Cloning and Heterologous Expression of an α1,3-Fucosyltransferase Gene from the Gastric PathogenHelicobacter pylori *

Helicobacter pylori is an important human pathogen which causes both gastric and duodenal ulcers and is also associated with gastric cancer and lymphoma. This microorganism has been shown to express cell surface glycoconjugates including Lewis X (Lex) and Lewis Y. These bacterial oligosaccharides are structurally similar to tumor-associated carbohydrate antigens found in mammals. In this study, we report the cloning of a novel α1,3-fucosyltransferase gene (HpfucT) involved in the biosynthesis of Lex within H. pylori. The deduced amino acid sequence of HpfucT consists of 478 residues with the calculated molecular mass of 56,194 daltons, which is approximately 100 amino acids longer than known mammalian α1,3/1,4-fucosyltransferases. The ∼52-kDa protein encoded byHpfucT was expressed in Escherichia coli CSRDE3 cells and gave rise to α1,3-fucosyltransferase activity but neither α1,4-fucosyltransferase nor α1,2-fucosyltransferase activity as characterized by radiochemical assays and capillary zone electrophoresis. Truncation of the C-terminal 100 amino acids of HpFuc-T abolished the enzyme activity. An approximately 72-amino acid region of HpFuc-T exhibits significant sequence identity (40–45%) with the highly conserved C-terminal catalytic domain among known mammalian and chicken α1,3-fucosyltransferases. These lines of evidence indicate that the HpFuc-T represents the bacterial α1,3-fucosyltransferase. In addition, several structural features unique to HpFuc-T, including 10 direct repeats of seven amino acids and the lack of the transmembrane segment typical for known eukaryotic α1,3-fucosyltransferases, were revealed. Notably, the repeat region contains a leucine zipper motif previously demonstrated to be responsible for dimerization of various basic region-leucine zipper proteins, suggesting that the HpFuc-T protein could form dimers.

Helicobacter pylori is a spiral, microaerophilic Gram-negative bacterium which has been recognized as an important human pathogen causing gastritis and both peptic and duode-nal ulcers (1)(2)(3). This bacterium is also associated with an increased risk for the development of both gastric adenocarcinoma (4,5) and primary gastric lymphoma (6). This pathogen is highly adapted to colonize human gastric mucosa and may remain in the stomach with or without causing symptoms for many years (7). Although H. pylori elicits local as well as systemic antibody responses (8), location in such a specific niche can permit it to escape elimination by the host immune response. Another mechanism by which H. pylori may protect itself from the action of the host immune response is the production of surface antigens mimicking those in the host.
Cell surface ␣(1,3)and ␣(1,2)-fucosylated oligosaccharides, that is, Lewis X (Le x ) 1 and Lewis Y (Le y ), are present on both eukaryotic and microbial cell surfaces. In mammals, Le x is a stage-specific embryonic antigen, and Le x , sLe x , and Le y are all regarded as tumor-associated markers (9 -11). sLe x directly mediates cell to cell adhesion through the interaction with selectins (12,13). It has been proposed that Le x plays a similar function during physiological and pathological processes (14). Bacterial Le x and Le y were first identified in H. pylori by Aspinall and Monteiro (15) and Aspinall et al. (16) who demonstrated that Le x and Le y structures are present in the Oantigen regions of the lipopolysaccharides purified from H. pylori cells using one-and two-dimensional nuclear magnetic resonance spectroscopy. Expression of Le x in H. pylori was further confirmed by examination using immunoelectron microscopy and enzyme-linked immunosorbent assay (17)(18)(19). Recently it was reported that production of Le x and Le y is related to cagAϩ H. pylori isolates which are associated with an increased risk for the development of gastric cancer (20). The biological functions of these bacterial oligosaccharide structures are not fully understood. It has been suggested that such glycoconjugates produced by H. pylori may mimic host cell antigens and could mask the bacterium and thus reduce the immune response (19,20). It is also possible that these bacterial Lewis antigens could down-regulate the host T-cell response (17). Therefore, production of such antigens may contribute to colonization and long-term infection of the stomach by H. pylori.
