Chemoenzymatic synthesis of GDP-l-fucose and the Lewis X glycan derivatives

Abstract

Lewis X (Le^x)-containing glycans play important roles in numerous cellular processes. However, the absence of robust, facile, and cost-effective methods for the synthesis of Le^x and its structurally related analogs has severely hampered the elucidation of the specific functions of these glycan epitopes. Here we demonstrate that chemically defined guanidine 5′-diphosphate-β-l-fucose (GDP-fucose), the universal fucosyl donor, the Le^x trisaccharide, and their C-5 substituted derivatives can be synthesized on preparative scales, using a chemoenzymatic approach. This method exploits l-fucokinase/GDP-fucose pyrophosphorylase (FKP), a bifunctional enzyme isolated from Bacteroides fragilis 9343, which converts l-fucose into GDP-fucose via a fucose-1-phosphate (Fuc-1-P) intermediate. Combining the activities of FKP and a Helicobacter pylori α1,3 fucosyltransferase, we prepared a library of Le^x trisaccharide glycans bearing a wide variety of functional groups at the fucose C-5 position. These neoglycoconjugates will be invaluable tools for studying Le^x-mediated biological processes.

Lewis X (Le^x), a fucosylated trisaccharide glycan epitope distributed throughout eukaryotes and certain bacteria, is a determinant of many functional glycoconjugates that play central roles in numerous physiological and pathological processes. For example, sialyl Le^x (sLe^x), a tetrasaccharide constitutively expressed on white blood cells such as granulocytes and monocytes, governs leukocyte rolling and extravasation (1–3); up-regulation of this glycan is strongly correlated with the transformed phenotype of tumors of diverse tissue origin, including pancreas, breast, colon, and lung tumors (4, 5). Le^x-bearing glycans are also found in the infectious bacterium Helicobacter pylori and the parasite Schistosoma mansoni (6, 7). In the former case, these glycans are hypothesized to mask the pathogenic bacterium from the host immune surveillance, while in the latter situation, the Le^x-containing glycans are found to down-regulate the host's protective immune responses against the parasite, largely via induction of anti-inflammatory cytokine interleukin-10.

As revealed by x-ray crystallographic and NMR analyses, the Le^x trisaccharide assumes a well-defined 3-dimensional structure, with its fucose ring stacking on top of the galactose residue (8–10). The exocyclic C-5 methyl group of the fucose forms key van der Waals contacts with the galactose, stabilizing this highly compact structure. Removal of the methyl group leads to a 5-fold decrease in binding affinity of sLe^x to its target protein E-selectin (11). This highly conserved structure is equally important for the specific Le^x– dendritic cell-specific ICAM-3-grabbing nonintegrin (Le^x–DC-SIGN) recognition (8), a unique interaction that is responsible for inducing cellular immunity upon pathogen recognition by dendritic cells (12). A number of studies have demonstrated that the fucose C-5 methyl group may also directly participate in Le^x–lectin recognition. For example, the crystal structure of a Le^x-bound scavenger receptor C-type lectin (SRCL) revealed that the terminal fucose resides in the secondary binding site of the protein, where the methyl group forms tight van der Waals interactions with Ile-712, providing specificity for Le^x over other galactose-containing glycans (10).

