Enantioselective Radical Dearomative Conjugate Amination Enabled by Co(II)-Based Metalloradical Catalysis

Abstract

Delocalized radical systems present a challenging yet appealing ground to test the control of multiple selectivity in organic synthesis. Despite some recent advancements, the issue of regioselectivity in delocalized radical system has largely centered on allylic radicals. To explore larger delocalized radical systems, we report the catalytic generation of extensively delocalized 4-vinylphenoxyl radicals and their involvement as key intermediates in regioselective radical C–N bond formation. Guided by the mechanistic principles of metalloradical catalysis, we develop a Co(II)-based enantioselective radical system for dearomative 1,7-conjugate amination of readily available 4-vinylphenols with aryl azides. This can afford valuable chiral α-tertiary amino acid derivatives in high yields with excellent enantioselectivities for the newly-created tetrasubstituted stereocenters. Unlike prior systems, this amination involves hydrogen-atom abstraction from O–H bonds. As demonstrated with 1,6-conjugate addition with various nucleophiles, the resulting α-tertiary amino acid derivatives, which bear additional para-quinone methide (p-QM) functionality, may find useful synthetic applications.

Graphical Abstract

Introduction

The past decade has witnessed considerable growth in the development of new synthetic methods involving organic radicals as intermediates.(1) While manifesting the rich fundamental reactivity and attractive practical attributes of radical chemistry, this burgeoning exploration has also highlighted the long-standing challenges of controlling the reactivity and selectivity in radical reactions.(2) Among the diverse radical intermediates, delocalized radical system poses an added challenge with the control of regioselectivity, alongside chemoselectivity and stereoselectivity, while presenting exceptional opportunities for constructing complex molecular structures. Among advances in catalytic strategies,(3–11) metalloradical catalysis (MRC) has emerged as a general approach for controlling reactivity and selectivity of homolytic radical reactions.(12) MRC exploits the metal-centered radicals in open-shell metal complexes as one-electron catalysts for homolytic activation of substrates to generate metal-entangled organic radicals as the key intermediates to govern the reaction pathway and stereochemical course of subsequent radical processes.(13–31) To this end, Co(II) complexes of porphyrins, which represent a family of stable 15e-metalloradicals with a well-defined low-spin d⁷ configuration, have demonstrated remarkable ability for homolytic activation of organic azides to generate α-Co(III)-aminyl radicals that can serve as competent intermediates for catalytic C–H amination.(32, 33) With the modularly-designed D₂-symmetric chiral amidoporphyrins (D₂-Por*) as a versatile ligand platform,(34) Co(II)-based MRC has been successfully applied for the development of an asymmetric radical process that can catalyze direct allylic C–H amination of 1,2-disubstituted alkenes with aryl azides.(32) This process involves intermolecular hydrogen atom abstraction (HAA) from allylic C–H bonds by the initially-generated α-Co(III)-aminyl radicals, followed by radical substitution (RS) directly at the original allylic site (Fig. 1a). As a further application of Co(II)-MRC for controlling regioselectivity, we have recently achieved asymmetric radical 1,3-conjugate allylic C−H amination of trisubstituted alkenes with aryl azides, with concurrent control of diastereoselectivity and enantioselectivity (Fig. 1b).(33) The key to this success lies in the effective catalyst design through tailored ligand engineering, creating an optimal catalytic environment that maximizes the desired non-covalent attractive interactions in key intermediates to govern regioselectivity between direct and 1,3-conjugate allylic C–H amination. Prompted by these recent findings, we wondered whether Co(II)-MRC could be explored to control regioselectivity, in addition to chemoselectivity and stereoselectivity, in a larger and more extensively delocalized radical system than allylic radicals.

Fig. 1. Regioselectivity in intermolecular radical amination process involving delocalized radical intermediates.

a, Direct C–H amination via delocalized allylic radicals (prior work). b, 1,3-Conjugate C–H amination via delocalized allylic radicals (prior work). c, Conjugate amination via delocalized 4-vinylphenoxyl radicals (this work).

