Direct catalytic enantioselective amination of ketones for the formation of tri- and tetrasubstituted stereocenters

A Zn–ProPhenol catalyzed direct amination of α-branched and unbranched ketones produces α-amino carbonyl and β-amino alcohol products.


Introduction
Nitrogen-containing molecules are ubiquitous among natural products and pharmaceuticals. In this regard, the stereoselective formation and functionalization of amines has been an area of continuous interest in synthetic organic chemistry. 1 Despite considerable advances in this eld, the enantioselective synthesis of a-tertiary amines, a common motif in naturally occurring and synthetic bioactive molecules, 2 still represents a synthetic challenge. 3 In this context, we report a direct catalytic enantioselective amination of cyclic ketones for the construction of tri-and tetrasubstituted N-containing stereogenic centers. This transformation allows efficient and stereocontrolled access to a broad array of a-amino carbonyl and bamino alcohol motifs starting from widely available ketone precursors (Scheme 1).
Nitrogen atoms are usually incorporated within organic scaffolds by the addition of a nitrogen-based nucleophile to a carbon-based electrophile, oen with the intermediacy of a C]N bond which can undergo further reactivity to afford amines. In comparison, the electrophilic amination of carbonbased nucleophiles has received much less attention. [4][5][6] This latter strategy is particularly relevant for the enantioselective synthesis of a-tertiary amines, since the catalytic asymmetric addition of nucleophiles to ketimines is an extremely challenging process. 5 Indeed, the direct catalytic enantioselective aamination of carbonyl compounds 6 has proven to be a useful strategy to access molecules decorated with tetrasubstituted Ncontaining stereogenic centers. However, these reactions rely on the use of aldehydes 7 or activated carbonyl compounds such as 1,3-dicarbonyls, 8 a-cyanocarbonyls, 9 a-keto esters, 10 and 2oxindoles. 11 There are only a few reports of direct enantioselective amination of unactivated ketones 12,13 and only two examples describe the use of a-branched ketones for the synthesis of a-tertiary amines. Terada's initial report utilizing a chiral organosuperbase with cyclic aromatic ketones 14 (Scheme 1a) was followed by Toste's chiral phosphoric acid methodology with cyclic aliphatic ketones 15 (Scheme 1b). Both methods require the formation of a tetrasubstituted carbon alpha to the carbonyl to avoid epimerization of the nal product. To date, there is no general method allowing for the direct catalytic asymmetric amination of both a-branched and unbranched ketones.
Our group's aza-hemicrown family of ProPhenol ligands form dinuclear main group metal catalysts when treated with alkyl metal reagents such as Et 2 Zn. 16 The ability of the Zn-ProPhenol catalytic system to promote a dual-activation of both nucleophiles and electrophiles has allowed for the development of a wide range of enantioselective addition reactions, but the use of heteroatom electrophiles has so far not been achieved. 17 In this context, we wished to evaluate the potential of our Zn-ProPhenol catalysts for amination reactions of unactivated ketones using azodicarboxylates.

