Asymmetric Synthesis of 2-Substituted Oxetan-3-ones via Metalated SAMP/RAMP Hydrazones

2-Substituted oxetan-3-ones can be prepared in good yields and enantioselectivities (up to 84% ee) by the metalation of the SAMP/RAMP hydrazones of oxetan-3-one, followed by reaction with a range of electrophiles that include alkyl, allyl, and benzyl halides. Additionally, both chiral 2,2- and 2,4-disubstituted oxetan-3-ones can be made in high ee (86–90%) by repetition of this lithiation/alkylation sequence under appropriately controlled conditions. Hydrolysis of the resultant hydrazones with aqueous oxalic acid provides the 2-substituted oxetan-3-ones without detectable racemization.

T here is considerable current interest in the preparation of oxetanes for use in medicinal chemistry. 1,2 As a result of the pioneering work of Carreira and Rogers-Evans, 1 these 4membered oxygen heterocycles are increasingly being used as bioisosteric replacements for common functional groups such as gem-dimethyl or carbonyl groups. 2 Their introduction can induce profoundly beneficial effects on the aqueous solubility, lipophilicity, metabolic stability, and conformational preference of drug candidates. To date, most work has centered on the use of oxetanes devoid of substituents at C-2 and/or C-4 to avoid the introduction of additional stereocenters into the molecular scaffold. 3 In part, this is due to the limited number of methods for the synthesis of chiral, nonracemic oxetane derivatives. 4,5 In seeking to expand the number of readily accessible, chiral oxetane building blocks, we decided to explore the enantioselective synthesis of 2-substituted oxetan-3-ones using the SAMP/RAMP hydrazone methodology developed by Enders. 6 No general asymmetric route to this oxetane subclass has been established. However, Zhang has reported a single example of a Au-catalyzed oxidative cyclization of a chiral propargylic alcohol to the enantiomerically enriched 2substituted oxetan-3-one without racemization, 5c and Williams has produced an enantiomerically enriched 2,2,4-trisubstituted oxetan-3-one by DMDO epoxidation/rearrangement of a chiral allene. 4a Our proposed strategy is depicted in Scheme 1. At the outset of this study, it was unclear whether the high degree of ring strain within the metalated hydrazone intermediate might inhibit its formation. Indeed, as far as we are aware, there are no reports of enolate generation from oxetan-3-ones. 7 We were encouraged, however, by a report by Fadel and co-workers who have demonstrated that the SAMP/RAMP hydrazones of cyclobutanone can be successfully lithiated and alkylated. 8 In this paper, we demonstrate how a variety of chiral 2-substituted as well as 2,2-and 2,4-disubstituted oxetan-3-ones can be prepared by metalation/alkylation of the SAMP/RAMP hydrazones derived from oxetan-3-one, offering a practical route to these important medicinal chemistry building blocks.
The SAMP hydrazone (S)-1 was prepared in quantitative yield by treatment of SAMP with an excess of commercially available oxetan-3-one at 55°C without solvent ( Table 1). The corresponding (R)-enantiomer was made using RAMP in an identical fashion. In order to investigate the metalation of (S)-1, a screening of lithium bases was performed. Hydrazone (S)-1 was in turn treated with 1.1 equiv of LDA, n BuLi, and t BuLi and then quenched with deuterated methanol. The extent of deuterium incorporation into 2 was estimated by mass spectrometry. Use of LDA was found to give only 59% deuterium incorporation (entry 1, Table 1), whereas n BuLi and t BuLi proved to be more effective in forming the lithiated derivative with 90% deuterium incorporation in each case (entries 2 and 3). Having identified n BuLi and t BuLi as the most suitable bases for the metalation step, alkylation with a representative carbon-based electrophile, namely benzyl bromide, was explored. After deprotonation with n BuLi at −78°C, and subsequent trapping with this electrophile at the same temperature, the benzylated hydrazone 3 was obtained in an encouraging 45% yield (entry 4). Addition of the additive TMEDA did not lead to an improvement in yield (entry 5), and a change from THF to diethyl ether as the solvent resulted in no product formation (entry 6). Use of the stronger base t BuLi led to a higher yield (entry 7), with a modest additional improvement seen using a longer metalation time (entry 8).
Under these conditions, 3 was produced in 73% yield and 76% de (entry 8). 9 The alkylation conditions used in entry 8 were used in all subsequent studies.
Conversion of hydrazone 3 to enantiomerically enriched ketone 4 could be achieved by oxidation with ozone 10 or by hydrolysis with oxalic acid, 11 although the latter method was found to be both higher yielding and more convenient (Scheme 2). Initial attempts to determine the enantiopurity of the resulting ketone using chiral shift reagents, chiral HPLC, and chiral GC all proved unsuccessful. However, reduction of ketone 4 to the corresponding alcohol and further acetylation enabled determination of its enantiopurity by chiral GC analysis, and an ee of 74% was established (see the Supporting Information). 12 The racemic ketone 4 was prepared for comparative purposes from achiral hydrazone 5 using the same chemistry (Scheme 2). 13 The absolute configuration of the major enantiomer derived from (S)-1 was established by performing a Pictet−Spengler reaction on ketone 4 with L-tryptophan ethyl ester, using reaction conditions developed within our group (Scheme 3). 14 Two diastereoisomers, 6 and 7, were isolated from the reaction mixture in 67% and 9% yields respectively. The structures of both 6 and 7, and the (S)-configuration of the oxetane C2 stereocenter of the major product 6 were unambiguously determined by X-ray crystallography (see the Supporting Information). Importantly, the product ratio (6:7; 88:12) is in close agreement with the enantiomeric ratio (er = 87:13) of 4 determined by GC analysis, supporting the supposition that no epimerization occurs during the Pictet−Spengler cyclization.
