Asymmetric Radical Bicyclization for Stereoselective Construction of Tricyclic Chromanones and Chromanes with Fused Cyclopropanes

Abstract

Asymmetric radical bicyclization processes have been developed via metalloradical catalysis (MRC) to stereoselectively construct chiral chromanones and chromanes bearing fused cyclopropanes. Through optimization of a versatile D₂-symmetric chiral amidoporphyrin ligand platform, a Co(II)-metalloradical system can homolytically activate both diazomalonates and α-aryldiazomethanes containing different alkene functionalities under mild conditions for effective radical bicyclization, delivering cyclopropane-fused tricyclic chromanones and chromanes, respectively, in high yields with excellent control of both diastereoselectivities and enantioselectivities. Combined computational and experimental studies, including the electron paramagnetic resonance (EPR) detection and 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO) trapping of key radical intermediates, shed light on the working details of the underlying stepwise radical mechanisms of the Co(II)-catalyzed bicyclization processes. The two catalytic radical processes provide effective synthetic tools for stereoselective construction of valuable cyclopropane-fused chromanones and chromanes with newly generated contiguous stereogenic centers. As a specific demonstration of synthetic application, the Co(II)-catalyzed radical bicyclization has been employed as a key step for the first asymmetric total synthesis of the natural product (+)-Radulanin J.

Graphical Abstract

INTRODUCTION

One-electron radical reactions have been considerably less utilized than two-electron polar reactions in organic synthesis even though radical chemistry is known for being both rich in fundamental reactivities and attractive with practical attributes. This lag in development is attributed to the long-standing challenges associated with the control of reactivity and stereoselectivity of highly active free radicals, especially regarding enantioselectivity as inversion between two prochiral faces of radical intermediates is typically facile.¹ In order to harness the untapped potential of homolytic radical chemistry for stereoselective construction of molecular structures, it calls for formulation of new concepts and establishment of different strategies to command the control of both generation and reaction of active radical intermediates.² Within this context, metalloradical catalysis (MRC), which exploits metal-centered radicals as one-electron catalysts for homolytic activation of substrates to generate metal-entangled organic radicals as key intermediates, has emerged as a powerful catalytic approach to achieving the control of both reactivity and stereoselectivity in radical reactions.^3–5 Cobalt(II) complexes of porphyrins [Co(Por)], as a family of stable 15e-metalloradicals, have demonstrated an unusual aptitude for homolytic activation of diazo compounds to generate α-Co(III)-alkyl radical intermediates, a class of metal-bonded, carbon-centered organic radicals that can undergo common types of radical reactions but in a catalytic and controlled manner.⁶ With the introduction of D₂-symmetric chiral amidoporphyrins as the ligand platform,⁷ Co(II)-based metalloradical catalysts [Co-(D₂-Por*)] have proven highly effective in homolytically activating different classes of diazo compounds for asymmetric intermolecular cyclopropanation of various alkenes via stepwise radical mechanism.⁸ More recently, it was shown that both diazomalonates (acceptor/acceptor-substituted diazo compounds) and in situ-generated α-aryldiazomethanes (donor/hydrogen-substituted diazo compounds) could function as potent metalloradicophiles for Co(II)-based MRC to cyclopropanate alkenes in an intermolecular fashion, delivering chiral cyclopropanes with effective control of stereoselectivities.^8h–j Prompted by these recent findings, we were intrigued to the possibility of refashioning the catalytic transformations into intramolecular processes for radical bicyclization with the use of diazomalonates and α-aryldiazomethanes that contain dangling alkene functionalities.

