Highly enantioselective metallation–substitution alpha to a chiral nitrile

Nitrile anions do not necessarily lack stereochemical integrity and we show good results for stereospecific reaction with a simple magnesium base.


Introduction
Metallated nitriles are well-used intermediates in synthetic chemistry due to their excellent reactivity as nucleophiles. 1 The formation of the organometallic species and its reaction with an electrophile such as an alkyl halide or aldehyde allows a highyielding preparation of the desired substituted nitrile that can then be converted readily to other functional groups. The most common method to prepare the metallated nitrile is to treat the nitrile with a base such as lithium diisopropylamide (LDA) and this is known to give the lithiated nitrile in which the lithium ion normally resides on the nitrogen atom. [2][3][4][5] Although this gives a reactive nitrile anion, one of its drawbacks is that this provides an achiral organometallic species (e.g. Fig. 1), 3,4 even starting from a chiral, enantiomerically enriched nitrile. Therefore it would be expected that achiral products would result from using chiral enantiomerically enriched nitrile starting materials and this is typically the case. [6][7][8] Remarkably, however, Takeda and co-workers reported recently that it is possible in certain cases at low temperature with in situ reactive electrophiles to trap the intermediate anions to give enantioenriched products. 9,10 At about the same time, we began to explore this possibility but by using magnesiated nitriles. 11 The idea that magnesiated nitriles may allow asymmetric reaction through a chiral organometallic species 1 rather than 2 (Fig. 2) was based on results from several groups including that of Carlier and co-workers, who reported the rst metallated nitrile with macroscopic congurational stability, albeit a cyclopropyl derivative 3 (prepared by Br-Mg exchange). 12 In addition, Fleming and co-workers had found opposing selectivities for reactions of lithiated and magnesiated nitriles and surmised that the magnesium cation has a preference for location on carbon. 13,14 If the metal atom is located on the carbon atom, as illustrated by the contact ion pair 1 (or its related solvent-separated ion pair in which the metal cation is nearby), then the metallated nitrile is chiral and has the possibility to transfer its chirality to the product on reaction with an electrophile. However, very little is known about the rate of enantiomerisation of such nitrile anions. 11,15 The importance of nitrogen-containing heterocycles in natural products and medicinal compounds led us to explore the metallation of nitrile 4 with the aim to determine whether the deprotonation-electrophilic quench would be feasible and how fast the intermediate magnesiated nitrile undergoes racemisation. Herein, we describe the rst high yielding, highly enantioselective metallation-substitutions of a chiral a-aminonitrile by using a simple magnesium base.