The stereochemical outcome of the alkylation of SAMP hydrazone (S)-1 is in accordance with previous studies by Enders et al. and can be explained by preferential attack of a conformationally rigid and chelated E CC Z C−N azaenolate by the electrophile from the less sterically hindered Si face (Scheme 4). 8,15 The sense of asymmetric induction in the other alkylations reported herein was made by analogy.
Having established satisfactory yields for both the alkylation of SAMP hydrazone (S)-1 and the hydrolysis of 3 to 2benzyloxetan-3-one (4), we sought to establish the scope and stereoselectivity of the alkylation step. A representative range of a Determined by mass spectrometry. b Isolated yield after chromatography.
electrophiles including alkyl iodides, allyl bromides, and an aldehyde were screened (Table 2). Satisfyingly, in addition to benzyl bromide, both alkyl iodides and allyl bromides were found to react in both good yield and stereoselectivity (up to 84% ee) (entries 2−5). Interestingly, treatment of 10 with oxalic acid was found to lead to both hydrazone hydrolysis and TBS removal to give bicyclic hemiketal 14 (entry 4). Although benzaldehyde reacted in good yield with the lithiated SAMP hydrazone to give 11, the β-hydroxy ketones 15a and 15b formed on hydrolysis were found to have significantly different levels of enantiopurity (15a: 54% ee; 15b, 2% ee), and the dr for 15a:15b was found to be essentially 1:1 by 1 H NMR (entry 5). The relative stereochemistry within both 15a and 15b was established by X-ray crystallography (see the Supporting Information). The poorer facial selectivity seen in this reaction may be attributed to a breakdown in the coordination of the methoxy group of the chiral auxiliary to the lithium azaenolate due to competing coordination by the aldehyde oxygen. 16 Good selectivities for the aldol reaction of lithiated cyclic SAMP hydrazones have only been reported for much bulkier aldehydes which are less likely to affect lithium coordination within the auxiliary. 17 Having achieved good yields and selectivities for the monoalkylation of (S)-1 in the majority of cases, we next explored the feasibility of making disubstituted oxetan-3-one derivatives using this methodology. By treating (S)-1 sequentially with t BuLi, benzyl bromide, t BuLi, and allyl bromide in one-pot, 2,2-dialkylated hydrazone 16 was isolated in 33% yield (Scheme 5). Hydrazone cleavage provided ketone 17 in excellent enantiopurity (90% ee). As before, the ee was determined by GC analysis of acetate derivatives (see the Supporting Information). In this case, however, we were unable to prepare racemic ketone 17 from dimethyl hydrazone 5 so instead prepared the opposite enantiomer of 17 from RAMPderived hydrazone (R)-1 for comparison purposes. While the generation of a quaternary stereocenter at the α-position of a SAMP/RAMP hydrazone is known, 18 this is, to the best of our knowledge, the first example of the generation of a quaternary center from an α-CH 2 unit where there is no prior monoalkylation observed at the α′-CH 2 unit. The high activation energy for conversion of the (Z)-2-benzyl hydrazone 3 (formed from the first alkylation) to the corresponding (E)hydrazone explains the lack of α′-alkylation in this case. 19 One explanation for the higher enantioselectivity seen in the formation of 17 compared with 4 and 12−14 is that hydrazone 16 is unable to undergo C2-epimerization under the reaction conditions.
General Procedure for the Preparation of Acetates from Ketones 4, 12, 13, 17, and 19 for Chiral GC/HPLC Analysis. Sodium borohydride (1.5 equiv) was added to a stirred solution of the ketone (0.05 mmol) in anhydrous methanol (1 mL) at 0°C. After 1 h, the reaction mixture was partitioned between dichloromethane (20 mL) and brine (5 mL). The layers were separated, and the organic layer was dried (MgSO 4 ), filtered, and concentrated under reduced pressure to give pure diastereoisomeric alcohols, as confirmed by 1 H NMR (see the Supporting Information).
4-(Dimethylamino)pyridine (1 crystal) and acetic anhydride (3 equiv) were added to a stirred solution of the alcohol in anhydrous dichloromethane (1 mL). After 3 h at room temperature, the reaction mixture was filtered through a small plug of silica which was washed well with ethyl acetate. The filtrate was concentrated in vacuo to give pure acetate derivatives as confirmed by 1 H NMR (see the Supporting Information).
Procedure for the Preparation of Acetates from Alcohol 14 for GC Analysis. 4-(Dimethylamino)pyridine (1 crystal) and acetic anhydride (22 μL, 0.264 mmol, 3 equiv) were added to a stirred solution of 2,7-dioxabicyclo[4.2.0]octan-1-ol (14) in anhydrous dichloromethane (1 mL). After 3 h at room temperature, the reaction mixture was filtered through a small plug of silica which was washed well with ethyl acetate. The filtrate was concentrated in vacuo to give pure monocyclic acetate derivatives (13 mg, 89%) as confirmed by 1 H NMR (see the Supporting Information).

* S Supporting Information
Copies of 1 H and 13 C NMR spectra of compounds 1 and 3−19, X-ray crystal structures and data of compounds 6, 7, 15a, and 15b (CIF), copies of 1 H NMR spectra of alcohols and acetates prepared from 4, 12−14, 17, and 19 for ee determination, copies of chiral GC and HPLC chromatograms of acetates and alcohols prepared from 4, 12−14, 17, and 19 and alcohols 15a and 15b, and copies of 1 H NMR spectra of 3 before and after thermal isomerization. This material is available free of charge via the Internet at http://pubs.acs.org.