We were particularly interested in exploring the potential application of Co(II)-based MRC for asymmetric radical bicyclization of 2-vinylaryl diazomalonates 1 and in situ-generated α-(2-(allyloxy)aryl)diazomethanes 3′ to stereo-selectively construct the structurally related cyclopropane-fused chromanones 2 and chromanes 4, respectively, in view of their biological importance (Scheme 1).⁹ While metalloradical activation of both diazomalonates 1 and α-aryldiazomethanes 3′ was foreseen to proceed readily, it was uncertain whether the intramolecular radical addition of the initially generated α-Co(III)-malonyl radical intermediate I (Scheme 1A) and α-Co(III)-benzyl radical intermediate III (Scheme 1B) could be negatively impacted by the steric hinderance associated with the substituents. Mechanistically, there are two intramolecular pathways for both intermediates I and III to proceed the following step of radical addition: 6-exo-trig versus 7-endo-trig. It was unclear what would be the preferred pathway to undergo the radical cyclization. Conceivably, this regioselectivity issue would closely connect with the control of enantioselectivity for the overall catalytic process since the radical cyclization creates the first stereogenic center. To achieve the control of enantioselectivity, it was foreseeable that α-Co(III)-malonyl radical I (Scheme 1A) and α-Co(III)-benzyl radical III (Scheme 1B) might require different ligand environments in order to discriminate the two prochiral faces of the respective alkene unit. Furthermore, it was unsure if the resulting γ-Co(III)-alkyl radicals II and IV could smoothly undergo the desired 3-exo-tet radical cyclization to produce chromanones 2 and chromanes 4, respectively, while creating the contiguous stereogenic centers on the fused cyclopropane ring with stereoselectivity, considering the highly strained nature of the tricyclic structures. We envisioned to address these and related challenges through catalyst innovation by designing a D₂-symmetric chiral amidoporphyrin ligand with proper steric, electronic, and chiral environment that promotes the radical process while governing the stereochemical course. If successfully implemented, it would lead to the invention of new synthetic tools for stereoselective construction of chiral cyclopropane-fused chromanones and chromanes, which occur as the common structures in many biologically active molecules (see Figure S1 for examples).

Scheme 1.

Proposal for Stereoselective Construction of Tricyclic Structures through Radical Bicyclization via Co(II)-MRC

Metal-catalyzed intramolecular cyclopropanation of diazo compounds offers a potentially attractive unimolecular approach for stereoselective construction of tricyclic chromanones and chromanes. Among previous reports on both catalytic and thermal processes,^10–13 the intramolecular process has been less developed in comparison with the intermolecular counterpart, let alone with effective control of enantioselectivity. This significant underdevelopment is presumably ascribed to the highly strained transition state associated with existing catalytic systems that typically proceed by concerted mechanisms involving electrophilic metal-locarbene intermediates. To the best of our knowledge, Wood and co-workers reported the only example for stereoselective synthesis of one cyclopropane-fused chromanone derivative in 89% yield with 45% ee through asymmetric intramolecular cyclopropanation using a Rh₂-based catalyst.^12a In most other cases, the construction of chiral tricyclic chromanones and chromanes with fused cyclopropanes relied on stereoselective intermolecular cyclopropanation of pre-formed bicyclic structures of chromanones and chromanes. For example, Feng and co-workers reported Ni-catalyzed inter-molecular cyclopropanation of coumarins with phenyliodonium ylide for asymmetric synthesis of two tricyclic chromanones.^12b Additionally, Sakamoto and co-workers synthesized two enantioenriched cyclopropane-fused chromanones through diastereoselective intermolecular cyclopropanation of prepared chiral coumarins with sulfur ylides.^12c Furthermore, Zeng and co-workers described asymmetric synthesis of tricyclic chromanones by Lewis base-mediated intermolecular cyclopropanation of coumarin derivatives with 2-bromoacetates.^12d,e Meanwhile, Yuan, Zhao, and co-workers demonstrated the application of intermolecular cyclopropanation by organocatalysts for asymmetric synthesis of cyclopropane-fused chromanones containing additional spirooxindoles.^12f More recently, Yang and co-workers employed a chiral organocatalyst for asymmetric synthesis of chromanones through intermolecular reactions of α,β-unsaturated aldehydes with 2,4-dinitrobenzyl chloride.^12g Although these previous reports disclosed different synthetic approaches, stereoselective construction of cyclopropane-fused chromanones via asymmetric intramolecular cyclopropanation of diazo compounds has been largely undeveloped. There has been even much less development for stereoselective construction of cyclopropane-fused chromanes. In fact, there has been no previous report on catalytic systems for asymmetric intramolecular cyclopropanation of diazo compounds for stereoselective synthesis of tricyclic chromanes containing fused cyclopropanes. Even for asymmetric intermolecular cyclopropanation, only two examples were previously reported.¹³ Tang and co-workers demonstrated the synthesis of one chiral tricyclic chromane by Cu-catalyzed intermolecular cyclopropanation of 2H-chromene with ethyl α-nitrodiazoacetate.^13a Recently, Charette and co-workers reported the preparation of another chiral cyclopropane-fused chromane by intermolecular chlorocyclopropanation of a 2H-chromene-derived allylic alcohol with dichloromethylzinc carbenoid.^13b While this manuscript was in the process of submission, Zhuo and co-workers reported an innovative Mo(salen)-based catalytic system for asymmetric synthesis of tricyclic chromanes by deoxygenative intra-molecular cyclopropanation of 1,2-dicarbonyl compounds using tertiary phosphines as the terminal reductants at elevated temperature.¹⁴ As a new application of metalloradical catalysis (MRC), we herein report the development of a Co(II)-based metalloradical system that is highly effective for asymmetric radical bicyclization of both 2-vinylaryl diazomalonates and in situ-generated α-(2-(allyloxy)aryl)diazomethanes. We describe a significant ligand effect on the Co(II)-catalyzed bicyclization process, leading to the identification of two different D₂⁻symmetric chiral amidoporphyrin ligands for the two catalytic intramolecular cyclopropanation reactions. Supported by the two optimal ligands, the Co(II)-based catalytic system is applicable to wide-ranging diazomalonates and α-aryldiazomethanes, leading to stereoselective construction of cyclopropane-fused tricyclic chromanones and chromanes in high yields with excellent control of both diastereoselectivities and enantioselectivities. Additionally, we present experimental and computational mechanistic studies that offer the insight into the working details of the underlying stepwise radical mechanism for the Co(II)-based catalytic system.

