The Acceptor and Site Specificity of a 3-Fucosyltransferase V HIGH REACTIVITY OF THE PROXIMAL AND LOW OF THE DISTAL Gal b 1–4GlcNAc UNIT IN i-TYPE POLYLACTOSAMINES*

We report here on in vitro acceptor and site specificity of recombinant a 3-fucosyltransferase V (Fuc-TV) with 40 oligosaccharide acceptors. Gal b 1–4GlcNAc (LN) and GalNAc b 1–4GlcNAc (LDN) reacted rapidly; Gal b 1–3GlcNAc (LNB) reacted moderately, and GlcNAc b 1–4GlcNAc ( N , N * -diacetyl-chitobiose) reacted slowly yet distinctly. In neutral and terminally a 3-sialylated polylactosamines of i-type, the reducing end LN unit reacted rapidly three successive steps, involving a b 4-galacto- syltransferase reaction, a b 3-GlcNAc transferase reaction (30), and another b 4-galactosyltransferase reaction. The N , N 9 , N 0 , N 999 -tet- raacetyl-chitotetraose (Glycan 18 ) was purchased from Seikagaku (To-kyo, Neu5Ac a 2–3 9 LN b 1–2Man (Glycan 19 was obtained by incubating 600 nmol of GlcNAc b 1–2Man simultaneously with 1200 nmol of UDP-Gal and 75 milliunits of bovine milk b 4-galactosyltrans- ferase and with 1800 nmol of CMP-Neu5Ac as well as 12 milliunits of a 3-( N )-sialyltransferase (ST3 Gal III; Calbiochem, La Jolla, CA) in 150 m l of 50 m M MOPS, pH 7.5, 10 m M MnCl 2 , 0.02% sodium azide at room temperature for 66 h. The acidic reaction product was isolated by chromatography, as specified below, first on Superdex peptide HR gel filtration column, then on a Mono Q 5/5 ion exchange column, and finally again on the Superdex peptide HR column. Neu5Ac a 2–3 9 LN b 1– 6Gal (Glycan 20 ) was obtained from GlcNAc b 1–6Gal by using the combined b 4-galactosylation and a 3-sialylation process described above. specificity each transfected ly- sates was tested with LN, LNB, and Neu5Ac a 3 9 LN experiments. The data resembled closely those All Fuc-TV reactions with CHO cell lysates were performed and analyzed at least twice. During the work up, the acceptors were directly monitored by UV absorption in all reactions, and no significant acceptor degradation was observed. No glycosidase inhibitors were used to protect the acceptors. Quantitation of Oligosaccharides— Oligosaccharides were assayed by integration the peaks in The data were related to the of external equal H 2 O (99.996 atom %, Cambridge Isotope Lab-oratories, Woburn, MA), and the NMR experiments were carried out in a Varian Unity 500 spectrometer at 296 K. A modified WEFT sequence was used for water suppression (46). The chemical shift data are based on internal acetone signals at 2.225 ppm. MALDI-TOF Mass Spectrometry— MALDI-TOF mass spectrometry was performed as described previously (11, 36). The present experiments extend the long list of small oligosaccharides that show acceptor activity for Fuc-TV. We show here that Fuc-TV (GalNAc as leading which possesses a rigid conformation in solution similar to Lex 48). Both the full-length as well as the truncated recombinant Fuc-TV reacted slowly with N , N 9 -diacetyl-chito-biose, N , N 9 , N 0 -triacetyl-chitotriose and N , N 9 , N 0 , N 999 -tetra-ace-tylchitotetraose in the present experiments. Preparative scale experiments, to elsewhere, have revealed that the fucosylated chitin glycans represent the trisaccharide GlcNAc b the tetrasaccharide GlcNAc b and the pentasaccharide 4 In addition, the present data show some acceptors Fuc-TV.