Results and discussion
The carboxylic acid N-Boc-pipecolic acid is commercially available as the (S) enantiomer and this was converted to (S)-N-Boc-2- cyanopiperidine 4 in two steps (see ESI †). This method involved simple amide formation with ethyl chloroformate and ammonia to give the primary amide, followed by dehydration to give the nitrile (S)-4. The corresponding racemic nitrile could be prepared in the same way starting from racemic pipecolic acid by initial N-Boc protection with Boc 2 O, Et 3 N, CH 2 Cl 2 followed by using the same method.
Initially we investigated the deprotonation of the nitrile (S)-4 with LDA in THF at À78 C. Aer 10 min, the anion was quenched by addition of various electrophiles. Under these conditions we generally obtained good yields of racemic products (see ESI †), as determined by chiral stationary phase (CSP) GC or HPLC analysis. The formation of racemic products was expected under these conditions as lithiated nitriles are known to exist with the lithium cation on the nitrogen atom 2-5 and therefore the stereochemistry at the carbon centre would be readily lost.
We then turned to the asymmetric reaction by deprotonation and electrophilic quench, particularly on using a soer metal counterion where we hoped that the metal would remain on the carbon centre for long enough to maintain the conguration. We had previously obtained some success in this approach by using magnesium bases, 11 with one equivalent of i PrMgCl, TMPMgCl$LiCl, or TMPMgCl. All these bases resulted in signicant enantioenrichment aer leaving the magnesiated intermediate for 5 s at À107 C prior to addition of acetone (enantiomer ratio, er, up to 96 : 4 of the substituted product 5e). However the conversion in these reactions was low and this product could not be isolated.
To optimise the reaction we carried out in situ IR studies and found that only partial conversion occurs with one equivalent of TMPMgCl (Fig. 3a). However, almost complete conversion occurs with two equivalents and full conversion with three equivalents of TMPMgCl (Fig. 3b).
Several bases were tested under these optimised conditions (Scheme 1). The bases CuO t Bu, mesityl copper, or TMPZnCl$MgCl 2 were unsuccessful. The method developed by Takeda and co-workers 10 with NaHMDS (and 4-BrC 6 H 4 COCl in situ) did give the product 5g but the enantioselectivity was poor (71% yield, er 53 : 47). With the magnesium base iPrMgCl the yield was low (10% yield of product 5a with PhSSO 2 Ph as the in situ electrophile, er not determined). However the base TMPMgCl was much more successful and a good yield and er of the product 5g was obtained (68% yield 5g, er 81 : 19).
We therefore selected TMPMgCl as the most suitable base. The enantioselectivity was not optimal and we were aware that Carlier and co-workers had found that a magnesiated cyclopropylnitrile racemises more rapidly in THF than in Et 2 O. 15 Therefore we conducted kinetic experiments to determine the rate of enantiomerisation of this organomagnesium species in THF/Et 2 O (1 : 1) and in Et 2 O (see ESI † and Fig. 4). At À104 C the intermediate organomagnesium compound was trapped aer various time periods to give the product 5a or 5f and the er was measured by CSP HPLC or GC respectively. These gave good rst order plots and revealed rates for inversion k $ 6.5 Â 10 À3 s À1 in THF/Et 2 O (see ESI †) and k $ 4 Â 10 À3 s À1 in Et 2 O (Fig. 4).
The kinetic data demonstrate a slightly slower rate for inversion of the intermediate organomagnesium species in Et 2 O. In the case of the magnesiated nitrile 4, the enantiomerisation half-life t 1/2 $ 3 min in Et 2 O and only $2 min in THF/ Et 2 O, presumably as THF helps to solvate the magnesium cation. Therefore, we carried out the deprotonation in pure Et 2 O and were pleased to nd that this improved the enantioselectivity of the metallation-electrophilic quench reaction (Scheme 2). The optimised conditions involved rapid addition of three equivalents of TMPMgCl (prepared from i-PrMgCl and TMPH in Et 2 O) to the nitrile 4 in Et 2 O at À104 C, either with the electrophile added in situ pre-mixed with the nitrile 4 (in the case of the S-aryl benzenesulfonates) or with the electrophile added aer about 10 s (for the carbonyl electrophiles).
High enantioselectivities of the arylthio derivatives 5a-c were obtained by the in situ method. With the ortho-methoxy compound 5c, the electrophile was only partially soluble in pure Et 2 O, so this reaction was carried out with some THF and this may account for the reduced selectivity. The organomagnesium intermediate has sufficient congurational stability to allow its formation followed by electrophilic quench without the need for an in situ electrophile. Benzaldehyde provided racemic product 5d, possibly due to single electron transfer. However, acetone gave highly enantioenriched alcohol 5e and cyclobutanone gave the alcohol 5f also with excellent er. The electrophile p-bromobenzoyl chloride gave the nitrile 5g with er 83 : 17 aer 10 s quench and similar selectivity with in situ quench.
Recrystallisation of the nitrile 5g gave essentially enantiopure compound (er 99 : 1 by CSP-HPLC) and the absolute conguration was determined by single crystal X-ray analysis (Fig. 5). † This demonstrated that the electrophilic quench occurred with retention of conguration. To determine the absolute conguration of the suldes, we carried out sulfurmagnesium exchange with i-PrMgCl. 16 This transformation has not to our knowledge been reported with an enantioenriched sulde and it was intriguing to discover whether it would be possible to transfer chirality by this method. Addition of i-PrMgCl to the sulde 5c in Et 2 O at À104 C followed by addition of p-bromobenzoyl chloride gave the product 5g in moderate yield and only partial loss of enantioselectivity (Scheme 3). The major enantiomer of the product 5g had the same conguration as that obtained by the direct addition of p-bromobenzoyl chloride, thereby demonstrating that the sulde 5c (and hence also likely the suldes 5a and 5b) has the stereochemistry as shown, and was formed by reaction with retention of conguration. We have not been able to determine the absolute congurations of the alcohols 5e and 5f, but these are likely to be as shown with reaction by retention of conguration, and this would be in line with other known electrophilic quenches of metallated N-Boc-piperidines. 17 A similar reaction was carried out, in which sulfur-magnesium exchange was followed, aer 10 s, by addition of the electrophile PhSSO 2 Ph to give the product 5a (Scheme 3). This was formed in moderate yield without signicant loss of enantiopurity, together with what appeared to be an alkene by-product from elimination.
The impressive enantioselectivities that can be obtained with the simple N-Boc-2-cyanopiperidine 4 and base TMPMgCl demonstrate that this method has potential for asymmetric synthesis. The magnesium metal likely has a preference for attachment to the carbon atom at least initially. A possible  intermediate, supported by analogy to that proposed by Carlier and co-workers for magnesiated cyclopropylnitriles, 15 would have two magnesium atoms, one on the carbon atom and one on the nitrile nitrogen atom, connected by a bridging chloride. Dimeric magnesium amides with bridging chloride ligands are well known. 18,19 To investigate this further, Density Functional Theory (DFT) calculations were performed [6-311G(d,p) basis set with B3LYP functional: see Computational methods section below]. Our calculations show, as expected, that nitrile 4 should be present as two rotamers corresponding to different orientations of the Boc group with approximate equal probability. The rate of rotation of the Boc group is slow at the temperatures used for the reaction. Upon deprotonation this will lead to either the C]O group or the C-O t Bu group pointing towards the deprotonated carbon. Complexes of these rotamers, with a bridging chloride ligand between two magnesium ions, were found and are shown in the ESI. † These complexes have very different conformations for the two rotamers. In the case where the O t Bu group points towards the magnesium, a ring-ip occurs to provide a lower energy structure in which the Boc group sits on the opposite side of the ring from the magnesium.
The case where an additional Et 2 O coordinates at the proximal magnesium was also considered. In contrast to that of Carlier and co-workers, 15 this leads to an additional stabilization energy and opens up the possibility of structures without the bridging Cl atom, which could be expected to be less strained. The two rotamers with the lowest energy are shown in Fig. 6. Other orientations were attempted as well, but all lead to higher energy structures. Fig. 6a (structure 6) has chelation of the C]O group to the magnesium. Fig. 6b (structure 7) derives from the other rotamer without this chelation and this structure is 62 kJ mol À1 higher in Gibbs energy than 6. As the Boc group is not rotating at low temperature, both species should be present in solution and able to react with the electrophiles as shown earlier.
The organomagnesium species 6 and 7 could racemise by breakage of the C-Mg bond followed by carbanion inversion and reattachment of the magnesium to the opposite face. Alternatively racemisation could take place by formation of the N-magnesiated ketene imine type structure. However, by whatever mechanism racemisation occurs, the experimental data show that the C-magnesiated intermediates have sufficient lifetime at low temperature for addition of an electrophile and reaction to give highly enantiomerically enriched products.

