Purification, Kinetic Characterization, and Mapping of the Minimal Catalytic Domain and the Key Polar Groups of Helicobacter pylori α-(1,3/1,4)-Fucosyltransferases*

The minimal catalytic domain of α-(1,3/1,4)-fucosyltransferases (FucTs) from Helicobacter pylori strains NCTC11639 and UA948 was mapped by N- and C-terminal truncations. Only the C terminus could be truncated without significant loss of activity. 11639FucT and UA948FucT contain 10 and 8 heptad repeats, respectively, which connect the catalytic domain with the C-terminal putative amphipathic α-helices. Deletion of all heptad repeats almost completely abolished enzyme activity. Nevertheless, with only one heptad repeat 11639FucT is fully active, whereas UA948FucT is partially active. Removal of the two putative amphipathic α-helices dramatically increased protein expression and solubility, enabling purification with yields of milligrams/liter. Steady-state kinetic analysis of the purified FucTs showed that 11639FucTs possessed slightly tighter binding affinity for both Type II acceptor and GDP-fucose donor than UA948FucT, and its kcat of 2.3 s-1 was double that of UA948FucT, which had a kcat value of 1.1 s-1 for both Type II and Type I acceptors. UA948FucT strongly favors Type II over the Type I acceptor with a 20-fold difference in acceptor Km. Sixteen modified Type I and Type II series acceptors were employed to map the molecular determinants of acceptors required for recognition by H. pylori α-(1,3/1,4)-FucTs. Deoxygenation at 6-C of the galactose in Type II acceptor caused a 5000-fold decrease in α1,3 activity, whereas in Type I acceptor this completely abolished α1,4 activity, indicating that this hydroxyl group is a key polar group.

Sequence alignments of mammalian and H. pylori ␣-(1,3/1,4)-FucTs show that they share very weak sequence homology (23,41) that is limited to two small regions designated as ␣-(1,3)-FucT motifs (22). These motifs have been suggested to be most likely involved in the binding of donor or metal ions (i.e. manganese) (22,42). Mammalian ␣-(1,3/1,4)-FucTs are typical Golgi-resident type II membrane proteins with a short cytoplasmic N-terminal tail, transmembrane segment, and hypervariable stem domain, preceding the C-terminal catalytic domain. In human FucT III and V, deletion of the N-terminal 61 and 75 amino acids did not lead to significant loss of activity, whereas removal of one or more amino acids at the C terminus in FucT V completely abolished enzyme activity (43). H. pylori ␣-(1,3/1,4)-FucT sequences lack transmembrane domain (21)(22)(23). Instead, downstream of the catalytic domain, they contain a short hypervariable region followed by 2-10 heptad repeats (21)(22)(23)41) and a short segment, which is rich in hydrophobic and positive residues that meet the requirements to form two amphipathic ␣-helices (24). These helices are proposed to function as a membrane anchor with the hydrophobic face embedded in the membrane and the positive charges interacting with phospholipid head groups (24). In most identified H. pylori ␣-(1,3/1,4)-FucTs, the heptad repeat consists of the amino acids DDLRVNY (23,41). 11639FucT contains 10 heptad repeats with the conserved sequence DDLR(V/I)NY. UA948FucT has 5 internal repeats with this consensus sequence, but the first and the last two heptad repeats consist of the amino acid sequence DDLRRDH (23,41). It was suggested that the heptad repeat region contains a leucine zipper-like motif responsible for dimerization, which might be essential for enzyme function (21)(22)(23). As a result, the C-terminal amphipathic ␣-helices and the preceding heptad repeat region in H. pylori ␣-(1,3/1,4)-FucTs may be functionally equivalent to the N-terminal transmembrane domain and the stem region of mammalian counterparts, respectively (24). For H. pylori FucT from strain NCTC11639, deletion of the C-terminal 115 amino acids, which contain the entire heptad repeat region and its downstream C-terminal tail, including the putative amphipathic ␣-helices, completely eliminated enzyme activity (21).
Both mammalian and H. pylori ␣-(1,3/1,4)-FucTs display distinct Type I and Type II substrate preferences. Domain swapping and site-directed mutagenesis studies show that acceptor specificity is determined by hypervariable regions that connect the catalytic domain to the membrane anchor, either preceding or following the catalytic domain for mammalian and H. pylori FucTs, respectively (12,13,24). In particular, a single aromatic residue has been reported to be critical for Type I acceptor specificity for both mammalian and H. pylori FucTs (10,11,44).
For human ␣-(1,3/1,4)-FucTs, a panel of modified Type I and Type II series acceptors was used to determine the relative contribution of each hydroxyl group to acceptor binding. For human FucT III, IV, V (16), and FucT VI (8), every hydroxyl group except the 6-OH of galactose and 3-OH or 4-OH of GlcNAc in Type II and Type I acceptor, respectively, tolerated modification. So the 6-OH of galactose and the reactive hydroxyl groups of Type II and Type I acceptors were identified as key polar groups essential for recognition by human ␣-(1,3/1,4)-FucTs (8,(15)(16)(17). For H. pylori ␣-(1,3/1,4)-FucTs, no systematic key polar group mapping has been carried out, and information is limited to two studies of the tolerance to sialylation. The FucT from strain NCTC11637 is able to use 3Ј-sialyl-Type II but not 6Ј-sialyl-Type II as an acceptor (22). For H. pylori FucT from strain DMS6709, primarily an ␣-(1,4)-FucT, 3Ј-sialyl-Type I is an excellent acceptor (25). This suggests that the hydroxyl at C-3 of galactose is not required for the recognition by H. pylori ␣-(1,3/1,4)-FucTs.
