Robust perfluorophenylboronic acid-catalyzed stereoselective synthesis of 2,3-unsaturated O-, C-, N- and S-linked glycosides

A convenient protocol was developed for the synthesis of 2,3-unsaturated C-, O-, N- and S-linked glycosides (enosides) using 20 mol % perflurophenylboronic acid catalyst via Ferrier rearrangement. Using this protocol, D-glucals and L-rhamnals reacted with various C-, O-, N- and S-nucleophiles to give a wide range of glycosides in up to 98% yields with mainly α-anomeric selectivity. The perflurophenylboronic acid successfully catalyzed a wide range of substrates (both glucals and nucleophiles) under very mild reaction conditions.


Introduction
2,3-Unsaturated glycosides, also known as pseudo-glycosides or enosides, are an important class of natural products with many biological activities and capacity to serve as substrates for further reactions [1][2][3]. They are involved in biochemical processes such as molecular recognition, cell-cell interaction, immunological recognition and transmission of biological information [4][5][6]. They are easily transformed into important bioactive compounds such as oligosaccharides, glycopeptides, nucleosides, antibiotics, uronic acids and other natural products [1][2][3].
The Ferrier rearrangement is one of the most useful processes to synthesize pseudo-glycosides in a direct and stereoselective fashion. Several classes of catalysts have been successfully applied in the Ferrier rearrangement including Brønsted acids [7][8][9][10][11][12][13], Lewis acids [14][15][16][17][18][19], redox reagents [20] and metal catalysts [21][22][23]. However, many of these catalysts have limited substrate scope, give variable selectivities and yields, require harsh reaction conditions and an excess amount of catalysts that are typically expensive, toxic and moisture/air sensitive. The majority of the reported catalysts are metal-based, and in pharmaceutical manufacturing, traces of metals pose a major challenge for their removal to acceptable limits. Therefore, the discovery of efficient, metal-free and mild catalysts for the Ferrier rearrangement is still challenging and desirable especially if such catalysts work well with a wide range of C-, O-, Nand S-nucleophiles under mild conditions. We also noted that the use of organocatalysts to catalyze the Ferrier rearrangement is scarcely reported.
Recently, organoboron-catalysis emerged as a mild and effective strategy for activation of alcohols [24], epoxide opening [25,26], Friedel-Crafts alkylations [27], dehydrative glycosylation [28] and many other reactions [29][30][31]. The robustness and mildness of organoboronic acid catalysts in comparison to traditional strong Lewis and Brønsted acid catalysts inspired us to investigate them as promoters for the Ferrier rearrangement. We envisioned that organoboronic acids can activate the allylic acetate of glycals making them susceptible to nucleophilic attacks under conditions favoring a strong polarization of the allylic acetate moiety (see Figure 4).
Herein, we report a phenylboronic acid-catalyzed synthesis of 2,3-unsaturated C-, O-, Nand S-glycosides via Ferrier rearrangement under very mild conditions. We also demonstrate the scope of the reaction using a wide range of glycals and C-, O-, Nand S-nucleophiles.