Among large delocalized radical systems, we were particularly intrigued by the possibility of catalytic amination of 4-vinylphenols with organic azides via Co(II)-MRC, which may involve the intermediacy of 4-vinylphenoxyl radicals. This class of oxygen radicals is known for extensive spin delocalization in their π-conjugate system (Fig. 1c). Given that the bond dissociation energy (BDE) of phenolic O−H bonds (~87 kcal/mol) is similar to that of benzylic C−H bonds (~90 kcal/mol) and allylic C−H bonds (~89 kcal/mol), we hypothesized that the initially-generated α-Co(III)-aminyl radical I from metalloradical activation of organic azide 1 could engage HAA from the O−H bond in vinylphenol 2. This would generate ∞-Co(III)-amido phenoxyl radical intermediate II, comprising the delocalized para-vinylphenoxyl radical ^aII and the Co(III)-amido complex ^bII. (Fig. 1c). While this hypothesis seems tenable, HAA from O−H bonds has not been previously demonstrated with α-Co(III)-aminyl radicals I. Considering the potential radical addition (RA) of aminyl radicals to alkenes, it was also uncertain if α-Co(III)-aminyl radical I could chemoselectively react with vinylphenols 2 at the desired O–H bond through HAA rather than at the C=C bond through RA. Additionally, phenols are effective free radical scavengers and tend to undergo oxidative dimerization or polymerization via phenolic radicals, raising concerns about potential side reactions.(35) Furthermore, the formation of ∞-Co(III)-phenoxyl radical intermediate II introduces additional reactivity challenges due to the absence of a covalent linkage between the delocalized para-vinylphenoxyl radical ^aII and Co(III)-amido complex ^bII. If radical ^aII escapes from complex ^bII, it would terminate the desired catalytic cycle, potentially leading to radical chain reactions and other unwanted side processes. Moreover, the expected RS between radical ^aII and complex ^bII via S_H2 would be kinetically demanding because concentrations of both [^aII] and [^bII] are as low as the catalyst loading. Beyond reactivity concerns, the C–N bond-forming RS step would face significant challenges with controlling regioselectivity. Among the four potential reacting radical centers (O1, C3, C5 and C7) in the delocalized radical intermediate ^aII, what are the determining factors that differentiate them and direct a specific radical resonance to react with Co(III)-amido complex ^bII at the nitrogen center via S_H2 to control the regioselectivity of the amination? Would it be possible to govern the S_H2 process to occur preferentially at the remote C7 site over the nearer C3 and C5 sites without reacting at the original O1 site? This would constitute 1,7-conjugate amination with dearomatization, a transformation that has not been documented to date. In addition to the regioselectivity, the S_H2 process for C–N bond formation would also present the challenge of concurrently controlling enantioselectivity, a demanding topic that has been virtually unexplored until recently.

Catalytic asymmetric dearomative functionalization of widely available phenols represents a highly attractive approach for assembling three-dimensional molecular architectures while creating new stereogenic centers from the planar phenolics.(36, 37) Among different functionalization reactions, asymmetric dearomative amination of phenols is a particularly appealing strategy for the stereoselective synthesis of chiral α-tertiary amine derivatives with additional functionalities. Previous efforts have been mainly involved with the use of naphthols as enol nucleophiles for electrophilic amination. Using azodicarboxylates and other electrophilic aminating reagents, highly enantioselective catalytic systems have been developed for 1,3- and 1,5-conjuate amination of β- and α-naphthols, respectively, via Brønsted or Lewis acid catalysis,(38–41) transition metal catalysis,(42, 43) and NHC catalysis.(44) In addition to the limited success in achieving asymmetric dearomative amination of non-fused phenols, 1,7-conjugate amination of 4-vinylphenols remains to be demonstrated. A survey of the literature reveals no previous reports of catalytic systems for asymmetric dearomative amination via radical pathways. This is perhaps not surprise considering the inherent propensity of phenoxyl radicals towards dimerization and polymerization together with the lack of effective solution for controlling regioselectivity and enantioselectivity of reactions in an extensively delocalized radical system.

Guided by the mechanistic principle of MRC, we aimed to translate these and related challenging issues into a solvable problem of catalyst development through ligand design. This would involve fine-tuning the steric, electronic and chiral environment of the versatile D₂-Por* ligand platform to develop an effective Co(II)-based metalloradical catalyst. The key strategy would be to utilize multiple non-covalent weak interactions as attractive but reversible forces to bind, position and orient the substrates and subsequent intermediates within the pocket-like ligand environment to govern the reaction pathway and stereochemical course of the catalytic radical process. In this work, we report the successful development of a catalytic asymmetric system that is highly effective for enantioselective dearomative 1,7-conjugate amination of 4-vinylphenols with aryl azides. Through the identification of an optimal D₂-Por* as the supporting ligand, the Co(II)-based metalloradical system can activate aryl azides at room temperature for the regioselective 1,7-amination of phenols bearing diverse para-substituted alkenes, affording chiral α-tertiary amino acid derivatives in high yields while creating tetrasubstituted stereocenters in high enantioselectivities. We present combined computational and experimental studies that shed light on the underlying stepwise radical mechanism of the Co(II)-catalyzed 1,7-conjugate amination. To demonstrate the synthetic applications of the current catalytic radical process, we showcase a number of stereospecific transformations of the resulting α-tertiary amino acid derivatives. This study highlights the great potential of MRC for enabling asymmetric radical transformations involving increasingly complex and delocalized radical intermediates, thereby expanding the synthetic utility of radical chemistry in stereoselective molecular construction.