FucTs have significant applied interest for the enzymatic synthesis of complex Lewis antigens or as targets for inhibitors. Development of these applications is in great need of crystal structures, but at present, no crystal structure is available for any member of the FucT family. The first challenge to achieve this goal is to isolate large quantities of soluble and stable FucT. In this study, removal of the C-terminal putative amphipathic ␣-helices increased both protein expression and solubility yet did not significantly reduce the specific enzyme activity. The truncated forms of H. pylori ␣-(1,3/1,4)-FucTs were successfully purified at yields of milligrams/liter and enabled us to determine the detailed kinetic parameters, to map the key polar groups of acceptors essential for enzyme recognition, and to evaluate the potential of recombinant FucTs in enzymatic synthesis of Lewis X and Lewis A structures. These studies greatly expand our knowledge of H. pylori FucT function and further demonstrate the great similarity between the mammalian and H. pylori enzymes. Finally, the soluble truncated enzymes may also allow us to determine the crystal structures to understand the molecular basis for FucT enzyme catalysis.

EXPERIMENTAL PROCEDURES
Materials-Primers for constructing the truncated FucTs (⌬FucTs) were synthesized by Invitrogen. Pwo DNA polymerase and alkaline phosphatase were purchased from Roche Applied Science. pGEM-T vector was obtained from Promega (Madison, WI). Type II-R (Gal␤1,4GlcNAc-O-(CH 2 ) 8 CO 2 CH 3 ) (1), Type I-R (Gal␤1,3GlcNAc-O-(CH 2 ) 8 CO 2 CH 3 ) (13), and Type II series (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12) and Type I series (14)(15)(16)(17)(18) modified compounds were kindly provided by Dr. Ole Hindsgaul. Antipentahistidine monoclonal antibody and plasmid mini-preparation and midi-preparation kits were purchased from Qiagen (Mississauga, Ontario, Canada). GDP-Fuc and horseradish peroxidase-conjugated goat anti-mouse IgG were from Sigma. GDP- 3  Cloning the Truncated FucTs-Wild type (WT) futA gene of 11639FucT or UA948FucT was cloned into pGEM-T vector with a His 6 tag fused at the C-terminal end (24). Plasmid DNA harboring the futA gene from strains NCTC11639 and UA948 were used as template DNA for PCRs. Primers for the FucTs from NCTC11639 or UA948 were designed to delete different lengths of sequence at either the 5Ј or 3Ј end. Deletion of the entire (n) or n-1 heptad repeats but retaining the downstream C-terminal tail was achieved by primer overextension with two steps of sequential PCR (45). The sequences of the primers are available in the Supplemental Material. The codons for six histidine residues were included in the antisense primers when amplifying each truncated FucT. The PCR products were cloned into A/T cloning vector pGEM-T under the control of the T7 promoter. Cloned truncated futA genes were sequenced to ensure that no mutation had occurred during their construction.
Expression of Truncated FucTs-Full-length and each truncated FucT (⌬FucT) were expressed in E. coli HMS174DE3 cells as described previously (24).
Separation of Soluble and Membrane Fractions of WT and C-terminal Truncated FucTs-500 ml of cells were harvested, resuspended in HEPES buffer (20 mM, pH 7.0, containing 0.5 mM phenylmethylsulfonyl fluoride), and lysed by three passages through a French press (American Instrument Co., Silver Spring, MD) at 12,000 p.s.i. Cell debris was removed by centrifugation at 8,000 ϫ g for 10 min, and the supernatant was subsequently centrifuged at 40,000 ϫ g for 2 h to separate the soluble fraction (cytosol and periplasm) and the membrane fraction (pellet).
SDS-PAGE Immunoblot Analysis -Cell crude extracts and soluble or membrane fractions were separated by SDS-PAGE. The total protein concentration was determined by the BCA assay using bovine serum albumin as protein standard. FucT expression was detected by immunoblotting with mouse anti-pentahistidine monoclonal antibody (1:1000). The density of the FucT band was quantified and used to normalize the FucT activity as described previously (24).
Radiochemical Assays of ⌬FucTs-Enzyme activities of each truncated FucT in crude cell extracts was assayed with donor GDP-Fuc at 200 M, GDP-3 [H]Fuc at 0.2 M (ϳ60,000 dpm), and Type II-R or Type I-R acceptors at 1.8 or 7.5 mM, respectively (21,23,24). The reverse phase C 18 cartridges were used to isolate the products (containing the hydrophobic aglycone and the acquired radiolabeled fucose moiety) from the unreacted GDP- 3 [H]Fuc as described previously (46). One milliunit represents the amount of enzyme that converts 1 nmol of acceptor substrate to product/min. The specific activity (milliunit mg Ϫ1 ) was obtained by dividing the enzyme activity (milliunit) by the amount of total protein. The activity below 0.01 milliunit mg Ϫ1 was considered undetectable. The specific enzyme activity of each FucT was subsequently standardized by FucT expression level, which was determined by immunoblot.