RESULTS AND DISCUSSION

Reaction Development.

Our effort started with the investigation of the radical bicyclization reaction of 2-vinylphenyl diazomalonate 1a as the model substrate using Co(II)-based metalloradical catalysts (Table 1). It was shown that the simple catalyst [Co(P1)] (P1 = 5,10,15,20-tetraphenylporphyrin) was incapable of activating diazomalonate 1a as no reaction was observed (Table 1; entry 1). When an achiral metalloradical catalyst [Co(P2)] (P2 = 3,5-Di^tBu-IbuPhyrin)¹⁵ was applied, the desired cyclopropane-fused tricyclic chromanone 2a was observed in 8% yield as the allowed cis-ring conjunction (Table 1; entry 2). This encouraging result promoted us to evaluate asymmetric induction during the radical bicyclization with the use of a first-generation chiral metalloradical catalyst [Co(P3)] (P3 = 3,5-Di^tBu-ChenPhyrin).^7a Excitingly, [Co(P3)] could enable productive formation of the cis-fused tricyclic chromanone 2a in good yield (50%) with high control of enantioselectivity (89% ee) (Table 1; entry 3). It was noted that the enantioselectivity could be even further enhanced to 96% ee by changing the solvent from toluene to tert-butyl methyl ether (Table 1; entry 4). When the catalyst loading was increased to 3 mol %, the catalytic reaction in tert-butyl methyl ether delivered cis-fused 2a in high yield (88%) with excellent enantioselectivity (96% ee) (Table 1; entry 5). In view of this positive outcome, we then tested metalloradical catalyst [Co(P3)] for potential activation of α-(2-((E)-cinnamyloxy)-phenyl)diazomethane 3a′ as the model substrate for asymmetric radical bicyclization. To our delight, [Co(P3)] was indeed able to catalyze the radical bicyclization reaction of 3a′ that was in situ generated from sulfonyl hydrazone 3a in the presence of Cs₂CO₃, delivering the desired cyclopropane-fused tricyclic chromane 4a in high yield (86%) as the allowed cis-ring conjunction with excellent diastereoselectivity (exo/ endo = 99:1) and promising enantioselectivity (−34% ee) (Table 1; entry 6). When a second-generation metalloradical catalyst [Co(P4)] (P4 = 3,5-Di^tBu-QingPhyrin)^7c was used, however, cis-fused 4a was formed in a relatively lower yield (80%) with diminished enantioselectivity (23% ee) but still as the sole exo-diastereomer (Table 1; entry 7). It is worth mentioning that ligand P4 differs from ligand P3 by replacing one of the two methyl groups in P3 with a phenyl group in each of the four chiral amide units. Interestingly, subsequent use of an analogous chiral metalloradical catalyst [Co(P5)] (P5 = 2,6-DiMeO-QingPhyrin)¹⁶ bearing 2,6-dimethoxyphenyl groups instead of 3,5-di-tert-butylphenyl groups as the 5,15-diaryl substituents on the porphyrin ligand gave rise to cis-4a as the single exo-diastereomer in increased yield (90%) with elevated enantioselectivity (42% ee) (Table 1; entry 8). These results signify a positive effect of rigidification in a ligand environment on the control of stereoselectivities for the catalytic radical process. Aiming at further improving the catalytic system, we designed a new metalloradical catalyst [Co(P6)] (P6 = 2,6-DiMe-QingPhyrin) with an even more rigid ligand environment by installing 2,6-dimethylphenyl groups in replacing the 2,6-dimethoxyphenyl groups as the achiral meso-aryl units. Gratifyingly, [Co(P6)] could catalyze the formation of cis-4a with dramatically enhanced enantioselectivity (81% ee) without significantly affecting the high diastereoselectivity (exo/endo = 96:4) albeit in a much lower yield (56%) (Table 1; entry 9). Delightfully, changing the solvent to trifluorotoluene resulted in considerable improvements in catalytic reactivity as well as stereoselectivities, affording cis-4a in 94% yield as the only exo-diastereomer with 90% ee (Table 1; entry 10).