Results and discussion

Reaction development

To assess the feasibility of the proposed catalytic process involving delocalized 4-vinylphenoxyl radicals, we first examined the reaction of para-vinylphenol 2a with 4-trifluoromethyl-2,3,5,6-tetrafluorophenyl azide (1a) by Co(II)-metalloradical catalysts (Fig. 2a). The selection of phenol 2a as the substrate was based on the consideration that both ortho-positions are substituted with bulky ^tBu groups to prevent potential homodimerization,(45) while aryl azide 1a was chosen as the aminating reagent due to its effectiveness as a metalloradicophile in Co(II)-based metalloradical system.(32, 33) It was found that the simple metalloradical catalyst [Co(P1)] (P1 = tetraphenylporphyrin; Fig. 2b) was able to catalyze the reaction of vinylphenol 2a with azide 1a, producing the major product 3aa in 47% yield, along with several unidentified byproducts. To our delight, compound 3aa was fully characterized as the desired dearomative 1,7-conjugate amination product. Its structure was further confirmed by X-ray crystallography, revealing the formation of para- quinone methide (p-QM) from the dearomatization of the para-vinylphenol and the construction of α-tertiary amino acid ester via 1,7-conjugate amination. When first-generation chiral metalloradical catalyst [Co(P2)] (P2 = ChenPhyrin; Fig. 2b)(34) was employed for the reaction, the formation yield of amino esters 3aa dramatically increased to 91% while achieving good control of enantioselectivity (71% ee). Excitingly, subsequent use of the analogous catalyst [Co(P3)] (P3 = 3,5-Di^tBu-ChenPhyrin; Fig. 2b),(34) which differs from [Co(P2)] by the addition of two tert-butyl substituents at the 3,5-positions of both 5,15-diphenyl groups without altering the chiral amide units, resulted in significant improvements in both yield (97%) and enantioselectivity (92% ee). However, switching to another analogous catalyst [Co(P4)] (P4 = 2,6-DiMeO-ChenPhyrin; Fig. 2b),(34) which replaces the 3,5-di-tert-butyl substituents in [Co(P3)] with 2,6-di-methoxyl groups, led to dramatic decrease in both yield (20%) and enantioselectivity (70% ee), likely due to the more sterically encumbered ligand environment. Surprisingly, when second-generation chiral metalloradical catalyst [Co(P5)] (P5 = 3,5-Di^tBu-QingPhyrin; Fig. 2b) was employed, only a trace amount of 3aa (6% yield) was detected with low enantioselectivity (16% ee). Notably, ligand P5 differs from ligand P3 only by replacing one of the two distal methyl groups in P3 with a phenyl group in the four chiral amide units. Furthermore, it was interesting to find that another second-generation chiral metalloradical catalyst [Co(P6)] (P6 = 3,5-Di^tBu-Tao(^tBu)Phyrin; Fig. 2b), which bears tert-butyl ester-containing chiral amides, could restore most of the catalytic reactivity and stereoselectivity, producing 3aa in 80% yield with 93% ee. Together, these results demonstrate a remarkable ligand effect on the Co(II)-based metalloradical system, leading to the identification of [Co(P3)] as the effective catalyst for the enantioselective dearomative 1,7-conjugate amination of vinylphenol 2a with azide 1a. After further optimization of the reaction conditions (see Supplementary Table 1 for details), it was found that [Co(P3)] could effectively catalyze the transformation at room temperature in 6 h with only 1 mol % of catalyst loading, forming 1,7-conjugate amination product 3aa in 93% isolated yield with 94% ee. The absolute configuration of the newly-generated tetrasubstituted stereogenic center in 3aa was established as (R) by X-ray crystallography.

Fig. 2. Optimization of Co(II)-based metalloradical system for dearomative amination of 4-vinylphenol with aryl azide.

a, Ligand effect on [Co(Por)] (2 mol %) for dearomative amination of 4-vinylphenol 1a (0.15 mmol) with aryl azide 2a (0.10 mmol) in benzene (0.5 mL) at room temperature for 24 h; ¹H NMR yields with 1,1,2,2-tetrachloroethane as internal standard. b, Structures of Co(II)-metalloradical catalysts used in this study.

