Enzymatic transfer of a preassembled trisaccharide antigen to cell surfaces using a fucosyltransferase.

The Lewis alpha (1-->3/4)-fucosyltransferase (Le-FucT) is known to fucosylate both Type I (beta Gal(1-->3) beta GlcNAc) and Type II (beta Gal(1-->4) beta GlcNAc) sequences even when these are sialylated at OH-3 or fucosylated at OH-2 of the terminal Gal residues. These acceptor sequences are ubiquitous on mammalian cell-surface glycoproteins and glycolipids. The Le-FucT enzyme is therefore a potential candidate as a universal reagent for the modification of cell surfaces. We have found that a readily accessible, partially purified Le-FucT from human milk, which normally uses GDP-fucose (a 6-deoxy sugar) as the donor for the transfer of a single fucose residue, will also transfer a fucose residue substituted on C-6 by a very large sterically demanding structure, in this instance, a synthetic blood group antigen. As a demonstration of the ability of the Le-FucT to modify glycoconjugates in a mild and specific manner, we chemically synthesized the complex sugar-nucleotide alpha Gal(1-->3) [alpha Fuc(1-->2)]-beta Gal-O-(CH2)8COHN(6)-beta-L-fucose-GDP (13) which is a GDP-fucose analog where the human blood group B trisaccharide antigen is covalently linked to C-6 of fucose through an amino group. It is shown that, in enzyme-linked immunosorbent assays, the Le-FucT uses both immobilized beta Gal(1-->3) beta GlcNAc-bovine serum albumin conjugates and fetuin as acceptor substrates and renders them blood group B-active as detected by a monoclonal anti-B blood-grouping antibody. The fucose residue to which the B-trisaccharide is linked therefore becomes covalently attached to the acceptor oligosaccharide chains of those glycoproteins. Incubation of type "O" erythrocytes with the Le-FucT and complex donor 13 results in the covalent transfer of alpha Gal(1-->3) [alpha Fuc(1-->2)] beta Gal-O-(CH2)8COHN(6)-beta-L-Fuc to cell-surface acceptors since the cells become phenotypically "B" and are agglutinated by the same antibody. It is proposed that the Le-FucT represents a powerful new tool with the ability to label animal cell surfaces with preassembled oligosaccharide and possibly also other complex recognition markers.


Enzymatic Transfer of a Preassembled Trisaccharide Antigen to Cell
Surfaces Using a Fucosyltransferase* (Received for publication, May 14, 1992) Geeta Srivastava, Kanwal J. Kaur, Ole HindsgaulS, and Monica M. PalcicS From the Department of Chemistry, University of Alberta, Edmonton, Alberta, T6G 2G2 Canada The Lewis a( 1+3/4)-fucosyltransferase (Le-FucT) is known to fucosylate both Type I (@Gal( 1+3)@GlcNAc) and Type I1 @Gal( 1+4)@GlcNAc) sequences even when these are sialylated at OH-3 or fucosylated at OH-2 of the terminal Gal residues. These acceptor sequences are ubiquitous on mammalian cell-surface glycoproteins and glycolipids. The Le-FucT enzyme is therefore a potential candidate as a universal reagent for the modification of cell surfaces.
We have found that a readily accessible, partially purified Le-FucT from human milk, which normally uses GDP-fucose (a 6-deoxy sugar) as the donor for the transfer of a single fucose residue, will also transfer a fucose residue substituted on C-6 by a very large sterically demanding structure, in this instance, a synthetic blood group antigen. As a demonstration of the ability of the Le-FucT to modify glycoconjugates in a mild and specific manner, we chemically synthesized the complex sugar-nucleotide aGal( 1+3)[aFuc(1+2)]-~Gal-O-(CH2)~COHN~6,-@-~-fucose-GDP (13) which is a GDP-fucose analog where the human blood group B trisaccharide antigen is covalently linked to C-6 of fucose through an amino group. It is shown that, in enzyme-linked immunosorbent assays, the Le-FucT uses both immobilized @Gal( 1+3)@GlcNAc-bovine serum albumin conjugates and fetuin as acceptor substrates and renders them blood group B-active as detected by a monoclonal anti-B blood-grouping antibody. The fucose residue to which the B-trisaccharide is linked therefore becomes covalently attached to the acceptor oligosaccharide chains of those glycoproteins. Incubation of type "0" erythrocytes with the Le-FucT and complex donor 13 results in the covalent transfer

of ~G~~(~+~)[~Fuc(~+~)]@G~~-O-(CHZ)~COHN~~~-@-L-
Fuc to cell-surface acceptors since the cells become phenotypically "B" and are agglutinated by the same antibody. It is proposed that the Le-FucT represents a powerful new tool with the ability to label animal cell surfaces with preassembled oligosaccharide and possibly also other complex recognition markers.
