Optimizing Group Transfer Catalysis by Copper Complex with Redox-Active Ligand in an Entatic State

Summary

Metalloenzymes use earth-abundant non-noble metals to perform high-fidelity transformations in the biological world. To ensure chemical efficiency, metalloenzymes have acquired evolutionary reactivity-enhancing tools. Among these, the entatic state model states that a strongly distorted geometry induced by ligands around a metal center gives rise to an energized structure called entatic state, strongly improving the reactivity. However, the original definition refers both to the transfer of electrons or chemical groups, whereas the chemical application of this concept in synthetic systems has mostly focused on electron transfer, therefore eluding chemical transformations. Here we report that a highly strained redox-active ligand enables a copper complex to perform catalytic nitrogen- and carbon-group transfer in as fast as 2 min, thus exhibiting a strong increase in reactivity compared with its unstrained analogue. This report combines two reactivity-enhancing features from metalloenzymes, entasis and redox cofactors, applied to group-transfer catalysis.

Graphical Abstract

Highlights

• •We design a catalyst interfacing two reactivity-enhancing tools from metalloenzymes
• •This work merges redox-active cofactors and entatic state reactivity
• •The modifications in the coordination sphere lead to enhanced catalytic behavior
• •These results open perspectives in bioinspired catalysis and group-transfer reactions

Inorganic Chemistry; Molecular Inorganic Chemistry; Catalysis

Introduction

The biological world offers potential clues to solve the current pressing energy and resources-related issues (Adesina et al., 2017, Rudroff et al., 2018). The efficiency of biological catalytic systems faces chemists with the challenges of transferring enzymatic performance into synthetic small-molecule catalysts using non-noble metals. Metalloenzymes use earth-abundant 3d metals and amino acid-derived coordination spheres to achieve complex (multi)electronic transformations required for small-molecule activation and atom transfer reactions. To fulfill these stringent criteria of natural resources, metalloenzymes have evolved reactivity-enhancing strategies aimed at performing such difficult transformations through more favorable thermodynamic pathways. The entatic state concept stems from the observation that single molecule coordination complexes designed as metalloenzymes models often fail to exhibit properties similar to the natural metalloenzymatic active sites (Vallee and Williams, 1968, Williams, 1971, Williams, 1995, Stanek et al., 2018). It was thus suggested that secondary and tertiary structures induced by amino acid-based side chains in proteins provide distortions in the coordinative environment that endow the metal center with enhanced reactivity, to which Vallée and Williams refer as “a catalytically poised state” (Vallee and Williams, 1968). Strikingly, a strong link between local geometric frustration induced by tertiary structure effects around enzymatic catalytic sites and their catalytic activity has recently been unveiled (Freiberger et al., 2019). Frustration patterns were demonstrated to be evolutionarily highly conserved, more than the primary structure or residues forming the catalytic site itself, which establishes the central relevance of coordinative distortion to enzymatic catalytic activity.

Entasis is well known in copper proteins involved in electron transfer such as blue copper proteins and has traditionally been studied in the context of Cu^I/Cu^II oxidation states. When tetracoordinated, these two oxidation states have distinct geometric requirements from tetrahedral to square planar, and the entatic state exhibits an in-between distorted geometry that minimizes energetic penalties arising from coordinative reorganization (Figure 1A). Possible designs for coordination spheres inducing ligand misfit have been discussed by Comba (2000). The majority of reports involve two ditopic or one cyclic tetradentate ligand and include systems with biphenyl subunits (Malachowski et al., 1999, Müller et al., 1996, Müller et al., 1988), multiple bonds and rigid or substituted structures, carbohydrate backbones (Garcia et al., 2015), and guanidinoquinoline ligands (Hoffmann et al., 2013, Stanek et al., 2017). More recently, the entatic state concept was extended to photoinduced electronic transfers (Dicke et al., 2018) thus showing the continued relevance of this biomimetic feature in electron transfer chemistry. Although most studies focus on electron transfer involving a Cu^I/Cu^II redox couple, the seminal article by Vallée and Williams (Vallee and Williams, 1968) states that entatic behavior arising from ligand misfit also extends to enzymatic metallic sites performing “transfer of an atom, radical, or a group,” as well as higher metallic oxidation states such as Co^II/Co^III in corrin enzymes and Fe^II/Fe^III (Mara et al., 2017). A recent report on the hydroxylation of methane to methanol by heterogeneous catalysis performed by iron-containing zeolithes demonstrates that the zeolithe lattice induces a matrix strain that activates a square planar Fe(II) site into a Fe(IV) = O reactive center through entasis (Snyder et al., 2016), and a bispidine Fe(IV) = O species in an entatic state has been reported to perform oxygen-group transfer in homogeneous catalysis (Comba et al., 2016). These examples of oxygen-group transfer systems, which emulate natural systems' catalytic arsenal, provide great opportunities for the design of synthetic systems able to perform highly desired transformations such as multi-electronic small-molecule activation. However, to date, no synthetic catalytic system was demonstrated to perform group transfer other than oxygen through the entatic state model.