Substrate scope

With the optimized conditions in hand, we then investigated the scope of fluoroaryl azides 1 for the catalytic dearomative 1,7-conjugate amination by [Co(P3)] using 2a as the standard vinylphenol substrate (Fig. 3; entries 1−12). Like para-CF₃-substituted tetrafluorophenyl azide 1a, its derivatives bearing other electron-withdrawing substituents at para-position such as −CN (1b), −NO₂ (1c), −SO₃Ph (1d), and −CO₂Me (1e) could also be effectively activated by [Co(P3)] for dearomative 1,7-conjugate amination of the vinylphenol, generating the corresponding α-tertiary amino esters 3ba−3ea bearing p-QM functionality in high yields with high enantioselectivities (Fig. 3; entries 2−5). As well, tetrafluorophenyl azides with para-substituted halogen atoms and both electron-neutral and electron-donating groups such as −F (1f), −Br (1g), −H (1h), and −MeO (1i) could also function as excellent aminating reagents for the catalytic process, leading to high-yielding formation of amino esters 3fa−3ia with high control of enantioselectivities (Fig. 3; entries 6−9). It was shown that the 1,7-conjugate amination process could be scaled up on a 2.5 mmol scale for the efficient synthesis of enantiomerically-enriched amino ester 3ha in 92% yield with 92% ee with only 0.5 mol % of catalyst loading, despite requiring an extended reaction time of 12 hours. Furthermore, 4-tetrafluoropyridinyl azide (1j) could serve as a competent aminating reagent for enantioselective C−H amination of 2a, forming the desired amino acid derivative 3ja productively without complication from potential coordination of the pyridine unit to the cobalt center (Fig. 3; entry 10). Additionally, [Co(P3)] could be similarly effective in activating difluorophenyl azides such as 1k and 1l for the catalytic process, generating the desired dearomative 1,7-conjugate amination products 3ka and 3la (Fig. 3; entries 11 and 12). It should be noted that the current Co(II)-based catalytic system is limited to aryl azides bearing fluoro substituents at both 2,6-positions. For examples, when aryl azides such as 2-bromo-3,4,6-trifluorophenyl azide, 2-bromo-4,6-difluorophenyl azide, 3,4,5-trifluorophenyl azide, and 2,6-dichlorophenyl azide were used, no corresponding amination products were observed under standard conditions.

Fig. 3. Enantioselective dearomative conjugate amination of vinylphenols with fluoroaryl azides by [Co(P3)].

Conducted with 1 (0.15 mmol) and 2 (0.10 mmol) using [Co(P3)] (1 mol %) in PhCF₃ (0.5 mL) at room temperature for 6 h; Isolated yields; Enantiomeric excess (ee) determined by chiral HPLC; Diastereomeric ratio (dr) and alkene configuration E:Z ratio determined by NMR. ^aThe reaction was performed on 2.5 mmol scale using [Co(P3)] (0.5 mol %) at room temperature for 12 h.

We next sought to study the scope of vinylphenols 2 for the catalytic dearomative 1,7-conjugate amination by [Co(P3)] using azide 1h as the common aminating reagent under the standard conditions (Fig. 3; entries 13−29). Like para-vinylphenol 2a containing methyl methacrylate (Fig. 3; entry 8), para-vinylphenol 2b bearing ethyl methacrylate and related derivatives with α-propyl (2c) and even α-octyl (2d) substituents at the vinyl position could effectively undergo the Co(II)-catalyzed dearomative amination with azide 1h, resulting in the production of p-QM-containing α-tertiary amino esters 3hb−3hd bearing different α-ester and α-alkyl groups in high yields with high enantioselectivities (Fig. 3; entries 13−15). In addition to the derivatives with α-alkyl groups, para-vinylphenols containing various functional groups such as cyano (2e), ester (2f), and ketone (2g) were shown to be well tolerated and could serve as effective substrates for the catalytic process, forming the desired α-tertiary amino esters 3he−3hg with the additional functionalities (Fig. 3; entries 16−18). The above results indicate that the catalytic system by [Co(P3)] could homolytically activate aryl azides to chemoselectively react with para-vinylphenols 2a−2g at the desired O–H bond over the allylic C–H bonds as well as over the C=C bond. To further challenge the catalytic system, we designed and synthesized para-vinylphenols 2h−2j, which contain allylic C–H bonds that are also located at the benzylic or another allylic positions. Remarkably, it was found that the Co(II)-based metalloradical system remained highly chemoselective for the enantioselective dearomative 1,7-amination of these substrates, affording the corresponding α-tertiary amino esters 3hh−3hj without aminating the further weakened benzyl allylic and bisallylic C−H bonds or aziridinating the additional C=C bonds (Fig. 3; entries 19−21). Besides ester-containing para-vinylphenols 2a−2j, [Co(P3)] was shown to be equally effective for dearomative amination of para-vinylphenols bearing lactone (2k), ketone (2l), cyclic ketones (2m and 2n), and benzocyclic ketones (2o and 2p) with azide 1h, leading to productive synthesis of enantioenriched α-tertiary amine derivatives 3hk−3hp with diverse carbonyl functionalities (Fig. 3; entries 22−27). The absolute configuration of the newly-generated tetrasubstituted stereogenic center in cyclic compound 3hm was established to be the same (R) configuration as that in acyclic compound 3aa by X-ray crystallography. Moreover, para-vinylphenols derived from natural products, such as (–)-menthol (2q) and cholesterol (2r), could be applied as suitable substrates for [Co(P3)]-catalyzed dearomative amination, leading to high-yielding production of natural product-containing α-tertiary amino esters 3hq and 3hr with high diastereoselectivities (Fig. 3; entries 28 and 29). The results highlight the remarkable capability of the metalloradical catalyst [Co(P3)] to accommodate substrates of various sizes while effectively controlling stereoselectivity of the dearomative 1,7-amination process, irrespective of the existing stereogenic centers. Beyond the dearomative 1,7-conjugate amination, we further demonstrated that [Co(P3)] could catalyze radical 1,5- and 1,9-conjugate amination triggered by O−H HAA, as showcased for the successful generation of α-tertiary amines 3hs and 3ht from the dearomative amination of vinylphenols 2s and 2t, respectively, with azide 1h (Fig. 3; entries 30 and 31).