Results and discussion
We initiated our studies with 2-methyl-1-indanone 1a and ditert-butyl azodicarboxylate as a cheap and practical electrophilic nitrogen source (Table 1). Pleasingly, our standard Zn-ProPhenol catalyst promoted the desired amination in toluene at 40 C to afford 2a with promising yield and selectivity (entry 1). Following this initial hit, a screen of Lewis-basic additives was undertaken based upon the previous observation that such additives can signicantly impact the catalyst turnover and resulting enantioselectivity of the Zn-ProPhenol catalyst. 17 The use of triphenylphosphine oxide (TPPO) or tert-butyl-N-methylcarbamate (3) as additives resulted in modest increases in ee (entries 2 and 3). However, bis-phosphine oxide (R)-BINAPO provided a signicant boost in both yield (97%) and ee (89%) (entry 4). A small but measurable mismatched effect was observed when (S)-BINAPO was used, since the same enantiomer of 2a was obtained albeit with lower selectivities (entry 5). Keeping with bidentate phosphine oxides, additives 4 and 5 were screened, but both resulted in a signicant reduction in ee (entries 6 and 7). Finally, while the bis-phosphine oxide of 1,4bis(diphenylphosphino)butane (dppbO 2 ) proved no better than TPPO, the use of the bis-phosphine oxide of 1,2-bis(diphenylphosphino)ethane (dppeO 2 ) restored the selectivity and reactivity of the additive (R)-BINAPO with the added benet of being achiral (entries 8 and 9). Switching ligands from (R,R)-L1 to the non-C 2 symmetric (S,S)-L2 resulted in an improved enantioselectivity at the expense of yield (entry 10). However, a slight increase of catalyst loading to 15 mol% gave 2a in high yield and selectivity (entry 11).
With optimized conditions in hand (Table 1, entry 11), the scope of the process was explored (Scheme 2). Indanones having a linear alkyl chain at the 2-position all gave excellent results (2a-2f). Of note, functional groups such as methyl esters, alkenes and terminal alkynes are well tolerated (2d-2f). Introduction of a bulkier isopropyl group at the 2-position of indanone still afforded product 2g with good enantioselectivity, albeit with a lower yield. Substrates were not limited to 5- membered rings, as 6-membered aryl and even vinyl ketones showed excellent reactivity and selectivity (2h-2i). To further highlight the broad scope of our catalytic amination process, two 3-alkyl-2-oxindole substrates were aminated using a reduced catalyst loading at room temperature (2j-2k). We next turned our attention to the use of unbranched aromatic ketones as nucleophiles in our amination reaction. Importantly, these substrates are not compatible with either Toste's or Terada's protocols due to product epimerization from the strongly acidic or strongly basic catalysts required. We felt the mild reaction conditions provided by our bifunctional Zn-ProPhenol catalyst should be perfectly suited for such challenging substrates.
When 1-tetralone 1l was reacted with 2.0 equivalents of ditert-butyl azodicarboxylate in THF (0.4 M) at room temperature using 11 mol% of (R,R)-L1 and 20 mol% of diethylzinc, the desired amination product 2l was isolated in a promising 66% yield and 88% ee ( Table 2, entry 1). Lowering the equivalents of electrophile to 1.1 resulted in an increase in yield without impacting the ee (entry 2). The molecular sieves were found to be important to catalyst turnover but not selectivity (entry 3). Decreasing the reaction concentration to 0.25 M improved the selectivity to 94% ee without sacricing catalyst activity, while performing the reaction at 4 C signicantly impacted catalyst turnover (entries 4 and 5).
The scope of this amination process was then explored (Scheme 3). A range of 1-tetralones substituted at the 5-, 6-and 7-positions were successfully aminated in high yields and selectivities (2l-2q). Of note, tetralones 1o, 1p and 1q substituted with strongly electron-withdrawing groups showed no evidence of epimerization under the reaction conditions, showcasing the mild activation provided by the Zn-ProPhenol bifunctional catalyst. Importantly, such electron decient tetralones were poor substrates for previously reported amination methods. The use of 1-chromanone 1r was challenging and lower temperatures as well as partial conversion were required to prevent signicant epimerization of 2r. On the other hand, 1benzosuberone 1s gave excellent results using our protocol.
Finally, the 1-indanone class of substrates was studied. This class of substrates proved to be a signicant challenge for previous amination reactions because they afford highly epimerizable products which were isolated with poor enantioselectivities. 13c,e Using our mild Zn-ProPhenol catalyst and stopping the reaction at partial conversion led to the isolation of products 2t and 2u with great enantioselectivities and good yields. In the case of substrate 1u, the use of L2 was benecial.