Experimental
A representative method for the deprotonation and quench of nitrile 4 is given below. For further details and all data, see ESI. † TMPMgCl (1.6 mL, 0.75 mmol) was added to the nitrile 4 (54 mg, 0.25 mmol) in Et 2 O (1 mL) at À104 C. Aer 10 s, cyclobutanone (0.056 mL, 0.75 mmol) was added. Aer 30 min, saturated aqueous NH 4 Cl (0.3 mL) was added. The mixture was allowed to warm to room temperature and was extracted with Et 2 O (3 Â 1 mL), dried (MgSO 4 ) and the solvent was evaporated. Purication by column chromatography on silica gel, eluting with petrol-EtOAc (9 : 1), gave the alcohol 5f (47 mg, 68%);

Computational methods
All calculations were performed using the D.01 version of Gaussian 09. 20

Conclusions
In conclusion, the base TMPMgCl can be used at low temperatures to deprotonate a chiral nitrile without signicant loss of enantiopurity even in the absence of an in situ electrophile. The intermediate magnesiated nitrile can be trapped with a variety of electrophiles to give enantioenriched substituted nitrile products with overall retention of conguration. The organomagnesium intermediate racemises fairly rapidly and the half-life is slightly slower in the presence of the less polar solvent Et 2 O than in THF. In addition we have shown that sulfur-magnesium exchange can occur with retention of conguration. Calculations support the experimental that two magnesium ions are present in the intermediate complexes.
These results suggest that, despite their general lack of use for asymmetric synthesis, chiral nitrile anions can be valuable intermediates that do not always lose their conguration but can be converted to highly enantiomerically enriched products.