Elucidation of the biological roles of specific cell-surface glycosylation remains a subject of intense investigation. Specific changes in the detailed structures of cell-surface carbohydrates have been extensively described in, among other * This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada (to both 0. H. and 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. j: To whom correspondence and reprint requests should be addressed. cases, cell differentiation, development, and tumor progression (1-3). While changes in their structure can be documented with good accuracy, it remains much more difficult to establish a biological function or significance (if any) to apparently regulated appearances of specific cell-surface carbohydrate sequences. The potential importance of specific glycosylation is highlighted by recent reports describing the essential role of sialylated and fucosylated tetrasaccharide antigens (sialyl-LeX and sialyl-Le") in the adhesion of neutrophils to endothelial cells, an example of carbohydrate-mediated cell-cell recognition which, in this instance, appears to be the first step in initiating an acute inflammatory response Well characterized oligosaccharides, most reliably available through chemical synthesis, are important as tools in obtaining evidence for the function of a specific cell-surface carbohydrate sequence. The most common applications for such compounds are as inhibitors of the binding of proteins or cells with the natural glycoconjugate ligand. If inhibition is observed, then that particular carbohydrate sequence becomes a candidate as a physiologically relevant recognition marker, and further experiments can be designed to assess whether the recognition is in fact functional (7). In the work reported here, we decided to investigate whether similar synthetic oligosaccharides could not also be used in an alternate experiment where they could be added in a mild and specific manner to intact glycoconjugates or living cells. If this became feasible, then postulated complex carbohydrate recognition markers could be added in a controlled fashion to proteins or cells which are completely devoid of related structures. The resulting panel of "sugar-tagged proteins or cells should be unique and invaluable tools for dissecting the biological function of specific oligosaccharide sequences.
In principle, there are two major chemical approaches available for tagging soluble as well as cell-surface glycoproteins with synthetic or isolated oligosaccharides. In the first type of approach, reactive groups on the protein (usually side chain amino, carboxyl, or sulfhydryl) are covalently attached to oligosaccharides by reductive amination or using chemically activated derivatives of the oligosaccharides. This results in alkylation or acylation of the protein. Such methods can be useful but clearly suffer from the disadvantage that peptide chains are actually chemically altered. There can be multiple random substitutions if there are multiple reactive groups on the proteins, and all cell-surface proteins can react. The second approach involves oxidation of the sugar chains of cell-surface glycoproteins (or glycolipids) followed by reductive amination of an amino derivative of the oligosaccharide to be attached. The oxidation is usually effected with periodate or, in a milder procedure, treatment of cells with sialidase followed by galactose oxidase and then the reductive amination (8). Such methods, when applied with care, cause minimum alteration of peptides, but the chemistry remains vig- orous (both oxidizing and reducing) by biological standards. Also, existing carbohydrate sequences are either modified or destroyed. We chose to develop an alternate enzymatic approach using a glycosyltransferase to transfer potential oligosaccharide recognition markers. Glycosyltransferases have in the past been used effectively to add sugars to both glycoproteins and cellsurface glycoconjugates (9)(10)(11). In such experiments, single sugar residues are added, from the appropriate sugar nucleotides, in a stepwise fashion. To build up a complex oligosaccharide determinant, a series of glycosyltransferases is therefore required. A more important limitation of this stepwise approach, however, is that a sequential series of glycosylations cannot likely be driven to completion and, as a result, a series of partial structures will invariably be present on the target glycoproteins or cells that have been glycosylated with the required sequence of glycosyltransferases.