The final step in the synthesis of cell surface ␣1,3and ␣1,4-fucosylated oligosaccharides in mammals is catalyzed by ␣1,3or ␣1,3/1,4-fucosyltransferases (Fuc-Ts). The human FUT genes encoding at least five distinct Fuc-Ts generating Le x and related structures have been cloned and sequenced (21)(22)(23)(24)(25)(26)(27)(28)(29). Similar FUTs were also identified in other animals, including * This work was supported by funding from the Canada Bacterial Disease Network (Centers for Excellence Program) (to D. E. T.) and the Natural Sciences and Engineering Research Council (NSERC) (to M. M. P.). 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  bovine (30), mouse (31), and chicken (32). The primary sequences of these eukaryotic Fuc-Ts exhibit significant amino acid similarity (32,33). It has been demonstrated that the discrete peptide fragments of human Fuc-Ts III, V, and VI contribute to discrimination of acceptor substrates (34,35). Chan et al. (18) demonstrated that the Le x produced by H. pylori is synthesized by the addition of galactose from UDP-Gal to GlcNAc and then fucose from GDP-Fuc to Gal␤1-4GlcNAc (LacNAc), catalyzed by ␤1,4-galactosyltransferase and Fuc-T, respectively. This pathway is identical to that found in humans. However, the genetic basis for the bacterial Fuc-T was unclear. Characterization of a gene encoding the ␣1,3-fucosyltransferase would not only lead to elucidation of the pathogenic role of the Le x determinant in the H. pylori infection but would also provide new insights into the structure-function relationship by comparison between eukaryotic and prokaryotic Fuc-Ts. We now report the cloning and heterologous expression of the novel bacterial Fuc-T gene, designated HpfucT, obtained from H. pylori. The deduced amino acid sequence of HpfucT is similar to the highly conserved catalytic domain among known mammalian and chicken Fuc-Ts. Several sequence features unique to this bacterial Fuc-T were also identified. In addition, the HpFuc-T protein produced in Escherichia coli was characterized biochemically.

EXPERIMENTAL PROCEDURES
Materials-Bacteria were used in this study, including H. pylori strain NCTC11639 for cloning, E. coli strain DH10B for production of recombinant plasmids, and E. coli CSRDE3 containing a Plac-controlled T7 RNA polymerase (36) for overexpressing proteins of interest. H. pylori cells were cultured on BHI-YE agar or in BHI-YE broth under microaerobic conditions as described previously (36). LB medium (37), M9 medium (37), and a supplemented M9 medium (37) were used for growth of E. coli cells. Either ampicillin (100 g/ml) or rifampicin (200 g/ml) were added to the above media for growth of plasmid-containing cells and for expression of plasmid-encoded proteins. Plasmid vectors pBluescript II SK(ϩ/Ϫ) and pCR TM II were purchased from Stratagene  (38). Sep-Pak Plus C-18 reverse-phase cartridges were purchased from Waters (Mississauga, Ontario). [␣-32 P]dCTP was from DuPont Canada Inc. (Mississauga, Ontario). Protein concentrations were determined with the BCA protein assay kit using bovine serum albumin as a standard according to the supplier's instructions (Pierce). All restriction endonucleases and other modified enzymes were obtained from Life Technologies, Inc. unless specified.
DNA Manipulation Techniques-Plasmid and H. pylori chromosomal DNA was prepared as described previously (36). Standard molecular cloning procedures and Southern hybridization were detailed by Sambrook et al. (37). Both strands of recombinant plasmid DNA were sequenced using the Thermo sequenase sequencing kit following the manufacturer's instructions (Amersham Life Science, Inc.). [␣-32 P] dCTP-labeled probes were prepared using a random priming kit (Life Technologies, Inc.).