Figure 1.

Design of a Redox-Active Ligand Creating a Vacant Coordination Site at Metal through Strong Coordination Distortion

(A) The entatic state reactivity model for electron and group transfer in a tetracoordinated copper complex.

(B) Structure of complex 1 [CuBINap (SQ-BQ)]⁺.

(C) X-ray crystal structure of ligand 2 representing the associated strains.

(D) X-ray crystal structure of complex 1 exhibiting a pentacoordinated coordination sphere resulting from coordination strain.

Just as the entatic state principle provides a fruitful trail of inspiration for coordination chemists, other biomimetic strategies, related to electron transfer, focus on enzymatic redox co-factors that provide electrons to the active sites (Stubbe and van der Donk, 1998, Davidson, 2018). Such systems are known as redox-active ligands. This field of research is morphing into an area of intense development aimed at harnessing the potential of systems with rich redox flexibility (Albold et al., 2019, Blanchard et al., 2012, Broere et al., 2015, Chirik and Wieghardt, 2010, Grützmacher, 2008, Kaim, 2011, Luca and Crabtree, 2012, Lyaskovskyy and de Bruin, 2012, Praneeth et al., 2012, van der Vlugt, 2012, van der Vlugt, 2019). Recently, an entatic pair of copper complexes bearing redox-active guanidine ligands was shown to enable fine-tuning between metal- and ligand-based electronic transfers (Schrempp et al., 2017). This structural study was the first report interfacing the two concepts of entasis and redox-active ligands and was focused on the electronic behavior of the entatic pair. Using redox-active copper complex Cu(SQ)₂ (SQ: iminosemiquinone) originally developed as a galactose oxidase (GAO) model (Chaudhuri et al., 1998, Chaudhuri et al., 2001), we have shown that ligand-based redox activity influences the overall reactivity of the complex toward electron transfer (Jacquet et al., 2016a, Jacquet et al., 2017), C–N bond formation (Jacquet et al., 2016b), and nitrogen group transfer (Ren et al., 2018) while preserving a Cu(II) oxidation state. Here we report an acyclic tetradentate redox-active ligand exhibiting C₂ symmetry with overall three possible rotations around single bonds (degrees of freedom). This particular geometry induces enhanced catalytic reactivity of resulting complex 1 through entasis. This specific ligand design combines an atropisomeric backbone, which enforces chirality and exerts an outward strain, with a redox-active iminosemiquinone/iminobenzoquinone ONNO coordination sphere that pulls in the opposite inward direction toward metal chelation (Figure 1B). These conflicting steric demands induce a coordination stress while the bridging binaphthalene backbone installs full conjugation between the two redox-active units, resulting in enhanced electronic delocalization. Complex 1 performs catalytic nitrogen and carbon group transfer in as fast as 2 min on a model substrate and exhibits up to a 40-fold increase in synthetic efficiency compared with its unstrained analogue Cu(SQ)₂. Moreover, it can convert sterically hindered or unactivated alkenes. The present work shows that interfacing two reactivity-enhancing features from metalloenzymes, such as entasis and redox cofactors, can improve group-transfer catalysis.

