Evaluation of reaction conditions. To check the feasibility of the above scenario, we started this investigation of Ni-catalyzed arylamination of homoallylic amine substrate 1a containing a PA directing group with phenylboronic acid and piperidino benzoate as coupling partners (Table 1). After extensive optimization, we were pleased to find that the reaction in the presence of NiBr2·DME as the catalyst and K3PO4 in tert-butanol at 80 °C gave the desired 1,2-arylamination product 2a as a single regioisomer in 80% isolated yield, accompanying with ≈5% unreacted alkene and trace amounts of hydroarylation byproduct (entry 1). Ni(COD)2 also catalyzed the reaction in moderate yield (entry 2), while NiCl2 or NiBr2 catalyst did not form any product (entry 3). Replacement of nucleophile PhB(OH)2 by PhBpin led to a slight lower yield (entry 4). The use of DMF solvent led to essentially no desired product (entry 5), while use of dioxane or iPrOH solvent provided a lower yield (entries 6-7). Inferior results were obtained when using other bases instead of K3PO4 (entries 8-10), or conducting the reaction at 50 °C (entry 11). Benzoyl protected homoallylic amine and N-methylated picolinamide were subjected to the standard conditions, and both reactions did not occur, indicating that the pyridine N(sp2) and N-H moiety were both indispensable in this transformation. It merits to mention that, the putative 8-aminoquinoline-masked 3-butenoic acid,[43-47] which was widely used in alkene difunctionalization, did not afford any product under the optimized conditions.
Substrate scope. With the identified conditions in hand, we first sought to define the scope of the arylboronic acid partner (Fig. 2). In general, arylboronic acids bearing a wide range of electronically varied substituents reacted to produce the desired products in good yields. A variety of functional groups were accommodated well, including ethers (2b-c), dimethylamine (2h), trifluoromethyl (2i), fluoro (2j), chloro (2k) and bromo (2l). Notably, sterically hindered ortho-substituted arylboronic acid showed slightly increased reactivity, reacted to afford the desired product in very good yield (2d). In addition, arylboronic acids containing iodide, alkene, aldehyde and ketone (2m-p), which can be further derivatized, were also amenable to this reaction.
Subsequently, we examined the N-O electrophile scope of the reaction using PhB(OH)2 as the nucleophilic component (Fig. 3). The reactivities of disubstituted amine sources were firstly investigated in the arylamination, diethylamine and N-methylbenzylamine could be introduced under the optimized conditions (3a-b). In addition to acyclic amines, cyclic amines containing a series of heterocyclic scaffolds, including morpholine (3c), thiomorpholine (3d), piperazine (3e), ester group (3f), and acetal group (3g), which are prevalent in biologically active molecules, were competent substrates. Seven-membered hexamethyleneimine was also tolerated, affording the desired product 3h. The scope of monosubstituted amine transfer agents for the synthesis of secondary amines were explored, which usually led to a dramatic drop in reaction efficiency.[48-49] To our delight, sterically hindered secondary alkylamines are accessible with O-benzoylhydroxylamines derived with tertiary alkyl group (3i-l). However, no reaction occurred when primary- or secondary-alkyl-substituted amine transfer agents were used.
After evaluation of electrophile and nucleophile scope, we turned our attention to the scope of alkene substrates (Fig. 4). The α-substituted homoallylamine was first tested under the standard conditions, the terminal alkene bearing alkyl- or aryl-substitution at the α-position proceeded smoothly to afford the arylamination products (4a-b) in moderate yields with high diastereoselectivity. The relative stereochemistry of the diastereomer of 4b was established by X-ray crystallographic analysis, with the trans orientation of α and γ substituents. This result established the relative stereochemistry of two stereocenters remote from one another, which has been shown to be difficult. n-Butyl-substituted homoallylamine at the β-position could also be tolerated in the reaction but with lower diastereoselectivity (1.9:1, 4c), and a more sterically bulky group led to a lower yield (4d). 1,1-Disubstitued alkene was also tolerated and successfully afforded desired product containing a quaternary center (4e). We then explored alkene substrates that are typically challenging in alkene carboamination. Gratifyingly, Z- or E- internal substrates could be efficiently converted into the syn-diastereomer under the optimized conditions (4f-l), which was consistent with our proposed mechanism. Both diastereoisomers were accessible based on the cis/trans configuration of the alkene substrate (4f and 4g), suggesting the alkene does not undergo isomerization in the Ni-catalyzed process. In addition, phenyl-, benzyl-, and isopropyl-substituents were well tolerated in the reaction.