Cloning of the H. pylori Fucosyltransferase (fucT) Gene-To clone the fucosyltransferase gene from H. pylori NCTC11639, degenerate primers were generated from the several regions conserved by three mammalian ␣1-3 fucosyltransferases including human Fuc-TVI (29) the remaining cycles, 1 min at 94°C, 30 s at 50°C, and 40 s at 72°C, followed by extension at 72°C for 10 min. The PCR products were cloned into vector pCR TM II according to the supplier's instructions. Subsequently, the inserts in recombinant plasmids were sequenced with Thermo sequenase, and their nucleotide sequences and deduced amino acid sequences were used in the search for related proteins in data bases with the software of Blast included in the GCG package (version 8.0, Genetic Computing Group, Inc., Madison, WI). A clone, designated pCRHpfucT3, was demonstrated to contain the insert encoding the amino acid sequence homologous to known mammalian ␣1,3-fucosyltransferases.
To clone a putative intact fucT gene from H. pylori, chromosomal DNA from H. pylori NCTC11639 was digested with restriction endonucleases, including BglII, EcoRI, BamHI, BglII-EcoRI, EcoRI-BamHI, and BglII-BamHI and then separated in a 1% agarose gel. DNA fragments containing the putative fucT gene were demonstrated by Southern hybridization with a [␣-32 P]dCTP-labeled probe made from pCRH-pfucT3 DNA. The 2.2-kb EcoRl-BglII and 4.5-kb EcoRI-BamHI fragments were cloned into vector pBluescript II KSϪ which was digested with EcoRI and BamHI. Two clones, pBKHpfucT8 carrying a 2.2-kb EcoRl-BglII fragment and pBKHpfucT31 carrying a 4.5-kb, EcoRI-BamHI fragment, were selected for further characterization.
Plasmid Constructs and Expression of the H. pylori fucT Gene-To construct recombinant plasmids containing an intact or partial H. pylori fucT gene, three primers were generated from the nucleotide sequence in Fig. 1: ZGE37 corresponding to nucleotides 1-19; ZGE38 and ZGE39 complementary to nucleotides 1215-1233 and 1692-1710, respectively. ZGE37 contained a BamHI site, whereas ZGE38 and ZGE39 contained an EcoRI site. PCR products were amplified from pBKHp-fucT31 with a primer pair of either ZGE37/ZGE38 or ZGE37/ZGE39. These PCR-amplified DNA fragments were digested with EcoRl and BamHl and then cloned into pBluescript II KSϪ. The respective clones containing the H. pylori fucT gene of interest were screened by PCR with the corresponding pair of the above primers. Two clones, designated pBKHp763fucT38 and pBKHp763fucT39, contained a partial and an intact H. pylori fucT gene, respectively. The coding region of the H. pylori fucT gene was controlled under the T7 promoter. The sequence of the PCR-amplified DNA fragments in pBKHp763fucT38 and pBKHp763fucT39 was demonstrated by sequencing to be identical to that of the native template.
Recombinant plasmids pBKHp763fucT38 and pBKHp763fucT39 were introduced into E. coli CSRDE3 cells by electroporation. The proteins encoded by the recombinant plasmids were overexpressed in the presence of either [ 35 S]methionine or cold methionine as described previously (36). The cells expressing [ 35 S]methionine-labeled proteins were boiled for 5 min in 1 ϫ sample buffer and separated in a 13.5% SDS-polyacrylamide gel (36).
Fucosyltransferase Assays-E. coli CSRDE3 cells expressing nonradioactively labeled proteins encoded by pBKHp763fucT38 and pBKHp763fucT39 were harvested and suspended in Hepes buffer (20 mM Hepes, pH 7.0) supplemented with 0.5 mM phenylmethylsulfonyl fluoride (a proteinase inhibitor). Subsequently, membrane and soluble fractions of the cells were prepared after disruption with a French press as described previously (36). The membrane pellets were resuspended in the same Hepes buffer, frozen in liquid nitrogen, and stored at Ϫ70°C prior to use.
