FUT8: from biochemistry to synthesis of core-fucosylated N-glycans

Introduction

Posttranslational modifications are very important since they influence folding, stability, signaling, transport and life span of proteins [1], [2]. One major posttranslational modification is N-glycosylation, the alteration of which is usually involved in biological processes such as: cell adhesion, infections, immunogenicity, and tumor metastasis [3], [4], [5], [6]. N-glycans are also proven to influence protein biosynthesis, folding, molecular stability, solubility, and serum half-life [5], [7], [8], [9]. The study of human glycosyltransferases (hGTs) is important to the understanding of these important cellular event as well as their involvement in major diseases [2], [6], [10], [11], [12], [13], [14]. N-glycan structures in nature can be very complex and diverse due to the great degree of modification that takes place. Core-fucosylation is a major N-glycan modification in eukaryotic systems.

This review will focus on the biochemical properties of human α1,6 fucosyltransferase (FUT8) and its biosynthetic applications. FUT8 is the enzyme responsible for all α1,6 core fucosylation in mammals. This enzyme catalyzes the transfer of an L-fucose residue from GDP-β-L-fucose (GDP-Fuc) to the inner GlcNAc of asparagine linked complex glycopeptides to fucosylated with an α1,6-linkage (Fig. 1) [15], [16], [17], [18], [19], [20].

Fig. 1:

Sample reaction catalyzed by α1,6 fucosyltransferase.

FUT8 belongs to the fucosyltransferase family (EC 2.4.1.68) and it is encoded in humans by the FUT8 gene. FUT8 is a type II membrane protein on the Golgi apparatus [15]. Human FUT8 has 575 aa without a consensus N-glycosylation and its localized on chromosome 14. FUT8 is structurally and genetically different from other α1,2, α1,3, and α1,4 fucosyltransferases [21], [22]. The 3D structure of human FUT8 has been solved using recombinant proteins expressed by baculovirus infected Sf21 cells in 2007 [23]. From their study, FUT8 consist of three domains: an N-terminal coiled-coil, a catalytic, and a C-terminal SH3 domain. The C-terminal portion of the catalytic domain includes a Rossmann fold with three conserved regions in α1,6-, α1,2-, and protein O-fucosyltransferases. The study revealed FUT8 has a similar catalytic region to that of GT-B glycosyltransferases. The enzyme also contains a SH3 domain, though its significance is yet unknown [23].

Core-fucosylation modification of N-glycans is found and conserved in vertebrates and invertebrates such as Caenorhabditis elegans and Drosophila melanogaster [19], [24], [25], [26]. In contrast to the mammalian FUT8, C. elegans and D. melanogaster FUT8 has one non-conserved potential N-glycosylation site [19]. Many groups have studied the% identity of different animal species of FUT8 and they are listed in Table 1.

FUT8 biological functions

FUT8 has great biological significance, loss of core-fucosylation modification can lead to growth retardation, emphysema, schizophrenia-like behaviors and even death [20], [29]. Loss of core-fucosylation in brain tissue can also lead to decrease working memory by impairment of Hippocampal Long-Term Potentiation [30]. In addition, core-fucosylation is related to the regulation and function of many glycoproteins. One example is the lack of N-glycan core-fucosylation in human IgG1 leads to enhanced antibody-dependent cellular cytotoxicity (ADCC) [24], [31], [32], [33]. Core-fucosylation also regulates the functions of immunoglobulin, for example, in IgG B cell receptors, it is required for the antigen recognition in humoral immune response [34], [35]. Numerous studies have shown the overexpression of FUT8 contributes to the development, progression, malignancy, recurrence, invasive and metastatic properties of cancer cells [10], [12], [13], [14], [36]. Patients with inflammatory bowel disease have shown to have increased core fucosylation of T-cell receptors on T cells from intestinal tissues [37]. Increase core fucosylation has also been shown to be related to Alzheimer [38]. Modifications in core-fucosylation have been linked to human diseases including liver cancer, pancreatic cancer, lung cancer, breast cancer, etc [39], [40]. Furthermore, core fucosylation has great potential as a cancer biomarker as shown by α-fetoprotein, which is highly core-fucosylated in hepatocellular carcinoma, but not other liver diseases [12], [41], [42], [43], [44]. Other core-fucosylated glycoproteins such as E-cadherin and CA125 were also found to be potential cancer biomarkers depending on their core fucosylation level [45], [46]. Core α1,6-fucosylation is a major modification and there is strong evidence for its importance in biological events.

FUT8 specificity studies

Due to its importance and medical relevance many research groups have studied the substrate specificity for this enzyme [17], [24], [47], [48]. Some research groups have expressed soluble forms of C. elegans and D. melanogaster FUT8 in Pichia pastoris. They had also showed that invertebrate’s FUT8 has the same biochemical function as vertebrate FUT8 and recognizes biantennary N-glycans as substrates [19].