In light of the success in developing a regioselective arylamination of homoallylamines, we continued our survey by applying the protocol to PA protected allylic and bishomoallylic amines. To our delight, the reaction of allylamine with phenylboronic acid and piperidino benzoate under the optimized conditions furnished the arylamination product in moderated yield, with only a single regioisomer was detected by GC-MS analysis. Moreover, a variety of terminal and internal bishomoallylamines underwent arylamination to regioselectively provide δ-amino benzenepentanamine products. Likewise, both trans- and cis-internal alkenes were effective in this reaction, delivering the desired product with excellent diastereoselectivity. We assumed that four and six membered nickelacycles were formed and can be stabilized in the catalytic system.
Synthetic potential. We next performed the gram-scale reaction to illustrate the synthetic utility of this methodology (Fig. 4a). The reaction of 1a with phenylboronic acid and morpholine benzoate on a 5 mmol scale afforded 3c in 80% yield. We were able to remove the PA directing group with NaOH in EtOH at 100 °C, and the primary amine 5 could be generated in nearly quantitative yield. This methodology was also applied to the late-stage modification of complex, phamaceutically relevant compounds, as shown in Fig. 4b. Arylboronic acid and O-benzoylhydroxylamine derived from fenofibrate and loratadine independently underwent arylamination, affording corresponding desired products (6a and 6b).
Mechanistic consideration. To elucidate the mechanism, we first conducted the radical clock experiment (Fig. 5a). When the prepared alkene substrate 1b was subjected to standard reaction conditions, only cyclopropane remained product 4s was formed in 64% yield, implying that the cyclopropylmethyl radical intermediate known to ring rupture might not be generated in the catalytic cycle. The effect of radical inhibitors was next examined (Fig. 5b). As it turned out, the arylamination was not largely inhibited by the addition of TEMPO or BHT. This suggested that the arylamination probably did not involve a radical process, although the possibility of radical formation followed by a fast recombination with Ni within the solvent cage cannot be rulled out.
Regarding redox manifolds of Ni catalysis, Ni0/NiII and NiI/NiIII catalytic systems were considered as two possible pathways (Fig. 6). In Pathway A, a Ni(II) species 7 was generated from Ni(0) species and O-benzoylhydroxylamine via oxidative addition. Transmetalation followed by migratory insertion of the alkene into the Ni-carbon bond would form 8, which could yield the product by reductive elimination. However, we consider this pathway to be less likely because the selective insertion into the Ni-C rather than Ni-N bond was suspicious and the energy barrier of reductive elimination of NiII amido species is too high under similar catalytic system based on DFT calculations (>50 Kcal/mol).[50] An alternative pathway in which the alkene inserted into the Ni-N bond precedes transmetalation and C-C reductive elimination was also considered unlikely, because it would involve formation of thermodynamically unfavored larger nickelacycles, especially for bishomoallylic amine substrates (seven-membered nickelacycles). Alternatively in Pathway B, the reaction was initiated by a Ni(I) species (I),[51-52] which was formed probably from comproportionation between NiBr2 and Ni(0) species.[53-54] After further transmetalation with arylboronic acids and olefin migratory insertion, nickel-alkyl species (III) was generated. The species stabilized by bidentate PA directing group underwent oxidative addition with the aminating reagent much faster than protonation with the alcohol solvent or β-hydride elimination, forming NiIII amido species IV, which was believed to be able to undergo facile reductive elimination.[55-58] Finally, the active Ni(I)-X catalyst I was regenerated, and the desired product was furnished through the subsequent ligand exchange with the alkene substrate.