Table 1.

Ligand Effect on Co(II)-Based Catalytic Systems for Asymmetric Radical Bicyclization of Diazo Compounds

^aConducted with 1a (0.10 mmol) using [Co(Por)] (2 mol %) at room temperature for 24 h.

^bConducted with 3a (0.10 mmol) using [Co(Por)] (2 mol %) in the presence of Cs₂CO₃ (0.20 mmol) at room temperature for 16 h.

^cIsolated yields.

^dDiastereomeric ratio (dr) between exo- and endo-isomers determined by ¹H NMR analysis of the reaction mixture.

^eEnantiomeric excess (ee) of major diastereomer determined by chiral HPLC.

^fUsing 3 mol % [Co(Por)]. TBME: tert-butyl methyl ether. Tris (trisyl): 2,4,6-triisopropylbenzenesulfonyl.

Substrate Scope.

Under the optimized catalytic conditions, the scope of asymmetric radical bicyclization by [Co(P3)] was evaluated with different 2-vinylaryl diazomalonates 1 as the substrates (Table 2A). Similar to 2-vinylphenyl diazomalonate 1a, 2-vinylaryl diazomalonates 1b and 1c, which contain an electron-donating −OMe substituent at 6- and 4-positions of the aryl group, respectively, were shown to be effective substrates for [Co(P3)]-catalyzed radical bicyclization, delivering cis-fused tricyclic chromanones 2b and 2c in high yields with excellent enantioselectivities (Table 2; entries 2 and 3). The absolute configurations of cyclopropane-fused tricyclic chromanone 2c was determined as (1S,2S) by X-ray crystallography. In the same way, 2-vinylaryl diazomalonate 1d containing a methyl group at the 4-position of the aryl group was demonstrated as a suitable substrate for the radical bicyclization as well, furnishing tricyclic chromanone 2d in high yield as the cis-fused diastereomer with high enantioselectivity (Table 2; entry 4). Additionally, the [Co(P3)]-catalyzed asymmetric radical bicyclization could well be applied to 2-vinylaryl diazomalonates with a halogen atom (−F, −Cl, and −Br) at various positions of the aryl group as exemplified for the high-yielding reactions of halogenated diazomalonates 1e–1h to form chromanones 2e–2h as the cis-fused diastereomers with high enantioselectivities (Table 2; entries 5–8). Besides monohalogenated substrates, it was found that the Co(II)-based metalloradical system could also effectively catalyze radical bicyclization reactions of dihalogenated 2-vinylaryl diazomalonates 1i–1k bearing the two halogen atoms (−Cl, −Br, and −I) at 4- and 6-positions of the aryl group, affording the corresponding cis-fused tricyclic chromanones 2i–2k in good to high yields with high enantioselectivities (Table 2; entries 9–11). In addition to the above diazomalonates 1a–1k containing monosubstituted alkenes, it was found that 2-vinylaryl diazomalonate derivatives bearing 1,2-disubstituted alkenes could also serve as effective substrates for Co(II)-based asymmetric radical bicyclization as demonstrated with the productive reactions of diazomalonates 1l and 1m by [Co(P4)], generating corresponding cis-fused tricyclic chromanones 2l and 2m in high yields with varied enantioselectivities (Table 2; entries 12 and 13).

