Highly diastereo- and enantioselective olefin cyclopropanation via engineered myoglobin-based catalysts

Abstract

Using rational design, an engineered myoglobin-based catalyst capable of catalyzing the cyclopropanation of aryl-substituted olefins with catalytic proficiency (up to 46,800 turnovers) and excellent diastereo- and enantioselectivity (98–99.9%) was developed. This transformation could be carried out in the presence of up to 20 g / L⁻¹ olefin substrate with no loss in diastereo- and/or enantioselectivity. Mutagenesis and mechanistic studies support a cyclopropanation mechanism mediated by an electrophilic, heme-bound carbene species and a model is provided to rationalize the stereopreference of the protein catalyst. This work shows that myoglobin constitutes a promising and robust scaffold for the development of biocatalysts with carbene transfer reactivity.

Expanding the scope of engineered and artificial biocatalysts beyond the realm of chemical transformations catalyzed by natural enzymes lies at the forefront of the biocatalysis field.^[1] Olefin cyclopropanation is a particularly valuable transformation owing to the occurrence of cyclopropyl moieties in many bioactive natural and synthetic compounds. Furthermore, cyclopropanes constitute versatile intermediates for a variety of synthetically useful ring-opening transformations.^[2] A well established chemical approach to olefin cyclopropanation involves transition metal-catalyzed decomposition of diazo reagents followed by metallocarbenoid insertion into C=C double bonds.^[3] A wide range of transition metal-complexes have demonstrated utility in this respect, with the use of chiral ligands enabling these reactions to proceed in an asymmetric manner.^[3] Despite this progress, achieving high levels of both diastero- and enantioselectivity, also in combination with high catalytic activity, has remained a significant challenge in these processes, particularly in the context of intermolecular cyclopropanation reactions in the presence of acceptor-only carbene donors.^[4]

Pioneering studies by Callot, Kodadek, and Woo demonstrated the ability of metalloporphyrins to promote olefin cyclopropanation in the presence of diazoacetates.^[5] More recently, Arnold and coworkers reported that a similar reactivity is exhibited by P450_BM3, with engineered variants of this P450 enzyme catalyzing the cyclopropanation of styrene in the presence of ethyl diazoacetate (EDA) with good Z diastereoselectivity (up to 84% d.e.) and good to high enantioselectivity (90–99% e.e._(Z)).^{[1f, 6]} Our group recently discovered that, along with other heme-containing proteins, myoglobin is able to activate arylsulfonyl azides in intramolecular nitrene C—H insertion reactions,^{[1g, 7]} suggesting that this hemoprotein could be useful also to promote mechanistically related carbene transfer processes. Here, we report the rational design of engineered myoglobin-based catalysts that can support the cyclopropanation of a variety of aryl-substituted olefins with catalytic proficiency as well as excellent E-diasteroselectivity and enantioselectivity.

The oxygen-binding metalloprotein myoglobin contains a heme (iron-protoporphyrin IX) cofactor coordinated at the proximal side via a histidine residue. Because of its small size (17 kDa) and robustness toward mutagenesis and other structural modifications,^[8] we selected this protein as a potentially promising scaffold for developing biocatalysts to promote ‘non-native’ transformations such as nitrene^{[1g, 7]} and carbene transfer reactions. In initial studies, we tested the ability of sperm whale myoglobin (Mb) to catalyze the cyclopropanation of styrene (1a) in the presence of ethyl diazoacetate (2) as carbene source. Under reducing and anaerobic conditions, Mb was found to effectively promote this reaction supporting about 180 turnovers and leading to (E)-ethyl 2-phenylcyclopropanecarboxylate (3a–b) as the major product (86% d.e.) (Table 1). Notably, this cyclopropanation activity compares well with that reported for the P450_BM3-based variants in vitro (200–360 total turnovers)^[1f] under similar reactions conditions (0.02 mol% protein, 3:1 styrene:EDA ratio), while exhibiting complementary diastereoselectivity. Despite its promising activity, wild-type Mb showed no asymmetric induction in the cyclopropanation reaction, leading to a racemic mixture for both the Z or E product as observed for free hemin (Table 1).

Table 1.