Purification of 11639  and UA948 1-428 -The cell lysates from 1 liter of HMS174DE3 11639 1-441 and UA948 1-428 culture were prepared as described above except the harvested cells were resuspended in loading buffer (20 mM HEPES, pH 7.5, 50 mM imidazole, 300 mM NaCl, 0.5 mM phenylmethylsulfonyl fluoride). Following French press, the cell debris was removed by centrifugation at 10,000 ϫ g for 20 min. The supernatant was retained as a crude cell extract and was subsequently passed through a 0.45-m filter to remove any remaining particles. The filtered cell extract was loaded onto a HisTrap chelating HP column that had been equilibrated with load buffer (20 mM HEPES, pH 8.0, 300 mM NaCl) on an AKTA purifier (GE Healthcare) at 4°C. The column was washed with column buffer (20 mM HEPES, pH 8.0, 50 mM NaCl), and the FucT protein was eluted with a 20-column volume of a linear imidazole (0.05-1 M) gradient at a flow rate of 1 ml min Ϫ1 . Fractions with FucT activity were pooled and loaded onto a Source 30Q anion exchange column, which had been equilibrated with 20 mM HEPES, pH 8.0. The anion exchange column was washed with 20 mM HEPES, pH 8.0, and eluted with a 20-column volume of a linear gradient of NaCl (0 -1 M) at a flow rate of 1 ml min Ϫ1 . As the presence of imidazole in the eluted fractions interferes with the BCA assay, the protein concentration was determined using a Bio-Rad protein assay kit with bovine serum albumin as protein standard. The identity of the purified FucTs was assessed by SDS-PAGE stained with Coomassie Blue, and the purity was quantified by UN-SCAN-IT gel software (Silk Scientific Corp., Orem, UT).
Storage Conditions of Purified FucTs-To determine the best conditions to store purified enzyme (11639FucT 1-441 and UA948FucT 1-428 ), the purified enzyme preparations were stored at 4, Ϫ20, and Ϫ80°C, with or without the addition of 20 or 50% glycerol. In addition, a 100-l aliquot of purified enzyme was lyophilized and kept at Ϫ80°C. At 2-week and 1-, 2-, and 3-month time points, the specific enzyme activities using Type II-R (1.8 mM) from the 10 different storage conditions were determined as described above. Taking the ␣1,3 activity at day 1 as 100%, the relative levels (%) of enzyme specific activity from the 10 different conditions at the various time points were obtained.
Single Substrate Kinetics of WT and 11639  and UA948  in Cell Crude Extracts-The acceptor and donor kinetics for the fulllength and 11639FucT 1-441 and UA948FucT 1-428 in cell crude extracts were carried out as described previously (24,44). Kinetic parameters were obtained by fitting the initial rate data to the Michaelis-Menten equation using nonlinear regression analysis with Prism 4.0 software (GraphPad, San Diego).
Dual Substrate Kinetics with Purified 11639FucT 1-441 and UA948 1-428 -Steady-state dual substrate kinetics was carried out on purified 11639FucT 1-441 and UA948FucT 1-428 enzyme preparations. Seven different concentrations of the donor and acceptor were used, and GDP- 3 [H]Fuc (0.8 M) was included in each reaction. The amount of substrate consumed was less than 15% to ensure linear initial reaction rates. Data were analyzed as described previously (47) where [A] and [B] represent the concentration of acceptor and donor, respectively. K a is the Michaelis constant for acceptor; K b is the Michaelis constant for donor, and K ia is the dissociation constant for acceptor.
Acceptor Hydroxyl Group Mapping with 11639 1-441 and UA948 1-428 -Purified 11639FucT 1-441 and UA948 1-428 enzyme preparations were used to map the key polar groups of acceptors essential for enzyme recognition. The enzyme activities of using Type II-R (1), Type I-R (13), Type II series 2-12, and Type I series 14-18 modified acceptors were determined with donor GDP-Fuc at 200 M, GDP- 3 [H]Fuc at 0.2 M (ϳ60,000 dpm), and each acceptor at 2 mM. Taking the transfer rates with Type II-R (1) or Type I-R (13) as 100%, the relative enzyme activity using the Type II series-or Type I series-modified acceptors were obtained and expressed as a percentage, respectively. Synthesis of Lewis X or Lewis A Using Purified 11639 1-441 and UA948 1-428 -Type II-R (2 mg, 3.6 mol) and GDP-Fuc (3.5 mg, 5.4 mol) were used for Lewis X synthesis with purified 11639FucT 1-441 in 1.5 ml of HEPES buffer (20 mM, pH 7.0) containing 100 mM NaCl, 35 mM MgCl 2 , 5 mg ml Ϫ1 bovine serum albumin, 20 mM MnCl 2 . The reaction was started by the addition of 180 milliunits of purified 11639FucT 1-441 and 2 units of alkaline phosphatase. The reaction was gently mixed on an end-over-end shaker at room temperature for 20 h. It has been shown that GDP and GMP are strong inhibitors of human FucT V with K i values of 0.03 and 0.70 mM, respectively, whereas guanosine has a K i Ͼ10 mM (20). Alkaline phosphatase was therefore added to degrade GDP to GMP and guanosine to minimize the inhibitory effects of the former two compounds. In the same manner, purified UA948FucT 1-428 was incubated with 5.2 mg (9.4 mol) of Type II-R and 8.6 mg (13.5 mol) of GDP-Fuc or 3.2 mg (6 mol) of Type I-O-R and 5.7 mg (9 mol) GDP-Fuc for Lewis X or Lewis A synthesis, respectively. The conversion of substrates (Type II and Type I) to products (Lewis X and Lewis A) was monitored by TLC. The synthesized products, Lewis X and Lewis A, were characterized by 1 H NMR spectroscopic analysis.

Mapping of the Minimal Catalytic Domain-
In an effort to aid FucT overexpression, purification, and structure determination, the minimal catalytic domain of H. pylori FucTs was mapped by truncating either the N or the C terminus of 11639FucT and UA948FucT. A schematic diagram depicting the full-length (478 amino acids for 11639FucT and 462 amino acids for UA948FucT) and the ⌬FucTs is shown in Fig. 1A. 11639FucT is primarily an ␣-(1,3)-FucT with very low level of ␣1,4 activity (44), thus only the ␣1,3 activity of 11639FucT and its truncated constructs was determined. In contrast, both the ␣1,3 and ␣1,4 activities of UA948FucT and its truncated mutants were determined by radiochemical assays. The enzyme activity of WT and each ⌬FucT was standardized by FucT protein expression level, which was detected by immunoblotting.