Six human ␣3-fucosyltransferases (Fuc-Ts) 1 have been cloned (1)(2)(3)(4)(5)(6)(7)(8)(9)(10), and all of them convert N-acetyllactosamine units (LN) of glycoconjugates into Lewis x (Lex) determinants, Gal␤1-4(Fuc␣1-3)GlcNAc. Sialylated and/or sulfated versions of this trisaccharide are believed to decorate L-selectin ligands, which initiate extravasation of lymphocytes by mediating their tethering and rolling on endothelium during their recruitment into lymph nodes and into sites of inflammation (11,12). Other leukocytes extravasate with mechanisms similar to those of lymphocytes. In some cases, also bacterial adhesion to target cells at the onset of infections (13), as well as metastasis of malignant tumor cells via blood circulation (14,15), are known to be mediated by interactions between this family of saccharides and appropriate lectin proteins. Homotypic Lex-Lex binding has also been observed and may play a role in cell adhesion, e.g. in embryonal compaction (16 -20).
Emerging differences in substrate specificities of the human Fuc-Ts (10,(21)(22)(23) may help to define the biological roles of the individual enzymes. The substrate specificities of these enzymes are of interest also for enzyme-assisted in vitro synthesis of different types of Lex-containing glycans and glycoconjugates, many of which are capable of inhibiting important selectin-mediated cell adhesion processes. Among this group of enzymes, Fuc-TV has probably the broadest acceptor specificity. It transfers at significant rates to both type 1 and 2 acceptors, Gal␤1-3GlcNAc (LNB) and LN, respectively, and works well also with ␣2,3-sialylated and ␣1,2-fucosylated forms of these disaccharides; in addition, it transfers to 3Ј-sulfo-and 6-sulfo-LN, to lactose as well as to ␣2Ј-fucosylated lactose (5, 24 -26). This exceptional versatility of Fuc-TV as a synthesis catalyst prompted us to study its reactions with a large number of polylactosamines and other acceptor saccharides. Here, we report on the relative Fuc-TV reactivities of these glycans and their individual acceptor sites.
Transfected Cells and Lysates-The transfection and stable expression of human Fuc-TV in CHO cells has been described previously (5). For the enzyme assays, the transfected cells were lysed in 1% Triton X-100 on ice in the presence of protease inhibitors (22).
The reactions were halted by adding 10 l of ice-cold ethanol followed by 100 l of ice-cold water and were frozen and stored at Ϫ20°C until the time of analysis. The reaction mixtures obtained from neutral acceptors were desalted in a mixed bed of ion exchange resins, after which the mixtures of the unlabeled acceptor and the corresponding radiolabeled product(s) were isolated by gel filtration on a Superdex peptide HR column (Amersham Pharmacia Biotech).
The reaction mixtures obtained from sialic acid-containing acceptors were separated from neutral saccharides and from GDP-[ 14 C]Fuc as described (7). The monosialo-[ 14 C]fucosyl products were then isolated as a mixture with the acceptor. For this purpose, a process involving (i) gel filtration chromatography on a Superdex peptide HR 10/30 column, (ii) ion exchange chromatography on a Mono Q column, and (iii) another run on the Superdex peptide column was used. Alternatively, the reaction mixtures were directly applied to the three-step chromatographic separation process. Here, most of the GDP-[ 14 C]Fuc was eliminated from the products in the first Superdex peptide column run. The alternative process was always used for the work up of multisialylated products.
The substrate specificity of each batch of transfected CHO cell lysates was tested with LN, LNB, and Neu5Ac␣3ЈLN before the actual experiments. The data resembled closely those described (5). All Fuc-TV reactions with CHO cell lysates were performed and analyzed at least twice. During the work up, the acceptors were directly monitored by UV absorption in all reactions, and no significant acceptor degradation was observed. No glycosidase inhibitors were used to protect the acceptors.
Quantitation of Oligosaccharides-Oligosaccharides were assayed by integration of the peaks obtained in Superdex Peptide HR 10/30 chromatography with UV detection. The data were related to the peaks of external GlcNAc and Neu5Ac standards. GlcNAc and GalNAc revealed equal molar absorbances.
Chromatographic Methods-Gel filtration was performed in a column of Superdex peptide HR 10/30 (Amersham Pharmacia Biotech) with 50 mM NH 4 HCO 3 as the eluant; a flow rate of 1 ml/min was used. The effluent was monitored by UV absorption at 205 or 214 nm. Anion exchange chromatography on a Mono Q (5/5) column (Amersham Pharmacia Biotech) was performed as described (30). Paper chromatography of radiolabeled oligosaccharides was performed as described (32). High pH anion exchange chromatography with pulsed amperometric detection was carried out with a Dionex series 4500i high pressure liquid chromatography system (Dionex, Sunnyvale, CA) equipped with a Car-boPac PA-1 column (4 ϫ 259 mm). The column was equilibrated and run with 40 mM NaOH at a flow rate of 1 ml/min as described (41).

