Theoretical Investigation of Chemoselectivity between C–H Insertion and Amide Insertion in Intramolecular Rhodium-Carbene Reactions

Abstract

C–H insertion and amide insertion reactions using metal-carbene species provide a powerful synthetic method for direct functionalization of kinetically inert or thermodynamically stable chemical bonds. Our group previously developed an amide insertion reaction using a rhodium-dimer complex, constructing an array of nitrogen-bridged heterocycles. Another research group reported C–H insertion reactions using structurally related substrates and rhodium catalysts. Detailed mechanistic studies were not provided, however, and therefore, the origin of the chemoselectivity was ambiguous. Here we describe our theoretical investigation of the chemoselectivity between the amide insertion reaction and C–H functionalization. An energy gap of the identified transition states in the reaction coordinates could support the reported experimental results and the observed chemoselectivity. Moreover, frontier molecular orbital analysis revealed that functionalities adjacent to the metal-carbene species could affect orbital populations and their energy levels, resulting in the construction of a completely distinctive ring system.

Introduction

Chemoselective reactions are functional group-selective transformations of molecules with plural reaction sites. Controlling chemoselectivity in diverse reaction manifolds streamlines the synthetic processes for natural products and agrochemicals by eliminating the protection/deprotection sequences for reactive functionalities. In 2019, our group with collaborators reported a formal synthesis of Iboga-class alkaloids, (+)-catharanthine¹⁾ (Chart 1). The pentacyclic natural product could be used as the monomeric precursor to vincristine and vinblastine.^2–4) A key reaction in our synthetic strategy was an amide insertion reaction of metal-carbenes,^5,6) assembling a symmetric azabicyclo[2.2.2]octane ring system. Because metal-carbenes could lead to divergent reactions, taming the reactivity of the highly active chemical species is needed for synthetic applications.⁷⁾ In the reaction of 1 using the Rh₂(NHCO^tBu)₄ catalyst designed by our group, only the C–N bond insertion reaction was observed,⁸⁾ without isolation of the corresponding C–H insertion products.⁹⁾ Even though the metal elements or supported ligands of the catalysts were changed, the intramolecular C–H insertion reaction into methine and methylene was not observed in the previous studies. Metal-carbene species can be generated from the corresponding α-diazoketones or α-diazoesters, but systematic studies of the differences in their reactivities have not been performed.^10–12)

Chart 1. Amide Insertion Reaction and Its Synthetic Application

PMB represents p-methoxybenzyl. DIBAL-H represents diisobutylaluminum hydride.

Interestingly, in 2006, Wee and colleagues reported that the C–H insertion reaction proceeded when using diazoester 5 with a lactam unit structurally related to our substrate 1¹³⁾ (Chart 2). The research group successfully synthesized a core framework of the bioactive natural product from C–H insertion product 6.¹⁴⁾ An array of rhodium (II) complexes was screened to develop the key C–H insertion reaction, but the authors did not mention that an amide insertion product was provided. The detailed follow-up report also did not describe the amide insertion reaction,¹⁵⁾ and no computational analysis for the reaction mechanism was performed.

Chart 2. C–H Insertion Reaction and Its Synthetic Application

This background prompted our interest in the origin of the chemoselectivity. Here we describe our computational research efforts to shed light on the chemoselectivity switching between amide insertion and C–H insertion in rhodium-carbene reactions.

Results and Discussion

Quantum computations based on density functional theory (DFT) were initiated to elucidate the reaction mechanism¹⁶⁾ and its coordinate diagram for the two distinctive insertion reactions. Analysis of the intrinsic reaction coordinate (IRC) was applied for all transition structures to trace the minimum energy route from the transition state (TS) to the corresponding reactant and product states on the potential energy surface.¹⁷⁾ The structurally simplified formamidate ligands on the catalyst were used to save computational time, but the substrate structure was little changed so that the origin of the chemoselectivity could be correctly analyzed (Fig. 1). At the outset of the virtual investigations, we analyzed the process of C–H insertion reactions from rhodium-carbene CP1c generated from the corresponding diazoketone because a mechanism of the rhodium-carbene formation from diazo compounds and a dirhodium(II) catalyst was previously studied computationally.¹⁸⁾ The optimized conformer of rhodium-carbene possesses an N–H⋯O classical hydrogen bonding interaction between the diazoketone carbonyl group and the supporting ligand of the catalyst (2.31 Å). The C–H insertion reaction was found to proceed in a stepwise fashion via intramolecular hydride transfer and a subsequent Mannich-type addition reaction (CP1c→CP2c_C–H→PD1c_C–H).¹⁹⁾ The former redox reaction process required an activation energy of ΔG^‡=+9.27 kcal/mol via TS1c_C–H, which has an O–H⋯O hydrogen bonding interaction between the transferring hydrogen atom and the catalyst (2.57 Å), providing iminium cation CP2c_C–H. Subsequent nucleophilic addition of C-bound Rh enolate constructed a cyclopentanone ring through a steep endothermic process (ΔG=−52.37 kcal/mol, TS2c_C–H→PD1c_C–H). The amide insertion reaction also progressed in stages through the formation of N-ylide, tautomerization to O-bound Rh enolate, and an acyl-group transfer via CP2c_Amide and CP3c_Amide. In the previous report, an acyl transfer reaction directly proceeded from Rh-associated N-ylide to construct azabicyclo[3.2.1]octane framework. Small factors can influence the kinetic differences between the tautomerization and the acyl group transfer, given the slight differences in substrate structure. The activation energy of the first step to provide Rh-associated N-ylide CP2c_Amide was calculated to be ΔG^‡=+4.76 kcal/mol. Accordingly, the chemoselectivity-determining energy gap between TS1c_Amide and TS1c_C–H as a saddle point on the potential energy surface was ΔΔG^‡ = 4.51 kcal/mol. Therefore, the amide insertion reaction of rhodium-carbene derived from the corresponding diazoketone was the kinetically favored pathway over C–H functionalization.