Results

Complex Synthesis and Structural Studies

Condensation of two tert-butylcatechol with 1,1′-binaphthyl-2,2′-diamine yields easy access to 1,1′-binaphtyl-2,2′-diaminophenol (hereafter noted BINap) ligand (Figure 2A and Supplemental Information), and its structure was confirmed by X-ray crystallography (Figure 1C). Aerobic complexation of BINap ligand 2 with copper chloride affords complex Cu(BINap-SQ₂) 3 as a green solid. The spectroscopic signature of 3 (Figures 2C–2E) is very similar to those of Cu(SQ)₂ 4 (Figure 2B) and Cu(L_BiphenSQ₂), a copper complex reported by Chaudhuri and incorporating a biphenyl unit instead of the atropisomeric binaphthyl unit (Mukherjee et al., 2008). At 10 K, 3 displays an S = 1/2 signal centered on the copper (Figure S10), as attested by the strong anisotropy of the EPR spectrum and the presence of hyperfine coupling with the I = 3/2 copper nucleus. The doublet ground spin state of 3 is further supported by density functional theory (DFT) calculations (Table S15). The UV-vis spectrum of 3 (Figure 2D) presents the expected ligand-to-ligand (SQ to SQ) charge transfer band around 900 nm (800 nm in 4 and 880 nm in Cu(L_BiphenSQ₂)), which points toward 3 as a Cu(II) complex bearing a dianionic diradical BINapSQ₂ ligand. This description is in agreement with the DFT-calculated electronic structure of 3 (Figures S110 and S111). The cyclic voltammogram of 3 displays four reversible waves (two oxidative and two reductive, Figure S9) at redox potentials very close to those of the parent complexes, indicating similar electronic structures and ligand-based redox processes (Tables S1 and S2).

Figure 2.

Synthesis and Spectroscopic Studies for Complex 1

(A) Synthetic scheme for complex 1, reaction conditions: (a) CuCl (1 equiv.), Et₃N (4 equiv.), dry acetonitrile; (b) AgOTf (1 equiv.).

(B) Unstrained complex Cu(SQ)₂4.

(C) Cyclic voltammogram of complex 1.

(D) UV-vis-NIR spectrum of [Cu(BINap-SQ-BQ)]⁺1 (purple), Cu(BINap-SQ₂) 3 (green), and [Cu(BINap-BQ₂)]²⁺5 (see Supplemental Information for structure, red); 0.1 mM in DCM at rt.

(E) Circular dichroism spectrum of racemate and enantiopure [Cu(BINap-SQ-BQ)]⁺1 (C = 0.3 mM).

Oxidation of Cu(BINap-SQ₂) 3 with Br₂ followed by halide abstraction led to the formation of fully oxidized [Cu(BINap-BQ₂)]²⁺ 5. The singly oxidized complex [Cu(BINap-SQ-BQ)]⁺ 1 can be prepared either by oxidation of 3 with 1 equiv. AgPF₆ or by mediamutation between an equimolar mixture of complexes 3 and 5 (see Supplemental Information). Cyclic voltammograms of 1 and 5 are very similar to that of 3 and only differ by the onset potential, thus attesting the chemical reversibility of the redox processes. Complex 1 is X-band EPR silent at 10 K, whereas 5 displays an EPR spectrum in agreement with an S = 1/2 ground state centered on the copper (Figure S13). The doublet ground spin state of 5 has been also predicted by DFT calculations (Figures S113 and S114). The UV-vis spectrum of 1 displays around 1,200 nm an ligand-to-ligand charge transfer (LLCT) band associated with the oxidation of an SQ moiety into BQ, whereas no LLCT can be identified for fully oxidized 5 (Figure 2D). This is reminiscent of the parent complexes and confirms that the two oxidation processes are ligand centered. Enantiopure R or S complexes of 1 and 3 can be synthesized from enantiopure R or S forms of 2, as attested by circular dichroism and chiral HPLC analysis (Figures S3 and S5).

