Radical addition of glycals. As shown in scheme 2, we hypothesized that the nitrogen-centered benzenesulfonimide radical A, generated from NFSI by reduction with TEMPO, might regioselectively electrophilic add the alkenyl of glycal to give radical D. The radical D would then be oxidized by the TEMPO+ B to afford the oxocarbenium E and recycle the catalysis TEMPO.54, 55 We considered that the steric hindrance of benzenesulfonimide and/or the neighboring effects of sulfonyl group would direct the glycosylation in 1,2-trans manner to furnish the anomerically specific product F.
Our initial investigation of the nitrogen-centered radical involved amidoglycosylation was conduct between the tri-O-benzyl-D-glucal 1 and diisopropylidene-D-glucose 2. After treatment with NFSI and catalytic amount of TEMPO in DCE at room temperature, the glycosylation product 3 was obtained in 22% yield (Table 1, entry 1). The NMR analysis showed that the acceptor 2 was β-selectively coupled with the glucal 1 as we proposed. We noticed that the glycal 1 was not completely consumed. To accelerate the reaction and increase the yield, the reaction temperature was raised to 50 oC. The glycosylation was slight improved with 38% yield (entry 2). Screening the solvents, such as DCM, THF, MeNO2, and toluene (entries 3 - 6), identified MeCN as the optimal solvent with 56% yield (entry 7). Meanwhile, a byproduct was also isolated from the reaction mixture as results of Ferrier-rearrangement. According to the mechanism of Ferrier-rearrangement,56-58 the C-3 elimination was resulted from the HF formation in the reaction. However, attempts to neutralize the HF with base (NaHCO3, Na2CO3, Na3PO4 et al.) did not significantly suppress the Ferrier-rearrangement. Instead, we found that reducing the amount of NFSI considerably decreased the Ferrier-rearrangement byproduct and the glycosylation product 2 was afforded in 75% (Table 1, entry 8 and 9). The amount of catalysis TEMPO was also evaluated and 0.2 equivalent of TEMPO gave the highest yield (entries10 and 11). To demonstrate the practicability of this method, the reaction was conducted in 2 grams scale and 64% yield was achieved.
Scope of glycals. With the optimized reaction conditions established, various glycal donors were examined to explore the scope of the NCR-mediated amidoglycosylation (Scheme 3). Besides the benzyl ether, the methyl ether, benzylidene, and diisopropylidene were also suitable protecting groups and the corresponding disaccharides 4 – 6 were afforded in good yields as well as excellent selectivities (β/α > 20:1 or β only). However, we found that the glycal with acetyl protecting group only gave the product 7 in 37% yield. The low yield was due to the lower HOMO energies of the double bond.59 The D-galactals and L-rhamnals were also subjected for the amidoglycosylation. The related disaccharides 8 - 11 were harvested in good yields too. These results demonstrated that the nitrogen-centered radical was practical protocol for the preparation of β-selective 2-amino-2-deoxyglycosides using glycals as glycosylation donors.
Scope of acceptors. The scope of acceptors for amidoglycosylation was also evaluated (scheme 4). Different monosaccharide acceptors with primary alcohols were subjected for the radical glycosylation. The products 12 – 14 were obtained in good yields and excellent anomeric selectivities. The acceptors with secondary alcohols were then conducted for the glycosylation. But we found that the yields were obviously decreased (30% for 15 and 42% for 16). We originally considered that the low yield was due to formation Ferrier-rearrangement byproducts too. However only trace amount of the rearrangement byproduct was isolated from the reaction mixture. Instead, we found that a 1-fluorosaccharide was formed as the major byproduct. We reasoned that the formation of 1-fluorosaccharide was due to the competition of NFSI-derived fluoride anion with the bulky secondary saccharide acceptors.60, 61 It was found that addition of Na2HPO4 to the reaction mixture significantly avoided the formation of 1-fluorosaccharide and the yields were increased to 53% and 64%.
Besides the glycosylated serine 17, a salidroside analog 18 with protective activities against the hypoglycemia and serum limitation induced cell death in rat pheochromocytoma cells,62, 63 was also synthesized. Finally, the epiandrosterone was subjected as acceptor and the corresponding glycan 19 was furnished in moderate yield as a single anomeric product.
The disulfonyl group is a stable protecting group for subsequent reactions in polysaccharides synthesis. The benzenesulfonimide glycosides also can be easily converted to 2-N-acetamido-2-deoxyglycosides by simple treatment. As shown in scheme 5, the disulfonyl group of disaccharide 3 was efficiently removed by treatment with SmI2.64 Then the reaction mixture was treated with acetyl anhydride to introduce the acetyl group and gave the NAc-disaccharide 20 in 77% yield in one-pot manner.
In summary, we have established an efficient protocol for cascade synthesis of 1,2-trans 2-amino-2-deoxyglycoside, which gives a novel and useful example for radical-mediated glycosylation. The reaction involves a nitrogen-centered radical-mediated amination of glycals and followed by glycosylation. The utilities of this method was illustrated with different glycals and acceptors with highly anomeric selectivities. Moreover, the disulfonyl group was easily converted to acetyl group by one-pot treatment.