Despite the pathophysiological significance of the Le^x antigen-containing glycans, progress toward delineating these glycans' specific functions has been hampered by their complexity and heterogeneity. Like all oligosaccharides, Le^x-bearing glycans are products of template-independent biosynthetic pathways. The dynamic process of glycosylation orchestrated by glycosyltransferases and the organ-specific expression of these enzymes are responsible for the microheterogeneity of these fucosylated glycoconjugates obtained from the mammalian sources. There is currently no facile and cost-effective chemistry for synthesizing Le^x-bearing glycoconjugates and their derivatives on preparative scales for functional studies. Poor selectivity of fucoside coupling reactions and tedious protecting group manipulations are major challenges for synthetic chemists preparing these formidable targets. In contrast, enzymatic glycosylation by fucosyltransferases overcomes laborious and expensive chemical routes and produces Lewis antigen-containing glycans in a regio- and stereospecific manner (13). This approach requires fucose or its synthetic analogs in the nucleotide-activated form—guanidine 5′-diphosphate-β-l-fucose (GDP-fucose) analogs—as the substrate for fucosyltransferases. GDP-fucose, although commercially available, is prohibitively expensive for large-scale synthesis. While Wong and co-workers developed an enzymatic approach for converting GDP-mannose into GDP-fucose (14), the preparation was air sensitive and performed on an analytical scale only. Alternatively, GDP-fucose and its synthetic analogs can be produced in milligram quantities via a coupling reaction between guanosine 5′-monophosphomorpholidate and fucose-1-phosphate (Fuc-1-P) (15). However, the shortest synthetic route for generating Fuc-1-P requires six consecutive steps (16). The length of this route prevents its practical application. Therefore, there is an urgent need to develop new strategies for facile synthesis of GDP-fucose and its derivatives as tools for large-scale preparation of structurally defined fucosides, including the Le^x-bearing glycan epitopes.

Here, we report a chemoenzymatic method for the preparative-scale synthesis of a diverse array of GDP-fucose derivatives (Fig. 1). This method exploits l-fucokinase/GDP-fucose pyrophosphorylase (FKP), a bifunctional enzyme isolated from Bacteroides fragilis 9343, which converts fucose to Fuc-1-P and thence to GDP-fucose (17). This transformation is found in the salvage pathway of B. fragilis 9343 fucosylation and is conserved in all Bacteroides species (17). As revealed by sequence alignment, the N terminus (1–430) of FKP shares 20% amino acid identity to the human GDP-fucose pyrophosphorylase, while its C terminus (584–949) is similar to mammalian l-fucokinases (18). Connecting these two domains is a 150-aa linker, whose function is currently unknown. Wang and co-workers demonstrated recently that a His₆-tagged recombinant FKP, expressed in Escherichia coli, is a promiscuous enzyme with relaxed specificity toward fucose analogs bearing unnatural substituents at the C-5 position (19).

Fig. 1.

A chemoenzymatic approach for the synthesis of GDP-fucose and Le^x trisaccharide derivatives. A short 2-azidoethyl spacer is introduced to the accepter substrate N-acetyllactosamine to allow further modification.

GDP-fucose serves as the donor for fucosyltranferases, enzymes that attach the activated fucose to cell-surface glycoconjugates. On the basis of the site of fucose transfer, fucosyltansferases are classified as α1,2, α1,3/4, α1,6, and protein O-fucosyltransferases (20). In eukaryotes, the former three subfamilies of fucosyltransferases are type II transmembrane proteins, with an N-terminal cytosolic tail, a hydrophobic transmembrane domain, a variable length luminal stem region, and a C-terminal catalytic domain (21). Their prokaryotic counterparts, however, are usually soluble proteins without the transmembrane segment (20, 22, 23). Several α1,2, α1,3/4 fucosyltransferases have been identified in H. pylori, and α1,6 fucosyltransferase activities have been observed in various soil bacteria including Bradyrhizobium japonicum, Azorhizobium caulinodans, Mesorhizobium loti, and Rhizobium loti (22–25). We demonstrate herein that a recombinant α1,3 fucosyltransferase from H. pylori 26695 has broad substrate tolerance toward GDP-fucose analogs modified at the C-5 position and that the action of B. fragilis FKP and H. pylori α1,3 fucosyltransferase can be combined in one pot for the facile synthesis of the Le^x glycan epitope and its derivatives (Fig. 1). This chemoenzymatic approach creates Le^x derivatives with unnatural substituents introduced site specifically at the fucose C-5 position. When presented in a microarray format, these neoglycoconjugates will serve as powerful tools for high-throughput analysis of Le^x-mediated lectin recognition. Complementary to lectin mutants generated via site-directed mutagenesis, these unnatural Le^x derivatives may shed light on key structural features that are unique to Le^x–lectin interactions.

Results and Discussion

Determination of the Catalytic Efficiency of B. fragilis FKP.