Attempts to grow crystals of 3 led to very weakly diffracting samples, with maximum resolution of 1.29 Å (see Supplemental Information), which indicates that the copper center is in coordinence 4 (N₂O₂ of BINap). Crystals suitable for X-ray diffraction were obtained on racemate and enantiopure samples of complex 1 and revealed that the coordination sphere is completed by an aquo ligand to provide a square pyramidal environment in which an oxygen atom of the BINap occupies the axial position and not the water molecule (from a CCDC database search, of 126 structures of 5-coordinated Cu complexes bearing {(NCCO)₂ + OH₂} coordination sphere, only five do not display a square pyramidal geometry with water as apical ligand) (Figure 1D). This unusual coordination sphere can be explained by the coordination stress resulting from the design of ligand 2, and the singly oxidized SQ-BQ redox-active subunit, which bears increased electrophilicity compared with complex 3. The combination of these unique steric and electronic features induces a distorted pentacoordinated sphere with a newly occupied coordination site. The atropisomeric backbone exerts an outward strain associated to the large torsion angle between the two naphthalene rings (99.9° in free ligand 2, 66.7° in the complex) and opposes to N₂O₂ and Cu being in the same plane. The long axial Cu-O_BINap distance (2.309(2)Å) indicates a weak coordination. Detailed analysis of the bond distances in coordinated BINap ligand shows that this O atom belongs to the benzoquinone part of the ligand (bond lengths in this part of the ligand of d_C-O = 1.235(4)Å, d_C-C = 1.519(15)Å, and d_C-N = 1.289(5)Å), whereas the in-plane half of the ligand corresponds to the iminosemiquinone moiety (d_C-O = 1.288(4)Å, d_C-C = 1.455(5)Å, and d_C-N = 1.344(4)Å) (Chaudhuri et al., 2001). Such deviation from square planar geometry has been reported in the context of a salen-based Cu^II GAO functional mimic with an atropisomeric backbone by Stack and co-workers (Wang et al., 1998). DFT calculations provided insights into the electronic features of complex 1, which can be assigned as a Cu(II) center bound to a radical BINap ligand (Figures S113 and S114). The system is characterized by a triplet S = 1 ground spin state due to moderate ferromagnetic coupling between the metal center and the ligand radical moiety (Table S15). A singlet S = 0 state is close enough in energy to be thermally populated.

Nitrogen-Group Transfer Reaction: Aziridination

We have previously reported copper-catalyzed aziridination with unstrained complex 4 bearing two independent redox-active iminosemiquinone units and exhibiting a planar geometry around the copper center (Figure 2B) (Ren et al., 2018). Complex 4 can perform aziridination on a wide range of unactivated alkenes through a mechanism involving molecular spin catalysis, related to the multistate reactivity model encountered in metalloenzymes. Aiming to assess the effect of tetratopic ligand 2 in comparison with two independent ditopic redox-active units, the reactivity of complex 1 in aziridination was therefore investigated.

To examine the influence of ligand strain and redox state, we performed benchmarking reactivity tests under identical conditions on all three possible redox state combinations (SQ/SQ, SQ/BQ, and BQ/BQ) of the redox-active units for the unstrained (4, 6, 7) and strained (3, 1, 5) complexes (Table S10). Strikingly, complex 1 was found to outperform all other complexes, delivering the aziridination adduct 8 of 4-chlorostyrene in quantitative yield. These results suggest that the major reactivity improvement observed with complex 1 is linked to two factors: electronic conjugation between the SQ and BQ redox-active subunits and high steric strain induced by the ligand. Interestingly, the SQ-BQ redox state Cu(L_BiphenSQ-BQ)⁺ of a functional galactose oxidase model, was suggested to be involved in the catalytic activity. This model incorporates a biphenyl unit instead of the atropisomeric binaphthyl unit and thus retains full conjugation between the redox-active subunits but does not exert steric strain as rotation around the arylic C–C bond is possible in the biphenyl unit but not in the atropisomeric binaphthyl unit (Mukherjee et al., 2008). This provides strong grounds for the role of electronic communication installed between redox-active subunits through the conjugated backbone on the redox state of the reactive complex. The second factor for the major reactivity improvement is the high steric strain observed in complex 1. This observation is reminiscent of the entatic state model, defined as a strong steric distortion induced by ligands in a metal complex leading to improvement in catalytic activity (Vallee and Williams, 1968).