It has been widely elucidated regarding the substrate specificity of FUT8 in vivo and in vitro. Human FUT8 is active in the absence of metal cations and was shown to be inhibited by Ni⁺², Cu⁺², and Zn⁺². Optimal buffers were cacodylate and Mes when compared to Tris-HCl and sodium phosphate buffers at pH 7.0 [18]. A library of 77 well defined N-glycans synthesized by our group composed of three types of unlabeled N-glycans structure types: oligomannose, complex and hybrid as well as four simple glycan structures was used to reveal details of FUT8 substrate specificity. The results showed that recombinant human FUT8 expressed by our group recognized 11 out of the 77 substrates assayed [49]. Our results support the previous substrate specificity studies that suggest that GlcNAc at the α-3 mannose branch of the substrate is required for the catalytic activity of the enzyme in vitro [16], [17], [18], [19], [25], [47], [49], [50], [51], [52]. Human and C. elegans FUT8 have a relax substrate preference towards the α1,6-mannose branch, since it fucosylated asymmetric N-glycans [49], [52], [53]. Human FUT8 recognition towards asymmetric N-glycans with an α1,3-linked fucose to the terminal N-acetyllactosamine (LacNAc) or sialyl LacNAc structures on the α1,6-mannose branch is less efficient [49]. N-glycan without a α1,3-mannose branch was recognized by human FUT8, even though with a neglectable percentage conversion [49]. Bisecting N-glycans and galactosylated GlcNAc residue at α1,3-mannose branch prevent FUT8’s activity [16], [17], [49]. Reichardt’s group determined that C. elegans FUT8 can also core fucosylated xylose containing biantennary N-glycans relevant in invertebrates and plants [54]. FUT8 showed to have strict requirement towards the α1,3-mannose branch, and very relaxed requirements towards the α1,6-mannose branch [16], [17], [18], [19], [25], [47], [49], [50], [51], [52].

FUT8 substrate recognition in vivo varies slightly according to findings by various groups. Even though the presence of GlcNAc at the α3 mannose branch catalyzed by N-acetylglucosaminyltransferase I (Gn-T I) is considered to be a prerequisite for core fucosylation by FUT8, there are traces of high mannose core fucosylated glycans in vivo [55], [56]. Core-fucosylated high mannose structures were identified from lysosomal proteins such as rat liver alkaline phosphatase [55], porcine cathepsin D [56], human glucuronidase [57], and bovine mannosidase [58]. Therefore, these structures were thought to be generated by N-acetylglucosaminidase hydrolysis of core fucosylated hybrid type N-glycans. After various groups detected the presence of core fucosylated high mannose structures from natural proteins expressed in Gn-T I knockout CHO [59], and HEK293S [60] cell lines, the presence of a Gn-T I independent fucosylation pathway was confirmed.

Wang’s group was able to demonstrate that FUT8 is the sole enzymes responsible for α1,6 core fucosylation in the Gn-T I independent fucosylation pathway. They also suggested the level of core-fucosylation depends on the nature of the recombinant proteins. This was demonstrated by the expression of erythropoietin (EPO), GM-CSF and ectodomain of FcIIIA receptor recombinant proteins in HEK293S Gn-T I^−/− cell line with a core-fucosylation percentage of 50%, 30%, and 3%, respectively. They confirmed core fucosylation was the sole activity of FUT8 by the expression of EPO in stable HEK293 Gn-T I^−/− cell line with both knockdown or overexpressed FUT8. Their results showed knockdown cell line produced pure Man5GlcNAc2 glycoform, compared to EPO from FUT8 overexpressing cell line which produced completely core fucosylated Man5GlcNAc [61].

Traditional versus chemo-enzymatic synthesis of core fucosylated N-glycans

Synthesis of core-fucosylated N-glycans has been targeted from a variety of aspects. Since 1996 Unverzagt’s group achieved to synthesize the first core fucosylated N-glycan structure synthetically [62]. They synthesized a biantennary core-fucosylated octasaccharide [62], and nonasaccharide biantennary [63], as well as a pentaantennary dodecasaccharide with bisecting and core fucosylation modifications and an azide group for the generation of glycopeptides [64]. The core fucose was introduced at the last step of the chemical synthesis. For the α-fucosylation a p-methoxybenzyl (MPM) substituted thiofucoside was used so that it could be selectively removed by oxidation [65]. The thiofucoside was activated by Bu₄NBr/CuBr₂ for coupling to the desired oligosaccharide [62], [63], [64], [66]. They were also able to optimize the synthesis of a biantennary core-fucosylated N-glycan by using a key building block selectively functionalized trisaccharide that allows glycosylation in a modular way including fucosylation at any stage of the synthesis [67]. Huang’s group has also successfully synthesized an α-2,3 sialylated core-fucosylated biantennary N-glycan dodecasaccharide by combining three modules/blocks in one-pot [68].