Mechanistic investigations

Density function theory (DFT) calculations were conducted to examine the details of the catalytic pathways and associated energetics for asymmetric dearomative conjugate amination of vinylphenol 2a with azide 1a by catalyst [Co(P3)] (Fig. 4; see Supplementary Figures 1–3 for details). The calculations reveal the initial formation of intermediate A, which results from the binding of azide 1a to [Co(P3)] through the interplay of metal coordination and multiple H-bonding interactions. Subsequent spin translocation via TS1 homolytically activates azide 1a, generating α-Co(III)-aminyl radical intermediate B with the elimination of dinitrogen as the byproduct. This metalloradical activation, which is highly exergonic by 22.1 kcal/mol, is identified as the rate-determining step due to its relatively higher but readily accessible activation barrier (ΔG^‡_TS1 = 11.4 kcal/mol). As displayed in the spin plot for intermediate B, most of the spin density is transferred from the cobalt center to the nitrogen atom. After binding of α-Co(III)-aminyl radical B with vinylphenol 2a to form intermediate C through a network of non-covalent interactions, the O−H group of the vinylphenol is positioned in close proximity to the nitrogen-centered radical, allowing for a kinetically facile and thermodynamically favorable HAA via TS2 (ΔG^‡_TS2 = 3.6 kcal/mol; ΔG° = −9.7 kcal/mol). This step leads to the generation of ∞-Co(III)-amido phenoxyl radical intermediate D. As shown in the spin plot for intermediate D, the spin density is extensively delocalized across the large π-conjugate system of the para-vinylphenoxyl radical unit. According to the DFT calculations, the final step of radical substitution at the α-position of the ester functionality via ^1,7TS3^Re, which is exergonic by 8.0 kcal/mol, is found to be nearly barrierless, resulting in the formation of 1,7-amination product (R)-3aa, which remains bound to [Co(P3)] as intermediate ^1,7E^Re before dissociating as free amino ester 3aa while regenerating metalloradical catalyst [Co(P3)]. The activation barrier (ΔG^1,7_TS3^Si‡) associated with the formation of the minor enantiomer (S)-3aa is found to be 7.3 kcal/mol higher than that for the preferred formation of (R)-3aa. Although the calculated energy difference (ΔΔG^‡) quantitatively overestimated the experimentally observed enantioselectivity, this significant energy gap clearly rationalizes the experimental preference for the Re-face substitution. Further DFT calculations show that the activation barriers (ΔG^1,n_TS3^‡) associated with the transition states ^1,nTS3 of radical substitution at other possible sites are significantly higher than ΔG^1,7_TS3^Re‡ (ΔG^1,1_TS3^‡ = 29.8 kcal/mol; G^1,3’_TS3^‡ = 15.4 kcal/mol; ΔG^1,3_TS3^‡ = 14.2 kcal/mol; ΔG^1,5_TS3^‡ = 7.8 kcal/mol). These computational results align with the experimentally observed regioselectivity for exclusive 1,7-amination without complications from potential 1,1-, 1,3 and 1,5-amination reactions. As illustrated by the non-covalent interaction (NCI) plots of the DFT-optimized transition state structure ^1,7TS3^Re, it is the multiple attractive non-covalent interactions within the pocket of the catalyst that together hold the delocalized para-vinylphenoxyl radical in the proper orientation and conformation. This precise arrangement positions the C-7 site of the para-vinylphenoxyl radical in close proximity to the nitrogen center of the Co(III)-amido complex, facilitating the highly regioselective C–N bond formation while also enabling effective asymmetry induction. Among different types of non-covalent interactions in the transition state ^1,7TS3^Re, it is worth highlighting the unique three-point hydrogen bonding interactions between the C=O group of the para-vinylphenoxyl radical and the N–H, C–H, N–H groups in the Co(III)-amido complex, which is supported by D₂-symmmetic chiral amidoporphyrin P3 (Fig. 4). These specific interactions play a crucial role in stabilizing the transition state, guiding regioselectivity, and ensuring enantioselectivity in the catalytic process.