Table 2.

Scope of Co(II)-Based Catalytic Systems for Asymmetric Radical Bicyclization of Diazo Compounds

^aConducted with 1 (0.10 mmol) using the [Co(P3)] catalyst (3 mol %) in tert-butyl methyl ether at room temperature for 24 h; isolated yields; only the cis-ring junction product formed; enantiomeric excess (ee) of major isomer determined by chiral HPLC.

^bConducted with 3 (0.10 mmol) using the [Co(P6)] catalyst (2 mol %) in the presence of Cs₂CO₃ (0.20 mmol) in trifluorotoluene at room temperature for 16 h; isolated yields;only the cis-ring junction product formed; diastereomeric ratio (dr) between exo- and endo-isomers determined by ¹H NMR analysis of the reaction mixture; enantiomeric excess (ee) of major isomer determined by chiral HPLC.

^cAt 40 °C.

^dAbsolute configuration determined by X-ray crystallography.

^eUsing [Co(P4)] (3 mol %).

^fUsing [Co(P6)] (4 mol %).

Likewise, the scope of the [Co(P6)]-based catalytic system for asymmetric radical bicyclization was examined using different α-(2-(allyloxy)aryl)diazomethanes 3′ in situ generated from the corresponding sulfonyl hydrazones under the optimized conditions (Table 2B). Like α-(2-((E)-cinnamyloxy)phenyl)diazomethane 3a′ (Table 2; entry 14), its analogues containing substituents with varied electronic and steric properties on the phenyl group of the cinnamyl unit could be effectively used as metalloradicophiles for the Co(II)-based metalloradical system. For example, α-aryldiazomethanes 3b′ and 3c′ with both electron-donating (−OMe) and electron-withdrawing (−CN) groups could serve as suitable substrates for Co(II)-based asymmetric bicyclization, resulting in the high-yielding formation of chromanes 4b and 4c as the allowed cis-ring conjunction with both excellent diastereoselectivities and enantioselectivities (Table 2; entries 15 and 16). Similarly, the [Co(P6)]-catalyzed radical bicyclization process was applicable to α-aryldiazomethanes 3d′–3f′ bearing halogen atoms (−F, −Cl, and −Br), affording cis-fused tricyclic chromanes 4d–4f in high yields with excellent diastereoselectivities and high enantioselectivities (Table 2; entries 17–19). The absolute configurations of cyclopropane-fused chromanes 4e and 4f were determined as (1S,2R,7S) by X-ray crystallography (Table 2; entries 18 and 19). Moreover, the Co(II)-based metalloradical system proved to be similarly effective for asymmetric radical bicyclization of α-(2-((E)-cinnamyloxy)aryl)diazomethanes bearing different substituents on the α-aryl group. For example, α-aryldiazomethanes 3g′–3j′ with an electron-donating group (−OMe) and halogen atoms (−F, −Cl, and −Br) on the α-aryl rings could be efficiently bicyclized by [Co(P6)] to construct cis-fused tricyclic chromanes 4g–4j in high yields with excellent diastereoselectivities and good enantioselectivities (Table 2; entries 20–23). Furthermore, this Co(II)-based metalloradical system was shown to be effective for asymmetric radical bicyclization of α-(2-(allyloxy)aryl)diazomethanes containing trisubstituted alkenes as exemplified with the catalytic reaction of α-aryldiazomethane 3k′ by [Co(P6)], delivering cis-fused tricyclic chromane 4k in good yield with high control of both diastereoselectivity and enantioselectivity (Table 2; entry 24).Given that the resulting tricyclic cyclopropane-fused chromanones and chromanes occur as common structures in many biologically active molecules, including natural products (Figure S1), we have explored the synthetic application of the Co(II)-based metalloradical system for stereoselective synthesis of target molecules. To this end, we have successfully employed the Co(II)-catalyzed radical bicyclization as a key step for the first asymmetric total synthesis of the natural product (+)-Radulanin J, which was isolated from liverwort Radula javanica.^9b Starting from the commercially available 3,5-dimethoxybenzaldehyde, the relatively complex sulfonyl hydrazone S8 was synthesized in high yield through a multistep route (Table 2C; see the Supporting Information for details). Excitingly, the in situ-generated α-(2-(allyloxy)-aryl)diazomethane from hydrazone S8 could be employed as a suitable substrate for catalytic asymmetric radical bicyclization by [Co(P6)], leading to the asymmetric construction of (+)-Radulanin J in 51% yield with 43% ee. The synthetic (+)-Radulanin J is both spectroscopically and optically matched with the isolated natural product.^9b In summary, we have completed the first asymmetric total synthesis of (+)-Radulanin J from 3,5-dimethoxybenzaldehyde through six steps in overall 22% yield with 43% ee (Table 2C; see the Supporting Information for details).

