A chemoenzymatic synthesis of ceramide trafficking inhibitor HPA-12

A chemoenzymatic synthesis of the title compound has been developed using an efficient and highly enantioselective lipase-catalyzed acylation in a hydrophobic ionic liquid, [bmim][PF6], followed by a diastereoselective asymmetric dihydroxylation as the key steps for incorporating the stereogenic centers. The further conversion to the appropriate intermediates and subsequent acylation with lauric acid furnished the target compound.

To realize our objective, we paid particular attention to obtain 2 using reactions that are high-yielding and can be executed under simple reaction conditions with commercially available and inexpensive materials/reagents. In this regard biocatalytic reactions offer green and sustainable alternative routes to develop asymmetric syntheses of pharmaceuticals with varied stereochemical features [30][31][32][33]. Our group has been using lipases for the chemoenzymatic syntheses of several bioactive molecules [34][35][36][37][38][39]. The prevalence of PhCH(OH) in 2, and in many other biochemicals, attracted our attention to formulate an enantioselective lipase-catalyzed transacetylation strategy to obtain a suitable molecule bearing the designated chiral segment (compound 2, vide infra).

Results and Discussion
For the synthesis, commercially available benzaldehyde (3) was allylated using Zn/allyl bromide in moist THF following Luche's protocol to obtain the homoallylic alcohol (±)-4 (Scheme 1) [40]. We envisaged that the enantiomers of 4 were ideal substrates for the asymmetric synthesis of different stereoisomers of 2. Several biocatalytic protocols for the preparation of (R) or (S)-4 were reported earlier. The hydrolysis of the corresponding acetate with crude enzyme preparations from pig liver and chicken liver esterase proceeded with modest enantioselectivity [41,42]. Rhizopus arrhizus-mediated hydrolysis of the acetate furnished the enantiomerically pure alcohol (99% ee), however, the enantiomeric excess (ee) of the antipode acetate was very poor (5-9%) [43]. On the other hand, the Amano PS lipase-catalyzed trans-acylation of (±)-4 furnished (S)-4 with good ees, while the ee of the (R)-acetate was poor [44]. The acylase 1-catalyzed resolution protocol gave (S)-4 in a poor % ee [45]. A unified approach where both the enantiomers of 4 are obtained in good optical purities is rare. This is important as the individual enantiomers can be converted to the antipodes by Mitsonobu inversion [46], thereby maximizing the yield of the desired enantiomer. In addition, the availability of both the enantiomers of 4 would be useful for the synthesis of all diastereomers of 2 (as per our synthetic plan) and also several other pharmacologically important compounds. To the best of our knowledge, only the lipase PS-catalyzed acylation of (±)-4 proceeded with good to excellent ees for both the acetate and alcohol. However, details of the protocol are unavailable, and also the resolution was quite slow [47]. Hence, we screened different lipases for the trans-acetylation of (±)-4 for its effective resolution and the results are summarized in Table 1. Based on our past experience in lipase-catalyzed resolution of homoallylic alcohols [38], initially a Novozym 435 ® -catalyzed acetylation of (±)-4 with vinyl acetate in diisopropyl ether (DIP) was attempted. However, the yield and enantioselectivity of the desired alcohol (S)-4 were very poor ( . The products were isolated by filtering the insoluble enzyme from the reaction mixture, followed by concentration of the filtrate. The obtained residue was subjected to silica gel column chromatography to isolate unreacted alcohol and acetylated product. We have reused the recovered lipase at least three times without any significant loss of enzyme activity.
There are similar instances in literature [48] for a substantial slow-down of the transesterification reaction rate in organic solvents. This issue can be overcome by using room temperature ionic liquids, which not only substitute the environment damaging organic solvent, but also increase the reaction rate, and provide many other technological advantages [49].  6 ] medium with diethyl ether and the ethereal phase was concentrated. The residue was then subjected to silica gel column chromatography to obtain the unreacted alcohol and the acetate. We have reused the recovered lipase in ionic liquid at least three times without any significant loss of enzyme activity.