We report here that a partially purified (a1+-3/4)-fucosyltransferase (Le-FucT) from human milk (12, 13), which biosynthetically uses GDP-fucose as the donor to transfer single fucose residues, is capable of transferring large preassembled oligosaccharide antigens when these are covalently attached through C-6 of the fucose residue. Using this enzyme, the sugar chains of glycoconjugates, both free and membranebound, can thereby be tagged with completed carbohydrate recognition markers. The decision to investigate whether the human Le-FucT could be used to add large oligosaccharides to cell surfaces was based on three considerations. Firstly, the enzyme can be easily partially purified from human milk (12,13). It has also been cloned (14) which augers well for more ready availability in the near future. Secondly, fucosyltransferases are the only glycosyltransferases which transfer a sugar having the L absolute configuration. It therefore seemed at least plausible that fucosyltransferases in general might not have highly evolved and specific recognition sites for the sugar being transferred since the remaining mammalian sugars in the Leloir pathway all have the D configuration. The Le-FucT has indeed been shown (15) to transfer both 3deoxyfucose and arabinose, a fucose analog where the 6methyl group is missing. Sialyltransferases are also candidates in this regard since they are the only ulosonic acids transferred in mammalian biosynthesis. In general, sialyltransferases do accept modifications on the sugar nucleotide, CMP-sialic acid, and several modified sialyl residues have been added enzymatically to galactose-terminated glycoconjugates (16). The third and most important consideration, however, was that a large number of the carbohydrate structures commonly found on cell-surface mammalian glycoproteins and glycolipids can act as acceptors for the Lewis enzyme. The structures of these ubiquitous (1-3, 17, 18) acceptor sequences and the products that would form on their reaction with GDP-fucose catalyzed by the Le-FucT are shown in Fig. 1. This figure also summarizes our strategy for tagging cell-surface glycoconjugates. This strategy involves covalent attachment of synthetic oligosaccharides directly to the fucose residue of GDP-Fuc to produce a complex analog of that sugar nucleotide. The hope was that the FucT would still recognize this analog as a donor substrate thereby covalently attaching the oligosaccharide to the cell surface via a fucose "spacer." This proved to be the case.

Chemical Synthesis of GDP-Fuc Derivatives
The general synthetic scheme is modelled after a previously reported synthesis of GDP-Fuc and two of its analogs (15). Details on the chromatography and spectroscopic characterization of intermediates are discussed in detail in that publication. The structures of synthetic intermediates are shown in Fig. 2. The following solvents used here have been designated by letters: A , benzene/methanol, 9:l (v/v); B , benzenelmethanol, 12:l (v/v); C, ethyl acetate/hexane, 1:2; D, dichloromethane/methanol/water, 60:35:6; E , 2-propanol/water/ The abbreviations used are; BSA, bovine serum albumin; PBS, phosphate-buffered saline; ELISA, enzyme-linked immunosorbent assay.
Transfer of a Trisaccharide Antigen by a Fucosyltrarzsferase ammonium hydroxide, 7:4:1; F, 2-propanol/water/ammonium hydroxide, 1:l:O.Z; G, 2-propanol/water/ammonium hydroxide, 7:3:1. NMR spectroscopy was performed on a Bruker 360 WM spectrometer operating at ambient temperature. (6)-To anhydrous zinc chloride (2.2 g) in dry acetone (45 ml) was added concentrated sulfuric acid (72 pl). Then powdered anhydrous L-galactose (1.8 g, 10 mmol) was added quickly, and the reaction mixture was stirred for 10 min. A suspension of anhydrous sodium carbonate (3.6 g ) in water (6.3 ml) was then added in portions. The suspension was filtered with suction, and the precipitate was washed several times by suspension in acetone and filtering. The filtrate and washings were combined, and the solution and the acetone were evaporated under diminished pressure. The residue was extracted with ether (3 X 100 ml), dried over anhydrous sodium sulfate, filtered, and concentrated t o dryness. The crude product was purified by chromatography an silica gel (solvent A). Pure

1,2:3,4-Di-O-isopropylidene-6-azido-a-~-~alactopyranose (7)-
Compound 6 (1.1 g, 4.23 mmol) was dissolved in dry dichloromethane (20 ml), and dry pyridine (4 ml) was added. Trifluoromethanesulfonic anhydride (1.4 ml, 8.45 mmol) was then added at 0 "C. The reaction mixture was stirred for 30 min a t 0 "C by which time TLC indicated the complete conversion of 6 (R/ 0.15) to less polar product 8 (R/ 0.47, solvent B ) . The mixture was diluted with dichloromethane (100 ml) and washed with ice-cold water (2 X 100 ml), dried over sodium sulfate, filtered, and evaporated. The residue was dissolved in dry dimethylformamide (5 ml), sodium azide (1.37 g, 2 1 mmol) was added, and the reaction mixture was stirred for 15 h at room temperature. After dilution with dichloromethane (100 ml) and washing with water (2 X 100 ml), the solvent was dried and evaporated. The residual syrup was purified by chromatography on silica gel using solvent C a s eluent.