Analysis of [ 14 C]Fucosaccharide Isomers Generated in Fuc-TV Reactions of Polylactosamines That Contained Multiple Potential Acceptor
Sites-Analysis of neutral [ 14 C]fucosylated isomers generated by Fuc-TV from type 2 polylactosamines was performed as described in Fig. 1. Analogous analysis of [ 14 C]fucosylated type 1 glycans was performed by a treatment with ␤3-galactosidase of Xanthomonas manihotis (New England Biolabs, Beverly, MA), followed by paper chromatography of the resulting [ 14 C]oligosaccharides. Control experiments involving 2-nmol samples of type 1 reference oligosaccharides were monitored by MALDI-TOF mass spectrometry and revealed that the ␤3-galactosidase treatment cleaved completely the tetrasaccharide LNB␤1-3ЈL, whereas the pentasaccharide Gal␤1-3(Fuc␣1-4)GlcNAc␤1-3ЈL was not degraded at all (data not shown). Analysis of neutral [ 14 C]fucosylated i-type polylactosamines was performed as described (22). Corresponding analyses of type 1 and type 2 sialopolylactosamines were performed after enzymatic desialylation. Analysis of the Fuc-TV reaction products of Lex␤1-3ЈLN␤1-3ЈLN was performed by B. fragilis endo-␤-galactosidase digestion (45). 1 H NMR Spectroscopy-Prior to NMR experiments the saccharides were twice dissolved in 2 H 2 O and evaporated to dryness. The samples were then dissolved in 2 H 2 O (99.996 atom %, Cambridge Isotope Laboratories, Woburn, MA), and the NMR experiments were carried out in a Varian Unity 500 spectrometer at 296 K. A modified WEFT sequence was used for water suppression (46). The chemical shift data are based on internal acetone signals at 2.225 ppm.
Fuc-TV Reactivities of Distal, Monovalent LN, and Sialyl-LN Determinants-Two trisaccharides, mimicking branches of Nglycans showed widely different Fuc-TV reactivities. LN␤1-6Man␣1-OMe, representing a substituted ␤-methyl glycoside of LN, was a good acceptor (Table II). Concomitantly, LN␤1-2Man, where the LN and the mannose are joined through an axial bond that generates a more crowded unit, revealed an 8-fold smaller reactivity. The corresponding ␣3-sialoglycans reacted better than the asialoglycans.
Two galactose-containing trisaccharides, mimicking elements of polylactosamine backbones, were also studied. Even here, the isomer representing the substituted ␤-methyl glycoside of LN, i.e. LN␤1-6Gal, showed good reactivity, whereas the more crowded sequence LN␤1-3Gal␤1-OMe was a relatively poor acceptor. The corresponding ␣3-sialoglycans reacted 2-2.5 times better than the asialoglycans. The LN unit of LN␤1-4GlcNAc appeared to be a good acceptor, 3 and so were the two O-glycans of Core 2 type, both representing molecules of the type of substituted ␤-methyl glycosides of LN (Table II).
Fuc-TV Reactivities of LN and LNB Units in Neutral Polylactosamines-Relative Fuc-TV reactivities of individual LN units of multisite polylactosamine acceptors were determined using an approach described in Fig. 1, where an analysis of mono- The data in Table III show that the reactivity of GlcNAc␤1-3ЈLN was 3-fold enhanced relative to free LN and was further up-regulated at the reducing end of the tetrasaccharide LN␤1-3ЈLN (the reactive sites being underlined). Among the LN units of the hexasaccharide LN␤1-3ЈLN␤1-3ЈLN and the octasaccharide LN␤1-3ЈLN␤1-3ЈLN␤1-3ЈLN, the best reactivity was also found at the reducing end LN. However, the levels of reducing end reactivity in these large glycans were not quite as 3 The [ 14 C]fucosylated product generated by Fuc-TV from the acceptor LN␤1-4GlcNAc resisted a treatment with jack bean ␤-galactosidase, whereas almond meal fucosidase released 89% of the [ 14 C]fucose as free monosaccharide. Both features are characteristic to the sequence Gal␤1-4(Fuc␣1-3)GlcNAc␤1-OR (45,64,65), and neither of them is expected to be a property of the glycan Gal␤1-4GlcNAc␤1-4(Fuc␣1-3)GlcNAc. Hence, we conclude that LN␤1-4GlcNAc probably reacted with Fuc-TV mainly at the LN unit, and the reactivity of this unit was not appreciably inhibited by the presence of the proximal GlcNAc.  (4) 1.0 LDN (5) 0.82 LDN␤1-OMe (6) 1.2 LNB (2) 0.13 N,NЈ-diacetylchitobiose (7) 0.04 c Man␤1-4GlcNAc (3) 0.03 d a Data were obtained at 5 mM acceptor concentrations. b The data were obtained using 3 mM acceptor concentration and were compared with those of 3 mM LN.