As a first step toward the chemoenzymatic synthesis of Le^x derivatives, we expressed an N-terminal His₆-tagged B. fragilis FKP in E. coli, using the construct generated by the Comstock lab (17). To confirm the activity of the recombinant FKP toward GDP-fucose production, we incubated l-fucose with FKP in the presence of ATP, GTP, and MnSO₄. We analyzed the crude reaction mixture by qualitative thin layer chromatography (TLC), using commercially available Fuc-1-P and GDP-fucose (Sigma) as authentic standards (Fig. 2). FKP converts l-fucose into GDP-fucose in the presence of both ATP and GTP, while in the absence of GTP only the Fuc-1-P intermediate was formed. The identities of Fuc-1-P and GDP-fucose produced in the enzymatic reaction were further confirmed by high-resolution mass spectrometry (HR MS) analysis [supporting information (SI) Text]. Interestingly, we detected little accumulation of Fuc-1-P intermediate in the successive reaction supplied with both ATP and GTP, implicating formation of Fuc-1-P as the rate-limiting step. To fully characterize recombinant FKP activity, we determined the kinetic parameters for both reactions under steady-state conditions (Table 1).

Fig. 2.

TLC analysis of B. fragilis FKP-catalyzed synthesis of GDP-fucose. The plate was stained with p-anisaldehyde sugar stain. Lane 1, reaction mixture in the presence of GTP; lane 2, reaction mixture in the absence of GTP; lane 3, GDP-fucose; lane 4, Fuc-1-P; lane 5, l-fucose. Both reactions were performed in 50 μL Tris-HCl buffer (50 mM, pH 7.5) containing l-fucose (5 mM), ATP (5 mM), and MnSO₄ (5 mM). For GDP-fucose production, GTP (5 mM) and inorganic pyrophosphatase (1 unit) were included in the reaction mixture.

Table 1.

Kinetic parameters of B. fragilis FKP

Substrate	KM (mM)	Vmax (min−1)
Fucokinase activity
l-fucose	0.045 ± 0.010	365 ± 22
ATP	1.080 ± 0.21	448 ± 27
GDP-fucose pyrophosphorylase activity
Fuc-1-P	0.030 ± 0.002	878 ± 18
GTP	0.012 ± 0.005	800 ± 73

The fucokinase activity of FKP was measured using a coupled enzymatic assay and the GDP-fucose pyrophosphorylase activity was measured using EnzCheck Pyrophosphate Assay (Invitrogen). All kinetic measurements were performed in 50 mM Hepes-KOH, pH 7.5, at 37° C.

The kinetic parameters for the B. fragilis FKP are quite different from those of the Arabidopsis FKP, the only other bifunctional FKP with reported kinetic parameters (26). The B. fragilis enzyme exhibited 19- and 9-fold greater maximal activities compared to the Arabidopsis enzyme for the fucokinase- and GDP-fucose pyrophosphorylase-catalyzed transformations, respectively. Therefore, B. fragilis FKP is a better choice over the Arabidopsis enzyme for our intended preparative-scale production of GDP-fucose and Le^x derivatives. In addition, the K_M value determined for l-fucose for the B. fragilis FKP was 25-fold lower than that determined for the Arabidopsis FKP. In this regard, the Michaelis constants for l-fucose and ATP in the phosphorylation reaction were more similar to the values reported for the monofunctional fucokinase isolated from pig kidney (27). Consistent with the previously characterized mammalian fucokinase, GDP-fucose pyrophosphorylase, and the bifunctional FKP from Arabidopsis, the activity of B. fragilis FKP requires the presence of divalent metal cations at the active site. Importantly, we found that the fucokinase activity of B. fragilis FKP was dependent on the identity of the divalent metal cation used in the assay, with Mn²⁺ being preferred over Mg²⁺. The specific activity of FKP was determined to be 4.5 units mg⁻¹ protein (1 unit is defined as the amount of enzyme that is required to produce 1 μmol of GDP-fucose per minute at 37 °C).