6-Azido-1,2,3,4-tetra-O-acetyl-6-deoxy-a,@-~-galactopyranose (8)-
Compound 7 (750 mg, 2.63 mmol) was dissolved in trifluoroacetic acid/water (9:1, 20 ml), stirred for 15 min at room temperature, then neutralized with triethylamine, and concentrated, and toluene was added to the residue and re-evaporated. The residual syrup was acetylated with pyridine (5 ml) and acetic anhydride (5 ml) for 15 h at room temperature. The acetylated mixture was diluted with dichloromethane (100 ml) and washed with cold 1 N hydrochloric acid, saturated sodium bicarbonate, and finally with water (100 ml each). Solvent was evaporated and the residue was purified by chromatography on silica gel (solvent C) to provide 8 ( E f 0.43, solvent C) as an cu,fl-mixture (850 mg, 87%). 'H NMR (CDCI:,) 6 (9)-Titanium tetrabromide was added at 0 "C to a solution of 8 (335 mg, 0.90 mmol) in dichloromethane/ethyl acetate (10 ml, 9:1), and the reaction mixture was stirred for 48 h at room temperature. Sodium acetate (1 g) was then added, and stirring was continued for 15 min. The mixture was then diluted with dichloromethane (50 ml) and washed with ice cold water (5 X 50 ml), dried over sodium sulfate, filtered, and evaporated to dryness. The residue was dissolved in dry toluene (2 ml) and added to a stirring mixJure of tetrabutyl ammonium phosphate (609.2 mg, 1.8 mmol) and 4 A activated molecular sieves (1.0 g) in dry acetonitrile (3 ml) under nitrogen at 0 "C. Stirring was continued for 0.5 h at 0 "C and 3 h at room temperature. Solvent was evaporated, the residue was dissolved in water (100 ml) and washed with ethyl acetate (3 X 100 ml), and the solvent was evaporated. The residue was deacetylated by adding 28% ammonium hydroxide (25 ml). Solvent was evaporated, the residue was dissolved in water (200 ml), and the solution was neutralized with 1 N HCI. The residue was purified loading onto Dowex 2-X8 (3.7 X 30 cm, HCO;) ion exchange resin in water and washing with water. The compound was eluted with 0.5 N NH4HC03 (200 ml), solvent was evaporated, and the residue was converted to the triethyl ammonium salt by passage through Dowex 50-X8 (Et,N+) ion exchange resin.  (10)"Compound 9 (221 mg, 0.57 mmol) was dissolved in a mixture of pyridine (3 mi), water (0.5 ml), and triethylamine (0.035 ml), and hydrogen sulfide was bubbled through it. After 15 h, the reaction mixture was taken to dryness, and the residue was converted into the triethyl ammonium salt by passage through Dowex 50-X8 (Et:lNH+) ion exchange resin (10 ml (12)"Amine 10 (104 mg, 0.29 mmol) was dissolved in buffer (35 mM KHC03, 8 mM Na2B407, pH 9.0, 11.4 ml) and added directly to the acyl azide of 11 (prepared from 328 mg, 0.50 mmol of 11 as previously described (26)). More buffer (10 ml) and saturated sodium bicarbonate (10 ml) was then added to maintain a pH of 9.0. The reaction mixture was stirred at 4 "C for 15 h. Solvent was evaporated, and the residue was purified by chromatography on Iatrobeads (solvent F). Carbohydrate-containing fractions were combined and passed through a C,, Sep-Pak cartridge (27). After washing with water, product was eluted with  (13)kCompound 12 as the triethylammonium salt (0.054 mmol) and guanosine 5'-monophosphate morpholidate (46.5 mg, 0.064 mmol) in anhydrous pyridine (10 ml) were concentrated to dryness in vacuo. The process of dissolution and concentration was repeated twice, dry nitrogen being admitted into the flask after each concentration. A solution of the residue in anhydrous pyridine (2 ml) was kept at room temperature for 10 days, then concentrated in uacuo to 0.5 mi, diluted with water (10 ml), and loaded to a column (1.5 X 45 cm) of reverse phase silica gel (C-18). The column was washed with water (200 ml) and finally with methanol (200 ml). Fractions containing the products were pooled and concentrated, and the residue was passed through a column (1.5 X 45 cm) of Bio-Gel P-4 using 10% aqueous ethanol as eluent. The carbohydrate fractions were pooled, concentrated, and lyophilized to provide 13 ( R f 0.37, solvent G) as a white powder (20 mg, 30%);  (2) The synthesis of 2 began with the 0-allylation of 6 and reduction of the allyl ether to the propyl ether by catalytic hydrogenation, and the sequence of reactions then paralleled exactly those for the conversion of azide 7 to the phosphate 9 and its coupling with guanosine

Comparison of Transfer of GDP-Fuc with GDP-6-0-propyl Gal
Preparative fucosylation was performed essentially as previously described (21). Briefly, to a mixture of acceptor 3 (1.4 pmol), GDP -Fuc (1, 1.4 pmol), and GDP-6-0-propyl-~-Gal (2, 1.4 pmol), ADP (0.8 pmol), and NaNB (4 pmol) was added 1.7 milliunits of Le-FucT in 0.5 ml of sodium cacodylate buffer (25 mM, pH 6.5) containing MnC12 (5 mM). After 28 h a t 37 "C, the sample was diluted with water, and the hydrophobic products were isolated on a C-18 Sep-Pak cartridge (21). The 'H NMR spectrum of the crude product confirmed that complete fucosylation of disaccharide 3 had occurred, and integration of the signal for the known (21) Le" trisaccharide uersus the new signals for the trisaccharide containing an 0-propyl group indicated that the ratio of 4 to 5 was approximately 2:l.
The Use of ELISA to Detect the Transfer of Fucose Analogs to Immobilized Glycoconjugates: ELISA Plate Coating with Type I BSA

Conjugates (~Gal(l+3)(3GlcNAc-O(CHJ&O-NH-)~~-~~-BSA) and
Fetuin Microtiter plates were coated by incubating with 100 pl of BSAglycoconjugate (20 pg/ml) or fetuin ( 50 pg/ml) in 50 mM potassium phosphate buffer, pH 7.5, containing 5 mM MgC12 and 15 mM NaN3. After 16 h at ambient temperature (4 "C for the fetuin plates), the solution was aspirated and replaced with 100 p1 of 5% BSA in PBS. After 4 h, this solution was removed, and the wells were washed three times with 200 p1 of PBS and two times with H20, air-dried for 1 h, and stored at 4 "C. Plates were washed again with 200 pl of H,O immediately before use.

ELISA Incubations
Assays were performed by adding 9 microunits of (u(l-+3/4) f'ucosyltransferase and 52 p M donor 13 substrate in 100 p1 of 50 mM sodium cacodylate buffer, p H 6.8, containing 7 mM MnClZ to the coated microtiter wells. The microtiter plates were incubated at 37 "C for 2, 6, and 21 h (Type I plates) or 16 h for fetuin-coated plates. Control reactions lacking either enzyme or donor were carried out simultaneously. After incubation, the reaction mixtures were removed by aspiration and wells were washed (2 X 200 pl of HzO) and 2 X 200 pl of PBST). Anti-B antibody (Synaff) was diluted 1:lOOO in 1% BSA/PBST, and 100 pl was added to the microtiter wells. After 2 h at ambient temperature, wells were aspirated, washed with 5 X 200 p1 of PBST, and then incubated with 100 pl of the alkaline phosphatase-conjugated goat anti-mouse antibody (1:lOOO dilution in 1% BSA/PBST) for 2 h a t ambient temperature. Solutions were aspirated, and the wells were washed 3 x 200 p1 of PBST and once with 300 p1 of H,O before adding p-nitrophenyl phosphate (1.0 mg/ml in 1 M diethanolamine-HC1 buffer, pH 9.8, containing 1% BSA and 0.5 mM MgC1,). The increase in absorbance at 405 nm with background correction a t 650 nm was monitored with a Molecular Devices Thermomax microplate reader. Data acquisition was controlled by a MacIntosh SOFTmax program, and the absorbance readings reported were taken after 120 min of color development.