Mechanistic Studies.

Combined experimental and computational studies were performed to shed light on the underlying stepwise radical mechanism of the Co(II)-based metalloradical system for the two bicyclization processes. The density functional theory (DFT) calculations were carried out to elucidate the working details of the catalytic pathways and associated energetics for asymmetric bicyclization of diazomalonate 1a by the optimal catalyst [Co(P3)] (A) (Scheme 2A; see the Supporting Information for details). The DFT calculations reveal the binding of diazo compounds 1a by catalyst A through a network of noncovalent attractions, including multiple H-bonds and π-interactions. This binding event, which is exergonic by 15.6 kcal/mol, results in the formation of intermediate B in which these noncovalent interactions position the α-carbon atom of 1a in close proximity to the Co(II)-metalloradical center of catalyst A (C···Co: ~2.673 Å) for further activation. The subsequent activation of 1a by the catalyst [Co(P3)], which is exergonic by 5.6 kcal/mol, generates α-Co(III)-malonyl radical intermediate C with the release of dinitrogen as the byproduct. The metalloradical activation, which is associated with a relatively high but accessible energy barrier (ΔG_TS1^‡ = 16.1 kcal/mol), is found to be the rate-determining step. According to the DFT calculations, intermediate C proceeds facile 6-exo-trig cyclization through intramolecular radical addition to the C=C bond with a very low activation barrier (ΔG_TS2^‡ = 2.5 kcal/mol), which is also exergonic by 12.9 kcal/mol, delivering γ-Co(III)-alkyl radical D. As illustrated in the optimized structure of TS2, the low barrier is attributed to the multiple attractive H-bonds and π-interactions. By contrast, the alternative pathway of 7-endo-trig radical cyclization of intermediate C is found to have a significantly higher activation barrier (ΔG_TS2′^‡ = 7.9 kcal/ mol) although the formation of γ-Co(III)-alkyl radical D′ is even more exergonic (ΔG_D′⁰ = −22.2 kcal/mol). The large difference in activation barriers between the two intramolecular radical addition pathways is presumably due to the less ring strain of the 6-membered TS2 associated with 6-exo-trig radical cyclization than that of the 7-membered TS2′ associated with 7-endo-trig radical cyclization. The DFT computations indicate that γ-Co(III)-alkyl radical D then undergoes very facile 3-exotet cyclization through intramolecular radical substitution with an exceedingly low activation barrier (ΔG_TS3^‡ = 1.8 kcal/mol), forming cyclopropane-fused tricyclic chromanone 2a as the final product while regenerating the metalloradical catalyst [Co(P3)].

Scheme 2.