Recently, Wang et al. showed that the His₆-tagged recombinant FKP from B. fragilis has relaxed specificity toward fucose analogs bearing unnatural substituents at the C-5 position (19). To quantitatively evaluate the specific activity of the recombinant FKP toward C-5 substituted fucose analogs, we determined the fucokinase activity of the recombinant FKP toward a panel of unnatural substrates, using a simple coupled enzymatic assay with a spectophotometric readout. The production of ADP by B. fragilis FKP is coupled to the oxidation of NADH in the presence of pyruvate kinase and lactate dehydrogenase. The consumption of NADH, and accordingly the fucokinase activity, is monitored by the change in absorbance at 340 nm. All compounds tested as substrates were competent for catalysis with V_max values within 3-fold of the value determined for l-fucose (Table 2). Substitutions were generally well tolerated as determined from the ratios of the second-order rate constant of FKP-catalyzed Fuc-1-P formation to those of the corresponding C-5 substituted analogs. However, removal of the C-5 methyl group resulted in the largest decrease in catalytic efficiency (≈43-fold), which corresponds to an energetic penalty of 2.3 kcal mol⁻¹ for this catalytic process. The main factor contributing to this result was the 54-fold increase in the K_M value, suggesting that interaction between the enzyme active site and the C-5 methyl group of l-fucose is critical for catalytic efficiency. Remarkably, the alkyne-, alkene-, and fluorine-substituted analogs, 3, 4, and 5, were phosphorylated by FKP almost as efficiently as the natural substrate l-fucose.

Table 2.

Fucokinase activity and specificity of B. fragilis FKP with C-5 substituted fucose analogs

B. fragilis FKP fucokinase activity and specificity were determined in the presence of 5 mM ATP under standard assay conditions (see Materials and Methods).

Enzymatic Synthesis of GDP-Fucose Derivatives.

GDP-fucose derivatives are key intermediates for the synthesis of fucosylated oligosaccharides and glycoconjugates. Having characterized the activity and promiscuity of FKP, we embarked on the synthesis of a panel of GDP-fucose derivatives. Briefly, seven monosaccharides with unnatural substituents at the C-5 position were chosen as fucose surrogates. The substituents, which vary in stereoelectronics and hydrophobicity, were chosen to test the scope of this method. Importantly, the monosaccharides used in our synthesis were either commercially available or readily synthesized in gram scales using well-established chemistry (28–32). In a typical enzymatic reaction, a fucose analog was incubated with the recombinant FKP in the presence of one equivalent each of ATP and GTP. An inorganic pyrophosphatase from Saccharomyces cerevisiae (Sigma), containing Mg²⁺ to maintain its maximal activity, was included to hydrolyze pyrophosphate generated in the reaction and to drive the reaction to completion. The syntheses were performed on preparative scales (30–50 mg) and the reaction progress was followed by TLC analysis. After quenching the reaction with ethanol, the crude products were purified by gel filtration using BioGel P-2 resin (Bio-Rad). The purity and identity of the GDP-fucose derivatives were confirmed by NMR and HR MS analyses. The typical isolated yield of the desired GDP-fucose derivatives was >75% (Table 3).

Table 3.

Synthesis of GDP-fucose and the Le^x trisaccharide derivatives

Enzymatic Synthesis of the Le^x Derivatives.

α1,3 fucosyltransferases are responsible for the last steps in type II Lewis antigen biosynthesis (20). Preparative scale synthesis of the Le^x glycans using a chemoenzymatic approach relies on the availability of these enzymes in milligram quantities. Toward this end, we generated a DNA construct encoding a truncated α1,3 fucosyltransferase derived from H. pylori 26695 fucosyltransferase (GenBank accession no. AAD07710), using the commercial pET-24b (+) vector (Novagen). Given that high-level expression of a homologous α1,3 fucosyltransferase isolated from H. pylori (NCTC 11639) was impeded by exposed residues at the C terminus (33), we removed the potentially problematic C-terminal tail (residues 434–478), which is rich in hydrophobic and positively charged amino acids, including 6 of the 10 heptad repeats. The truncated protein was highly soluble without a significant change in overall structure compared to the full-length enzyme and its enzyme-specific activity was determined to be ≈12 units mg⁻¹ protein.

