Application of chiral 2-isoxazoline for the synthesis of syn-1,3-diol analogs

The asymmetric cycloaddition of TIPS nitronate catalyzed by “Cu(II)-bisoxazoline” gave the 2-isoxazoline product in 95% yield, which was converted into tert-butyl (3S,5R)-6-hydroxy-3,5-O-isopropylidene-3,5-dihydroxyhexanoate in 14 steps through a β-hydroxy ketone.


Results and Discussion
Our synthesis commenced with a chiral 3,5-disubstituted-2isoxazoline 3 or 4, which were prepared from silyl nitronate through an asymmetric 1,3-dipolar cycloaddition developed in our lab (Table 1) [49]. The synthesis of the triisopropylsilyl nitronate was initially attempted starting with 3-nitropropionic acid methyl ester but no desired product was observed. How-Scheme 1: Accesses to tert-butyl 3,5-O-isopropylidene-3,5-dihydroxyhexanoates. (a) Previous methods using Claisen condensation. (b) Our new method using cycloaddition. CH 2 Cl 2 catalyzed the cycloaddition between N-acryloyl-1,3oxazolidin-2-one and the silyl nitronate at −50 °C to give 1 in 95% isolated yield, which subsequently generated 3,5-disubstituted isoxazoline 4 in 80% ee. Decreasing the amount of the chiral Lewis acid catalyst led to a decrease of both the ee and the yield. Desilylation of the 2-isoxazolidine 1 was effected in CHCl 3 using catalytic amounts of p-toluenesulfonic acid (PTSA). Though the yield of the in situ-generated 2-isoxazoline 2 bearing the 1,3-oxazolidin-2-one auxiliary was perfect, purification of 2 by silica gel chromatography was problematic due to decomposition. No pure product was isolated from crude 2 by chromatography on silica gel. Decomposition occurred to a compound similar to 2, in which the 3-substituent was CH 2 OH [49]. To overcome this problem, the crude reaction mixture containing 2 and PTSA was concentrated before excess Et 3 N was added followed by CH 3 OH as the solvent. These operations removed the 1,3-oxazolidin-2-one auxiliary while preserving the THP group, and afforded the corresponding methyl ester 3 (Table 1), which was stable and could be subjected to silica gel chromatography. Compound 4 was used to determine the stereoselectivity of the cycloaddition step as well as for oxidation.
Oxidation of the 2-isoxazoline 4 with Jones' reagent gave a complicated mixture, in which the desired carboxylic acid was not observed (Scheme 2). The stepwise oxidation of the free hydroxy to the carboxy group via intermediary aldehyde was then examined. Swern or pyridinium chlorochromate (PCC) oxidation of 4 also gave a complicated mixture without the desired aldehyde detected. These failed reactions indicated that the 2-isoxazoline moiety could not survive oxidation conditions. Based on this assumption, the corresponding silyl nitronate from 3-nitropropanal or its acetal were not tried for cycloaddition.
We then set to liberate the β-hydroxy ketone synthon by ring opening of the isoxazoline 3 (Scheme 3). Raney-Ni-catalyzed hydrogenolysis in the presence of boronic acid had been widely utilized to disconnect the N-O bond as well as to hydrolyze the resulting imine into a ketone [52]. We applied this method to deprotect the isoxazoline 3. However, the desired β-hydroxy ketone was never obtained. In one instance, the methyl ketone from a retro-aldol reaction of the desired β-hydroxy ketone was observed. In our experience, the hydrogenolysis of a 2-isoxazoline having a 5-ester group was troublesome. Thus, the 5-ester group was reduced with NaBH 4 to give 5. The hydroxy group was subsequently protected with benzoyl (Scheme 3), which also worked as a chromophore facilitating HPLC analysis. Afterwards, we tried oxidations once again. After removal of THP from 6, the resulting compound 6' was subjected to oxidation with various reagents (Scheme 4) [53][54][55]. The expected carboxylic acid or aldehyde was not observed, which further verified the intolerance exemplified in Scheme 2. These results prompted us to try the oxidation in a later stage.
The racemic sample of 15 was prepared from racemic diethyl malate following known methods [26,27]. Finally, K 2 CO 3 -catalyzed methanolysis gave 16 in 87% yield [26,27]. The absolute stereochemistry of 16 was confirmed by crystal structure analysis [72] and the specific rotation [28] of 17. Centimeter-long prismatic single crystals of 17 were obtained by slow evaporation of a petroleum solution.
Starting from 9, we tested several reactions in order to selectively protect the internal hydroxy groups (Scheme 5). Though not fruitful, these results deserve some comments. The PTSAcatalyzed acetonization of 9 using 2.0 equiv DMP gave the acetonide 18 in a quantitative yield. Treating 18 with a catalytic amount of PTSA in methanol gave 10, with the protecting groups removed except benzoyl. PTSA-catalyzed acetonization of 10 using 2.0 equiv DMP gave a mixture of two acetonides 19 and 13, which are separable by silica gel chromatography (Scheme 5a). In another trial (Scheme 5b), acylation of the two hydroxy groups in 9 yielded 20 in a quantitative yield. PTSAcatalyzed removal of THP in 20 in methanol did occur. However, concomitant monodeacylation as well as further an acyltransfer reaction also took place, resulting in a mixture. These results indicated THP, isopropylidene or Ac protection to primary or secondary hydroxy groups did not well tolerate PTSAcatalyzed methanolysis.