To establish if the two putative amphipathic ␣-helices at the C terminus were essential for enzyme activity, we made deletion constructs lacking these helices and either 5 residues (11639FucT 1-447 , UA948FucT 1-434 ) or 11 residues (11639FucT 1-441 , UA948 1-428 ) of the 19-residue linker (Fig. 1A, small white bar) that connects the helices to the heptad repeats (Fig. 1A). The enzyme activities of these four ⌬FucTs were about 3-8 times higher than those of the full-length proteins; however, because the protein expression levels were about 10 -37 times higher (Fig. 2, A and B), their specific enzyme activity turned out to be lower than that of the full-length proteins (Fig. 1, B and C). Protein expression of these four ⌬FucTs was increased dramatically in both soluble and membrane fractions as shown in Fig. 2, C and D. To quantify the distribution of FucT enzyme in the soluble and membrane fractions, we determined the total enzyme activity in each fraction. With fulllength FucTs, only 28% (11639FucT) and 20% (UA948FucT) of total enzyme activities were localized in the soluble fraction. For 11639  and UA948  , this increased to 45 and 47%, respectively, indicating that truncation of the C-terminal putative ␣-helices increased FucT solubility. Nevertheless, a significant amount of active protein was still observed in the membrane fractions when the amphipathic helices were deleted (11639FucT 1-447 , 11639FucT 1-441 , UA948 1-434 , and UA948FucT 1-428 ) (Fig. 2, C and D), suggesting that the C-terminal amphipathic ␣-helices may not be the sole determinants for membrane association.
To determine whether the entire heptad repeat region was required for FucT activity, 11639 1-363 , UA948 1-364 , and 11639 0-HepRep were constructed. After removal of the entire heptad repeat region, these three ⌬FucTs exhibited extremely low levels of enzyme activity, although the protein expression was either moderately higher than (11639 1-363 , UA948 1-364 ) or similar to (11639 0-HepRep ) those of full-length FucTs (Fig. 2). This indicated that the heptad repeat region is essential for enzyme activity.
To determine whether a shorter heptad repeat region was sufficient for conferring enzyme activity, constructs 11639 1-370 and UA948 1-371 were made, which are truncated immediately after the first heptad repeat. In addition, 11639 1-HepRep and UA948 1-HepRep that contain one heptad repeat plus the downstream 19 amino acids linker and the two putative amphipathic ␣-helices were also made (Fig. 1A). Constructs 11639 1-370 and UA948 1-371 had little enzyme activity (Fig. 1, B and C). Remarkably, 11639 1-HepRep displayed enzyme activity comparable with that of the full-length protein. This suggests that a much shorter "stem" region of just one heptad repeat is sufficient for full activity of 11639FucT (Fig. 1B). In contrast, UA948FucT with one heptad repeat (UA948 1-HepRep ) exhibited partial activity compared with that of the full-length enzyme (Fig. 1C). We consistently observed that UA948 1-HepRep displayed a lower protein expression level than the fulllength protein and the other constructs (Fig. 2B), suggesting that this construct was either poorly expressed or the protein was unstable. UA948 0-HepRep was also made; however, it not possible to characterize this mutant because of the lack of protein expression. It is possible that deletion of the entire heptad repeat region of UA948FucT causes folding or degradation problems that prevent proper expression.
Finally, to determine whether or not the N terminus of H. pylori FucTs is essential for function, 11639 10 -447 and UA948 10 -434 were constructed by deleting 10 amino acids at the N terminus of 11639  and UA948 1-434 , respectively. These two truncated constructs completely lost enzyme activity (Fig. 1, B and C), despite the fact that both were expressed at a similar level as full-length enzymes (Fig. 2, A and B). This demonstrates the necessity of the N-terminal residues for enzyme function.
Purification of 11639 1-441 and UA948 1-428 -Removal of the C-terminal helices greatly improved protein expression and solubility while maintaining enzyme activity at a significant level (Figs. 1 and 2). Therefore, 11639FucT 1-441 and UA948FucT 1-428 with a His 6 tag at the C terminus were purified by a sequence of immobilized metal affinity chromatography (IMAC) and anion exchange chromatography (AIEX). The calculated molecular masses of 11639FucT 1-441 and UA948FucT 1-428 are 52.5 and 51.2 kDa, respectively. The FucT bands were visible at the predicted position but did not stand out on a Coomassie Blue-stained SDS-polyacrylamide gel prior to purification (Fig. 3, lanes 1 and 4). The two-step purification resulted in 78.2 and 79.7% purity for 11639 1-441 and UA948 1-428 , respectively (Fig. 3). A minor doublet band is sometimes observed in constructs 11639  and 11639 1-441 either before (Fig. 2C) or after purification (Fig. 3, lane  2). The nature of this band is unknown, but because it runs at higher molecular weight than the main band, it is likely because of a posttranslational change (e.g. oxidation of cysteine). As there was only small amount of this material, any effects on our measurements should be negligible.