In the more recent years’ different approaches have been developed to generate more complex core fucosylated N-glycan structures in more efficient ways. As described by Yu’s review on thioglycosides, there are many advantages and disadvantages to using thioglycosides (such as thiofucoside) for the chemical synthesis of these complex glycan structures [69]. Some of the more known advantages are that they can be activated under extremely mild conditions, stable under wide range of conditions, and easily accessible. However, these compounds resulting waste and their promoters have very unpleasant smell and are toxic. The requirement for promoters also leads to the generation of many side reactions. These major disadvantages are a real problem to the environment and human health, therefore not a favorable route for industrial applications [69].

Several groups have merge chemical synthesis with enzymatic synthesis in order to develop more efficient synthetic ways of producing large quantities of core-fucosylated N-glycans. Wang’s group was one of the pioneers in chemically synthesize a core fucosylated N-linked disaccharide to test the specificity of several endo-β-N-acetylglucosaminidases such as Endo A, Endo M, Endo D, etc. Endoglycosidases from Flavobacterium meningosepticum Endo-F2 and Endo-F3 were the only ones that recognized Fucα1,6GlcNAc-N-Fmoc/peptide substrate. The substrate and donor were chemically synthesized. The substrate was synthesized to have an α1,6-fucosylation. Using Endo-F2 and Endo-F3 they were able to synthesize both sialylated and asialylated core fucosylated N-glycans. With this approach they generate a full sized CD52 glycopeptide antigen both with sialic acid and core fucose (Fig. 2) [70].

Fig. 2:

Chemo-enzymatic synthesis of core-fucosylated glycopeptide and glycoprotein using Endo-F3 and mutated Endo-F3 D165A to transfer biantennary and triantennary complex N-glycans [70], [71].

Wang’s group years later demonstrated that by mutating Endo-F3 to Endo-F3 D165A it is able to transfer not only sialylated and asialylated biantennary N-glycans oxazolines, but also triantennary N-glycan oxazolines to form a complex triantennary core fucosylated glycopeptide or glycoprotein. The donor triantennary oxazoline was semisynthesized by digestion of fetal bovine fetuin using Endo-F3 followed by purification and desialylation using sialidase. Conversion into triantennary glycan oxazoline was done in a single step using excess of 2-chloro-1,3- dimethylimidazolinium chloride and triethylamine in water [72], [73], [74]. This new endoglycosidase based glycosynthase can transfer biantennary and triantennary complex N-glycans to glycopeptides and glycoproteins such as rituximab and porcine fibrinogen (Fig. 2). Making it possible to remodel therapeutic antibody glycosylation pattern including those with α1,6 core-fucosylation [71].

Reichardt, used an 18 glycan array to test the specificity of Caenorhabditis elegans α1,6 fucosyltransferase (CeFUT8) expressed in Pichia pastoris. The specificity of the enzyme toward the different glycan structure was assayed by four lectins: Lens culinaris agglutinin (LCA), Aleuria aurantia lectin (AAL), Pisum sativum agglutinin (PSA), Aspergillus oryzae lectin (AOL), and anti-horseradish peroxidase antibody (anti-HRP). AAL was the only lectin that bound with good selectivity and sensitivity to all α1,6 core-fucosylated structures. CeFUT8 recognized A8–A17 (10 out 18 substrates assayed) (Fig. 3a). After specificity assays for the CeFUT8 the findings were applied to synthesize two core fucosylated N-glycans in solution. A9 and A14 were used as substrates for the synthesis of core fucosylated N-glycans in solution. The final product Compound 5 and 7 were confirmed by NMR and MS (Fig. 3c, d). All the substrates applied for the specificity studies and scale up synthesis for core α-1,6 fucosylation were chemically synthesized [51].

Fig. 3:

(a) CeFUT8 specificity assay using glycan array. (b) All substrates were chemically synthesized with linker. (c&d) Compound 5 and 7 were synthesized using CeFUT8 from A9 and A14, respectively, in solution and the structure was confirmed by NMR and MS [51].

Reichardt’s group also determined that N-glycan structures containing xylose can be α1,6 core-fucosylated by CeFUT8. Using three chemically synthesized N-glycan structures containing xylose (G3, G5, G6), 5 glycosyltransferases (α1,6- fucosyltransferase (CeFUT8), α1,3-fucosyltransferase (AtFucTA/CeFUT1), β1,4-galactosyltransferase (GalT), β1,4-N-acetyl-galactosyltransferase (GalNAcT), LeX-type α1,6-fucosyltransferase (CeFUT6)), and 1 N-acetyl-glucosaminidase they were able to synthesize by chemo-enzymatic extension 14 α1,6 core fucosylated invertebrate and plant N-glycans containing xylose Fig. 4 [54].