Mechanistic Studies on Co(II)-Catalyzed Bicyclization Reactions of Diazomalonates and α-Aryldiazomethanes

The same type of computational studies was also conducted to examine the catalytic mechanism of asymmetric radical bicyclization of α-aryldiazomethane 3a′ by the optimal catalyst [Co(P6)] (E) (Scheme 2B; see the Supporting Information for details). The results from the DFT calculations unveil a similar catalytic pathway for the construction of cyclopropane-fused tricyclic chromane 4a that consists of four fundamental steps involving three key intermediates. First, the binding of diazo compounds 3a′ by catalyst E results in the initial formation of intermediate F, which is exergonic by 5.5 kcal/ mol. In intermediate F, the α-carbon atom of 3a′ is located only 2.262 Å away from the Co(II)-metalloradical center of catalyst E. Second, the further activation of the bound diazo compound 3a′ by the catalyst [Co(P6)] generates α-Co(III)-benzyl radical intermediate G, which is exergonic by 18.6 kcal/ mol. Although it has a relatively low activation barrier (ΔG_TS4^‡ = 9.2 kcal/mol), the step of metalloradical activation turns out to be the rate-determining step as the following steps have even lower activation barriers. Third, intermediate G proceeds facile 6-exo-trig cyclization through intramolecular radical addition to the C=C bond with a low activation barrier (ΔG_TS5^‡ = 4.0 kcal/mol) to deliver γ-Co(III)-benzyl radical intermediate H, which is also highly exergonic by 23.4 kcal/ mol. By comparison, the alternative 7-endo-trig radical cyclization of intermediate G has a much higher activation barrier (ΔG_TS5′^‡ = 14.9 kcal/mol), which is endergonic by 2.0 kcal/mol rather than exergonic. Lastly, intermediate H undergoes 3-exo-tet radical cyclization with a relatively low activation barrier (ΔG_TS6^‡ = 4.3 kcal/mol) to produce chromane 4a while regenerating the catalyst [Co(P6)]. To investigate the origin of stereoselectivity for asymmetric radical bicyclization of α-aryldiazomethane 3a′ by [Co(P6)], the energetic details associated with the generation of corresponding major and minor diastereomers as well as enantiomers were also calculated (Figures S5 and S6; see the Supporting Information for details). The energy profiles are correctly correlated with the observed trends in asymmetric induction and diastereoselectivity. The overall low activation barriers of the two catalytic systems from the DFT calculations are also consistent with the experimental results that both of the catalytic radical bicyclization processes could proceed effectively at room temperature.