From 1-liter cultures, 6.7 and 4.4 mg of FucT protein was isolated, and the enzyme-specific activity was increased from 0.05 to 3.7 units mg Ϫ1 (77-fold) and 2.8 units mg Ϫ1 (52-fold) for 11639FucT 1-441 and UA948FucT 1-428 , respectively. It is worth noting that UA948 1-428 was rather unstable after the IMAC column and would precipitate if the fractions were left at 4 -8°C for 30 -60 min. Therefore, the FucT fractions eluted from the IMAC column were loaded immediately onto the The hypervariable loop that confers Type I acceptor specificity is shown in a narrow black vertical block. B and C, the specific enzymatic activities of the full-length (FL) and truncated FucTs derived from 11639FucT (B) and UA948FucT (C), respectively, were standardized by FucT expression levels, which were detected by immunoblotting as shown in Fig. 2, A and B. The ␣1,3 activity is shown as gray bars, and the ␣1,4 activity is shown as black bars. Each value represents the average of three independent determinations with the standard deviation indicated. 11639FucT is primarily an ␣-(1,3)-FucT with very low amount of ␣1,4 activity (44); thus ␣1,4 activity was not determined. mU indicates milliunits.
AIEX column. Even so, a significant drop in both total and specific enzyme activity was observed during the AIEX purification procedure.
Storage and Stability-To determine the best storage method for the purified FucTs, the ␣1,3 activity was monitored by radiochemical assays following enzyme storage after 2 weeks and 1, 2, or 3 months at three different temperatures (4, Ϫ20, and Ϫ80°C) with different concentrations of glycerol (0, 20, or 50%). In addition, one sample was lyophilized and stored at Ϫ80°C. Lyophilized 11639 1-441 and UA948 1-428 lost 38 and 23% activity after 3 months, respectively. Addition of glycerol (20 or 50%) significantly stabilized the enzyme preparations at both 4 and Ϫ20°C, with little loss of activity after 3 months. FucT preparations were most stable at Ϫ80°C with or without glycerol supplementation; thus this condition has been adopted as our storage protocol.
Single Substrate Kinetics of WT and 11639  and UA948  in Crude Extracts-Removal of the C-terminal putative amphipathic helices facilitated the overexpression and purification of H. pylori ␣-(1,3/ 1,4)-FucTs; however, it was not known whether the truncation affected the kinetic parameters for the enzyme. The single substrate kinetics of full-length and truncated constructs 11639 1-441 and UA948 1-428 in cell crude extracts were characterized and compared.
The acceptor kinetic data showed that truncation of the C terminus did not cause any modification of the Type II acceptor kinetics for 11639FucT; however, although the acceptor K m remained unchanged for UA948FucT, the V max decreased by 50% (Table 1). With Type I acceptor, UA948 1-428 possessed a 2-fold tighter acceptor binding affinity, which exactly compensated for the 2-fold reduction of the V max value when calculating the catalytic efficiency (V max /K m ) of the enzyme. Notably, the C-terminal truncation improved donor binding by about 3-fold with both Type II and Type I acceptors for both FucTs.
Dual Substrate Kinetics of Purified 11639 1-441 and UA948 1-428 -The success in purifying milligrams/liter of 11639 1-441 and UA948 1-428 allowed us to perform dual substrate kinetics with seven different concentrations of the donor GDP-Fuc and seven different concentrations of acceptor (Type II or Type I). GDP- 3 [H]Fuc at 0.8 M was included in each reaction. The kinetic parameters are shown in Table 2. The acceptor and donor K m values for fucose transfer to Type II acceptor by 11639FucT 1-441 and UA948FucT 1-428 are similar, with those of 11639FucT 1-441 just slightly lower ( Table 2). The K m values of 11639FucT 1-441 and UA948FucT 1-428 are 2.3 and 1.1 s Ϫ1 , respectively, which give rise to the catalytic efficiency parameters (k cat /K a and k cat / K b ) of 11639FucT 1-441 , with Type II acceptor being ϳ3-fold those of UA948FucT 1-428 . UA948FucT possessed the same k cat values (1.1 s Ϫ1 ) for Type II and Type I acceptors. However, the much lower K m value for the Type II acceptor resulted in a 28-fold higher k cat /K a with Type II compared with that of the Type I acceptor; therefore, UA948FucT strongly favors Type II over Type I acceptor. For the donor GDP-Fuc, UA948FucT 1-428 displayed a slightly tighter binding when fucose was transferred to Type I versus Type II acceptor. The acceptor and donor K m values obtained from dual substrate kinetics with the purified enzyme ( Table 2) are lower than those obtained from single substrate kinetics with cell crude extracts (Table 1), except for the K m using Type I acceptor for UA948  .
To determine the kinetic mechanisms of 11639 1-441 and UA948 1-428 , a pattern of double-reciprocal plots for velocity and concentration of either substrate at a series of fixed concentrations of the second substrate was produced. The resulting family of intersecting lines (data not shown) excludes a double displacement (ping-pong) mechanism. Therefore, we can conclude that the catalytic mechanism of H. pylori FucTs is sequential and that both substrates bind to the enzyme before any product is released.  Acceptor Hydroxyl Group Mapping with H. pylori FucTs-To map the molecular determinants in Type II and Type I acceptors that are essential for recognition by H. pylori ␣-(1,3/1,4)-FucTs, 16 synthetic acceptors with modifications at the hydroxyl and N-acetamido groups were employed. Because of the limited availability of the compounds, the enzyme activities were determined using 2 mM of each acceptor. In comparison to the transfer rate with unmodified Type II acceptor (1) or Type I acceptor (13), the relative rates (%) of transfer for each modified acceptor were obtained (Table 3).