Transfer of the R-actiue Trisaccharide to Red Cells Detected by Agglutination: Cell Incubations
Fresh red blood cells type 0 Le"-Leh-, were collected by venipuncture and treated with 1/10 volume of 3.8% sodium citrate as an anticoagulent. 100 pl of whole red cells were washed twice with 1 ml of cell incubation buffer by suspending the cells in buffer in 1.5-ml Microfuge tubes, centrifuging for 15-30 s, and removing supernatant buffer with a pipette. Packed cells (100 pl) were incubated with 62 microunits of n(l+3/4)fucosyltransferase and 116 p~ donor 13 in 120 p1 of cell incubation buffer. Control reaction mixtures contained only enzyme or donor in 120 p1 of cell incubation buffer. After 15 h at 37 "C, the cells were washed twice with PBS, spun in Microfuge tubes for 15 s after each wash, and resuspended in 1 ml of PBS. For hemagglutinations, 100 p1 of a 2% cell suspension were mixed with 100 pl of Syntype anti-B antibody serially diluted in PBS, the mixtures were immediately spun a t 1000 X g for 15 s, and scored by the method of Marsh (25).

The Le-FucT Can Transfer a Fucose Residue
Chemically Substituted at C-6"The first experiments performed were to establish that GDP-Fuc (1) could be substituted by a sterically demanding group without destroying its activity as a donor for the Lewis FucT. For ease of chemical synthesis, only substitution at C-6 was examined and GDP-(g-O-pro-py1)-L-galactose (2) was synthesized. The Type I disaccharide @Gal(l+3)@GlcNAc-OR (3) (19) was used as the acceptor since we have previously demonstrated (13, 15) its conversion to the Le" trisaccharide @Gal (1)(2)(3) [aFuc( l-A)]@GlcNAc-OR (4) using an identical preparation of the FucT (Fig. 3). A competition experiment was set up where equal amounts of GDP-Fuc (1) and its 6-0-propyl analog 2 were incubated with acceptor 3 and the FucT. The ratio of the trisaccharide products (4 and 5 ) was determined to be near 2:l from the 'H NMR spectra of the crude products. Analog 2 therefore remains very active as a donor for the FucT, despite the introduction of a large 0-propyl group on C-6 ( Fig. 3), and the preparation of a more complex analog could therefore be justified.
Covalent Attachment of the Blood Group B-active Trisaccharide,aGal(l"t3)[cuFuc(l~2)]@Gal-, to C-6 of the Fucose Residue in GDP-fucose Does Not Destroy Its Donor Activity-Since a substitution as large as an 0-propyl group could be made on C-6 of the Fuc residue of GDP-Fuc, the conclusion was that this part of the molecule did not interact with the enzyme active site. It thus seemed probable that much larger groups could also be attached to this position. To examine by guest on March 24, 2020 http://www.jbc.org/ Downloaded from this possibility, we synthesized 6-amino-P-fucose-1-phosphate (10) where the 6-amino group could now easily be derivatized by simple acylation. The 8-methoxycarbonyloctyl derivative of a blood group B active trisaccharide, aGal(l+3)[aFuc(l+ 2)]PGal-0-(CH2)RCOOMe (1 l), available from previous work (20), was then used to acylate the 6-amino group following established procedures. Pyrophosphate formation to provide the GDP-"B"-Fuc structure 13 (Fig. 3) was effected as previously described (15) for other analogs of GDP-fucose.