To provide experimental evidence for the existence of the Co-supported organic radical intermediates in the catalytic pathways, the spin trapping reagent N-tert-butyl-α-phenylnitrone (PBN) and stable radical 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO) were employed to trap them for EPR detection and structural characterization, respectively (Scheme 2). We first added the spin trapping reagent PBN into the reaction mixture of diazomalonate 1a with the catalyst [Co(P3)] and then monitored it by X-band electron paramagnetic resonance (EPR) spectroscopy at room temperature (Scheme 2C; see the Supporting Information for details). The observed isotropic EPR spectrum displays the characteristic triplet of doublet signals at the g-value of ~2.00, which could be fittingly simulated as two well-defined triplet of doublet signals by involving two PBN-trapped radical species 5 and 6 that are originated from α-Co(III)-malonyl radical I_[Co(P3)]/1a and γ-Co(III)-alkyl radical II_[Co(P3)]/1a on the basis of the hyperfine couplings by ¹⁴N (I = 1) and ¹H (I = 1/2): 79% of O-centered radical 5 (g = 2.00632, A_(N) = 41.5 MHz,and A_(H) = 5.2 MHz) and 21% of O-centered radical 6 (g =2.00645, A_(N) = 39.7 MHz, and A_(H) = 5.8 MHz). We also added the spin trapping reagent PBN into the reaction mixture of α-aryldiazomethane 3a′ with the catalyst [Co(P1)] for EPR detection of the corresponding radical intermediates (Scheme 2D; see the Supporting Information for details). The observed EPR spectrum exhibits strong signals with the distinctive splitting pattern of the triplet of the doublet at the g-value of ~2.00. This was taken as the evidence for the formation of a mixture of O-centered radicals 7 and 8 that resulted from PBN trapping of the initially generated α-Co(III)-benzyl radical III_[Co(P1)]/3a and subsequently formed γ-Co(III)-benzyl radical IV_[Co(P1)]/3a. The observed signals could be agreeably simulated on the basis of hyperfine couplings by ¹⁴N (I = 1) and ¹H (I = 1/2): 79% of O-centered radical 7 (g = 2.00639, A_(N) = 41.1 MHz, and A_(H) = 6.7 MHz) and 21% of O-centered radical 8 (g = 2.00649, A_(N) = 39.5 MHz, and A_(H) = 6.7 MHz). In addition to PBN trapping for EPR detection, significant efforts were made to trap the catalytic radical intermediates by stable radical TEMPO with an aim at isolating the TEMPO-trapped product for structural characterization. After many experimentations, we were excited to be able to successfully isolate a new product 9 in ~7% yield from the reaction mixture of α-aryldiazomethane 3c′ with the catalyst [Co(P1)] (2 mol %) in the presence of TEMPO (3.0 equiv) (Scheme 2E). Compound 9 was characterized and further confirmed by X-ray crystallography as a bis-TEMPO-trapped product that contains the two TEMPO units at the α-and γ-positions, respectively. The formation of compound 9 evidently implies the involvement of both the initially generated α-Co(III)-benzyl radical III_[Co(P1)]/3c and the subsequently formed γ-Co(III)-benzyl radical IV_[Co(P1)]/3c intermediates. Conceivably, the resulting γ-Co(III)-benzyl radical IV_[Co(P1)]/3c was trapped by the first molecule of TEMPO through radical recombination to generate intermediate V_[Co(P1)]/3c. The succeeding radical substitution of V_[Co(P1)]/3c with another molecule of TEMPO was likely responsible for the final formation of bis-TEMPO-trapped product 9 upon the homolytic cleavage of the weak Co–C bond. The successful trap of γ-Co(III)-benzyl radical IV seems in conformity with the finding from the DFT calculations that the radical substitution of intermediate IV has a relatively higher activation barrier than the radical addition of intermediate III (Scheme 2B). Together, the results from these experimental and computational studies are taken as convincing evidence in support of the proposed stepwise radical mechanism for Co(II)-based asymmetric radical bicyclization.

CONCLUSIONS

In summary, we have successfully applied Co(II)-based metalloradical catalysis (MRC) for asymmetric radical bicyclization of both stable diazomalonates and in situ-generated α-aryldiazomethanes that contain dangling alkene units to construct valuable tricyclic compounds. We have demonstrated the power and versatility of the D₂-symmetric chiral amidoporphyrin ligand platform in support of the Co(II)-based metalloradical system for molecular construction by homolytic radical chemistry. The key to the success in controlling the stereochemical course of the two new radical processes is the judicious modulation of the D₂-symmetric chiral amidoporphyrin to create a suitable ligand environment that maximizes cooperative noncovalent attractive interactions. With the support of optimized ligands, the Co(II)-based metalloradical system can effectively activate 2-vinylaryl diazomalonates and in situ-generated α-(2-(allyloxy)aryl)-diazomethanes for asymmetric radical bicyclization to construct cyclopropane-fused tricyclic chromanones and chro-manes, respectively, in excellent yields with excellent control of both diastereoselectivities and enantioselectivities for the newly created contiguous stereogenic centers. As a demonstration of synthetic application, the Co(II)-catalyzed radical bicyclization has been employed as a key step for the first asymmetric total synthesis of the natural product (+)-Radulanin J. Our combined computational and experimental studies have suggested that the two catalytic bicyclization processes proceed with a similar stepwise radical mechanism that comprises 6-exo-trig cyclization of initially generated α-Co(III)-alkyl radicals and 3-exo-tet cyclization of subsequently formed γ-Co(III)-alkyl radicals. We believe these two Co(II)-based metalloradical systems for asymmetric radical bicyclization will find useful applications in organic synthesis as the resulting chiral chromanones and chromanes are common structures in many biologically active compounds and pharmaceuticals.