Capillary Electrophoresis Assay-The incubation mixtures contained 16 l of the membrane fraction containing the intact HpFuc-T protein, 100 M LacNAc-TMR, 100 M GDP-fucose in a total volume of 20 l of 20 mM Hepes (pH 7.0) containing 20 mM MnCl 2 and 0.2% bovine serum albumin. Incubation was done at 37°C for 30 min. Subsequently, the sample was prepared and analyzed by capillary electrophoresis by injecting 12 pl onto an electrophoresis column (60 cm long) at 1 kV for 5 s as described previously (18). The electrophoretic separations were performed at a running voltage of 400 V/cm. ␣-Fucosidase treatment was done by incubating the sample (10 M total TMR) with 4 microunits of almond meal ␣-fucosidase in a total volume of 40 l of 50 mM sodium citrate buffer, pH 5.0, at 37°C for 90 h. Products were isolated and analyzed by capillary electrophoresis as described above.

RESULTS
Cloning and Nucleotide Sequence of a H. pylori Fucosyltransferase Gene-Three recombinant plasmids, pCRHpfucT3, pBKHpfucT8, and pBKHpfucT31 containing the intact or partial sequence of the fucT gene from H. pylori NCTC11639 were obtained as described under "Experimental Procedures" (Fig.  1). The nucleotide sequences of these recombinant clones were sequenced from both strands using nested primers. Fig. 2A gave rise to the nucleotide sequence of 1710 base pairs derived from clone pBKHpfucT31. A major open reading frame (ORF), starting at nucleotide 145 and ending at nucleotide 1578, was predicted in this region. An unusual sequence feature of this ORF was 10 direct repeats of 21 nucleotides (Fig. 2A). An AA to GG transition at positions 12 and 13 of this repeat has occurred in repeat copies III, V, and VIII. An SD sequence, a ribosomal binding site in prokaryotes (39), precedes the predicted translation initiation codon AUG. In addition, the sequence "AC-CATGT," which is similar to the Kozak's consensus context "ACCATGG" (a common ribosomal binding site in eukaryotes) (40), is also present at the beginning of the ORF. Putative transcription elements including Ϫ10 and Ϫ35 regions immediately upstream of the ORF and a stem-loop structure following the stop codon of the ORF, which probably act as a transcription promoter and -independent transcription terminator (41), were identified ( Fig. 2A). An asymmetric inverted repeat sequence was found, encompassing 18 nucleotides and containing the putative Ϫ10 region ( Fig. 2A). Another ORF downstream from the major ORF, in the opposite orientation as indicated in Fig. 2A, encodes the amino acid sequence similar to the corresponding region of the glutamate dehydrogenase identified in Corynebacterium glutamicum (nucleotide sequence accession number S32227).
Features of the Deduced Amino Acid Sequence of the H. pylori fucT-A protein consisting of 478 amino acids with a calculated molecular mass of 56,194 daltons was predicted from this ORF. A hydropathy profile calculated by the method of Kyte and Doolittle (42) indicates that the deduced amino acid sequence is primarily hydrophilic and does not contain a potential transmembrane segment (Fig. 2B). The predicted protein carries 10 direct repeats of seven amino acid residues proximal to the C terminus. There is a conservative replacement of valine with isoleucine at position 5 found in repeats III, V, and VIII, which results from the corresponding AA to GG mutations as mentioned above. Five putative N-linked glycosylation sites were predicted, two of which are proximal to the N terminus similar to those identified in mammalian Fuc-Ts. However, the remaining three such sites are close to the C terminus. This latter feature is similar to the sites identified in rabbit and human ␣1,2-fucosyltransferases (43). Comparison of this polypeptide sequence with other proteins in the data bases using the Blast search revealed significant sequence similarity (40 -45% identity) to ␣1-3 and 1-3/1-4 fucosyltransferases from mammalian sources including human Fuc-Ts III to VII (21)(22)(23)(24)(25)(26)(27)(28)(29), bovine Fuc-TIII (30), mouse Fuc-TIV (31), and CFT1 from chicken (32) within an approximately 72-amino acid stretch. As denoted in Fig. 3A, this region is located in the proposed C-terminal catalytic domains of Fuc-Ts (44,45). Therefore, we designated this gene as HpfucT.
