Automated Flow Synthesis of Artificial Heme Enzymes for Enantiodivergent Biocatalysis

Abstract

The remarkable efficiency with which enzymes catalyze small-molecule reactions has driven their widespread application in organic chemistry. Here, we employ automated fast-flow solid-phase synthesis to access catalytically active full-length enzymes without restrictions on the number and structure of noncanonical amino acids incorporated. We demonstrate the total syntheses of iron-dependent Bacillus subtilis myoglobin (BsMb) and sperm whale myoglobin (SwMb). The synthetic enzymes displayed excellent enantioselectivity and yield in carbene transfer reactions. Absolute control over enantioselectivity in styrene cyclopropanation was achieved using synthetic L- and D-BsMb mutants, which delivered each enantiomer of cyclopropane product in identical and opposite enantiomeric enrichment. BsMb mutants outfitted with noncanonical amino acids were used to facilitate detailed structure–activity relationship studies, revealing a previously unrecognized hydrogen-bonding interaction as the primary driver of enantioselectivity in styrene cyclopropanation. We anticipate that our approach will advance biocatalysis by providing reliable and rapid access to fully synthetic enzymes possessing noncanonical amino acids.

Graphical Abstract

INTRODUCTION

Enzymatic catalysis has exerted a significant impact on the field of synthetic organic chemistry. Current research has focused on identifying and optimizing enzymes that catalyze small-molecule organic reactions,¹ pointing to a wide range of opportunities for improving chemistry catalyzed by both natural and engineered proteins. Toward this goal, directed evolution approaches have excelled, leading to a generalizable platform for the development of enzymes for a variety of enantioselective transformations.^2,3 Indeed, evolved enzymes^3–7 produced using recombinant expression techniques have been deployed on industrial scales, including in the syntheses of small-molecule therapeutic compounds such as sitagliptin,⁸ molnupiravir,⁹ and islatravir.¹⁰ The presence of noncanonical amino acids (ncAAs) in the primary sequence of enzymes has in several cases been demonstrated to exert beneficial effects on reaction performance.^11,12 Genetic code reprogramming^13–16 techniques have expanded access to proteins containing ncAAs, however, current technologies remain somewhat laborious and a majority are only able to simultaneously incorporate up to two distinct ncAAs.^{11,13,17–23} These limitations highlight the importance of developing general strategies to install multiple ncAAs into the primary amino acid sequence of proteins.

As a complement to recombinant expression techniques, we considered the total chemical synthesis of enzymes which would allow for straightforward incorporation of ncAAs without constraints on their structure and number. Rapid synthetic access to protein catalysts outfitted with ncAAs would allow for exploration of under-investigated noncanonical sequence space, providing enzymes with potential nonbiological reactivity or enhanced efficiency.^13,14 Furthermore, installation of ncAAs by chemical synthesis could be used to enhance mechanistic understanding of biocatalytic reactions with precise control over steric and electronic properties of key amino acid residues.

Recent efforts in our laboratories have resulted in the development of automated fast-flow peptide synthesis (AFPS) technology to produce full-length single domain proteins, and is well suited to those with primary sequences exceeding the length limit of conventional solid-phase peptide synthesis (SPPS).²⁴ Key to the efficacy of AFPS is the execution of coupling reactions in a flow format at elevated temperature, which results in fast (~2.5 min) and highly efficient (typically >99.7% yield) coupling/deprotection cycles. Using AFPS, full-length proteins (up to 214 AAs)²⁵ and abiotic mirror-image D-proteins²⁶ can be synthesized on the time scale of hours through successive amino acid couplings on a solid support.

In this work, we describe the first application of AFPS to the total chemical synthesis of full-length artificial heme enzymes and their use in catalytic carbene transfer reactions (Figure 1A). Our results demonstrate that AFPS provides a new tool for accessing artificial enzymes, paving the way for exploring noncanonical sequence space in enzyme engineering.

Figure 1.

Streamlined approach to obtain artificial heme enzymes using AFPS. (A) The AFPS technology uses a sequence input to provide full-length artificial proteins that are competent in biocatalytic reactions. (B) Amino acid sequences of BsMb (PDB 1UX8)²⁷ and SwMb (PDB 1A6K).²⁸ All proteins were synthesized as C-terminal primary amides.

RESULTS AND DISCUSSION

Heme Enzymes in Biocatalysis.