Fig. 1. Reaction Coordinate Diagram of Rhodium-Carbene Species Possessing a Ketone Functionality, Based on rb3pw91-D3BJ/6-311++G**/SDD//rb3lyp/6-31G*/LanL2DZ Levels of Theory in Dichloroethane

Bond lengths are shown in Å.

Figure 2 shows the calculated potential energy profiles of the insertion processes of rhodium-carbene CP1o generated from the corresponding diazoester. Similarly, the redox-neutral C–H functionalization from CP1o proceeded by way of the intermediate CP2o_C–H, furnishing cis-fused lactone PD1o_C–H. The activation barrier of C–H cleavage was predicted to be +5.08 kcal/mol, which is 4.19 kcal/mol lower than that of TS1c_C–H in the formation of the carbocycle. For the amide insertion reaction that did not actually proceed, three elementary reaction processes were found in theory. Nucleophilic attack of amide nitrogen into the carbene carbon required an activation energy of ΔG^‡=+5.79 kcal/mol via TS1o_Amide, giving a Rh-associated N-ylide CP2o_Amide. The Rh C-bound enolate could evolve the amide insertion products PD1o_Amide through a transfer of the acyl group without a tautomerization to Rh O-bound enolate. The difference in the activation energy between TS1c_Amide and TS1o_Amide is, therefore, ΔΔG^‡ = 0.71 kcal/mol; thus, a C–H insertion reaction proved to be an advantageous pathway in reactions using rhodium-carbene derived from the corresponding diazoester.

Fig. 2. Reaction Coordinate Diagram of Rhodium-Carbene Species Possessing an Ester Functionality, Based on rb3pw91-D3BJ/6-311++G**/SDD//rb3lyp/6-31G*/LanL2DZ Levels of Theory in Dichloroethane

Bond lengths are shown in Å.

All the TSs have reasonable activation energies below 30 kcal/mol and the energy gap supports the experimentally observed reaction outcome. The classical and nonclassical hydrogen bonding interactions (N−H⋯O and C−H⋯O, respectively) in the transition states and the intermediates contribute to stabilization of the reactive species, preventing the generation of uncontrollable free-ylide or dipole.

To further elucidate the origin of the chemoselectivity, frontier molecular orbital (FMO) analysis describing a chemical reactivity was carried out and the results are shown in Fig. 3.²⁰⁾ The lowest unoccupied molecular orbital (LUMO) was spread over the carbenoid carbon and the adjacent carbonyl group along with rhodium catalyst in each case (a) and (b). Rhodium-carbene with the ketone has orbitals filled with electrons as the highest occupied molecular orbital (HOMO)−4 at the lactam moiety (c), while rhodium-carbene with the ester has orbitals filled with electrons as HOMO−5 at the lactam moiety (d). In addition, the difference in the orbital energy level of rhodium-carbene with the ester was greater than that of rhodium-carbene with the ketone (3.61 eV, 3.22 eV, respectively). Given the reaction outcome and the computational insight, one possible interpretation is that reaction of rhodium-carbene possessing the ketone functionality would favor an orbital-controlled process to react with an amide nitrogen as a soft nucleophile. On the other hand, reaction of rhodium-carbene possessing the ester functionality would favor a charge-controlled process leading to the hydride-abstraction based on hard/soft acid/base (HSAB) principles. These different natures of the carbene species could affect each activation barrier, resulting in the experimentally observed chemoselectivity.

Fig. 3. Distribution of Frontier Molecular Orbitals and Energy Levels of Rhodium-Carbene Species

In summary, we delineated two distinctive reaction coordinates from rhodium-carbene possessing a ketone functionality and an ester functionality by virtual simulation. The energy gap of the key transition states and FMO analysis based on DFT supports the reported experimental results for an amide insertion reaction and a C–H insertion reaction. If differences in the ester carbonyl group or ketone carbonyl group at the adjacent position of the carbene species significantly affect the chemoselectivity, this would lead to the synthesis of totally distinctive molecular skeletons and natural products. We anticipate that the revealed chemical profile of metal-carbene species will contribute to further development of synthetic organic chemistry using transition-metal catalysis and pharmaceutical science.