The % ee of the (R)-5 and (S)-4 were determined from chiral HPLC analyses (150 mm × 4.6 mm, 5 μm, chiral AD-H column, 5% isopropanol/hexane @ 1.0 mL min −1 , UV detection at 254 nm). The absolute configurations of (R)-5 and (S)-4 were assigned by comparison of the chiroptical data with those reported [45]. The stereochemical outcome of the reaction is consistent with Kazlauskas' empirical rule [51]. The conversion (% c) and the enantiomeric excess (E) values were calculated according to the method described by Sih et al. [52]. Besides the high yield and % ee, the biocatalytic protocol is operationally simple and convenient. We carried out the transformation using inexpensive vinyl acetate as the acyl donor due to its volatility that would assist easy isolation of the products.
All the reactions were carried out at least 3-4 times and the best results are presented in Table 1. To the best of our knowledge, this is the first attempt towards the resolution of (±)-4 in a room temperature ionic liquid at an elevated temperature.
To make the synthesis enantio-convergent, and also to offset the limitations of a resolution-based protocol, (R)-5 was hydrolyzed with alcoholic KOH to furnish (R)-4 (Scheme 1). Its inversion under the Mitsunobu conditions (Ph 3 P/DIAD/pnitrobenzoic acid/THF; KOH/EtOH/25 °C/8 h, 91% over two steps) gave (S)-4 [46]. The benzylation of the hydroxy function in (S)-4 with benzyl bromide (BnBr) and Bu 4 NI in the presence of NaH produced compound 6 (Scheme 2). This was subjected to asymmetric dihydroxylation (ADH) using AD mix-β [K 2 OsO 2 (OH) 4 and (DHQD) 2 -PHAL]. The reaction proceeded predominantly from the α-face, resulting in the formation of the 1,3-anti diol 7a and 1,3-syn diol 7b in a 91:9 ratio (based on the isolated yields of 7a and 7b, separated by column chromatography). Previously the ADH reaction of a homologue of 6, bearing a methyl substitution and a hydroxy group (instead of benzyloxy group) with AD mix-β also produced the corresponding α-alcohol [53]. However, unlike in our case, the reaction proceeded with poor diastereoselectivity irrespective of the dihydroxylating agent used. To confirm the 1,3-anti diol stereochemistry of 7a, it was debenzylated using DDQ/CH 2 Cl 2 -H 2 O to furnish the trihydroxy compound 7a'. The 1 H and 13 C NMR spectra, and the optical rotation of 7a' were in conformity with those reported [54]. In addition, the 1,3-anti stereochemistry of the diol 7a was confirmed by converting it to the target compound 2, and comparing its chiroptical data with the reported values, as described afterwards.
For the synthesis of 2, the primary hydroxy function of 7a was benzoylated to get 8. Compound 8 was mesylated with methanesulfonyl chloride (MsCl)/Et 3 N and the product reacted with NaN 3 /DMF at 90 °C to obtain the azide 9 (Scheme 2). We first attempted to convert 9 to the target compound 2 by i) converting the azide group to the amine using LiAlH 4 with concomitant debenzoylation, followed by the acylation of the amine with lauric acid to afford 9a, and finally, ii) reductive cleavage by hydrogenolysis using Pd-C/H 2 leading to debenzylation (Scheme 3). However, during hydrogenolysis, the elimination of the -OBn group led to product 9b, which was undesirable. A similar elimination was earlier observed by Sharf et al. during hydrogenolysis of dibenzyl ether [55]. To avoid this, an oxidative debenzylation of 9a using DDQ/CH 2 Cl 2 -H 2 O was carried out. However, this led to a very poor yield of the target compound 2 (Scheme 3). Hence, we decided to debenzylate compound 9 prior to its reduction to the amine and subsequent acylation. Towards this (Scheme 4), oxidative debenzylation of 9 using DDQ yielded 10 in 84% yield. Treatment of 10 with LiAlH 4 led to the reduction of the azide group to amine along with debenzoylation to furnish an intermediate aminodiol, which, without further purification, was acylated with lauric acid in the presence of dicyclohexylcarbodiimde (DCC) and 4-dimethylaminopyridine (DMAP) to give the target compound HPA-12 (1R,3S)-2. The optical and spectroscopic data of compound 2 were in accordance with those reported [18].

Conclusion
In summary, we have demonstrated an efficient protocol for the synthesis of HPA-12 using a lipase-catalyzed resolution of the alcohol (±)-4 in an ionic liquid and a diastereoselective ADH reaction as the key steps. The synthesis was accomplished by employing reactions that use inexpensive reagents, are operationally simple and proceed with good to excellent yields and excellent stereoselectivities. The target compound was obtained in 23% overall yield starting from (S)-4. The protocol can also be used to access different HPA analogues and derivatives.