In nature, heme-dependent enzymes are involved in an extraordinary number of catalytic oxidative transformations and have proven to be highly successful scaffolds for synthetic small-molecule biocatalysis.²⁹ Significant efforts have been directed toward engineering heme enzymes for biocatalysis, with a particular focus on the chemistry performed by iron-carbene intermediates.³⁰ In 2013, Arnold demonstrated that cytochrome P450 enzymes from Bacillus megaterium could be evolved to catalyze highly enantioselective cyclopropanation reactions on styrene substrates with ethyl diazoacetate.³¹ Since then, a variety of other heme-dependent enzymes have been shown to be competent for enantioselective styrene cyclopropanation,^32,33 including those harboring abiological noble metal cofactors.³⁴ Beyond cyclopropanation, enzyme-supported iron carbene intermediates derived from diazo compounds have also been leveraged to perform insertion reactions into C–H, C–N, and Si–H bonds.³⁵

Total chemical syntheses of heme enzymes are rare and can be challenging, requiring multiple native chemical ligation steps or the use of ligation auxiliaries.^36–38 We hypothesized that the AFPS technology could provide rapid access in a single-shot format to artificial heme enzymes that are active in biocatalytic reactions. To test this hypothesis, we targeted the syntheses of two heme-dependent enzyme mutants, Bacillus subtilis myoglobin (BsMb)³⁹ and sperm whale myoglobin (SwMb),³³ which have previously been evolved to catalyze enantioselective olefin cyclopropanation and N–H insertion reactions, respectively (Figure 1B). We began with the synthesis of the triple mutant BsMb Y25L T45A Q49A, which was reported by Arnold to catalyze the reaction of 3,4-difluorostyrene and ethyl diazoacetate to provide the corresponding cyclopropane product in high yield (79%) and enantiomeric ratio (e.r.) (99:1 e.r._trans).³⁹

Synthetic Heme Enzymes Are Produced with AFPS.

Access to catalytically active synthetic heme enzymes was accomplished through a four-step protocol consisting of: (1) synthesis of the primary amino acid sequence by AFPS, (2) high-performance liquid chromatography (HPLC) purification, (3) refolding by dialysis, and (4) reconstitution of the folded protein with the heme cofactor (Figure 2A). Enzymes were synthesized by AFPS according to our previously reported automated flow methods²⁴ utilizing a solid support (24 μmol typical scale), in situ activated amino acids, and Fmoc deprotection. The syntheses of BsMb and SwMb proceeded in 6.5 and 7.5 h, respectively, with an average time of 2.5 min for the incorporation of each amino acid residue. Following acid-catalyzed cleavage and deprotection of the polypeptides from the resin and purification of the resulting material by preparative reverse phase HPLC, the target compounds were isolated in multimilligram quantities (~2 to 5 mg isolated, see Section 3 in the Supporting Information). The identity and homogeneity of the amino acid sequences of apo-BsMb and apo-SwMb were characterized by HPLC and liquid chromatography mass-spectrometry (LC-MS) methods (Figure 2B) (for SwMb, see Section 3.4 in the Supporting Information).

Figure 2.

Synthetic heme enzymes can be refolded and reconstituted with a heme cofactor. (A) Summary of the production workflow to obtain catalytically active heme-based enzymes. (B) Analytical HPLC chromatogram (bottom), mass-to-charge (left inset), and deconvoluted (right inset) mass spectra of purified BsMb (detailed methods can be found in Section 2 of the Supporting Information). (C) Analytical size exclusion chromatography (SEC) trace of refolded synthetic BsMb (highlighted in green) compared to a molecular weight standard (shown in gray) confirms the monomeric form of BsMb. (D) Circular dichroism (CD) spectrum of apo-BsMb obtained in 10 mM TRIS, pH 8.0 at 25 °C supports α-helical character. (E) UV–vis spectrum of synthetic BsMb reconstituted with heme cofactor shows a red-shift relative to hemin only, supporting cofactor incorporation.

A two-step protocol consisting of refolding and subsequent reconstitution with the heme cofactor was applied to access functional synthetic enzymes from the purified primary amino acid sequences. Ultimately, we found that dialysis-based methods for refolding provided optimal results in accessing the apo forms (see Section 4.1.2 and 4.1.3 of the Supporting Information). The synthetic polypeptides were dissolved in a denaturing buffer (typically 6 M guanidinium chloride, 50 mM TRIS·HCl, pH 8.0) and dialyzed for 16 h against an aqueous buffer at 4 °C (20 mM TRIS·HCl, 200 mM NaCl, pH 7.5). Following these procedures, soluble monomeric forms of folded apo-BsMb and apo-SwMb were obtained as verified by analytical size exclusion chromatography (SEC) (Figures 2C and S71). Circular dichroism (CD) spectra of the apo forms of BsMb and SwMb obtained following dialysis protocols displayed minima at approximately 210 and 220 nm, consistent with largely α-helical secondary structures, matching previously reported data for recombinant BsMb⁴⁰ and SwMb³⁴ (Figures 2D and S76).