Activity and selectivity of wild-type myglobin and its variants toward styrene cyclopropanation with ethyl diazoacetate.^[a]

Catalyst	Conv.[b]	TON	d.e.(E) (%)	e.e.(E) (%)[c]	e.e.(Z) (%)[c]
Hemin	29%	145	74	0	0
Mb	36%	180	86	6	0
Mb(L29A)	38%	190	82	−1	1
Mb(H64V)	73%	365	92	2	−1
Mb(V68A)	56%	280	96	68	−1
Mb(V68F)	52%	260	98	>99.9	26
Mb(F43V)	41%	205	88	44	15
Mb(F43W)	45%	225	88	34	3
Mb(F43V,V68A)	91%	455	36	67	71
Mb(F43V,V68F)	99%	500	78	>99.9	13
Mb(H64V,V68A)	99%	500	99.9	>99.9	−6
Mb(H64V,V68A)	99%[d]	10,000[d]	99.9	99.9	−6
Mb(H64V,V68A)	47%[e]	46,800[e]	99.9	99.9	−6

^[a]Reactions conditions: 20 μM Mb (or hemin), 30 mM styrene, 10 mM EDA, 10 mM dithionite, 16 h.

^[b]% conversion based on GC analysis and relative to the limiting reagent.

^[c]trans = (1S,2S); cis = (1R,2S) as determined by chiral GC.

^[d]Reaction conditions: 20 μM protein, 0.2 M styrene, 0.4 M EDA, 10 mM dithionite,1 h.

^[e]Reaction conditions: 2 μM protein, 0.2 M styrene, 0.4 M EDA, 10 mM dithionite, 16 h.

Control experiments showed that the absence of reductant (dithionite) or the presence of air resulted in no cyclopropanation product, indicating that ferrous myoglobin is the catalytically active species and that O₂ is deleterious to this reactivity, likely due to competition with the diazo reagent for binding to the heme. Based on these results and previous studies with metalloporphyrin catalysts,^{[4d, 5b, 5c, 9]} we hypothesized the Mb-catalyzed cyclopropanation reaction to involve a heme-bound carbene intermediate formed upon reaction of EDA with the protein in its reduced, ferrous state (Figure 1b). ‘End-on’^{[4d, 5c, 9b, 10]} attack of the styrene molecule to this heme-carbenoid species would then lead to the cyclopropanation product. While the E-selectivity of the Mb-catalyzed reaction clearly indicated that a ‘trans’ heme-carbene / styrene arrangement is preferred, the lack of enantioselectivity also suggested that the native Mb scaffold is unable to dictate facial selectivity for the styrene approach to the heme-carbene intermediate. Accordingly, we reasoned that mutation of the amino acid residues lying at the periphery of the porphyrin cofactor could provide a means to improve the diastereo- and enantioselectivity of this catalyst, possibly by imposing only one modality of attack of the styrene molecule to the heme-carbene group.

Figure 1.

a) Active site of sperm whale myoglobin (pdb 1A6K). The residues targeted for mutagenesis are highlighted in orange. b) Proposed mechanism for myoglobin-catalyzed styrene cyclopropanation with diazo esters.

Upon inspection of sperm whale Mb crystal structure,^[11] residues Phe43, His64, and Val68, were selected as promising targets for mutagenesis due to their close proximity to the distal face of the heme (Figure 1a). Specifically, a Mb variant where the distal histidine (His64) is mutated to Val, Mb(H64V), was considered as this mutation was previously found to increase the C—H amination activity of this protein on bulky arylsulfonyl azides.^[7] On the other hand, positions 43 or 68 were substituted with amino acids carrying a larger (i.e., Mb(F43W), Mb(V68F)) or smaller apolar side chain (i.e., Mb(F43V), Mb(V68A)), in an attempt to affect the catalyst selectivity in the cyclopropanation reaction by varying the steric bulk on either side of the heme (Figure 1a). Finally, a Mb(L29A) variant, which contains a mutation at a remote position not expected to directly interact with the heme-bound carbene during catalysis, was used as a negative control.