GDP-"B"-Fuc (13) was shown to be a donor for the Le-FucT using an ELISA assay system. First, the synthetic conjugate of /~G~~(~+~)PG~cNAc-O-(CH~)RCO)~~NH-BSA (21), previously shown to be an acceptor for the Le-FucT in a n ELISA using GDP-Fuc as the donor and a monoclonal anti-Le" as the detecting antibody (21), was coated on microtiter plates. The coated plates did not bind a commercial anti-B blood-typing antibody, as expected. When these coated plates were incubated with the FucT in the presence of GDP-"B"-Fuc (13), the plates acquired B activity, as detected by the anti-B antibody (22)) despite repeated washings (A405 = 0.70, 1.33, and 2.63 for 2-, 6-, and 21-h incubations, respectively). Omission of either the FucT or donor 13 led to Binactive plates (A405 = 0.063). The conclusion was therefore reached that the B-trisaccharide was covalently attached to the conjugate on the plate. A detailed investigation of the donor properties of 13 was beyond the scope of the present investigation. When using the immobilized @Gal( 1-3)PGlcNAc-BSA conjugate as the acceptor, however, 10 times more enzyme (LeFucT) was required with donor 13 than with GDP-Fuc (1) to give the same rate of product formation. Since the product formation was estimated by ELISA using two different detecting antibodies, an anti-B and an anti-Le", this observation that GDP-Fuc is 10 times more active than 13 as a donor should only be considered a rough estimate. Next, fetuin was coated on microtiter plates. The triantennary N-linked oligosaccharide chains of fetuin are known to terminate in structures a-d (Fig. l), as well as in structures which contain a(2+6)-linked sialic acid and additional 0linked oligosaccharides (23,24). Sequences a-d should provide acceptor structures for Le-FucT. The fetuin-coated plates were incubated with the enzyme and donor 13. As before, the plates acquired strong B activity (A405 = 0.97 after 16 h compared with 0.074 when either donor or enzyme was omitted) indicating that the covalent addition of the B-active trisaccharide, attached through fucose, to these sequences had occurred.
Transfer of the Blood Group B Trisaccharide Antigen to the Surface of Blood Group 0 Le"-b-Red Cells Detected by a Monoclonal Anti-B Antibody-The demonstration that the Lewis FucT could use GDP-"B"-Fuc (13) as a donor and human cell surfaces as the acceptor was achieved using type 0 red cells. Blood group 0 activity is conferred by the H-Type I1 antigen, structure f in Fig. 1, a known substrate for the Lewis FucT which converts it to the LeY structure ( Fig. 1) (1, 15). Additional ubiquitous acceptor sequences similar to those of fetuin, as well as potential acceptor structures at the termini of glycolipids, are undoubtedly also present on erythrocytes. The blood group 0 red cells were, of course, not agglutinated by the anti-B antibodies. Incubation of these cells with the FucT and GDP-"B"-Fuc (13), followed by washing by suspension and aspiration of the supernatant, converted these to B-active cells that were agglutinated as strongly as the cells from a natural group B blood donor with a score of 56 by the method of Marsh (25). Again, if either the enzyme or donor 13 were omitted, the cells remained Bnegative in this agglutination assay. This experiment is pic-torially summarized in Fig. 4 where the expected structures of the enzymatically produced B-active cell-surface carbohydrates are also shown. A photograph of the actual agglutinations is reproduced in Fig. 5. The locations of attachment of the B-active trisaccharides to the cell surfaces were not investigated in this study.

DISCUSSION
The chief finding presented in this report is that large molecules, a t least the size of a trisaccharide, can be synthetically attached to the 6-position of the fucosyl residue of GDP-fucose and subsequently transferred to acceptor oligosaccharides using the Lewis a( 1+3/4)fucosyltransferase. Most glycoproteins, and possibly all cell surfaces, possess some of the FucT acceptor sequences shown in Fig. 1, and it should consequently be possible to label these glycoproteins and cells in this very simple and direct manner. Clearly, the new linkage, via a fucose "spacer," is unnatural, and the labeling strategy reported here should not be considered as a viable method for remodeling the sugar chains of glycoconjugates for in vivo use. We expect that the major use of this labeling strategy will be in rapidly assessing whether specific carbohydrate sequences are sufficient to target glycoproteins or cells to carbohydrate-recognizing receptors and to assist in the discovery of such receptors. The labeling strategy should aFuc(l-12)~Gal(l-rl)~lcNAc---Oligosaccharide .... Duplicates are shown. In the right-hand pair, the 0 cells were incubated with Le-FucT, but the donor (13) was omitted. In the center pair, the cells were incubated with the donor (13), but the Le-FucT was omitted. In the pair to the left, the cells were incubated with both the Le-FucT and donor (13). Only the latter pair was agglutinated by the anti-B antibody. be attractive since, once the GDP-Fuc analog has been synthesized, the procedure involves only incubation under neutral conditions of the FucT, donor, and glycoconjugate to be derivatized. The major impediment to this labeling approach remains the synthesis of the complex sugar nucleotide analogs which we are attempting to simplify. The exciting possibility that peptides, or even large proteins, attached to the fucose residue of GDP-fucose might be transferred is also under investigation.