c The data were obtained with the full-length Fuc-TV; slightly higher values were obtained with purified truncated recombinant Fuc-TV. N,NЈ,NЈЈ-Triacetylchitotriose and N,NЈ,NЉ,NЈЉ-tetra-acetylchitotetraose reacted also with both samples of Fuc-TV.
d The data were obtained with commercial purified truncated recombinant Fuc-TV.

The Acceptor and Site Specificities of Fuc-TV
high as in the tetrasaccharide LN␤1-3ЈLN, possibly because of the presence of higher amounts of total LN units. The high reactivity of the proximal LN unit in the unconjugated saccharides was preserved in the ␤-methylglycoside of the tetrasaccharide LN␤1-3ЈLN.
The second LN in the tetrasaccharide LN␤1-3ЈLN and and its ␤-methyl glycoside, LN␤1-3LN␤1-OMe reacted very slowly. This property of the second LN was related to its most distal position, because the second LN in the hexasaccharide LN␤1-3LN␤1-3ЈLN and its analog in the octasaccharide LN␤1-3ЈLN␤1-3ЈLN␤1-3ЈLN reacted fairly well (Table III). This view is also supported by the data showing that the third LN unit of the hexasaccharide LN␤1-3ЈLN␤1-3ЈLN reacted extremely slowly, but the third LN unit in the octasaccharide LN␤1-3ЈLN␤1-3ЈLN␤1-3ЈLN reacted fairly well. The most distal LN units of the linear polylactosamines LN␤1-3ЈLN␤1-3Gal␤1-OMe and LN␤1-3ЈLN␤1-3ЈLN␤1-3ЈLN also reacted very slowly. Considered together, the data in Table III show that the Fuc-TV reactivity of the proximal LN of neutral i-type chains was consistently high, whereas the distal LN units always reacted poorly. The midchain LNs of the chains possessed intermediate reactivities.
To rule out the possibility that the apparent poor distal fucosylation was due to the presence of site specific ␣3-fucosidases in lysates of the transfected CHO cells, a preparative experiment was performed using purified soluble recombinant Fuc-TV as the enzyme. Here, 29 mol % of the the pentasaccharide LN␤1-3ЈLN␤1-3Gal␤1-OMe (Glycan 23) was converted to a monofucosylated fraction according to MALDI-TOF mass spectrometry. The monofucosylated product was isolated by high pH anion exchange chromatography on a Dionex Car-boPac PA-1 column, and was identified as LN␤1-3ЈLex␤1-3Gal␤1-OMe by 1 H NMR spectroscopy (22) as well as by exoglycosidase degradation monitored by MALDI-TOF mass spectrometry (data not shown). Hence, the poor relative reactivity with distal LN units of neutral i-type polylactosamines, observed with Fuc-TV of the CHO cell lysates, is a property of the purified Fuc-TV too and does not represent an artifact caused by fucosidase action.
The Fuc-TV present in CHO cell lysates reacted well with the inner LN unit of the pentasaccharide LNB␤1-3ЈLN␤1-3Gal␤1-OMe (LNB, Gal␤1-3GlcNAc); the reaction at the distal LNB unit was much weaker, yet significant (Table III). Indeed, the Fuc-TV reactivity of LNB␤1-3ЈLN␤1-3Gal ␤1-OMe (at the underlined unit) was similar to that of free LNB, contrasting strongly with the dissimilarity of the reactivities of the analogous pair of LN␤1-3ЈLN␤1-3Gal␤1-OMe and free LN (cf. Tables I and III). In the branched i-type hexasaccharide LN␤1-3Ј(LN␤1-6Ј)LN, the other LN units reacted poorly, but the 3-linked distal LN unit reacted faster than the LN units at the distal end of i-type chains.
