In the process of screening the reaction conditions, we found that compound 3A can be obtained with trifluoromethanesulfonic acid as a reagent and water as an additive in the solvent of dichloromethane (Table S1, Supporting Information) and then, more optimizations were carried out (Table 1). By investigating the amount of water (Entry 1–4), we found that one equivalent of water gave the highest yield of 3A (40% yield, Entry 3). Then the amount of trifluoromethanesulfonic acid was studied (Entry 5–7) and two equivalents were found to give 3A in the highest yield as of 74% (Entry 6). In addition, changing the temperature (Entry 8–9) and reaction time (Entry 10–11) did not obtain higher yield of 3A. Several other solvents were also screened but no 3A was observed (Entry 12–14).
According to the optimized reaction conditions, different benzonitrile substrates were then investigated to look for which have good stereoselectivity of products (Table 2). The yields of the compounds with substituents of electron-donating group are between 57% and 90% (3B-3E). However, the yields of which with electron-withdrawing substituents dropped significantly, to less than 40% (3F-3G). The stereoselectivity of product 3E with the electron-donating substituent is much better than other compounds, similar to the un-substituted 3A. The yields and stereoselectivity of compounds 3H-3I with the halogen substituent at the para position of the benzene ring are obviously weaker than the compounds 3J-3L with the halogen substituent at the ortho position of the benzene ring. When adding other electron-donating substituents on the benzene ring substituted by the ortho-halogen, there is no rule to follow in the yields and configuration control (3M-3T). However, most products have good stereoselectivity, such as 3M, 3N, 3Q, 3S and 3T. In addition, when the heterocyclic cyanide substrate is used, the high product yield would be obtained but with poor stereoselectivity (3U-3V). Finally, we also tried a nitrogen-containing cyanide substrate, but no product was obtained, which is probably due to the protonation of nitrogen atoms under acidic conditions (3W-3X). This also shows that the nucleophilicity of the cyano group is closely related to the reaction yields. In order to determine the structure of the product, we removed the acetyl protecting group of compound 3E to obtain 3Y, and then determined the X-ray single crystal structure of 3Y.
Based on the above results, we have selected the benzonitrile substrates with good stereoselectivity of products and continue to screen various saccharides (Scheme 2). Different six-membered ring saccharides can successfully obtain the corresponding products (3Ea-b, 3Sa). The C-N configuration at 1-position of the product is controlled by the configuration of the ortho group at 2-position on the saccharide ring. For example, glucose gives the alpha products 3E and 3S, while mannose gives the beta products 3Eb and 3Sa, which indicated that we can further obtain 1-α or 1-β glycosylamines originated from different saccharides. Moreover, both disaccharide and trisaccharide substrates can give the corresponding products (3Ec-d). In addition, five-membered ring saccharides can also generate oxazolinoses in high yields and with good stereoselectivity (3a-g). We have compared the 1H-NMR spectra of several oxazolinoses and found visible differences of the hydrogen chemical shift at 1-position between 1-α and 1-β C-N configurations. The chemical shifts of hydrogen atom on C-1 for 1-α oxazolionses are around at 6.10 ppm. Electron-donating or withdrawing substitutes can make the chemical shifts fluctuate slightly. Meanwhile, the chemical shifts of hydrogen atom on C-1 for 1-β oxazolinoses are around at 6.40 ppm. Therefore, the difference of two isomers can be recognized via 1H-NMR spectra (Fig S1, Supporting Information).
The reaction mechanism is proposed in Scheme 3a. Acetyl glucose 1A generates carbocation compound T under the action of trifluoromethanesulfonic acid. Then T is attacked by the benzonitrile 2A, which might give 1-S and 1-R products. However, the experiment results mainly obtained 1,2-cis product 3A (C1). In order to gain a further insight into the intramolecular cyclization mechanism and the preference of stereoselectivity for the oxazoline ring of target product C1 (3A), a computational study using the density functional theory (DFT) method at M06-2X/6-311G(d) level was performed with Gaussian 16 B.01 [11, 12] for geometry optimization, frequency analysis and transition state search. The solvent effects of dichloromethane (DCM) were taken into consideration by using solvation model based on density (SMD) [13] during the calculations.
As shown in Scheme 3b, the intermediate A1 was predicted to be more stable than A2 by 5.0 kcal/mol, indicating that A1 is the dominant configuration produced from the nucleophilic attack of benzonitrile. A1 could undergo an intramolecular nucleophilic addition of its oxygen atom of the acetoxy group attacking the carbon atom of the imide bond to afford B1, via transition state TS1 with the free energy barrier of 18.8 kcal/mol. On the other hand, A2 might undergo a similar reaction path via transition state TS2 to afford B2 with the free energy barrier of 23.4 kcal/mol, 4.6 kcal/mol higher than TS1, which is consistent with the experimental observation that C2 almost cannot be detected at room temperature. The water existing in the environment promoted the departure of the B1 acetyl group to generate target product C1, which is 14.1 kcal/mol more stable than C2, suggesting again that C1 should be the main product of the reaction (Supporting Information).
We further investigated the potential application of the oxazolinoses in organic synthesis (Scheme 4A). For example, 3A can perform ring-open reaction under acidic conditions to give the 1-α-amino compound 4A with the benzoate group at 2-position. On the other hand, the acetyl protecting groups of 3A can be removed to generate 5. Then the benzyl protecting groups were introduced to give 6 which can be treated with acid to produce 4B. After diazotization treatment, the 1-α-amino group of 4B can be converted into a 1-β-hydroxyl group to obtain 7 with selective protecting group at 2-position. The 1-α-amino carbohydrate 4B can smoothly undergo the synthesis of glycosylamines 10a-c. Compound 5 can also undergo the simple conversion of the protective group on the saccharide ring through 8 and 9 to obtain glycosylamine 10d with differentiated protecting groups at the 2,3-positions on the saccharide ring. Thus, the N-glycoside derivatives similar to 10a-c can be further prepared via 10d. In addition, 3S can undergo Ullmann-coupling reaction under the action of copper powder to generate the axial chiral compound 11 with P configuration (Scheme 4B). It shows that the oxazolinoses have the potential as the chiral blocks in organic synthesis.