The scope of the reaction was evaluated (Scheme 1) and revealed the ability of complex 1 to perform aziridination on a wide range of substrates including mono-, di-, and trisubstituted double bonds, as well as electrophilic functions (11 and 14) with yields from 76% up to quantitative (8–10). Trans-stilbene gave the corresponding trans diastereoisomer 12 in 90% yield, whereas cis-stilbene provided a mixture of cis- and trans-aziridines (12 and 13), thus pointing toward a probable radical mechanism accounting for cis-trans isomerization toward the most stable trans isomer. Also, conversion of unactivated alkenes (15 and 17) could be performed in up to 80% yield, whereas aziridination of a deactivated substrate afforded the expected product in 38% yield. Tetrasubstituted substrate 18, unreactive with complex 4, was transformed in 86% yield within an hour. Reaction on a geraniol derivative led to products 19 and 20 in higher yield compared with the unstrained complex, and regioselectivity was reinforced toward the less hindered site in accordance with the high steric hindrance in complex 1. The most striking results are observed when comparing catalysis with strained complex 1 and the most efficient unstrained analogue from our benchmarking reactivity test complex 4 Cu(SQ)₂. Reaction time could be divided by up to a factor 8 for compounds 10, 11, and 16–20, and yields increased up to 42-fold (compound 14), thus demonstrating the efficiency of complex 1. Based on the substrate scope, average yields of 90% for styrene substrates and 73% for unactivated or hindered substrates are obtained with complex 1, whereas the unstrained analogue 4 provides much lower average yields (38% and 22%, respectively). The calculated average gain in yield observed is 59% and 51%, respectively (calculated as the mean value of subtracted yields). This indicates that steric strain in complex 1 is accountable for a major reactivity improvement. TON (turnover number) was calculated and found to be significantly and consistently higher in the case of complex 1 versus its unstrained analogue (Scheme 1). The maximum TON value of 20 was found for substrate 8, along with a TOF (turnover frequency) of 600 h⁻¹. In the case of styrene, we checked that, once the aziridination reaction was complete with either complex 1 (after 2 min) or complex 4 (after 12 min) as a catalyst, addition of new equivalents of styrene and PhINTs led to the formation of the aziridination product in 95% yield for 1 and 82% yield for 4: thus, little or no deactivation of the catalyst occurs during the reaction. Since the BINap ligand exhibits axial chirality, the two enantiomers of complex 1 were prepared from commercial enantiopure 1,1′-binaphthyl-2,2′- diamine and the S-isomer was tested in the reaction with 4-chlorostyrene. However, only modest enantiomeric excesses were observed with a maximum 35% ee (Table S8). This could be related to the fact that the chirality-inducing atropisomeric motif is remote from the coordination sites involved in the reaction, which disfavors enantioselectivity.

Scheme 1.

Scope for Nitrogen-Group Transfer with Complex 1 and Comparison with Unstrained Analogue 4

Aziridination reaction performed with ^acomplex 1 [CuBINapSQ-BQ]⁺ (dark blue), ^bcomplex 4 Cu(SQ)₂ (pale blue), ^cTON calculated for complexes 1 and 4^d ratio of products (numbers in gray). Reaction conditions: 4-chlorostyrene (1 equiv.), PhINTs (2 equiv.), complex (5 mol%), acetonitrile, rt. Yields were determined by ¹H NMR using trimethoxybenzene as internal standard and calculated considering olefin as limiting reactant.

Carbon-Group Transfer Reaction: Cyclopropanation

Having established the efficiency of complex 1 in N-group transfer, we investigated its ability to perform C-group transfer with ethyl diazoacetate (EDA), a carbene precursor used in metal-based cyclopropanation reactions. Optimized reaction conditions (Tables S11–S14) led to the isolation of the cyclopropanated adduct 21 of 4-chlorostyrene in quantitative yield, and the scope of the cyclopropanation reaction was studied (Scheme 2). A wide range of substrates including diverse mono- and disubstituted styrene derivatives (21–26) and polycyclic (27–28) substrates as well as unactivated or deactivated tri- and tetrasubstituted scaffolds (29-30) was efficiently converted. Similarly as before, complex 1 was found to systematically outperform the most efficient unstrained analogue (complex [Cu(SQ-BQ)]⁺ in this case) and deliver average yields of 95% and 97% on styrene derivatives and unactivated or hindered substrates, respectively, whereas the unstrained complex reached 67% and 81% average yields, respectively. This translated into average gains in yield of 21% and 16% for the two substrate scope categories. The lower gain observed with complex 1 in cyclopropanation compared with aziridination can be explained by the reaction conditions. Reaction yields are assessed over a fixed period of time set by the best performing catalyst, and the reaction is stopped upon completion of maximum conversion by the best catalyst (complex 1). Although the aziridination reaction is very fast and can be performed in as fast as 2 min, the cyclopropanation conditions require that the carbene source is added slowly over 20 min to the reaction mixture to avoid unproductive homo-coupling of EDA as side product. Although this accounts for both longer reaction times and decreased gains in yield, complex 1 remains more efficient than the unstrained complex, and this is further confirmed by the calculated TON values close to 100 and a calculated TOF of 300 h⁻¹ for complex 1 with substrate 21.