Purified 11639 1-441 and UA948 1-428 FucTs exhibited variable levels of enzyme activity using 12 different Type II series acceptors (Table 3). Substitution of the hydroxyl group at C-3 of the galactose with a sulfate group (2) and replacement of the hydroxyl group at C-2 of the galactose with fucose (3) resulted in enhanced or comparable rates of transfer for 11639 1-441 and UA948 1-428 , respectively, relative to the unmodified acceptor (1). Addition of a bulky sialic acid moiety to the C-3 of galactose (4) greatly reduced the rate of transfer, particularly for UA948FucT (Table 3).
Modifications at the C-4 position with deoxygenation (5) reduced the activity about 4-and 13-fold for 11639FucT and UA948FucT, respectively ( Table 3). The most dramatic impact was observed by modifications at C-6. When this hydroxyl was deoxygenated, compound (7) became a very poor substrate for both 11639 1-441 and UA948 1-428 with 0.02% of transfer rate relative to the unmodified Type II acceptor (1). Replacement of the C-6 hydroxyl group by a sulfate group (8) gave a slightly improved transfer rate at 2 and 0.04%, compared with compound (7) for 11639 1-441 and UA948 1-428 , respectively. Such dramatic decreases in activity qualify the hydroxyl at C-6 of galactose in Type II acceptor as a key polar group required for enzyme recognition (Fig. 4). In general, 11639 1-441 tolerated modifications on the terminal galactose moiety of Type II acceptor (compounds 2 and 4-6) better than UA948  . This may be due to the lower K m for Type II acceptor of 11639FucT than that of UA984FucT (Table 1 and Table 2).
Modifications of the GlcNAc moiety in Type II acceptor greatly reduced the transfer rate for both enzymes. Phosphorylation of the hydroxyl group at C-6 of the GlcNAc (8) caused a large decrease in activity for both 11639 1-441 and UA948  . The lack of an N-acetamido group (9) also severely impaired activity. Modification of compound (9) by adding a galactose at C-3 (10) resulted in a better acceptor. Apparently, the protein can sterically accommodate additions as large as a galactose to the C-3 position, and the enhanced enzyme activity likely results from favorable interactions of this moiety with the FucT protein. Modification of compound (9) by adding a galactose at C-4 (11) did not show the same increase in activity as compound (10), but neither did it reduce activity completely. Again, addition at the C-4 position appears to be sterically tolerated, suggesting that the C-4 hydroxyl points away from the FucT protein.
We also deoxygenated the C-3 of the GlcNAc of Type II substrate to evaluate if compound (12) could function as an inhibitor. When 4.4 mM of compound (12) and 55.6 M of substrate (1) were included in the reaction, 42.5 and 28.3% inhibition was observed for 11639 1-441 and UA948 1-428 , respectively. The limited availability of compound (12) precluded us from performing a complete kinetic study to obtain an accurate K i . Nevertheless, assuming that the inhibition is competitive, a K i of 5.0 and 9.8 mM could be calculated for 11639       acceptor concentration, and K m is the acceptor K m for compound (1). The high K i value of compound (12) argues for an important role for OH-3 of Type II acceptor in recognition by 11639FucT and UA948FucT. Such weak inhibition has also been observed for FucTs isolated from human milk when 4-OH of GlcNAc in Type I acceptor was deoxygenated (15,17). Five modified Type I series analogues were also utilized to map the hydroxyl groups in Type I acceptor that are required for recognition by UA948   (Table 3). Substitutions of the hydroxyl group at C-3 of the galactose with a sulfate group (14) and the C-2 of the galactose with a fucose group (15) were relatively well tolerated. In contrast, the replacement of C-3 of the galactose with sialic acid (16) created a very poor acceptor. Notably, deoxygenation at C-6 of the galactose (17) rendered the compound completely inactive, indicating that the hydroxyl group at C-6 of the galactose in Type I acceptor is absolutely required for recognition by UA948  . As expected, deoxygenation at C-4 of GlcNAc in Type I structure (18) made the compound unable to function as an acceptor. When Type I analogue (18) at 4.4 mM was evaluated as an inhibitor with (13) being included in the reaction at 55.6 M as an acceptor, no inhibition was observed. This indicates that the reactive hydroxyl in Type I acceptor is crucial for enzyme recognition.

Type II-R b Type I-R c K m for GDP-Fuc
Synthesis of Lewis X and Lewis A Using Purified 11639 1-441 and UA948 1-428 -Purified 11639 1-441 and UA948 1-428 were used for enzymatic synthesis of Lewis X and Lewis A. The conversion rate from substrate to product in each reaction was monitored by TLC (CHCl 3 / MeOH/water ϭ 70:35:2). When TLC showed the complete disappearance of the substrate and the formation of a single product, the reaction was terminated. The product was isolated by column chromatography on a reverse-phase C 18 cartridge and was subsequently lyophilized, with yields ranging from 87 to 94%. The products (Lewis X and Lewis A) and the starting substrates (Type II and Type I) were analyzed by 1 H NMR spectroscopy, and the parameters of the feature peaks are given in the Supplemental Material. The chemical shifts for products are identical to literature values for Lewis X and Lewis A (48,49).
Removal of the putative ␣-helices at the C terminus did not significantly reduce the specific enzyme activity, indicating that the amphipathic helices are not absolutely required for conferring activity. Remarkably, the truncation greatly increased protein expression and protein solubility (Fig. 2), which enabled us to purify a significant amount of soluble active enzyme for detailed kinetic characterization and key polar group mapping. Previously, similar increases of protein expression have been observed by deleting the C-terminal putative amphipathic helices in ␣1,4-galactosyltransferase LgtC from Neisseria meningitidis (50) and the sialyltransferase CstII from Campylobacter jejuni (51). It is worth noting that the increase in protein solubility after removal of the putative amphipathic helices supports their role as a membrane anchor. Nevertheless, a substantial portion of protein still localized to the membrane fraction when they were deleted (Fig. 2, C  and D). This remained the case even when the preceding heptad repeats were also removed (data not shown). It is possible that the catalytic domain possesses an additional membrane-attachment region, which remains to be identified. Alternatively, protein overexpression may have caused membrane association because of the exposed hydrophobicity of some partially unfolded proteins.
