Amphiphilic π-Allyliridium C,O-Benzoates Enable Regio- and Enantioselective Amination of Branched Allylic Acetates Bearing Linear Alkyl Groups

Abstract

The first examples of amphiphilic reactivity in the context of enantioselective catalysis are described. Commercially available π-allyliridium C,O-benzoates, which are stable to air, water and SiO₂ chromatographically, and are well known to catalyze allyl acetate-mediated carbonyl allylation, are now shown to catalyze highly chemo-, regio- and enantioselective substitutions of branched allylic acetates bearing linear alkyl groups with primary amines.

Graphical Abstract

Since the discovery of the Tsuji-Trost reaction,¹ diverse catalytic systems for metal catalyzed allylic substitution have emerged.² Among the many useful transformations based on this pattern of reactivity, enantioselective iridium catalyzed aminations figure prominently.^3–11 Largely due to the pioneering work of Takeuchi,^3,4 Helmchen,^5,6 Hartwig,^7,8 Carreira⁹ and You,^10,11 a broad range of amine nucleophiles can now be accommodated in highly regio- and enantioselective reactions of linear or branched allylic acetates (and in certain cases allylic alcohols).^3–11 Two classes of iridium catalysts may be distinguished on the basis of their ability to promote regio- and enantioselective reactions of either linear or branched allylic acetates (Figure 1). For type I catalysts, linear allyl proelectrophiles are used under basic conditions. For type II catalysts, branched allyl proelectrophiles are used under acidic conditions. While type I catalysts are modified by one phosphoramidite and one external diene ligand, type II catalysts incorporate an internal mono-olefin ligand and two phosphoramidites, which may account for their exceptional electrophilicity. The internal olefin of type II catalysts may enhance enantioselectivity in reactions of branched allyl proelectrophiles by displacing the π-bond of (σ+π)-allyl (enyl) iridium intermediates,¹² thus enabling rapid π-facial interconversion in otherwise stereospecific substitutions.¹³

Figure 1.

Phosphoramidite modified iridium complexes and amphiphilic π-allyliridium C,O-benzoates for regio- and enantioselective allylic amination.

Both type I and II catalysts bear π-acidic ligands and a formal positive charge at the metal center. These features enforce electrophilic properties of the allyliridium intermediate. In contrast, we have developed π-allyliridium C,O-benzoates, which are neutral and incorporate relatively strong σ-donor ligands (Figure 1).^14,15 As demonstrated by their ability to promote diverse allylative carbonyl additions, the supporting ligands of this catalyst confer nucleophilic character onto the allyl moiety.16 However, as observed in our initial studies,^14b carbonyl allylation is often accompanied by small quantities of O-allylation, suggesting amphiphilic character of these π-allyl complexes. Pursuant to an accumulation of similar observations by coauthors at Genentech, a systematic attempt to evoke electrophilic behavior was undertaken in the context of allylic amination. Here, we show that the commercially available π-allyliridium C,O-benzoate modified by SEGPHOS catalyzes asymmetric allylic amination. This catalyst overcomes a significant limitation in scope across all known catalytic systems – the ability to engage branched allylic acetates bearing linear alkyl groups with uniformly high levels of regio- and enantioselectivity.^3–11,17,18

In an initial set of experiments, α-methyl allyl acetate 1a (100 mol%) was exposed to benzyl amine 2a (200 mol%) in the presence of the SEGPHOS modified π-allyliridium C,O-benzoate (5 mol%) in THF (1 M) at 40 °C (Table 1). In the absence of base, products of amination were not observed. However, after screening various heterogeneous bases, it was found that reactions conducted in the presence of CS₂CO₃ (120 mol%) provided the desired product of allylic amination 3a exclusively as the branched regioisomer with high levels of enantiomeric enrichment. After further optimization, the allylic amine 3a could be obtained in 90% isolated yield as a single regioisomer in 90% enantiomeric excess. Wet THF solvent is required, perhaps to partially solubilize CS₂CO₃. Using distilled THF under otherwise optimal conditions, 3a was obtained in only 61% yield. Optimal yields were restored using distilled THF in combination with water (250 mol%). The absolute stereochemistry of 3a was determined by comparison of its optical rotation to that of an authentic sample reported in the literature (and x-ray analysis of 3j′).¹⁹

Table 1.

Influence of base in the amination of α-methyl allyl acetate 1a to form allylic amine 3a.^a

^aAll reactions were performed on a 0.44 mmol scale. Enantioselectivities were determined by chiral stationary phase HPLC analysis. See Supporting Information for further experimental details.

