Scalable Synthesis of Fmoc-Protected GalNAc-Threonine Amino Acid and T_N Antigen via Nickel Catalysis

Abstract

The highly α-selective and scalable synthesis of the Fmoc-protected GalNAc-threonine amino acid and T_N antigen in gram scale (0.5 – 1 gram), is described. The challenging 1,2-cis-2-amino glycosidic bond is addressed through a coupling of threonine units with C(2)-N-ortho-(trifluoromethyl)benzylidenamino trihaloacetimidate donors mediated by Ni(4-F-PhCN)₄(OTf)₂. The desired 1,2-cis-2-amino glycoside was obtained in 66% (3.77 g) with α-only selectivity and subsequently transformed into the Fmoc-protected GalNAc-threonine and T_N antigen. This operationally simple procedure no longer requires utilization of the commonly used C(2)-azido donors and overcomes many of the limitations associated with the synthesis of 1,2-cis linkage.

Graphical abstract

Protein glycosylation can be generally divided into two major classes: N-linked and O-linked. In N-linked glycoproteins, an N-acetyl-glucosamine (GlcNAc) unit is β-linked to the amide nitrogen of an asparagine amino acid side chain.¹ In O-linked glycoproteins, an N-acetyl-galactosamine (GalNAc) unit is α-linked to the hydroxyl group of serine or threonine to generate a core structure 1 (Figure 1), commonly referred to as the T_N antigen.² Branching of this core structure 1 can take place at the C(3)- and/or C(6)-hydroxyl groups of GalNAc to give rise to a diverse array of structural motifs (e.g. TF antigen 2 and ST_N antigen 3, Figure 1). These antigens are widely distributed on cell-surface mucin glycoproteins, which participate in cell adhesion events associated with cancer metastasis.³ The T_N antigen 1, in particular, has been found to be highly expressed by mucins on most epithelial cancers.⁴ As a result, this T_N antigen has been investigated extensively as a biomarker and a therapeutic target for cancer vaccine therapy.⁵

Figure 1.

Structure of T_N, TF, and ST_N antigens.

In the development of cancer vaccines, well-defined and pure T_N antigen as a single tumor antigen or as a component of a polyvalent vaccine is required. However, acquiring adequate amounts of T_N antigen from natural sources in homogenous form is challenging. In many cases, high purity T_N antigen can only be obtained by chemical and/or enzymatic synthesis.⁶ In the chemical synthesis strategy, C(2)-azido donors are the most commonly used substrates for generating the T_N antigen. Early work utilized a C(2)-azido halo donor 4 (Scheme 1) in the presence of the reagent combination of Ag₂CO₃ and AgClO₄ as a promoter to ensure α-selectivity (α:β = 4:1) in the glycosylation reaction.⁷ Another efficient synthesis of T_N antigen employed a C(2)-azido thioglycoside donor 5 (Scheme 1), and the Ph₂SO/Tf₂O system is employed to promote α-glycosylation reaction.⁸ Compound 7 was further converted into Fmoc-protected GalNAc-threonine amino acid 8 (2 steps, for use in the production of full-length glycosylated proteins) and T_N antigen 1 (3 steps). A number of efficient strategies were subsequently developed for generating glycopeptides containing the T_N antigen moiety.^9,10

Scheme 1.

Previous Methods for Synthesis of T_N Antigen

Both glycosyl amino acid 8 and T_N antigen (1) are readily available, but they are expensive to purchase (8: $303.50/25 mg and 1: $250/mg from Sigma-Aldrich). Although high purity T_N antigen can be chemically prepared, it cannot be easily and reproducibly obtained in large quantities. Most of existing glycosylation procedures require stoichiometric amounts of the activating agents to sufficiently activate donors, resulting in excessive waste materials.^7–9 Some of these reagents can be air- and moisture-sensitive (e.g. Ph₂SO/Tf₂O)⁸ and potentially explosive (e.g. AgClO₄).⁹ In addition, the synthesis of the commonly used C(2)-azido donors 4 and 5 (Scheme 1) is not trivial. Lemieux’s azidonitration method for preparing 4 and 5 is not very diastereoselective,¹¹ depending on the nature of the protecting groups on glycal starting material.¹² Alternatively, diazotransfer reaction can be utilized to prepare donors 4 and 5 through direct conversion of galactosamine by the action of either trifluoromethanesulfonyl azide or imidazole-1-sulfonyl azide,¹³ which are potentially explosive reagents. Although the diazotransfer method is frequently used for preparing 4 and 5, it is unlikely to be suitable for large scale synthesis.¹⁴ Herein, we report a scalable and reproducible protocol for the synthesis of the glycosyl amino acid 8 and T_N antigen (1) via nickel-mediated α-glycosylation of threonine amino acids with C(2)-N-ortho-(trifluoromethyl)benzylidenamino trihaloacetimidate donors. This operationally simple procedure no longer requires the utilization of C(2)-azido glycosyl donors and is suitable for a gram-scale synthesis of 1 and 8.