With a panel of fucose analogs, a promiscuous FKP, and a recombinant α1,3 fucosyltransferase in hand, the stage was set for the one-pot synthesis of the Le^x trisaccharides. This task was accomplished by combining a fucose analog and the acceptor disaccharide N-acetyllactosamine with the recombinant FKP and α1,3 fucosyltransferase. A short 2-azidoethyl spacer was introduced to the acceptor substrate N-acetyllactosamine to allow fast click modification via Cu(I)-catalyzed azide–alkyne cycloaddition (CuAAC) for future applications such as fabricating glyco-microarrays (34, 35). Using this method, we synthesized a series of the Le^x trisaccharide derivatives bearing a variety of substituents at the fucose C-5 position (Table 3). A number of these fucose analogs were functionalized with elongated hydrocarbon groups that should enforce different conformational constraints on the final Le^x trisaccharides. The resulting Le^x derivatives bearing these functionalities are valuable tools for probing the van der Waals contacts between the fucose and the galactose and the importance of this interaction in Le^x–lectin binding. Fucose analogs 7 and 8 were modified with hydroxyl and methoxyl groups, respectively. These groups may form additional hydrogen bonds with the neighboring galactose, thereby twisting the natural conformation of Le^x. As previously reported, in a number of cases the C-5 methyl group of the fucose residue is directly involved in lectin binding (10). Thus, the Le^x derivatives decorated with unnatural functionalities at this position may provide a new platform for examining intermolecular contacts during receptor binding. The unnatural trisaccharide glycans in Table 3 were typically isolated in 70% yield; in a few cases nearly quantitative formation of the desired trisaccharide was achieved.

Conclusion

The chemoenzymatic method described here offers a practical and versatile approach for the synthesis of GDP-fucose and the Le^x trisaccharide glycan and their derivatives bearing neosubstituents at the fucose C-5 position. The procedure requires only simple cloning steps to generate the necessary enzymes for fucose installation. Many unnatural fucose analogs and acceptor glycans are commercially available or can be easily synthesized using well-established chemistry. Thus, this procedure can be readily extended for large-scale synthesis using these building blocks as substrates. Without any optimization, we obtained a panel of C-5 substituted GDP-fucose derivatives in 90% yield using recombinant FKP. We expect that further engineering of B. fragilis FKP will produce a bifunctional enzyme that is capable of accepting an even broader spectrum of unnatural fucose analogs. Toward this end, we have embarked on two complementary approaches—structure-based engineering and directed evolution—to generate a promiscuous FKP that can serve as a general tool for synthesis of GDP-fucose libraries encompassing diverse structures. In nature, fucosylated glycans are produced by the concerted action of α1,2, α1,3/4, and α1,6 fucosyltransferases. We expect that FKP can be readily coupled with these enzymes in vitro to produce a diverse array of mono- or difucosylated glycans.

Structurally defined fucosides are invaluable tools for studying fucose-mediated cellular processes. Access to chemically defined Le^x derivatives combined with the emerging glyco-microarray technology provides a powerful, rapid means to profile Le^x–lectin interactions and to identify key structural features contributing to binding. Fabrication of glycodendrimer microarrays using the Le^x trisaccharide derivatives for high-throughput screening of the Le^x-binding lectins is currently underway.

Materials and Methods

Measurement of the l-Fucokinase Activity of FKP.

The production of ADP was measured using a coupled assay system, monitoring the reaction progress by absorbance at 340 nm. Standard conditions were 50 mM Hepes-KOH, pH 7.5, 10 mM MnCl₂, 5 mM ATP, 100 μM NADH, 250 μM phosphoenol pyruvate, 0.7 unit of pyruvate kinase, and 1 unit of lactate dehydrogenase (from a stock solution in 50% glycerol) in a 1-mL reaction at 37 °C. All components except fucose were mixed in the cuvette and allowed to equilibrate for at least 2 min. Reactions were initiated by the addition of fucose. The amount of coupling enzymes was sufficient to not limit the rate of reaction.