The remaining sequences beyond this conserved region between HpFuc-T and eukaryotic Fuc-Ts are relatively divergent.
HpFuc-T appears to lack a transmembrane segment which is common to eukaryotic Fuc-Ts and which is usually located in their N-terminal region. On the other hand, these eukaryotic Fuc-Ts do not contain the ϳ100-amino acid region encompassing ten "DDLR(V/I)NY" repeats. Sequence similarity searches revealed that this region is significantly similar to the domain potentially forming a leucine zipper structure within several homeobox-leucine zipper proteins (HD-Zip protein) including ATHB-1, ATHB-5 to -7 from Arabidoposis thaliana (46 -48), and tomato (nucleotide sequence accession number X94947) (Fig. 3B). The conserved leucine residues are also colinear to those present in the leucine zipper motif of the group of the basic region-leucine zipper (bZip) proteins found in eukaryotes including yeast, higher plant, animals, and recently in a bacterium (49 -52).
Characterization of the HpfucT Gene Product in E. coli Cells-To investigate whether or not the predicted HpfucT gene represents a complete locus, a modified maxicell system using CSRDE3 (36) was applied for the HpfucT expression. Two recombinant plasmids, pBKHp763fucT38 carrying the partial HpfucT gene and pBKHp763fucT39 carrying the intact HpfucT gene, were constructed (Fig. 1). The HpfucT genes in these two plasmids were controlled by a T7 promoter. The expression results were shown in Fig. 4. pBKHp763fucT39 gave rise to a specific product of ϳ52 kDa (Fig. 4, lane 2) which is close to the predicted molecular mass of 56 kDa. In addition, a protein of ϳ41 kDa was produced from pBKHp763fucT38containing cells (lane 1). The size of this product is consistent with the predicted 42 kDa of the truncated HpFuc-T in which the C-terminal 115 amino acids of HpFuc-T was removed. In contrast, these HpfucT-encoded proteins were not produced in cells containing either a vector without the insert or with no plasmid (Fig. 4, lanes 3 and 4). Two strong protein bands of ϳ35 kDa and 29 kDa, respectively, were present in all the samples, indicating that they were encoded by host genes. Therefore, the above evidence demonstrates that the cloned HpfucT represents a complete locus.
Biochemical Assay of the Overproduced HpFuc-T Protein- The partial sequence of this bacterial Fuc-T is homologous to the catalytic domain of mammalian Fuc-Ts, suggesting that HpFuc-T is a fucosyltransferase. The nature of this enzyme activity was investigated. To delineate the cellular location of the enzyme activity, membrane and cytoplasmic fractions of E. coli cells producing the HpFuc-T proteins were prepared. The ␣1,3-HpFuc-T activity was quantitated using LacNAc-R as an acceptor and GDP-fucose as the donor. Approximately 85% of the total enzyme activity was associated with the membrane fraction containing the intact HpFuc-T protein expressed from pBKHp763fucT39, whereas the remaining portion was present in the cytoplasmic fraction (Table I). There was no detectable activity of either ␣1,2-Fuc-T or ␣1,4-Fuc-T associated with the same samples. The Triton X-100-solubilized membrane fraction gave rise to slightly higher ␣1,3-Fuc-T activity than the untreated extract (Table I). No ␣1,3-Fuc-T activity was obtained from either the membrane or the cytoplasmic fractions prepared from cells producing the truncated HpFuc-T protein encoded by pBKHp763fucT38. This result indicated that the C-terminal 115 amino acids of HpFuc-T is crucial for this activity.