Analysis of these Mb variants revealed an important effect of the active site mutations on the activity and/or selectivity of the hemoprotein toward styrene cyclopropanation with EDA (Table 1). In particular, the H64V mutation resulted in a two-fold increase in TON, the highest among this set of single mutants, while having marginal effect on diastereo- and enantioselectivity. Conversely, all the mutations at the level of Phe43 and Val68 dramatically improved the enantioselectivity of the Mb variant as compared to wild-type Mb, resulting in formation of the (1S,2S) stereoisomer (3a) with e.e._(E) values ranging from 44% to 99.9%. The V68 substitutions also resulted in an appreciable increase in both catalytic activity (TON) and E-diastereoselectivity (Table 1). In contrast, the L29A mutation had essentially no effect on either the cyclopropanation activity or diastero- and enantioselectivity of the protein. Thus, in accord with our design strategy, the H64V mutation was particularly effective in enhancing Mb-dependent cyclopropanation activity, whereas the mutations at the level of V68 and F43 were beneficial toward tuning its diastereo- and enantioselectivity. To combine the beneficial effects of these mutations, a series of Mb double mutants were prepared (Table 1). Gratifyingly, variant Mb(H64V,V68A) was found to exhibit high activity as well as excellent E-diastereoselectivity (>99.9%) and (1S,2S)-enantioselectivity (>99.9%), and it was thus selected for further investigations.

Mb(H64V,V68A)-catalyzed cyclopropanation was determined to follow Michaelis-Menten kinetics, with estimated K_M values of ~2 mM and ~5 mM for styrene and EDA, respectively (Figure S1). To further optimize this transformation, the impact of the olefin : diazoester ratio on the efficiency of the reaction was first examined. These experiments revealed an increase in TON as the EDA to styrene ratio was raised from 1:5 to 6:1, with significant amounts (37% of total products) of dimerization byproducts (diethyl maleate and fumarate), accumulating only in the presence of a large (6-fold) excess of the diazo compound (Figure S2). Overall, a two-fold excess of EDA over styrene was found to be optimal for maximizing cyclopropanation turnovers while minimizing dimerization (<1% of total products). Notably, using this reagent ratio and a catalyst loading of 0.01 mol%, quantitative conversion of the olefin could be achieved in the presence of up to 0.2 M styrene (20 g L⁻¹) within one hour (Table S2). Furthermore, despite the reaction being biphasic at this reagent concentration, excellent levels of diastereo- and enantioselectivity (99.9% e.e. 99.9% d.e.) were maintained, indicating that the Mb variant is stable under these conditions (release of hemin from the protein would indeed lead to racemization). Quantitative conversion of the olefin in these reactions also suggested that the TON supported by the Mb catalyst are in excess of 10,000. To examine this aspect, reactions at high substrate loading (0.2 M styrene, 0.4 M EDA) were repeated using decreasing amounts of the hemoprotein (20 to 1 μM, Table S2). At a catalyst loading of 0.001 mol%, Mb(H64V,V68A) was found to support about 30,000 turnovers after 1 hour and over 46,000 total turnovers (TTN) after overnight incubation with styrene and EDA. Notably, high total turnover numbers were maintained using stoichiometric amounts of reductant relative to the Mb catalyst (TTN: 10,600). Time-course experiments also revealed that Mb(H64V,V68A)-catalyzed cyclopropanation proceeds very rapidly, with an initial rate of 1,000 turnovers min⁻¹ over the first 10 minutes and an average rate of 500 turnovers min⁻¹ over the first hour of the reaction. Overall, the catalytic efficiency of this engineered Mb rivals that of some of the most active transition metal catalysts reported to date for similar transformations (11–98,000 TTN),^{[4d, 12]} while offering greater diastero- and stereocontrol (cp. to 75–94% d.e. and 83–98% e.e.)^{[4d, 12]}. Furthermore, unlike the latter, no slow addition of the diazo reagent was required in the Mb-catalyzed reactions to minimize dimer formation.