Fuc-TV Reactivities of LN Units in ␣3-Sialylated Polylactosamines-The data of Table III show that all reliably assessed LN units in ␣3-sialylated polylactosamines reacted almost twice as rapidly as their counterparts in neutral polylactosamines. The very low reactivity of the distal, sialylated LN unit in Neu5Ac␣2-3ЈLN␤1-3ЈLN and Neu5Ac␣2-3ЈLN␤1-3ЈLN␤1-3Gal␤1-OMe compared with that of Neu5-Ac␣2-3ЈLN is noteworthy; it shows that a proximal ␤1-3Јbonded LN neighbor reduces strongly the Fuc-TV reactivity of a LN unit in ␣3-sialylated i-type polylactosamines in the same way as in neutral polylactosamines. Some reduction of the reactivity in the Neu5Ac␣2-3ЈLN was caused already by a proximal, ␤1-3Јbonded Gal␤1-OMe neighbor. The low reactivity of the 1,6-linked branch, compared with the 1,3-linked main chain terminus, in the doubly sialylated i-type polylactosamine reveals the presence of a high level of branch selectivity in Fuc-TV.
The Fuc-TV reactivities of the reducing end LN units were high in all i-type polylactosamines. In striking contrast, the reactivities of the nonreducing end LN units were very low in all i-type polylactosamines. Our data are in agreement with the recent observation that Fuc-TV transfers more rapidly to the inner than the distal LN unit of LN␤1-3ЈLN␤1-3R (49). Even in the terminally ␣3-sialylated polylactosamine Neu5Ac␣2-3ЈLN␤1-3ЈLN, the reducing end LN unit reacted quite rapidly in the present experiments, whereas the distal (sialyl)LN group reacted very slowly. Koeller and Wong (50)   The data on neutral glycans was obtained at 5 mM acceptor concentrations and were compared with data of 5 mM LN. The data on sialoglycans were obtained at 3 mM acceptor concentrations and were compared with data of 3 mM LN.
a Initial transfer rates were measured at 5 mM acceptor concentrations and were compared with those of 5 mM LN. b Initial transfer rates were measured at 3 mM acceptor concentrations and were compared with those of 3 mM LN.
The Acceptor and Site Specificities of Fuc-TV than the K m reported for LN by Murray et al. (51). Accordingly, both K m and k cat effects may be responsible for the activity differences between different acceptors observed in our experiments. Several other ␣3-Fuc-Ts such as Fuc-TIV (22), Fuc-TIX (49), and the ␣3/4Fuc-Ts of human milk (52,53), which probably represent a mixture of Fuc-TIII and Fuc-TVI (21,54), are known to work well with LN units of i-type polylactosamines. However, the present data reveal major differences between Fuc-TV and the other human Fuc-Ts working on polylactosamines. Whereas the Fuc-TV reactivities of LN units are strongly influenced by the presence or absence of neighboring GlcNAc, Gal, and LN units in i-type chains, this is not the case for the analogous Fuc-TIV reactivities (22). Another striking difference is provided by the very poor ability of Fuc-TV to fucosylate at the distal ends of neutral i-type polylactosamines, which is the preferred site of action of Fuc-TIX (49). The different site specificities of Fuc-Ts V, IV, and IX are summarized in Fig. 2, which shows their preferred sites of action at the composite acceptor LN␤1-3ЈLN␤1-3ЈLN. The present data show that Fuc-TV reacts most rapidly at the reducing end LN unit of this acceptor, and the data suggest that the tetrasaccharide determinant LN␤1-3ЈLN at the reducing end of the acceptor may be involved in the accelerated reaction. For Fuc-TIV, in turn, the optimal acceptor site of the hexasaccharide appears to be the tetrasaccharide determinant GlcNAc␤1-3ЈLN␤1-3Gal in the middle of the acceptor. The enzyme is known to prefer the middle LN of the hexasaccharide acceptor, and the distal as well as the proximal LN units of the hexasaccharide are known to have rather similar acceptor activities (22). Clearly, both Fuc-TIV and Fuc-TV appear to be particularly well fitted to work with i-type polylactosamines. By contrast, Fuc-TIX works best with the distal LN unit of i-chains; whether it, too, reacts preferentially with elongated elements of i-type polylactosamines remains to be elucidated.