The reaction products synthesized by the H. pylori ␣1,3fucosyltransferase were characterized by capillary electrophoresis with laser-induced fluorescence detection using tetra-methylrhodamine (TMR)-labeled acceptors (18). The reaction mixture containing the membrane fraction of cells harboring pBKHp763fucT39, GDP-fucose, and LacNAc-TMR produced a FIG. 3. A, representative sequence alignment of the HpFuc-T with eukaryotic ␣1,3-fucosyltransferases using the program of Pileup (the GCG package, version 8.0). bFuc-TIII, hFuc-TVI, mFuc-TIV, and cFuc-TI represent Fuc-Ts cloned from bovine, human, mouse, and chicken, respectively. Underlined residues indicate the proposed transmembrane segment within the respective Fuc-Ts. Identical residues within all the aligned proteins are denoted by both asterisks and bold type, whereas corresponding residues partially conserved by HpFuc-T and other Fuc-Ts were indicated by bold type alone. B, sequence comparison of the direct repeat region of HpFuc-T with the leucine zipper motifs within the chicken EAP-300 protein, HD-Zip proteins, and bZip proteins. Conserved leucines among all the compared proteins are marked by asterisks and bold type. ATHD-Zip, A. thaliana homeobox-leucine zipper proteins (46 -48); EAP-300, a developmentally regulated embryonal protein (51); TAF-1, a tobacco transcription activator 1 (49); CPRF1, a common plant regulatory factor isolated from parsley (50); TodS, a histidine kinase in Pseudomonas putida F1 (52) .   FIG. 4. Overexpression of the H. pylori fucT gene in E. coli CSRDE3 cells. Proteins were synthesized as described under "Experimental Procedures," and plasmid constructs were shown in Fig. 1. Equal amounts of protein extracts determined by the absorbance (A 620 nm ) of the cultures were separated on a 13.5% polyacrylamide gel. Lanes: 1, pBKHp763fucT38; 2, pBKHp763fucT39; 3, pBluescript II KSϪ; 4, no plasmid. The protein bands of interest and molecular mass makers (Life Technologies, Inc.) are indicated by arrowheads and lines on the left, respectively. a Membrane and cytoplasmic fractions were prepared from cells grown in 300 ml of LB broth as described under "Experimental Procedures." b A milliunit of enzyme activity is expressed as the amount of the enzyme fraction that catalyzes the conversion of 1 nmol of acceptor to product per min. Numbers represent total milliunits obtained from each enzyme fraction.
c Specific activity (milliunits per mg of protein) is obtained by dividing the total activity by the total protein content. d 100% activity is total milliunits obtained from both the cytoplasmic and membrane fractions. new peak (Fig. 5a, Lewis-X peak) which co-migrated with a synthetic Le x -TMR in the electropherogram (Fig. 5c, peak 3), indicating that this new peak represents the Le x product synthesized by this bacterial ␣1,3-fucosyltransferase. The synthesis of Le x with this enzyme was further demonstrated by digestion with fucosidase which cleaved the Le x product and released LacNAc-TMR. As a result, the concentration of Lac-NAc in the reaction mixture increased (Fig. 5b, LacNAc peak), whereas the concentration of the Le x product decreased (Fig.  5b, Lewis-X peak).

DISCUSSION
In this study we have cloned and sequenced the H. pylori ␣1,3-fucosyltransferase gene HpfucT. There are two unique features found in the DNA sequence of HpfucT as well as its flanking regions. First, in addition to a prokaryotic ribosomal binding site, the Kozak's consensus context (ribosomal binding site) also exists around the translation codon AUG. It will be of interest to determine whether or not this Kozak's context is functional in eukaryotic cells. Second, 10 direct 21-nucleotide repeats are present in the region close to the 3Ј end of the HpfucT gene, which encode 10 corresponding 7 amino acid repeats in the HpFuc-T protein. Sequence analysis and expression of HpfucT in E. coli indicated that this gene coded for a protein of ϳ56 kDa comprising 478 amino acids. Fucosyltransferase activity of HpFuc-T produced in E. coli cells was assayed by in vitro biochemical characterization, which demonstrated that this enzyme strictly utilizes LacNAc as an acceptor. No fucose transfer was detected when either Gal␤1-3GlcNAc-R or Phenyl-Gal was used as an acceptor. Although the activity and the substrate specificity of HpFuc-T determined in vitro may or may not represent the in vivo situation of this enzyme, the lines of evidences presented here confirmed that the HpFuc-T represents the novel bacterial ␣1,3-fucosyltransferase.