To examine the substrate scope of the Mb variant, a variety of styrene derivatives and other olefin substrates were subjected to Mb(H64V,V68A)-catalyzed cyclopropanation in the presence of EDA. Using a catalyst loading of 0.07 mol%, efficient cyclopropanation of para- (1b–1e), meta- (1f), and ortho-substituted (1g) styrenes could be achieved with yields ranging from 69 to 92% (Table 2). Importantly, excellent levels of E-diastereoselectivity and, when measurable, of (1S,2S)-enantioselectivity were observed in each case, highlighting the broad scope of the Mb-based catalyst in terms of activity and selectivity across the substituted styrene derivatives. At lower catalyst loadings (0.001 mol%), Mb(H64V,V68A) was found to support TTNs ranging from 7,700 to 14,500 on these substrates. Analysis of reactions with α-methylstyrene (1h) and trans-β-methylstyrene showed efficient and highly diastero- and enantioselective cyclopropanation only in the case of the former, suggesting that β-substitutions on the alkene group are not tolerated by the Mb catalyst. Among alternative alkene substrates, N-methyl-3-vinyl-indole (1i) could be converted to the corresponding cyclopropanation product with high selectivity, albeit the efficiency of this reaction was compromised by the instability of this substrate in water. In contrast, no appreciable cyclopropanation activity was observed in the presence of 1-hexene or trans-penta-1,3-diene, evidencing the chemoselective reactivity of the Mb-based catalyst toward aryl-substituted olefins versus aliphatic ones.

Table 2.

Substrate scope for Mb(H64V,V68A)-catalyzed cyclopropanation.

Subst rate	Product	Conv. (TON)[a]	TTN[b]	% d.e.	% e.e.
1b	4a	77% (1,150)	7,760	99.6	n.a.[c]
1c	5a	89% (1,330)	11,670	99.9	n.a.[c]
1d	6a	92% (1,380)	12,280	99.8	99.9
1e	7a	69% (1,035)	8,660	99.9	99.9
1f	8a	73% (1,095)	14,500	99.8	99.9
1g	9a	85% (1,275)	11,700	99.4	99.9
1h	10a	86% (1,290)	8,510	97.2	96
1i	11a	10% (150)	1,790	97.4	n.a.[c]

^[a]Reactions conditions: 20 μM Mb, 30 mM styrene, 60 mM EDA. Yield is based on GC conversion.

^[b]Reactions conditions: 2 μM Mb, 200 mM styrene, 400 mM EDA.

^[c]Enantiomers could not be resolved.

To gain insights into the mechanism of Mb-catalyzed cyclopropanation, the relative rates (i.e., k_X/k_H ratios) for cyclopropanation of para-substituted styrenes (p-XC₆H₄CH=CH₂, 1b–1e) versus styrene were estimated from competition experiments in the presence of Mb(H64V,V68A) as the catalyst and methanol (20%) as cosolvent. Electron-donating substituents were found to accelerate the cyclopropanation reaction, while electron withdrawing substituents lead to reduced rates, a phenomenon consistent with cyclopropanations operated by electrophilic metal carbene intermediates.^{[4d, 5c, 9b, 12a, 13]} Furthermore, a plot of the log(k_X/k_H) values against the Hammett constants σ⁺ for the corresponding para substituents yielded a reasonably good (R² = 0.79) linear correlation with a small negative ρ⁺ of −0.34 ± 0.07 (Figure S3). This value is comparable to that measured for similar reactions with iron-porphyrin catalysts (ρ⁺ = −0.41^[9b]) and it is suggestive of a partial positive charge build-up at the level of benzylic carbon in the transition state. Unlike the latter, however, no significant secondary isotope effect (k_H/k_D = 0.96 ± 0.02 (Figure S4) cp. to 0.87^[5c]) was observed for the Mb(H64V,V68A)-catalyzed cyclopropanation of styrene vs. styrene-d₈. Thus, while these results point at subtle mechanistic differences between the two systems, the Hammett analyses support a general mechanism for the Mb-catalyzed cyclopropanation involving an electrophilic heme-carbene species analogous to that proposed for iron-porphyrin catalysts.^{[5c, 9b]}