Results and Discussion
We began our study by investigating the reaction of 3,4,6-tri-Oacetyl-D-glucal (1a) with benzyl alcohol (2) in the presence of 20 mol % of arylboronic acids in different solvents (Table 1, entries 1-6). Phenylboronic acid failed to promote the reaction in several solvents and the starting glucal 1a was recovered unchanged (Table 1, entry 1). This is attributed to its low acidity. Gratifyingly, the more acidic perflurophenylboronic acid successfully promoted the reaction to give 4,6-di-O-acetyl-2,3-unsaturated glucoside 3a in CH 3 CN or CH 3 NO 2 solvents (Table 1, entries 3 and 4). It gave better 92% yield of glucoside 3a in CH 3 NO 2 over a shorter reaction time (Table 1, entries 4  vs 3). Under the same conditions, the reaction did not proceed in THF or DCM due to the lower polarity of these solvents in comparison to CH 3 CN and CH 3 NO 2 ( Table 1, entry 2 vs 3 and 4). Attempts to reduce the amount of perfluorophenylboronic acid to 5 mol % resulted in a reduction of the yield of glucoside 3a despite various attempts to promote the reaction by increasing the temperature and time ( Table 1, entries 5 and 6). In all the cases, the α:β ratio of glucoside 3a was 90:10. Conditions in Table 1, entry 4 were considered as optimum. The structure of glucoside 3a was confirmed by the 1 H NMR spectra where the anomeric proton (H1) appeared at δ 5.16 ppm (for glucal 1a it appears at δ 6.47 ppm) and the protons of the new double bond (H2, H3) appeared at δ 5.90-5.88 ppm [19]. The corresponding protons in the β-isomer appeared at δ 5.22 (H1) and δ 6.01 (H2, H3) [19,21].
Using the optimized conditions in Table 1, entry 4, we then examined the substrate scope. Therefore, glucal 1a was reacted with various O-nucleophiles (using primary, secondary, tertiary, allyl, propargyl alcohols and sugars), C-nucleophiles (using trimethylsilyl cyanide and trimethyl(propargyl)silane), S-nucleophiles (using thiophenol and p-toluenethiol) and N-nucleophiles (methane sulfonamide and p-toluene sulfonamide) ( Figure 1). In all the cases, the reactions successfully gave the respective 2,3-unsaturated glycosides 3a-u in up to 92% yield with mainly α-anomeric selectivity (Table 1). Noteworthy, the reaction also gave disaccharide 3n and 3o smoothly with complete α-anomeric selectivity albeit in a moderate yield. Like- wise, reaction using Et 3 SiH gave the desired 2,3-unsaturated sugar 3i in 74% yield. These results testify to the robustness of the perflurophenylboronic acid as a versatile organocatalyst for the Ferrier rearrangement reaction. We noted that the yields of the disaccharides 3n and 3o and sulfonamides 3t and 3u can be increased with increase in the temperature (60 °C) and extension of the reaction time. The results in Table 1 are superior to the results obtained using boron trifluoride diethyl etherate [32].
Based on the excellent results obtained with the reactions of glucals 1a and 4a with O-, C-, N-, S-nucleophiles, we further extended the scope of this reaction to 3,4-di-O-acetyl-Lrhamnal (6a, Figure 3). As a demonstration, the reaction between 3,4-di-O-acetyl-L-rhamnal (6a) and selected alcohols and p-toluenethiol proceed smoothly and afforded the desired 2,3unsaturated L-rhamnosides (enosides) 7a-h in up to 89% yield with complete α-anomeric selectivity (except for 7a). Disaccharide 7g was also obtained smoothly with complete α-anomeric selectivity. The reactions using rhamnal 6a was completed at a much faster rate within 2 h at room temperature in comparison to glucals 1a and 4a which required ≈6 h at 40 °C to give the products.
A plausible pathway of the reaction is proposed in Figure 4. Coordination of perflurophenylboronic acid to the allylic acetate moiety of glucal 1a induces polarization (structure I) and leads to the formation of an allyloxycarbenium ion (structure II) in the preferred 4 H 3 conformation. Addition of the nucleophiles to C1 from the α-face gives the lower energy half- chair conformer and results in the observed α-selectivity of the 2,3-unsaturated glycosides III (Figure 4) [22]. However, the addition of the nucleophiles from the β-face gives the higher energy twist-boat conformer.

Conclusion
We developed a robust perfluorophenylboronic-acid-catalyzed protocol for the synthesis of a broad range of 2,3-unsaturated O-, C-, Sand N-linked glycosides (enosides) in high yields and mostly α-anomeric selectivity through the reactions of D-glucal 1a, 2-acetoxy D-glucal 4a and L-rhamnal 6a with various C-, O-, Nand S-nucleophiles. Application of this protocol using other glycals is underway in our laboratory.

Supporting Information
Supporting Information File 1 Experimental data and copies of 1 H and 13 C NMR spectra of glycosides 3a-u, 5a-d and 7a-h are provided.