Amino acid sequence comparison revealed that HpFuc-T exhibits a significant sequence similarity to the highly conserved eukaryotic Fuc-T C-terminal catalytic domain. Beyond this region, the sequences between this bacterial Fuc-T and the eukaryotic Fuc-Ts extensively diverge. This is not surprising since the N-terminal amino acid sequences among these eukaryotic Fuc-Ts are relatively variable (33). However, several unique structural features indicate that the HpFuc-T is distinct from eukaryotic Fuc-Ts. First, in comparison with the primary sequences of mammalian and chicken Fuc-Ts, the HpFuc-T protein contains an additional C-terminal ϳ100amino acid region including the 10 direct repeats. Our preliminary characterization indicated that this region is crucial for Fuc-T activity since the deletion of this portion from the HpFuc-T protein completely abolishes the Fuc-T activity (Table  I). This result is consistent with the previous finding that removal of one or more amino acids from the C terminus of Fuc-TV dramatically reduced or completely abolished enzyme activity (35). Second, HpFuc-T does not contain a N-terminal transmembrane segment, a typical feature for mammalian Fuc-Ts, primarily consisting of hydrophilic residues (Fig. 2B). It should be noted that the majority of the HpFuc-T activity is present in the membrane fraction, whereas Triton X-100 solubilization of the membrane fraction did not significantly increase such activity. This suggests that HpFuc-T could be as- sociated with the cytoplasmic membrane through electrostatic forces as was found for the E. coli phosphatidylserine synthase whose overall amino acid sequence is also hydrophilic (53). Third, five potential asparagine-linked glycosylation sites lie within the HpFuc-T sequence, three of which are located in the C-terminal region downstream of the Fuc-T catalytic domain. These three C-terminal glycosylation sites are not found in mammalian and chicken ␣1,3-Fuc-Ts. An unexpected finding was that sites III and IV are comparable to those in ␣1,2-Fuc-Ts from humans and rabbit (43).
It is interesting that the sequence of the region containing the 10 direct repeats is extensively similar to the leucine-zipper domain of the plant HD-Zip proteins (ATHB-1 and -5 to -7) found in A. thaliana (46 -48). These highly conserved leucine residues also correspond to those in the leucine motif of the eukaryotic bZip proteins (Fig. 3B). The leucine-zipper motif of plant HD-Zip proteins is closely linked to the homeodomain which consists of 61 amino acids with DNA-binding properties (54,55), whereas this motif in the bZip proteins is located immediately downstream of a basic region responsible for the sequence-specific DNA binding (49,50). In contrast, the putative leucine zipper domain in HpFuc-T follows the two potential glycosylation sites. Leucine zippers are responsible for dimerization in a separate class of transcription factors found in some eukaryotes (reviewed in Ref. 56). More recently, a bZip motif was identified in a bacterial histidine kinase TodS regulating toluene degradation in Pseudomonas putida F1 (52). The bZip motif-mediated dimerization of TodS for DNA binding was suggested by in vitro DNA binding assays using a short dimerized peptide derived from the N-terminal region of TodS to mimic the full-length protein; this short peptide dimer could bind to the 196-base pair DNA fragment PCR-amplified from the intergenic region between todH and todS (52). In addition, the leucine zipper repeats of HpFuc-T exhibit significant sequence similarity (31% identity) to the multiple leucine zipper motifs of EAP-300, a developmentally regulated embryonal protein found in chicken (51). This protein has been implicated in a role of neural development, and it was postulated that existence of the multiple leucine-zipper motifs might enable EAP-300 to form dimers or multimers (51). We hypothesize that this leucine-zipper domain may dimerize the HpFuc-T protein. Such a dimer could act as either a transcription regulator controlling Le x production or a functional enzyme complex or both. Construction of mutations in this region would test this hypothesis. Additional studies on relationships between these structures and the ␣1,3-fucosyltransferase functions, and transcriptional and translational regulation of this gene, should shed light on the pathogenic role of this gene product during H. pylori infection.