In H. pylori FucTs, the putative membrane anchor is connected to the catalytic domain by a series of 2-10 leucine zipper-like heptad repeats (41), which has been speculated to mediate formation of a functional dimeric enzyme (21)(22)(23). Deletion of the entire heptad repeat region (11639 1-370 , UA948 1-371 , and 11639 0-HepRep ) almost completely abolished enzyme activity for both 11639FucT and UA948FucT, suggesting that the heptad repeat region is essential for enzyme activity. Because as few as two heptad repeats have been observed in nature (41), it is not surprising to observe the full or partial enzyme activity present in the single heptad repeat constructs, 11639 1-HepRep and UA948 1-HepRep , respectively. Most interestingly, the amphipathic helices are necessary for function when the repeat region is reduced to one heptad repeat (Fig.  1). In addition, the 11639FucT heptad repeats truncations are typically more stable and have higher activity than the equivalently truncated UA948 FucT constructs (Fig. 1). These observations suggest a more complex model where the C-terminal amphipathic helices, heptad repeats, and perhaps the catalytic domain all contribute to dimerization; thus full activity is obtained as long as the net interaction energy is high enough to form a stable dimer. We are currently pursuing this hypothesis, and our preliminary dynamic light scattering data has confirmed that 11639 1-441 indeed forms a dimer (results to be published elsewhere). A functional requirement for dimerization has not yet been reported for any other ␣-(1,3/1,4)-FucTs, but dimeric forms of human FucT VII exist within cells (52).
The dual substrate kinetic analysis shows that H. pylori ␣-(1,3/1,4)- FucTs possess comparable kinetic parameters to their mammalian counterparts. Specifically, the donor GDP-Fuc K m values for H. pylori ␣-(1,3/1,4)-FucTs are similar to those for human FucT III, IV, V (16), and VI (8), and the K m values for Type II-R acceptor (RϭO(CH2) 8 CO 2 CH 3 ) are similar to that of FucT VI (8). With Type II and Type I acceptors that do not contain the hydrophobic linker (ϪO(CH2) 8 CO 2 CH 3 ), human FucT III and V display a low k cat ranging from 5.2 to 9.3 min Ϫ1 (13). The newly characterized H. pylori ␣-(1,3/ 1,4)-FucT from strain DMS6709, which is primarily an ␣-(1,4)-FucT, possesses a similar acceptor K m (313 M) for Type II-R acceptor, a lower K m (114 M) for Type I acceptor, and a relatively low K m for donor (5.73 M). Compared with UA948FucT, DMS6709FucT has 24-fold higher k cat for Type I acceptor (26 s Ϫ1 ) and 16-fold lower K m for Type II acceptor (0.067 s Ϫ1 ) (25). It is interesting that the Type II acceptor preference in UA948FucT is because of the 21-fold lower K m for Type II than that for Type I, but their k cat values are identical, whereas DMS6709FucT favoring Type I acceptor results from the 388-fold lower k cat but only Ͻ3-fold lower K m values for Type I compared with Type II acceptor.
Similar to human FucT V (20), the pattern of the double-reciprocal plots for velocity with concentration of either substrate at a series of fixed concentrations of the second suggests that H. pylori ␣-(1,3/1,4)-FucTs catalyze fucose transfer following a sequential mechanism. Our current study cannot distinguish whether the substrate binding is random or ordered, and if it is ordered, whether the acceptor or the donor binds first. In the process of optimizing the purification protocol, we observed that both 11639 1-441 and UA948 1-428 bind to GDP-hexanolamine-Sepharose affinity chromatography support columns but not to immobilized acceptor-based affinity columns, 4 indicating that H. pylori FucTs most likely binds to the donor first, as has been observed for most other glycosyltransferases (53). Further detailed kinetic and structural studies are required to fully resolve this issue.
It has been shown that mammalian ␣-(1,3/1,4)-FucTs are able to use 3Ј-sialylated Type II or 3Ј-sialylated Type I acceptors to synthesize 3Ј-sialylated-Lewis X or 3Ј-sialylated-Lewis A, respectively, but with variable efficiencies. Our hydroxyl group mapping data demonstrates that UA948FucT tolerates sialylation very poorly. Compared with mammalian ␣-(1,3/1,4)-FucTs, H. pylori 11639FucT and 11637FucT resemble human FucT VI (8) in their preference for neutral Type II acceptor and tolerance for 3Ј-sialylation, whereas UA948FucT behaves more like human FucT IV and V for its ␣1,3 activity and ␣1,4 activity, respectively. Human FucT IV transfers fucose much more efficiently to neutral Type II acceptors than those that are sialylated (6). FucT V prefers both fucosylated and sialylated Type II acceptor; however, it does not favor sialylated Type I structures (16). The ␣-(1,3/1,4)-FucT from strain DMS6709 is more like human FucT III as both of them prefer the sialylated Type I over the unmodified Type I acceptor (25).
