Single enantiomer synthesis of α-(trifluoromethyl)-β-lactam

The first synthesis of α-(trifluoromethyl)-β-lactam ((S)-1) is reported. The route starts from α-(trifluoromethyl)acrylic acid (2). Conjugate addition of α-(p-methoxyphenyl)ethylamine ((S)-3b), generated an addition adduct 4b which was cyclised to β-lactam 5b. Separation of the diastereoisomers by chromatography gave ((αS,3S)-5b). N-Debenzylation afforded the desired α-(trifluoromethyl)-β-lactam ((S)-1). The absolute stereochemistry of diastereoisomers 5 was determined by X-ray crystallographic determination of a close structural analogue, (αS,3S)-5c, and then 1H and 19F NMR correlation to the individual diastereoisomers of 5a and 5b.

Introduction β-Lactams (azetidin-2-ones) have played a prominent role in medicinal chemistry and many structural variants have been prepared and elaborated [1]. Similarly, the CF 3 group is an ubiquitous substituent in pharmaceutical research, where it is used to modify the activity of a drug candidate or to block adventitious metabolism and improve the pharmacokinetic profiles [2,3]. Likewise, the substituent is found widely distributed in agrochemical products [4]. The majority of CF 3 containing compounds reported in the literature are CF 3 -aryl, CF 3 -ether [5,6] or CF 3 -heteroaromatic in nature, and these substituents have contributed significantly to the fine chemicals and related industries. However, there is an increasing awareness and demand for molecular building blocks which carry the CF 3 group at a stereogenic centre [7,8] and some such motifs are emerging in new chemical entities licenced for the pharmaceuticals market, but also in organic materials area, e.g., liquid crystals [9]. Methodologies continue to emerge for the asymmetric introduction of the CF 3 group [10][11][12][13]. In this contribution, we report a synthesis of α-(trifluoromethyl)-β-lactam (1) which allows access to its enantiomerically pure forms, (R)-1 and (S)-1 as illustrated in Figure 1.

Results and Discussion
The synthesis shown in Scheme 1 follows a strategy developed [14] for enantiomers of the methyl analogue of 1, and starts from α-(trifluoromethyl)acrylic acid (2) which offered a Scheme 1: Synthetic route involving a diastereoisomeric separation to α-(trifluoromethyl)-β-lactam ((S)-1) from α-(trifluoromethyl)acrylic acid 2 and (S)-α-(p-methoxyphenyl)ethylamine (3b). Yields are quoted for the conversion of 4b. commercially available source of the CF 3 group. It was envisaged that conjugate addition of an enantiomerically pure amide such as (R)-or (S)-3 would generate the addition adducts, carboxylate salts 4, as a mixture of diastereoisomers. Cyclisation to the N-substituted β-lactams 5 would give a mixture of two stereoisomers which might be separated into their individual diastereoisomers 5 by chromatography. N-Benzyl deprotection of the individual diastereoisomers would then provide β-lactam 1 as a single enantiomer. At the outset, (S)-(αphenyl)ethylamine (3a) was explored as the enantiomerically pure amine. In the event 4a was generated after aza-Michael addition as a mixture of two stereoisomers, without any obvious diastereoisomeric bias (1:1, 0% de) as judged by both 1 H and 19 F NMR. β-Lactam ring closure, using thionyl chloride and triethylamine gave the N-methylbenzyl-β-lactam 5a with a modest diastereoisomeric bias (~50% de) indicating some epimerisation of the α-trifluoromethyl stereogenic centre, to a thermodynamic product ratio. This was supported in an analytical reaction by the observation of deuterium exchange at this stereogenic centre after the addition of D 2 O to the reaction on work up. The major and minor diastereoisomers of 5a were separated by careful chromatography on silica gel into single stereoisomers. Completion of the syntheses required hydrogenolysis of the (S)-N-methylbenzyl moiety of 5a. A range of conditions and catalysts were explored for the hydrogenolysis of 5a, however cleavage of the C-N bond proved very difficult and a satisfactory method could not be found. Therefore, (S)-α-(p-methoxyphenyl)ethylamine (3b) was explored as an alternative amine for the aza-Michael reaction, as removal of this amine using ceric ammonium nitrate (CAN) oxidation offered a milder deprotection method [15]. The aza-Michael reaction proved straightforward to generate 4b and then cyclisation again using thionyl chloride and triethylamine gave β-lactams 5b in a 40% de, presumably again a thermodynamically biased isomer ratio. The diastereoisomers of 5b could be separated by careful chromatography over silica gel and this led to the recovery of each isomer as a major and a minor product. Finally, oxidative scission of the major β-lactam stereoisomer (αS,3R)-5b, using CAN was relatively straightforward generating β-lactam (S)-1 in 67% yield. This novel β-lactam is a crystalline solid, and a suitable crystal was subjected to X-ray structure analysis. The structure of 1 is shown in Figure 2a.
Accordingly, (S)-α-(p-methoxyphenyl)ethylamine (3b) emerged as the more satisfactory amine over 3a for the preparation of 1, due to the straightforward benzylic cleavage. Enantiomeric purity analysis of the resultant β-lactam 1 was evaluated by 19 F NMR using a europium chiral shift reagent [Eu(hfc) 3 ]. The comparison of a racemic sample of 1 and then a sample after diastereoisomer separation of 5b as described above, indicated that β-lactam 1 was prepared in an enantiomerically pure form. There was no evidence that benzylic cleavage of the (S)-α-(pmethoxyphenyl)ethyl moiety with CAN resulted in epimerisation at the stereogenic centre of the β-lactam.
Finally, it was necessary to determine the absolute configuration of the resultant β-lactam. To this end, X-ray crystallography of the major diastereoisomers of 5a and 5b was attempted after chromatographic separation. Despite considerable effort however we could not obtain crystals of single isomers of 5a or 5b suitable for X-ray structure analysis. Thus, a preparation of 5c was carried out as illustrated in Scheme 2. The presence of the naphthyl ring in amine (R)-3c rendered the resultant β-lactam diastereoisomers 5c more crystalline, and a single crystal X-ray structure was solved for the major-diastereomer revealing the (αR,3R)-5c configuration. The structure is shown in Figure 2b.
With this structural information in hand it was necessary to correlate to the stereoisomers of 5b. The 1 H and 19 F NMR spectra of the major and minor diastereoisomers of 5a-c were now compared. Clear chemical shift trends are observed as illustrated and tabulated in Table 1 and Table 2. For example, in all cases, the 1 H NMR chemical shifts of the C-3 hydrogens (H a ) of the β-lactam are shifted downfield in the major relative to the minor diastereoisomers. The non-equivalent faces of the planar β-lactam ring differentiates the two diastereotopic hydrogens at C-4 (H b and H c ).
It is clear that for all three major isomers the 1 H NMR signals for H b and H c have a larger chemical shift difference than that between the H b and H c signals of the minor isomers. Also the chemical shifts for H b and H c of the minor isomers lie within those of the major isomers. For the 19 F NMR spectra shown in Table 2 the major isomers all have their trifluoromethyl signals downfield of the minor diastereoisomers.