Scheme 2.

Scope for Carbon-Group Transfer with Complex 1 and Comparison with Unstrained Analogue

Cyclopropanation reaction performed with ^acomplex 1 [CuBINapSQ-BQ]⁺ dark green, ^bcomplex 6 [Cu(SQ-BQ)]⁺ pale green, ^ctrans/cis ratio. Reaction conditions: 4-chlorostyrene (1 equiv.), EDA (2 equiv.), copper complex (1 mol%), DCM, rt. EDA was added slowly in 20 min, reaction time: 20 min. Yields were determined by ¹H NMR using trimethoxybenzene as internal standard and calculated considering olefin as limiting reactant. ^dCalculated TON for complexes 1 and 6.

Mechanistic Studies

Mechanistically, metal-catalyzed aziridination and cyclopropanation rely on the formation of transient and highly reactive metal-carbene (Dzik et al., 2010, Dzik et al., 2011) and metal-nitrene (Suarez et al., 2013, Goswami et al., 2015, Corona et al., 2016, Kuijpers et al., 2017, Fujita et al., 2018) species, which have been actively investigated in the context of ligand noninnocence. High-resolution mass spectrometry performed on an aliquot of a mixture of complex 1 and excess nitrene (respectively, carbene) source at −80°C evidences the presence of a species at m/z 920.3522 (respectively, m/z 837.3693) corresponding to the formation of mono-nitrene adduct [1-NTs] (respectively, mono-carbene adduct [1-C(H)CO₂Et]) (Figures S23 and S26). UV-vis studies of complex 1 with the nitrene or carbene sources (Figures S24, S27, and S28) show rapid disappearance of the intervalence band around 1,200 nm, in agreement with an electron transfer from the BINapSQ-BQ ligand to the nitrene upon coordination thus generating [Cu(BINap-BQ₂)]²⁺ 5, which does not present any LLCT band. To further assess the efficiency of complex 1, reaction kinetics were studied by ¹H NMR, and it was observed that aziridination with complex 1 is complete in less than 2 min (Figure 3A), whereas cyclopropanation yields the final product in 20 min upon slow addition of EDA over the course of the reaction (Figure 3B).

Figure 3.

Spectroscopic Investigations of Reaction Mechanism

Operando ¹H NMR and EPR studies for aziridination (A and C) and cyclopropanation (B and D) of 4-chlorostyrene with complex 1 and kinetic UV-vis studies for aziridination of trans-chalcone with complexes 1 and 4 (E).

The group transfer reactions were studied by continuous wave EPR spectroscopy and operando mode, as our previous work has shown that these techniques can monitor the redox state changes of the complexes and provide structural and kinetic insights into the intermediate species (Figures 3C and 3D) (Jacquet et al., 2016b). Regarding the aziridination reaction (Figure 3C), the first important observation is the absence of spectral differences between the beginning and the end of the reaction (2 h), indicating that the catalytic reaction is extremely fast. X-band EPR spectra of complex 1 alone shows that the starting complex is EPR silent from 5 K to room temperature. However, when the aziridination reaction with complex 1 is followed by operando studies (Figure 3C), two signals are immediately observed: one typical of a Cu²⁺ species (g_iso = 2.12 and A_iso = 82 G) and the other an organic radical displaying five lines with a coupling of 12.2 G and an intensity ratio of 1:2:3:2:1. The latter can be attributed to the formation of an organic radical coupled to two magnetically equivalent nitrogens. This suggests that nitrene incorporation instantaneously leads to a Cu²⁺ species with formation a ligand-based radical, these two species not being magnetically coupled. These EPR results indicate fast electron transfer from the starting complex to the nitrene part upon coordination and are consistent with the ¹H NMR monitoring of the reaction showing completion of the reaction in 2 min. Operando EPR studies attempted on the cyclopropanation reaction were not conclusive as the experimental conditions implying slow addition of EDA could not be reproduced under operando conditions (Figure 3D). Following the reaction by UV-vis spectroscopy on trans-chalcone as substrate showed an initial instantaneous reaction rate at least two times higher using complex 1 compared with complex 4 (Figures 3E and Figures S20–S22). This initial rate enhancement can be directly related to the difference in geometries of the two complexes and points toward a behavior similar to the entatic state, defined as a reactivity improvement arising from strong steric distortion induced by ligands in a metal complex.