Although it appears most H. pylori FucTs can utilize sialylated acceptors, only very few H. pylori isolates (2 out 94) were found to express 3Ј-sialylated-Lewis X (54), such as strain P466 (55). This is most likely because of the absence of a functional ␣-2,3-sialyltransferase, which has not yet been identified in any H. pylori strains. It would be of interest to determine whether H. pylori ␣-2,3-sialyltransferase is able to use the 2Ј-fucosylated Type II or 2Ј-fucosylated Type I as acceptors, like the ␣-2,3-sialyltransferase from myxoma virus (51), or if sialylation has to precede to fucosylation as has been reported for mammalian ␣-2,3sialyltransferases (56).
Our hydroxyl group mapping data show that the 2-fucosylated Type II (H-Type II (3)) is an excellent acceptor for both H. pylori FucTs, and 2-fucosylated Type I (H-Type I (15)) is also a fairly good acceptor for UA948  . This challenges the notion that the difucosylated Lewis antigens (Lewis Y and Lewis B) in H. pylori are synthesized predominantly via subterminal fucosylation (in ␣1,3or ␣1,4-linkage) followed by terminal fucosylation (in ␣1,2-linkage) (57). Martin et al. (22) showed that 11637FucT was unable to use Fuc(␣1,2)Gal␤1,4Glc as an acceptor. Because the lack of the N-acetamido group in Type II structure severely impaired enzyme activity (Table 3), one cannot conclude that 11637FucT is unable to use Fuc(␣1,2)Gal␤1,4GlcNAc (H-Type II) as an acceptor. The ␣1,2-FucTs from H. pylori were reported to use both the unfucosylated (Type II and Type I) and the fucosylated (Lewis X and Lewis A) structures as acceptors (28). Therefore, we propose that the difucosylated Lewis antigens in H. pylori can be synthesized via either route as follows: ␣-(1,3/1,4)-fucosylation followed by ␣1,2-fucosylation or vice versa. The route used predominantly in vivo will depend on activity and expression levels as well as the substrate preferences of ␣-(1,3/1,4)-FucTs and ␣1,2-FucT in a particular H. pylori strain.
In H. pylori strains NCTC11639 and UA948, this is not an issue because their futC gene does not encode a functional ␣1,2-FucT protein (23,57).
The key polar group mapping data also show that phosphorylation of the GlcNAc at 6-C position (8) caused a dramatic reduction in activity for both enzymes, and the lack of the N-acetamido group (9) also generated a dramatic decrease in activity, by 250-fold for 11639FucT and by 100-fold for UA948FucT. This implies that the 6-OH and the acetamido group in GlcNAc of Type II acceptor contribute to optimal enzyme activity. It has been shown previously that the acetamido group in Type II acceptor is required for optimal activity of human FucT III, IV, and V (16).
Our key polar group mapping data suggest that the hydroxyl at C-6 of galactose in Type I and Type II acceptors is a key polar group essential for recognition by H. pylori ␣-(1,3/1,4)-FucTs (Fig. 4). For the reactive hydroxyl group in Type II acceptor, when it is deoxygenated, the compound functions as a weak inhibitor, but the relatively high K i value suggests that this hydroxyl does play an important role in enzyme binding. When OH-4 of the GlcNAc moiety in Type I acceptor is deoxygenated, we did not observe any inhibition. As UA948 1-428 possesses a relatively high K m value for Type I acceptor, the actual K i value is hard to quantify. However, if we assume competitive inhibition and a 5% detection limit of the assay, we estimate the K i to be 83 mM or greater. A similar sensitivity pattern has been reported for mammalian ␣-(1,3/ 1,4)-FucTs (8,(15)(16)(17). This suggests that both mammalian and H. pylori ␣-(1,3/1,4)-FucTs bind the Type II and Type I compounds in a similar manner. In our previous work, we have proposed that the GlcNAc moiety in Type I structure is rotated by 180°relative to its location in Type II compound, bringing the 4-OH of GlcNAc in Type I to the same position as the 3-OH of GlcNAc in Type II (44). The 6-CH 2 OH and acetamido groups in Type I are also rotated 180°relative to those in Type II structures. This hypothesis is strengthened by NMR studies of Lewis X and Lewis A trisaccharides, where the fucose and galactose in both compounds occupy very similar relative positions with the major difference being the opposite orientation of their GlcNAc moiety (58). We noticed that the absence of the acetamido group (9) is the only modification that causes a less dramatic effect on the transfer rate observed by UA948FucT as compared with 11639FucT (Table 3). This suggests that the N-acetamido group contributes less to recognition by UA948FucT than 11639FucT. This may at least partially explain why Type I acceptor cannot be properly accommodated by 11639FucT because of the flipped GlcNAc moiety.
In conclusion, H. pylori ␣-(1,3/1,4)-FucTs and their mammalian counterparts seem to share striking functional similarities despite a very low level of sequence homology. The functional similarities that have been shown previously with respect to domain architecture and the region and residues that discriminate Type I from Type II acceptors (24,44) have now been extended to include the minimal catalytic domain, kinetic parameters, and key polar groups of acceptors that are essential for enzyme recognition. Our data suggest that all ␣-(1,3/1,4)-FucTs very likely share a conserved mechanistic and structural basis for fucose transfer. This makes H. pylori ␣-(1,3/1,4)-FucTs an attractive model system for studying the entire FucT family, and the success in purifying soluble, stable, and active truncated forms of H. pylori ␣-(1,3/1,4)-FucTs allows structural studies to be pursued. In addition, H. pylori ␣-(1,3/1,4)-FucTs have great potential to be exploited in the enzymatic synthesis of fucosylated glycoconjugates. Such glycoconjugates have generated broad pharmaceutical interest for the prevention of bacterial and viral infections, the neutralization of toxins, and in immunotherapy for cancer (59 -62).