Measurement of the GDP-l-Fucose Pyrophosphorylase Activity of FKP.

The production of pyrophosphate was measured using the commercially available EnzCheck Pyrophosphate Assay Kit (Invitrogen). Standard conditions were 50 mM Hepes-KOH, pH 7.5, 2 mM MgCl₂, 200 μM 2-amino-6-mercapto-7-methylpurine ribonucleoside, 1 unit nucleoside phosphorylase, and 0.03 unit inorganic pyrophosphatase. All components except FKP were mixed in the cuvette and allowed to equilibrate for at least 2 min. Reactions were initiated by the addition of enzyme.

Analysis of Kinetic Parameters.

To determine the basic kinetic parameters for each substrate, initial velocities were determined at various concentrations of one substrate while the concentration of the other substrate was held constant at saturated levels (5–10 times the K_M value) and fit to the Michaelis–Menten equation (Eq. 1). A stands for a substrate [data were analyzed using KaleidaGraph (Synergy Software)]:

General Procedure for Preparative Scale (30–50 mg) Synthesis of the GDP-Fucose Derivatives.

Reactions were typically carried out in a 15-mL centrifuge tube with 5.0 mL Tris-HCl buffer (100 mM, pH 7.5) containing l-fucose or its C-5 substituted analogs (7.5–10.2 mg, 0.05 mmol), ATP (1.0 eq), GTP (1.0 eq), MnSO₄ (10 mM), inorganic pyrophosphatase (90 units, lyophilized form containing MgCl₂), and FKP (9 units). The reaction mixture was incubated at 37 °C for 5–6 h with shaking (225 rpm). The reaction was monitored by TLC analysis, using 10 mM tetrabutylammonium hydroxide in 80% aqueous acetonitrile as the developing solvent (p-anisaldehyde sugar stain). After adding the same volume of ice-cold ethanol to quench the reaction, the alcoholic mixture was incubated on ice for 30 min. Insoluble material was removed by centrifugation (5,000 × g, 30 min) and the supernatant was concentrated in vacuo to remove volatile ethanol. Crude reaction products were purified by Bio-Gel P2 gel filtration chromatography (1.5 × 75 cm) and eluted with H₂O. Only the fractions containing the product were collected, lyophilized, and further purified using Bio-Gel P2 gel filtration chromatography (1.5 × 120 cm) and eluted with NH₄HCO₃ (50 mM). Lyophilized GDP-fucose derivatives were characterized by NMR and HR MS.

General Procedure for Preparative Scale (15–30 mg) Synthesis of the Le^x Trisaccharide Derivatives.

One-pot reactions were performed in 15-mL centrifuge tubes with 5.0 mL Tris-HCl buffer (100 mM, pH 7.5) containing l-fucose or its C-5 substituted analogs (2.0 eq, 0.05 or 0.1 mmol), 2-azidoethyl N-acetyllactosamine (1.0 eq, 0.025 or 0.05 mmol), ATP (2.0 eq), GTP (2.0 eq), MnSO₄ (20 or 40 mM), inorganic pyrophosphatase (75 units, lyophilized form containing MgCl₂), FKP (9 units), and α1,3 fucosyltransferase (2 units). The reaction mixture was incubated at 37 °C for 2–3 h with vigorous shaking (225 rpm). The reaction was monitored by TLC analysis using EtOAc:MeOH:HOAc:H₂O (6:3:3:1) as the developing solvent (H₂SO₄ stain). After adding the same volume of ice-cold ethanol to quench the reaction, the alcoholic mixture was incubated on ice for 30 min. Insoluble material was removed by centrifugation (5,000 × g, 30 min) and the supernatant was concentrated in vacuo to remove volatile ethanol. The aqueous residues were lyophilized to dryness. Crude product was purified by Bio-Gel P2 gel filtration chromatography (1.5 × 75 cm) and eluted with H₂O. Lyophilized Le^X derivatives were characterized by NMR and HR MS.