To rationalize the activity and selectivity enhancements brought about by the mutations in Mb(H64V,V68A), a model of this protein was generated based on the available structure of the closely related Mb(H64V) variant.^[14] Inspection of the model revealed a wider opening leading the distal cavity due to the H64V mutation (Figure S5a–b), which is likely to increase the accessibility of the heme center to the diazoester and olefin substrate. The V68A mutation, on the other hand, expands the size of the distal cavity above the ‘N2’ nitrogen atom of the heme group (Figure S5c–d). In the crystal structure of Fe-(porphyrin)-carbene complexes^[9b], the carbene moiety is roughly aligned (15–20° deviation) with the diagonal N—Fe—N bonds of the porphyrin ring. Assuming a similar geometry is adopted by the heme-bound carbene, four orientations of this group are possible, i.e. two projecting the ester moiety toward the protein core (i.e. above heme atoms ‘N2’ or ‘N3’, Figure 1) and two projecting it toward the solvent-exposed face of the heme cofactor (i.e., above heme atoms ‘N1’ or ‘N4’). Among the possible arrangements for an end-on attack of styrene to this intermediate (Figure S6), the one featuring the carbene ester group above heme N2 and the styrene phenyl group extending toward the solvent (geometry I, Figures 2 and S6) is consistent with the experimentally observed E-(1S,2S)-selectivity of the catalyst and appears sterically feasible. While leading to the same cyclopropane stereoisomer, geometry VII (Figures 2 and S6) imposes steric clashes between the styrene ring and Phe43. Thus, the V68A mutation could favor I by better accommodating the carbene ester group in proximity of N2, which could explain the dramatic effect of this mutation toward improving (1S,2S)-enantioselectivity (6→99.9% e.e.). This scenario would also explain the remarkable tolerance of Mb(H64V,V68A)-induced selectivity to variations at the level of the aryl group of the olefin-—which is solvent-exposed in I—but not at the β-position, due to steric clashes between the β-substituent and the carbene ester group and/or heme porphyrin ring. Another implication of this model is that an increase in steric bulk at the level of the alkyl ester group or at of the α-carbon in the diazo reagent is expected to cause a decrease in diastero- and enantioselectivity, as these changes would disfavor I over other geometries (Figure S6). In agreement with these predictions, Mb(H64V,V68A)-catalyzed styrene cyclopropanation with tert-butyl diazoacetate (12) or ethyl diazopropanoate (13) yielded the corresponding (1S,2S) cyclopropane products (14a, 15a) with lower diastereoselectivity (82% d.e. and 74% d.e., respectively) and drastically reduced enantioselectivity (58% e.e. and 1% e.e., respectively). Thus, while further studies are clearly necessary to fully substantiate it, the proposed model can justify the stereochemical outcome of the Mb(H64V,V68A)-catalyzed cyclopropanation reactions and qualitatively predict the effect of structural modifications at the level of the diazo reagent.

Figure 2.

Proposed geometries for styrene approach to the putative heme-carbene leading to the (1S,2S)-ethyl 2-phenylcyclopropanecarboxylate stereoisomer. The orientation of the heme rings is the same as in Figure 1.

In summary, this work demonstrates that myoglobin constitutes a versatile and robust scaffold for the development of highly active and selective olefin cyclopropanation catalysts. By rational design, an engineered Mb variant capable of catalyzing the cyclopropanation of a variety of aryl-substituted olefins with an unprecedented combination of catalytic proficiency (10–46,800 TON) and excellent E-diastereo- and enantioselectivity (>99%) was obtained. The practical utility of this biocatalyst is further highlighted by its ability to operate at high reagent concentration (i.e., 0.2–0.4 M) and in presence of organic cosolvents (e.g., 20% MeOH). Mutagenesis and Hammett analyses support the intermediacy of an heme-bound electrophilic carbene species in these reactions, analogous albeit not identical to that operating in cyclopropanation reactions catalyzed by iron-porphyrins in organic solvents. Importantly, the much greater reactivity and selectivity offered by the Mb catalyst as compared to free hemin highlights the key role of the protein matrix in modulating the catalytic efficiency and stereochemical outcome of the reaction. Finally, a model was presented for rationalizing the selectivity of the Mb-based catalyst which could be useful for further tuning of this scaffold, e.g., to access other cyclopropane stereoisomers. Based on the present results, we anticipate that myoglobin-derived catalysts can prove useful for a variety of other synthetically valuable carbene transfer reactions.