Experimental General
All reagents were obtained from commercial sources and were used without further purification unless otherwise stated. Airand moisture-sensitive reactions were carried out under a positive pressure of argon in flame-dried glassware using standard Schlenk-line techniques. Dry CH 2 Cl 2 was obtained from the Solvent Purification System MB SPS-800. Room temperature (RT) refers to 20-25 °C. Reaction temperatures of 0 °C were obtained in an ice/water bath. Reaction reflux conditions were obtained using an oil bath equipped with a contact thermometer. Solvent evaporations were carried out under reduced pressure on a Büchi rotary evaporator. Thin layer chromatography (TLC) was performed using Macherey-Nagel Polygram Sil G/UV254 plastic plates. Visualisation was achieved by inspection under UV light (255 nm). Column chromatography was performed using silica gel 60 (40-63 micron). NMR spectra were recorded on Bruker AVANCE 300, 400 or 500 MHz instruments. 1 H and 13 C NMR spectra were recorded in CDCl 3 as solvent. 19 F NMR spectra were referenced to CFCl 3 as the external standard. Chemical shifts are reported in parts per million (ppm) and coupling constants (J) are given in Hertz (Hz). IR spectra were recorded on a Nicolet Avatar 360 FT-IR from a thin film (either neat or combined with nujol) supported between NaCl plates. Optical rotations [α] D are given in 10 -1 deg cm 2 g -1 and were measured using a Perkin Elmer Model 341 polarimeter. Mass spectrometric (m/z) data was acquired by electrospray ionisation (ESI). High resolution mass analyses were recorded on a Micromass LCT TOF mass spectrometer using ES ionisation in positive ion mode.
General aza-Michael procedure NaHCO 3 (840 mg, 10.0 mmol) was added to a solution of α-(trifluoromethyl)acrylic acid (2, 1.41 g, 10.0 mmol) in methanol (10 mL), and then after 30 min stirring, a single enantiomer amine 3 (10.0 mmol) was added. The reaction was stirred at RT for 24 h. The solvent was evaporated under reduced pressure to afford the aza-Michael product as a colourless oil (quant) which solidified upon standing. Optionally, the sodium salt was converted to the amino acid HCl salt by treatment with aq HCl, followed by evaporation of the water and redissolving the residual in DCM.