Fig. 4. DFT study of catalytic pathways for asymmetric dearomative conjugate amination of vinylphenol 2a with azide 1a by metalloradical catalyst [Co(P3)].

The computed free energy profile (ΔG° at 298 K in kcal/mol) illustrates a stepwise radical mechanism, including (i) Metalloradical activation of azide 1a by [Co(P3)], proceeding through intermediate A to generate α-Co(III)-aminyl radical intermediate B via transition state TS1; (ii) Hydrogen atom abstraction (HAA) from the O–H bond of vinylphenol 2a by intermediate B, passing through intermediate C to afford ∞-Co(III)-amido phenoxyl radical intermediate D via transition state TS2; (iii) Radical substitution (RS) within intermediate D, involving the phenoxyl radical and the Co(III)-amido complex to form intermediate ^1,7E^Re via transition state ^1,7TS3^Re; and (iv) Product liberation from intermediate ^1,7E^Re to release the final α-tertiary amino ester (R)-3aa. Insets: Relative Gibbs free energy barriers (ΔG^‡) for competing radical substitution pathways at different positions (upper right); Spin density plots for intermediates B and D, respectively (middle and bottom left); Non-covalent interaction (NCI) plots and schematic illustration of transition state structure ^1,7TS3^Re (bottom center and right). All calculations were carried out at the SMD(Benzene)-BP86-D3(BJ)/def2TZVP//BP86/def2SVP level of theory.

In addition to the computational studies, significant efforts were devoted to detecting and probing the radical intermediates involved in the stepwise radical mechanism of the Co(II)-catalyzed 1,7-conjugate amination (Fig. 5). To directly detect the α-Co(III)-aminyl radical intermediate I, an isotropic X-band electron paramagnetic resonance (EPR) spectrum was recorded at ambient temperature for the reaction mixture of [Co(P3)] with azide 1a in benzene without the vinylphenol substrate (Fig. 5a). The spectrum displays distinct signals characteristic of α-Co(III)-aminyl radicals.(32) Both the isotopic g-value of 2.05687 and the hyperfine couplings of the observed peaks are consistent with the generation of α-Co(III)-aminyl radical I_[Co(P3)]/1a from metalloradical activation of azide 1a by [Co(P3)]. Upon adding vinylphenol 2a into the reaction mixture, the relatively weak and broad signals associated with I_[Co(P3)]/1a disappeared, being replaced by strong and sharp signals. The new EPR signals, with an isotopic g-value of 2.00271, indicates the generation of the para-vinylphenoxyl radical ^aII_2a as part of ∞-Co(III)-amido vinylphenoxyl radical intermediate II_{[Co(P3]/1a/2a}, along with Co(III)-amido complex ^bII_[Co(P3)]/1a, through HAA from the O–H bond of vinylphenol 2a by the initially-generate α-Co(III)-aminyl radical I_[Co(P3)]/1a. Consistent with the spin delocalization in vinylphenoxyl radical ^aII_2a, the well-defined EPR spectrum could be nicely simulated by taking into account its various resonance structures (see Supplementary Figures 4 and 5 for details). To further probe the O–H HAA step and the resulting delocalized radical intermediate, cyclopentanone-based bis-vinylphenol 2u was synthesized and employed as the substrate in the catalytic reaction with azide 1h by [Co(P3)] (Fig. 5b). It was anticipated that the ethano-bridged bisallylic oxyally biradical ^aII_2u would form via double HAA from the two O‒H bonds of substrate 2u by the initially-generated α-Co(III)-aminyl radical I_[Co(P3)]/1h. Instead of undergoing radical substitution for C–N bond formation, biradical ^aII_2u was expected to proceed via intramolecular radical recombination, yielding the corresponding ethano-bridged cyclopropanone 4 through C–C bond formation.(46) Although no amination product was observed, the catalytic reaction resulted in the isolation of product 5 in 91% yield, along with tetrafluoroaniline 1h’ as the byproduct (Fig. 5b). X-ray crystallography confirmed that compound 5 is a tricyclic diketone bearing four p-QMs, which was evidently formed through the dimerization of biradical ^aII_2u. Interestingly, when compound 5 was dissolved in benzene at room temperature, notable EPR signals at the isotopic g-value of 2.00489 were detected, indicating the generation of biradical ^aII_2u from homolytic cleavage of the two elongated weak C(4°)–C(4°) bonds (see Supplementary Figure 6 for details). In a separate experiment to probe the stepwise radical mechanism, 2,2,2-trichloroethoxycarbonyl azide (TrocN₃) was used in place of aryl azides in the catalytic reaction of vinylphenol 2a with [Co(P2)] as the catalyst (Fig. 5c). Rather than forming the corresponding amination product, the catalytic reaction led to the isolation of compound 7 in 90% yield. X-ray crystallography revealed that compound 7 possesses a spiro[6.6]bicyclic structure, featuring an additional quaternary carbon bonded to the shared quaternary carbon. Evidently, compound 7 is a dimer of the dehydrogenation compound 6 resulted from a Diels-Alder reaction. Compound 6 likely originated from secondary HAA from the substantially weakened α-C–H bond of the initially-generated para-vinylphenoxyl radical ^aII_2a, a competitive process that became favored over the radical substitution for C–N bond formation. Analogous to TrocN₃, arylsulfonyl azides were also found to exclusively promote the dehydrogenation pathway under the standard reaction condition, completely suppressing the formation of amination products.

