resistant to NEM inhibition (62). 5 Interestingly, the conserved motif contains a lysine residue ( Fig. 2A), possibly a candidate for the so far unidentified GDP-fucose-protected lysine residue identified by pyridoxal phosphate labeling of a human fucosyltransferase (63). Further work is clearly needed to test these speculations, but in this respect the lack of overall similarity between the H. pylori and mammalian transferase sequences may be advantageous. The relatively small number of conserved residues inside and outside the ␣(1,3)-Fuc-T motif may provide a useful focus for mutagenesis experiments to probe structural and mechanistic aspects of the ␣(1,3)-fucosyltransferases. The dissimilarity of H. pylori and human Fuc-T enzymes would also seem to auger well for the design of selective inhibitors of the bacterial enzyme.
The H. pylori enzyme, which lacks a transmembrane domain and is, presumably, nonglycosylated, is devoid of some of the features which make eukaryotic Fuc-Ts difficult to work with. The possibility of bacterial expression also makes this enzyme a promising candidate for chemoenzymatic glycoconjugate synthesis. The same features may simplify the task of structural determination. Given the conserved motif, it seems reasonable to assume that this enzyme shares at least some structural features in common with its mammalian counterparts which have so far resisted structural elucidation.
Mounting evidence appears to point to a role for Lewis antigen mimicry in H. pylori pathogenesis. Identification and cloning of a Fuc-T gene from this organism will allow us to probe the biosynthesis of Le x by H. pylori in vivo via disruption of fucT and may make it possible to test the role of Le x directly in models of H. pylori pathogenesis.