Experimental

General Methods

All DFT calculations were performed with Gaussian 16 program and the ball and stick models were drawn with Avogadro software. The molecular structure optimizations were carried out using the hybrid density functional method based on Becke’s three-parameter exchange function and the Lee–Yang–Parr nonlocal correlation functional (rB3LYP)²¹⁾ and the LanL2DZ basis set for Rh, and the 6-31G* basis set for H, C, N, and O. The vibrational frequencies were computed at the same level to check whether each optimized structure is at an energy minimum (no imaginary frequency) or a transition state (one imaginary frequency) and to evaluate its zero-point vibrational energy (ZPVE) and thermal corrections at 298.15 K. Single point energies were calculated at the rb3pw91-D3BJ level using the SDD basis set for Rh and the 6-311++G** basis set for H, C, N, and O in DCE solvent (i-pcm model).

Acknowledgments

This work was supported by Futaba Electronics Memorial Foundation, Ube Industries Foundation, The Sumitomo Foundation, Suzuken Memorial Foundation, The Research Foundation for Pharmaceutical Sciences, JSPS KAKENHI (Grant Numbers 18H02550 and G21K06471). Numerical calculations were carried out on SR24000 at the Institute of Management and Information Technologies, Chiba University of Japan.

Conflict of Interest

The authors declare no conflict of interest.

Supplementary Materials

This article contains supplementary materials.

References

• 1) Kono M., Harada S., Nozaki T., Hashimoto Y., Murata S., Gröger H., Kuroda Y., Yamada K., Takasu K., Hamada Y., Nemoto T., Org. Lett., 21, 3750–3754 (2019).
• 2) Ishikawa H., Colby D. A., Boger D. L., J. Am. Chem. Soc., 130, 420–421 (2008).
• 3) Ishikawa H., Colby D. A., Seto S., Va P., Tam A., Kakei H., Rayl T. J., Hwang I., Boger D. L., J. Am. Chem. Soc., 131, 4904–4916 (2009).
• 4) Beatty J. W., Stephenson C. R. J., J. Am. Chem. Soc., 136, 10270–10273 (2014).
• 5) Harada S., Kono M., Nozaki T., Menjo Y., Nemoto T., Hamada Y., J. Org. Chem., 80, 10317–10333 (2015).
• 6) Harada S., Kato R., Nemoto T., Adv. Synth. Catal., 358, 3123–3129 (2016).
• 7) Nakayama H., Harada S., Kono M., Nemoto T., J. Am. Chem. Soc., 139, 10188–10191 (2017).
• 8) Hashimoto Y., Kono M., Harada S., Nemoto T., Chem. Pharm. Bull., 66, 1041–1047 (2018).
• 9) Davies H. M. L., Manning J. R., Nature (London), 451, 417–424 (2008).
• 10) Jinnouchi H., Nambu H., Takahashi K., Fujiwara T., Yakura T., Tetrahedron, 75, 2436–2445 (2019).
• 11) Nakamura S., Hirata Y., Kurosaki T., Anada M., Kataoka O., Kitagaki S., Hashimoto S., Angew. Chem. Int. Ed., 42, 5351–5355 (2003).
• 12) Deng Y., Jing C., Arman H., Doyle M. P., Organometallics, 35, 3413–3420 (2016).
• 13) Fan G.-J., Wang Z., Wee A. G. H., Chem. Commun., 35, 3732–3734 (2006).
• 14) Pearson W. H., Bergmeier S. C., Williams J. P., J. Org. Chem., 57, 3977–3987 (1992).
• 15) Wee A. G. H., Fan G.-J., Bayirinoba H. M., J. Org. Chem., 74, 8261–8271 (2009).
• 16) Uchiyama M., J. Syn. Org. Chem, 73, 119–130 (2015).
• 17) Fukui K., Acc. Chem. Res., 14, 363–368 (1981).
• 18) Nakamura E., Yoshikai N., Yamanaka M., J. Am. Chem. Soc., 124, 7181–7192 (2002).
• 19) Lamb K. N., Squitieri R. A., Chintala S. R., Kwong A. J., Balmond E. I., Soldi C., Dmitrenko O., Castiñeira Reis M., Chung R., Addison J. B., Fettinger J. C., Hein J. E., Tantillo D. J., Fox J. M., Shaw J. T., Chem. Eur. J., 23, 11843–11855 (2017).
• 20) Fukui K., Yonezawa T., Shingu H., J. Chem. Phys., 20, 722–725 (1952).
• 21) Lee C., Yang W., Parr R. G., Phys. Rev. B Condens. Matter, 37, 785–789 (1988).