Fig. 5. Mechanistic studies on Co(II)-based metalloradical system for 1,7-conjugate amination of vinylphenols with azides.

a, Detection of α-Co(III)-aminyl radical and ∞-Co(III)-amido vinylphenoxyl radical intermediates by EPR. b, Probing of delocalized para-vinylphenoxyl radical intermediate through formation and homodimerization of bisallylic oxyally biradical. c, Probing of delocalized para-vinylphenoxyl radical intermediate through formation and Diels-Alder dimerization of conjugate diene.

Synthetic applications

On top of being a common structural motif in many biologically active natural products, para-quinone methides (p-QMs) are known for their diverse chemical reactivities, including their effectiveness as Michael acceptors in conjugate addition reactions with nucleophiles.(47) To demonstrate the synthetic applications of the p-QM-containing α-tertiary amino esters produced from the Co(II)-catalyzed 1,7-conjugage amination, a series of further synthetic transformations were performed using 3ha as a model compound (Fig. 6). It was shown that α-tertiary amino ester 3ha could serve as a versatile building block to access highly functionalized chiral amino esters 8 while creating additional stereogenic center adjacent to the original one through 1,6-conjugate addition reactions with a variety of nucleophiles (Fig. 6a). For example, electron-rich aromatic compounds like indole could act as a carbon-based nucleophile to react with 3ha in the presence of BF₃·Et₂O as catalyst, affording β-aryl-α-tertiary amino ester 8a in excellent yield, although with moderate diastereoselectivity. Additionally, 3ha could undergo Sc(OTf)₃-catalyzed 1,6-conjugate addition with 1,1-diphenylethylene, forming β-vinyl-α-tertiary amino ester 8b in high yield with good diastereoselectivity. Under Cu(OTf)₂ catalysis, 3ha could also react with both nitrogen-based nucleophile TMSN₃ and oxygen-based nucleophile ethanol, leading to high-yielding syntheses of β-azido-α-tertiary amino ester 8c and β-ethoxy-α-tertiary amino ester 8d, respectively, despite with moderate diastereoselectivities. Similarly, triflic acid-catalyzed 1,6-conjugate addition of 3ha with sulfur-based nucleophile PhSH produced β-thiol-α-tertiary amino ester 8e in high yield with excellent diastereoselectivity for the newly-created adjacent stereogenic center. Furthermore, copper-catalyzed 1,6-conjugate addition of 3ha with B₂(pin)₂ provided an attractive route for the synthesis of β-boryl-α-tertiary amino ester 8f in high yield, albeit with moderate diastereoselectivity. Moreover, reduction of 3ha using NaBH₄ resulted in the formation α-tertiary amino ester 8g in quantitative yield. In essence, the combination of the 1,6-conjugate addition with the Co(II)-catalyzed 1,7-conjugate amination enables the difunctionalization of the trisubstituted alkene units of the starting vinylphenols 2, facilitating the stereoselective synthesis of highly-functionalized chiral amino esters 8 with creation of two adjacent stereogenic centers. As a demonstration for the synthesis of α-methyltyrosine derivative, it was shown that the two ortho-tert-butyl groups in the phenol unit of the α-tertiary amino ester 8g could be readily removed through AlCl₃-mediated de-tert-butylation, yielding α-methyltyrosine methyl ester 9 (Fig. 6b). Without further purification, ester 9 was directly subjected to hydrolysis, affording metirosine derivative 10, an inhibitor of the tyrosine hydroxylase enzyme, in 89% overall yield from the starting material 8g. In a separate demonstration for the synthesis of α-tertiary primary amino acid derivative (Fig. 6c), N-aryl amino esters 3ma, which was synthesized from [Co(P6)]-catalyzed 1,7-conjugate amination of vinylphenol 2a with 2,6-difluoro-4-methyoxyphenyl azide (1m) in 91% yield with 84% ee, was shown to undergo CAN oxidation followed by in situ acid hydrolysis, furnishing the unprotected α-tertiary primary amino ester 11. However, the low yield of this de-N-arylation process suggests that further optimization is needed.