The Fuc-TV worked well with the ␤-methyl glycoside of LN and with the proximal LN unit of LN␤1-3LN␤1-OMe, showing that the high reactivity at proximal LN units in i-type polylactosamines is not due to mutarotational freedom. Fuc-TV worked well also with LN␤1-6Gal and LN␤1-6Man and their ␣3-sialylated derivatives, implying that methylene group-generated conformational freedom between the acceptor LN and its proximal substituents is important for a good reactivity. In line with these observations were the good reactivities of the O-glycans of core 2 type, LN␤1-6(Gal␤1-3)GalNAc, and its ␣3Ј-sialylated form, as well as the poor reactivities of LN␤1-2Man and the biantennary N-glycan (Glycan 37), where the axial linkage between LN and the proximal Man generates considerable steric hindrance. However, very poor reactivities were also observed at the 6-linked arms in LN␤1-3Ј(LN␤1-6Ј)LN and its ␣3Ј-sialylated form. Hence, it appears that in the I branched polylactosamines, the properties of linear acceptors may be dramatically changed. The very poor reactivity at the 6-linked arm in Neu5Ac␣2-3ЈLN␤1-3Ј(Neu5Ac␣2-3ЈLN␤1-6Ј)LN suggests that Fuc-TV is definitely not a tool of choice for in vitro synthesis of oligovalent sialyl Lewis x polylactosamines, representing branched structures that are potent antagonists of lymphocyte L-selectin (55)(56)(57).
A characteristic feature of Fuc-TV is its ability to react at significant rates with two distinct, isomeric disaccharides, i.e. LN and LNB. Our present data show that Fuc-TV transfers to the LNB unit of the type 1 pentasaccharide LNB␤1-3ЈLN␤1-3Gal␤1-OMe almost as well as it transfers to the free LNBdisaccharide (cf . Tables I and III). By contrast, the enzyme fails to transfer to the distal LN of the type 2 pentasaccharide LN␤1-3ЈLN␤1-3Gal␤1-OMe nearly as well as it transfers to unconjugated LN-disaccharide. This suggests that the GlcNAc units of LN and LNB interact with different amino acid residues on the Fuc-TV surface. This observation lends support to the conclusion drawn from a series of the mutagenesis and domain shuffling experiments, establishing amino acid differences in modified Fuc-TVs that are able to react only with free LNB, only with free LN, or with both unconjugated disaccharides (58 -60).
ESL-1 is a glycoprotein ligand for E-selectin, and it is believed to be N-glycosylated in its functional form (61). Tetraantennary N-glycans with a Neu5Ac␣2-3ЈLex␤1-3ЈLex␤1-4Man branch have been reported to bind E-selectin (62). One of the reported properties of Fuc-TV is its ability to convert nonfunctional ESL-1 of CHO cells into a functionally active form that is isolatable by affinity chromatography on a column of immobilized E-selectin (63). The present data on Fuc-TV reactivities of N-glycan acceptors suggest that the ␤1,4-linked, equatorial LN branch reacted much faster than the ␤1,2linked, axial LN branches. Considered together, these sets of  (22) and FucTIX (49) on a composite hexasaccharide acceptor LN␤1-3LN␤1-3LN. Optimally reactive determinants for each of the enzymes are indicated by solid horizontal lines above the hexasaccharide formula, and the best acceptor GlcNAcs are marked by arrows. Whether Fuc-TIX also binds to elongated elements of i-type polylactosamines remains to be elucidated; part of the optimally reactive epitope for this enzyme is therefore marked by a dotted line.
observations suggest that (i) Fuc-TV may participate in ␣3fucosylation of the ␤1,4-linked (sialyl)LN and (sialyl)polylactosamine branches of tri-and tetra-antennary N-glycans with some selectivity and that (ii) this may be important for generation of functional forms of ESL-1.