Fig. 4:

Compound 2 before deprotection of R₁=Bn by hydrogenation. In red, incorporated 13C isotopes. Scheme of chemo-enzymatic synthesis of CF-glycans [6], [53]. (a) CeFUT8, (b) GalT, (c) AtFucTA/CeFUT1, (d) GalNAcT, (e) CeFUT6, (f) N-acetyl-glucosaminidase.

Reichardt’s group also used similar chemo-enzymatic strategy, reviewed before, to generate a library of 16 ¹³C-labeled N-glycans useful as internal standards for absolute quantification by MALDI-TOF. 8 out 16 N-glycans synthesized by their group were core-fucosylated N-glycans and only three were asymmetrical. The chemically synthesized core structure (Compound 2) that’s partially protected and isotope labeled (Fig. 5) was used for the extension using three enzymes (β-1,4-GalT, β-1,4-N-acetyl-glucosaminidase, α-2,3-SialylT) to generate the N-glycan library. In order to generate the eight core-fucosylated N-glycans Compound 2 was deprotected by hydrogenation before α1,6 fucosylation [53].

Fig. 5:

Scheme of chemo-enzymatic synthesis of 14 α1,6 core fucosylated invertebrate and plant N-glycans containing xylose [54]. (a) β-1,4-GalT, (b) 10% Pd/C, H2 (c) α1,6 FucT (d) N-acetyl-glucosaminidase, (e) α-2,3-SialylT.

Our group has also contributed to the preparation of N-glycans using a chemo-enzymatic strategy. 72 N-glycans were chemo-enzymatically synthesized following extension starting from 8 core chemically synthesized N-glycans to generate 62 isomers and 4 core fucosylated glycans using recombinant human α-1,6 fucosyltransferase (FUT8). The core fucosylated structures generate (N6030, N6000, N6211, N6212) were synthesized from N030, N000, N211, and N212 (Fig. 6). N030 and N000 were chemically synthesized as part of the core structures used for extension. N211 and N212 were chemo-enzymatically synthesized [52]. The core-fucosylated N-glycans generated were the same as previously reported by Reichardt’s group.

Fig. 6:

CF-glycans generated by Wang lab [52].

A complete study on the specificity of recombinant human FUT8 was done by our group using the 77 N-glycan library that was mentioned above. Using the specificity of the enzyme a chemo-enzymatic strategy was developed for the synthesis of a panel of core-fucosylated asymmetric N-glycans. The generation of 14 core-fucosylated N-glycans was achieved by following the scheme in Fig. 7 [49].

Fig. 7:

(A) Scheme for chemo-enzymatic synthesis of asymmetric α1,6 core fucosylated N-glycans starting with N110 as a substrate. (B) Scheme for chemo-enzymatic synthesis of asymmetric α1,6 core fucosylated N-glycans starting with N210 as a substrate [49]. (a) FUT8, (b) β4GALT1, (c) 30 % ammonium hydroxide : H₂O (1 : 10), (d) PmST1m, (e) Pd2,6ST, (f) Hpα1,3FT.

The scheme used two chemically synthesized starting N-glycans N110 and N210 as substrates. For the enzymatic extension of asymmetric core-fucosylated N-glycans N110 and N210 had a protected GlcNAc at the α6 or α3 mannose branch, respectively. All the core-fucosylated N-glycans were confirmed by MALDI-MS and HPLC retention time. Core fucosylated isomers N6122, N6123, N6244, N6222, N6223, and N6144 were confirmed by ¹H-NMR as well [49].

Conclusion

In conclusion, FUT8 the only mammalian enzyme responsible for core α1,6 fucosylation, has been widely studied by many research groups. The crystal structure has been determined as well as its metal, buffer and pH preferences. The substrate specificity for the enzyme was studied in detailed and the results concluded that there is a strict requirement towards the α1,3-mannose branch, but very relax requirements towards the α1,6-mannose branch. Many ways to synthesize core α1,6 fucosylated N-glycans have been explored both chemically and enzymatically. These structures are valuable as standards for glycan analysis, and glycan-binding studies. New approaches to chemo-enzymatically synthesize of complex α1,6 core fucosylated N-glycans have allowed for the rapid accessibility to well-defined structures. Future focused arrays containing core-fucosylated N-glycans will allow for identifying prospective antigens for vaccine design. These new chemo-enzymatic strategies will play a crucial role in the synthesis of potential antigenic fucosylated N-glycans for glycoconjugate vaccines [51].