Fig. 6. Synthetic applications of resulting α-amino acid derivatives from 1,7-conjugate amination of vinylphenols with azides.

a, Further transformations via 1,6-conjugate addition of p-QM functionality with diverse nucleophiles. b, Synthesis of α-methyltyrosine derivative through de-tert-butylation and ester hydrolysis. c, Synthesis of α-tertiary primary amino acid derivative by deprotection of N-aryl group.

Conclusions

In summary, we have successfully developed the first catalytic system via Co(II)-based MRC that is highly effective for enantioselective radical dearomative conjugate amination. Facilitated by an optimal D₂-symmetric chiral amidoporphyrin ligand, the Co(II)-based metalloradical system can homolytically activate aryl azides for dearomative 1,7-conjugate amination of para-vinylphenols with concurrent control of chemoselectivity, regioselectivity, and enantioselectivity. This Co(II)-catalyzed radical amination, which operates at room temperature with low catalyst loading without the need for oxidants or other additives, provides an attractive methodology for the stereoselective synthesis of valuable chiral α-tertiary amino acid derivatives, while generating N₂ as the sole byproduct. We have demonstrated that the resulting amino acid esters, bearing a para-quinone methide (p-QMs) functionality, can undergo 1,6-conjugate addition with various nucleophiles, enabling the stereoselective synthesis of highly-functionalized α-tertiary amino derivatives bearing two adjacent stereogenic centers. Unlike Co(II)-based C–H amination, this dearomative conjugate amination proceeds through an extensively delocalized 4-vinylphenoxyl radical intermediate, formed via uncommon HAA from O−H bonds of phenols by the initially-generated α-Co(III)-aminyl radicals, followed by regioselective radical substitution to achieve remote 1,7-amination. The success in concurrently controlling multiple selectivities in this challenging radical process is attributed to the catalyst innovation by leveraging a network of non-covalent weak interactions within the pocket-like environment of the D₂-symmetric chiral amidoporphyrin ligand platform. Along with our preliminary demonstration of dearomative 1,5- and 1,9-conjugate amination, we anticipate that this work will inspire further research endeavors aimed at exploring the untapped synthetic potential of extensively delocalized radical systems.

Methods

In a general procedure for Co(II)-catalyzed enantioselective dearomative conjugate amination, an oven-dried Schlenk tube was charged with Co(II)-metalloradical catalyst [Co(P3)] (1 mol %, 1.4 mg) and vinylphenol 2 (0.10 mmol, 1.0 equiv). The Schlenk tube was evacuated and backfilled with nitrogen three times. After replacing the Teflon screw cap with a rubber septum, aryl azide 1 (0.15 mmol, 1.5 equiv) and trifluorotoluene (0.5 mL) were added via a gas-tight syringe. The resulting mixture was then purged with nitrogen for 30 seconds, resealed with the Teflon screw cap, and stirred at room temperature for 6 hours. Upon completion, the reaction mixture was concentrated by rotary evaporation under reduced pressure. The residue was subject to purification by flash chromatography on silica gel to afford the pure dearomative amination product 3.

Data availability

Crystallographic data for the structures reported in this article have been deposited at the Cambridge Crystallographic Data Centre, under deposition numbers CCDC 2411816 (3aa), 2411818 (3hm), 2411819 (5) and 2411820 (7). Copies of the data can be obtained free of charge via https://www.ccdc.cam.ac.uk/structures/. All other data that support the findings of this study, which include experimental procedures and compound characterization, are available within the paper and its Supplementary Information. Data is available from the corresponding authors upon request.