In light of our combined mechanistic studies, the following general mechanistic scheme can be proposed (Scheme 3). Insertion of the nitrene or carbene group on starting complex 1 generates intermediates 36a-b (both detected by high-resolution mass spectroscopy [HRMS]), in which the high coordinative strain and torsion angles in complex 1 are released upon reactive group coordination. Intermediates 36a-b subsequently undergo alkene insertion to yield species 37a-b (Figure S115 and Table S15) and release the group-transfer product upon ring closure. DFT calculations (Figure 4) conducted on intermediate 36a show that one singly occupied molecular orbital (SOMO) is perfectly aligned with the Cu-nitrene axis, which reflects the presence of an unpaired electron occupying a metal-based orbital having a Cu 3d_z² character. The other SOMO appears mainly distributed over the N-SO₂ group confirming the formation of a nitrene radical moiety (Figures S115 and S116). This supports the triplet ground spin state of 36a due to the orthogonal character of the two SOMOs preventing magnetic coupling as observed by EPR spectroscopy. When looking at the electronic structure of 37a, we observe that the nature of the first SOMO remains metal centered, whereas the second SOMO is now a delocalized π-orbital distributed over the styrene moiety with the spin density being displaced from the nitrogen to the aryl group upon styrene insertion (Figures S117 and S118). DFT calculations conducted on intermediate 36b reveal that one SOMO features a dominant Cu-character and the other one is a ligand-based orbital almost exclusively centered on the carbon atom of the -CHCO₂Et ligand, which supports the formation of a carbene adduct (Figures S119 and S120). Upon styrene insertion on 36b, the spin density is shifted toward the carbon center adjacent to the aryl group and the two SOMOs of 37b appear similar to those of 37a with one SOMO being metal centered and the other one being ligand based delocalized on the styrene moiety (Figures S121 and S122). Interestingly, a ΔΔG calculation comparing the relative stability of the species obtained upon nitrene and alkene insertion on strained and unstrained complexes 1 and 4 shows that intermediates 36a and 37a are stabilized by 2.5 and 7.1 kcal.mol⁻¹, respectively, in the case of complex 1 compared with complex 4 (Tables S16 and S17). These findings highlight the energetic gain observed when using the strained BINap ligand instead of the unstrained SQ and also reflect that the highly distorted structure induced by coordination of this ligand to the metal center facilitates insertion reactions. Comparison of catalytic efficiency shows that complex 1 compares well with other reported catalytic systems in terms of both reaction time and yield (Table S18).

Scheme 3.

Proposed Catalytic Cycle for N- and C-Group Transfer Reactions with Complex 1

a: Y = -NTs, b: Y = -CHCO₂Et.

Figure 4.

DFT-Calculated Spin Density Plots for Intermediates 36a-b and 37a-b

In conclusion, this report combines the entatic reactivity model and redox cofactors, which are staples of metalloenzymes' reactivity toolbox, for enhanced group-transfer catalytic transformations. We have demonstrated that an atropisomeric tetratopic redox-active ligand imposing high steric and coordinative strain on a copper center can perform group transfer reactions up to eight times faster than related unstrained systems. The redox-active atropisomeric backbone establishes electronic communication between the two redox-active subunits, thus favoring intramolecular electron transfer during catalysis and maintaining a Cu^II redox homeostasis. This system bridges the gap between small-molecule enzymatic structural models focused on the design of the first coordination sphere and top-down approaches achieving control over the amino acid secondary and tertiary coordination spheres by directed enzymatic evolution (Arnold, 2017, Farwell et al., 2015, Coelho et al., 2013, Knight et al., 2018). This work shows that the combination of enzymatic reactivity-enabling strategies offers a promising approach to reactivity control in bioinspired catalysis and should open the door toward more challenging synthetic reactions.

Limitations of the Study

Enantioselective catalysis could not be achieved with this system. Furthermore, the cyclopropanation procedure requires addition of ethyldiazoacetate through syringe pump over a set period of time as to avoid dimerization of the carbene source.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Published: March 27, 2020

Data and Code Availability

The accession number for the compounds 1 and 2 reported in this paper is CCDC 1906978 and 1906977 (1, C2/c and P1 forms) and 1908272 (2). These data are provided free of charge by The Cambridge Crystallographic Data Centre.