^bConversion was determined by ¹H NMR.

^cCs₂CO₃ (200 mol%).

^dYield of material isolated by silica gel chromatography.

Under these optimized conditions, primary benzylic, allylic and aliphatic amines serve as nucleophilic partners in aminations of diverse branched allylic acetates (Table 2). Complete branched-regioselectivity was accompanied by high levels of enantioselectively (¹H NMR & HPLC). Additionally, the levels of chemoselectivity in favor of primary amine allylation were sufficiently high that N-benzyl ethylene diamine and tryptamine could be converted to the respective amination products 3l and 3a′ without competing allylation of the secondary amine. As demonstrated by the formation of 3z, 3g′, 3h′, 3i′, primary amines that are branched at the α-position are effective nucleophilic partners. Further, as illustrated by the formation of 3b vs 3c, 3e vs 3f and 3h vs 3i, excellent levels of catalyst-directed stereoinduction are observed. Perhaps the notable feature of this catalytic system, however, resides in the ability to engage branched allylic acetates bearing linear alkyl groups in highly regio- and enantioselective aminations – a capability that complements the scope of previously reported iridium catalysts for allylic amination.^3–11,17,18

Table 2.

Regio- and enantioselective iridium-catalyzed amination of branched allylic carboxylates.^a

^aYields of material isolated by silica gel chromatography. Standard conditions: (R)-IrLn (5 mol%), amine (200 mol%), Cs₂CO₃ (200 mol%), “wet” THF (1 M), 50 °C. Enantioselectivities were determined by chiral stationary phase HPLC analysis. See Supporting Information for further experimental details.

^bH₂NCH₂CH₂NHBn (400 mol%), ee% determined at the stage of 4l (eq. 1).

^cThe cyclometallated iridium catalyst modified by the Roche ligand was used.¹⁸

To illustrate the utility of the reaction products a series of transformations were performed. The N-benzyl ethylene diamine adduct 3l was subjected to reductive amination with glyoxal to form the piperazine 4l (eq. 1).²¹ Adducts 3x and 3j′ were converted to 1-(N-benzyl-amino)-2-cyclohexene 4x (eq. 2)²² and the cyclopropyl substituted lactam 4j′ through ring-closing metathesis (eq. 3).^5g

(eq. 1)

(eq. 2)

(eq. 3)

While numerous reports of “umpoled allylations” exist,²³ the amphiphilic properties displayed by π-allyliridium C,O-benzoates appear quite unique.²⁴ The same catalyst, (R)-IrLn (5 mol%), will promote both nucleophilic and electrophilic allylation under very similar conditions. For example, under standard amination conditions, 4-aminomethyl-benzyl alcohol and α-methyl allyl acetate are directly converted to compound 7 in 36% yield. Alternatively, compound 5 can be converted to compound 7 in a step-wise manner with higher levels of stereoselectivity using the same iridium catalyst (Scheme 1).

Scheme 1.

Identical π-allyliridium C,O-benzoate complexes promote both nucleophilic and electrophilic allylation.^a

^aSee Supporting Information for further experimental details.

The nature of the C-N bond forming event merits discussion. For related enantioselective iridium catalyzed allylic aminations, an outer-sphere mechanism is postulated.^3–11 While it is likely such pathways are also operative in aminations catalyzed by π-allyliridium C,O-benzoates, an inner-sphere mechanism cannot be excluded. The more stringent steric constraints of an inner-sphere mechanism may account for the fact that primary amines engage in allylation while secondary amines do not. Furthermore, enantiofacial selectivity for amination through an inner sphere mechanism matches that observed in nucleophilic allylations of α-substituted allylic acetates (Figure 2). Computational studies are underway to evaluate outer vs inner sphere pathways.

Figure 2.

Stereochemical models accounting for equivalent π-facial selectivity in crotylation and amination.

In summary, highly tractable and commercially available π-allyliridium C,O-benzoates, which are well known to catalyze nucleophilic carbonyl allylation, are now shown to catalyze chemo-, regio- and enantioselective electrophilic aminations of branched allylic acetates bearing linear alkyl groups. These processes broaden access to chiral N-containing building blocks²⁵ and establish unique amphiphilic properties of π-allyliridium C,O-benzoates, which should inform the design of new catalytic process. More broadly, this work demonstrates how academic-industrial collaboration accelerates the discovery of robust, innovative methods for chemical synthesis.