To showcase the scalability and practicality of the process, we performed some amination reactions on a 1 to 2 mmol scale with reduced catalyst loadings (Scheme 4). Amination of oxindole 1j was scaled to 2.1 mmol using 2 mol% catalyst without any signicant change in outcome. Similarly, the amination of 2-methyl-1-indanone 1a could be run on a 1 mmol scale using 10 mol% of catalyst without a signicant impact on the efficiency of the reaction. Finally, the amination of 7-uorotetralone 1o was performed on a 1 mmol scale with 3 mol% catalyst  and 2o was obtained with comparably excellent yield and enantioselectivity. While some of these amination processes require relatively high catalyst loadings, these loadings can be mitigated by the easy recovery of the ProPhenol ligand through column chromatography and a subsequent acid-base extraction. To showcase this feature, the amination of 1-benzosuberone 1s was performed on a 0.6 mmol scale using our conditions from Scheme 3 (Scheme 5). Aer workup and purication, 83% of ligand L1 was recovered. This recovered ligand was then reused in a subsequent reaction with no loss of reactivity or selectivity.
A requirement for any useful amination technique utilizing dialkyl azodicarboxylate electrophiles is to show that the resulting hydrazines can be readily transformed into synthetically desirable amino products. In this respect, the use of di-tertbutyl azodicarboxylate represents a signicant improvement over related methods requiring use of the less readily deprotected dimethyl azodicarboxylate (DMAD) or diethyl azodicarboxylate (DEAD). 13a,c,e To highlight the utility of our products, oxindole 2j was converted to Boc-amine 6 in a two-step sequence involving an alkylation with methyl bromoacetate followed by an E1cB elimination promoted by cesium carbonate (Scheme 6a). 18 In addition, both syn and anti amino alcohol derivatives could be accessed from ketone 2s with no erosion of ee (Scheme 6b). Bulky L-selectride delivers hydride from the more sterically accessible face to form the syn-product while the use of lithium aluminium hydride results in an internal delivery of the hydride to form the anti-product. 19 We then developed a one-pot procedure for the subsequent alkylation-elimination cleavage of the N-N bond to access compounds syn-7 and anti-8 in 3 and 2 steps respectively. Overall, this approach allows a stereodivergent synthesis of cyclic compounds incorporating a phenylethanolamine motif. Such a motif can be found in many bioactive natural products, including the Ephedra alkaloids. 20 It is also a common motif in marketed and clinical drugs as exemplied by salbutamol, a bronchodilator used to treat asthma, and naxagolide, Merck's antiparkinson compound. 21,22 A proposed catalytic cycle is depicted in Scheme 7 using (S,S)-L1 instead of the non-C 2 symmetric (S,S)-L2 for purposes of simplicity. The initially formed dinuclear Zn-ProPhenol catalyst enters the catalytic cycle through a deprotonation of the ketone to form complex I. Bi-dentate coordination of the electrophile produces complex II which sets the electrophile up for attack from the re-face of the enolate. The resulting zinc-hydrazine product (III) serves as a source of base to deprotonate another equivalent of ketone while simultaneously releasing the product. Importantly, the stereochemical outcome of the reaction is in perfect agreement with our recent work on related Mannich reactions. 23 The source of selectivity in complex II is made clearer when viewed down the axis of the phenol (IIa, bottom of Scheme 7). From this orientation, steric interactions are minimized when the bulk of the reactants is directed, as shown in IIa, towards the western pseudo-equatorial phenyl group (green) and would be maximized if the bulk of the Scheme 5 Recovery of ligand. reactants is instead positioned towards the eastern pseudo-axial phenyl group (purple).

Conclusion
In conclusion, we have developed a mild catalytic method for the direct amination of aromatic and vinyl ketones as well as of oxindoles. The procedure is unique in its ability for a single catalyst to form tri-and tetrasubstituted N-containing stereocenters with high enantioselectivities, even those whose products contain readily epimerizable indanone or electronwithdrawing tetralone scaffolds that do not fare well under previously reported conditions. 13 This method negates the need for cryogenic temperatures and multi-day reaction times characteristic of previous reports. 14,15 The reactions can be scaled up with reduced catalyst loadings without impacting the efficiency of the process and the ligand can be recovered and reused, making the effective catalyst loadings very low. Finally, the resulting hydrazine products are readily converted into valuable amines utilizing a simplied one-pot alkylation-elimination procedure.

Conflicts of interest
There are no conicts of interest to declare.