Regioselectivity in the Ring Opening of Epoxides for the Synthesis of Aminocyclitols from D-(−)-Quinic Acid

Efficient syntheses of four aminocyclitols are reported. Each synthesis is accomplished in eight steps starting from D-(−)-quinic acid. The key step involves a highly regioselective ring opening of epoxides by sodium azide.

Previously, we have synthesized three aminocyclitols from D-()-quinic acid in nine to ten steps via stereoselective dihydroxylation as a key step [18] (Figure 2). These quercitol-like structures of aminocyclitols are also called as deoxyinosamines [4]. We described herein an alternative synthesis of two known aminocyclitols 5 and 6 along with two new aminocyclitols 10 and 11. The synthesis was accomplished in eight steps via a regioselective ring opening reaction of epoxides.

Results and Discussion
Unlike the strategy we previously used in the synthesis of aminocyclitols (Figure 2), we started from the epoxides 1, 2 and 7, which were prepared from D-()-quinic acid in six steps, respectively [19]. When compounds 1 and 2 were treated with sodium azide in DMF under reflux conditions, they underwent a highly regioselective opening at the C4 position to afford 3 and 4, respectively (Scheme 1). Scheme 1. Synthesis of aminocyclitols 5, 6, 10 and 11.
The yields were mediocre but no other regioisomers were detected by TLC or isolated from column purification [20]. Interestingly, the TMB-protected compound 7 was treated with NaN 3 to afford 8 in 71% yield and its epimer 9 in 14% yield. The azide directly attacked the least hindered side of 7 at the C4 position to give 8. However, a plausible mechanism for the formation of the minor component 9 results from the C5 hydroxide group of 7 being attacked at the C4 position to give intermediate 12a ( Figure 3). Instead of attack at the least hindered side at the C5 position of 12a by azide, known as the Payne rearrangement [21], the hydroxide group at C3 of 12a internally removed the proton of HN 3 (intermediate 12b). That allowed the resulting azide to attack the vicinal C4 position of 12b to give 9. This resulted in the retention of configuration of epoxide 7. This observation was very unusual and in contrast to the results that occurred in the 2,3-epoxy rearrangement [22]. Based on the Chem3D simulation, the cyclohexane core of 7 was in a boat-like conformation ( Figure 4). The trans-diaxial attack at C4 in 7 by azide leading to 8 as the major compound was energetically favorable. However, we could not rule out the possibility in formation of 12a which was derived from the trans-axial attack of the epoxide by C5-OH in 7. The lower yield of 9 was probably due to the half-chair like structure 12a that was less favorable than 7 for allowing by azide attack (Figure 4).  The Payne rearrangement of epoxide 7 intrigued us as an interesting issue when no rearrangement product 13 was found when compound 1 was treated with NaN 3 ( Figure 5). According to the Chem3D simulation, the conformation of cyclohexane core of 1 is a slightly twisted boat form. However, compound 13 was in a boat conformation if the Payne rearrangement occurred. The reason was probably due to the steric congestion in the formation of 13 because the distance between epoxide and the C2 acetal oxygen atom of 13 is around 3.054 Å. On the contrary, the distance between the C5-OH and C2 oxygen atom of 1 is about 3.328 Å. Therefore, the trans-diaxial attack at C4 of 1 by azide might be kinetically or sterically controlled to lead to the major component 3. In order to obtain better yields of final products 5, 6, 10, and 11, we determined that azido compounds 3, 4, 8, and 9 should be hydrogenated first over Pd/C, followed by deprotection under acidic conditions. The one pot reaction conditions (H 2 /Pd/C/HCl) afforded low yields of target compounds accompanied by a more complicated mixture. It is worth noting that our strategy was much shorter than the reported method in the syntheses of molecules 5 and 6 which involved sixteen steps starting from D-mannitol [23]. The structure determinations were based on a series of NMR experiments (COSY, 2D-NOESY, HMBC, HMQC, and HRMS) and the selected NMR data were listed on Tables 1 and 2.

General Procedure of Ring Opening
Compound 1 (0.838 g, 4.0 mmol), for example, was dissolved in DMF (30 mL). To this mixture was added NaN 3 (2.3 g, 36.0 mmol) and a catalytic amount of 15-crown-5 and heated under reflux for 56 h. At the end of the reaction time, the mixture was diluted with H 2 O (100 mL) and extracted with Et 2 O (2x100 mL). The organic layer was dried (MgSO 4 ) and purified by column chromatography.

General Procedures of Hydrogenation and Deprotection
Compound 3 (0.079 g, 0.29 mmol) for example, was dissolved in MeOH (2 mL). To this mixture was added 10% Pd/C (10 mol%) and it was hydrogenated under one atmosphere at ambient temperature for 2 h. The resulting mixture was filtrated through a pad of Celite and washed with MeOH. The organic layer was concentrated and 80% TFA was added (2 mL), then stirred for 11.5 h, at the end of which time, the solvent was evaporated and the residue purified by column chromatography.

Conclusions
In conclusion, aminocyclitols are a very important class of aminocarbasugars. We have synthesized two known and two new aminocyclitols in an efficient manner from D-()-quinic acid. Especially, our method provided a short alternative in syntheses of 5 and 6 than the literature. The ring opening of epoxide in 1, 2 and 7 by sodium azide to provide moderate to good yields of 3, 4, and 8, respectively, was highly regioselective owing to the steric effect. The studies of the biological activities of these compounds are currently ongoing and will be reported in due course.