In recent years, our group has introduced nickel-catalyzed α-stereoselective glycosylation reaction as a general platform for preparations of a variety of 1,2-cis-2-amino glycosides.¹⁵ Additionally, we have illustrated that Ni(4-F-PhCN)₄(OTf)₂ effectively promoted a coupling of Cbz-protected threonine residue 10 with C(2)-para-(trifluoromethyl)benzylidenamino trichloroacetimidate donor 9 to afford glycosyl amino acid 11 (Scheme 2a) in 81% yield with α:β = 15:1.^15b We postulated that an analogous nickel-catalyzed α-selective coupling would be possible with Fmoc-protected amino acid 13 (Scheme 2b). Of two standard methods for the solid-phase peptide synthesis (SPPS) of glycopeptides containing T_N antigen unit, Fmoc-based chemistry is more utilized than Boc-based chemistry.^1b Unfortunately, employing 5 mol% of Ni(4-F-PhCN)₄(OTf)₂ to promote the coupling of 13 with donor 9 only resulted in a 32% yield of 14 (Scheme 2b) with α:β = 2:1. Alternatively, use of C(2)-N-ortho-(trifluoromethyl)benzylidenamino donor 12 (Scheme 1b) improved both the yield (32% → 80%) and α-selectivity (α:β = 2:1 → 7:1). Although α-trichloroacetimidate donor 12 acted as an effective donor, it was a minor anomer resulting from the reaction of hemiacetal with Cl₃CCN and DBU (α:β = 1:3). Unfortunately, reaction of β-anomer of 12 with 13 resulted in no reaction.¹⁶

Scheme 2.

Route to GalNAc-Threonine Residue via Nickel Catalysis

On the basis of our recent successful results with the use of N-phenyl trifluoroacetimidates as effective donors,^15d–f we hypothesize that triacetyl galactosamine donor 16 (Scheme 3), bearing the C(2)-N-ortho-(trifluoromethyl)benzylidene group, is a suitable starting material for the rapid and gram-scale synthesis of glycoside 15, its corresponding Fmoc-protected threonine amino acid 8, and T_N antigen (1). In contrast to our existing systems (Schemes 2a–b),^7–10 this process can promote the glycoyslation with both α- and β-anomers of 16¹⁶ and only relies on substoichiometric amounts of the nickel catalyst.

Scheme 3.

Reproducible and Gram-Scale Synthesis of Glycosyl GalNAc-Threonine Compound 15

While it was known that Ni(4-F-PhCN)₄(OTf)₂ effectively promoted the glycosylation of a wide variety of carbohydrate acceptors with C(2)-ortho-(trifluoromethyl)benzylidenamino N-phenyl trifluoroacetimidate donors,^15e–f,17 it was unclear if the reaction of Fmoc-protected threonine amino acid 13 with substrate 16 would proceed with high yield and α-selectivity. Importantly, it was still unclear if the nickel method can be utilized in a large scale preparation of glycosyl amino acid 15 (Scheme 3). We were delighted to find that by employing only 10 mol% Ni(4-F-PhCN)₄(OTf)₂ the coupling reaction reached completion in 12 h at 35 °C to afford the desired product 15 in 74% yield with exclusive α-anomeric selectivity (Scheme 3a). Purification of the glycosyl amino acid 15, however, was tedious due to closeness in R_f value of the threonine acceptor 13 to the desired product 15. In the second trial, we glycosylated 13 with N-phenyl trifluoroacetimidate donor 14 on a similar scale (Scheme 3b) and obtained a comparable yield and selectivity (63%, α only). The yield in this second run was slightly lower because we tried two different purification methods (manual and automated chromatography) to separate 15 from unreacted threonine donor 13. Unfortunately, it was not successful. Anticipating that this problem would be exacerbated in a larger scale, we made the threonine acceptor 13 the limiting reagent (Scheme 3c) and isolated 3.77 g of pure product 15 in 66% yield with α-only selectivity.¹⁸ Overall, the results obtained in Scheme 3 have illustrated the high α-selectivity and scalability of the nickel-catalyzed glycosylation reaction under mild and operationally simple conditions.

Further investigation of scope showed that a glycosylation reaction could be realized using 10 mol% of nickel catalyst, Ni(4-F-PhCN)₄(OTf)₂, with other donors and a number of Fmoc-protected threonine amino acids to afford the desired 1,2-cis-2-amino glycosides 17 – 21 (Figure 2) in good yields (61 – 86%) with excellent α-selectivity (α:β = 14:1 – α only). The terminal alkyne of product 18 is capable of conjugating to biorthogonal azide, via click chemistry,¹⁹ for incorporation into a wide variety of biomolecules.²⁰ This alkyne can also conjugate to a linker possessing the azide functionality to form the corresponding polymerizable monomer, which can then undergo ring-opening metathesis polymerization²¹ to generate highly clustered T_N antigens for potential use as antitumor vaccine candidates.²² On the other hand, both glycosyl amino acids 19 and 20 can be further functionalized to generate the ST_N antigen (3, Figure 1). The Cbz-protected threonine amino acid was also compatible with this nickel system, providing the desired glycoside product 21 (Figure 2) in 67% yield as a single α-anomer.

Figure 2.

Scope of the reaction with threonine amino acids

Since the Fmoc-protected GalNAc-threonine amino acid 8 (Scheme 4) is a versatile building block required for SPPS of mucin-type glycopeptides,^1b we next investigated the mild conditions for converting glycosyl amino acid 15 into 8. The previous conditions (2N – 5N HCl, acetone, 25 – 50 °C)¹⁵ for the exchange of C(2)-N-benzylidenamino functionality with N-acetyl group to form 22 (Scheme 4) may not be suitable for use in a large scale synthesis. Using the product 15 from Scheme 3c, we found that the benzylidene group could be removed with acetyl chloride (1.6 equiv) in methanol at 25 °C. Subsequent acetylation of the amine salt intermediate provided the desired Fmoc-protected GalNAc-threonine 22 in 70% yield (Scheme 4). One-half of 22 from batch C (Scheme 4) was utilized to synthesize 1.02 g of Fmoc-protected GalNAc-threonine amino acid 8 (97% yield) using Pd(Ph₃P)₄ in THF and NMA at 25 °C for 1 h. Global hydrolysis of the other half of 22 from batch C with sodium hydroxide in methanol provided 0.4 g of the T_N-antigen (1) in 82% yield. While a high yield of 1 was obtained, we found these conditions to be insufficient in fully deprotecting Fmoc group.

Scheme 4.

Gram-Scale Synthesis of T_N Antigen and Fmoc-Protected GalNAc-Threonine Amino Acid: First Conditions

We hypothesized that addition of triethylamine alongside sodium hydroxide in methanol would facilitate quantitative global deprotection of the intermediate 22 to produce the T_N antigen (1). To test our hypothesis, batches A and B of 1,2-cis-2-amino glycoside 15 from Scheme 3a and 3b were combined and transformed to 1.18 g of GalNAc-threonine amino acid intermediate 22 (Scheme 5). We are delighted to report that global deprotection of 22 in the presence of triethylamine and sodium hydroxide occurred with almost quantitative yield (99%, Scheme 5).

Scheme 5.

Gram-Scale Synthesis of the T_N Antigen (1) - Second Conditions

In summary, we have illustrated a highly α-selective 1,2-cis-2-amino glycosylation reaction utilizing substoichiometric amount of Ni(4-F-PhCN)₄(OTf)₂ to mediate the coupling of a number of Cbz- and Fmoc-protected threonine amino acids with C(2)-ortho-(trifluoromethyl)benzylideneamino N-phenyl trifluoroacetimidate donors. This methodology demonstrates the utility of our catalytic, selective glycosylation method for a gram scale preparation of glycosyl 1,2-cis-2-amino acids and their subsequent transformation into the corresponding Fmoc-protected GalNAc-threonine amino acid and T_N antigen. This operationally simple procedure no longer requires utilization of the commonly used C(2)-azido donors, which are often prepared via the potentially explosive diazotransfer reaction. As the scalable, catalytic, and stereoselective synthesis of complex oligosaccharides and glycoconjugates continues to develop, we anticipate that this nickel-catalyzed glycosylation methodology will have an impact on the strategies used for the preparation of biologically active carbohydrate molecules.