With access to the apo enzymes, we examined methods for reconstitution with the heme cofactor, exploiting its tight binding with the globin scaffold.⁴¹ Following incubation of Feprotoporphyrin IX possessing an axial chloride (hemin chloride) with apo-BsMb, successful reconstitution was verified with analytical SEC and in-line UV–vis spectroscopy (Figure S77). Importantly, the UV–vis spectrum of reconstituted BsMb displays a significant (~50 nm) red-shift of the Soret band (λ_max = 412 nm) relative to the heme cofactor alone, consistent with productive binding (Figure 2E).²⁷ SwMb was prepared through a similar protocol using the corresponding AFPS-generated polypeptide, with the resulting synthetic enzyme displaying a comparable red-shift in its UV–vis spectrum (Figure S78). Additionally, a negative control experiment involving incubation of bovine serum albumin (BSA) with hemin chloride exhibited negligible changes in the UV–vis profile, supporting highly specific binding between the heme proteins and their cofactor (Figure S78).⁴²

Synthetic Heme Enzymes Catalyze Carbene Transfer Reactions.

With access to reconstituted synthetic enzymes, we assessed their performance in catalytic carbene transfer reactions. We began by subjecting BsMb to similar conditions reported by Arnold for the reaction of 3,4-difluorostyrene 1 and ethyl diazoacetate 2. Employing a low catalyst loading of BsMb (0.2 mol %) in aqueous potassium phosphate buffer, the cyclopropyl product (R,R)-3 was formed in good yield and with excellent enantioselectivity (72% and 95:5 e.r._trans) in 2 h at room temperature (Figure 3A, entry 1). These values compare well with results reported by Arnold obtained with whole-cell lysates harboring mutant BsMb (79% yield and 99:1 e.r._trans).³⁹ In an effort to avoid potential methionine S-oxidation⁴³ or modification by carbene transfer,⁴⁴ we synthesized a BsMb mutant (BsMb^Nle) in which all five methionine residues were substituted with the ncAA L-norleucine (Nle), a close structural analog of methionine that features a methylene group in place of the sulfur atom (Figure 3B). Such a modification would not be expected to dramatically impact the overall conformation of BsMb. Refolding and reconstitution were performed using identical protocols (Figure S70) and the CD spectrum of BsMb^Nle confirmed the overall α-helical character of the protein (Figure S76). UV–vis spectroscopy was used to demonstrate successful incorporation of heme (Figure S78). When assayed under standard reaction conditions, BsMb^Nle displayed comparable enantioselectivity relative to BsMb, delivering (R,R)-3 in 99:1 e.r._trans with good yield (74%) (Figure 3A, entry 3). Taken together, these data support that both BsMb and BsMb^Nle can be accessed in their active conformations and display similar catalytic activity.

Figure 3.

Artificial heme enzymes catalyze stereoselective olefin cyclopropanation and N–H bond insertion. (A) Enantioselective cyclopropanation catalyzed by BsMb. Reaction conditions: 20 μM of Mb mutant, 30 mM of 1, 10 mM of 2, 10 mM of Na₂S₂O₄, 10% MeOH, 2 h, r.t., 250 μL scale, reaction performed under N₂ atmosphere. *Data for recombinant BsMb (entry 1) was obtained from reported data in the literature (TON was not reported).³⁹ (TON = turnover number, n.d. = not determined, d.r. = diastereomeric ratio, e.r. = enantiomeric ratio). (B) Sequence of BsMb^Nle with mutations highlighted in green (PDB 1UX8). BsMb^Nle was obtained as C-terminal amide. (C) UV–vis spectrum of reconstituted CR-BsMb^Nle relative to the spectrum of the hemin cofactor. (D) BsMb^Nle catalyzed cyclopropanation employing α-methylstyrene. (E) SwMb-catalyzed N–H bond insertion reaction. Reaction conditions: 30 μM of SwMb variant, 10 mM of 4, 5 mM of 5, 10 mM of Na₂S₂O₄, 5% MeOH, 12 h, r.t., 250 μL scale, reaction performed under N₂ atmosphere. All yields, d.r. and e.r. values were evaluated based on GC analysis using dodecane as an internal standard, relative to 2. All reactions were performed in technical triplicate.

Direct Folding Bypasses HPLC Purification.

With the aim of reducing the time and resources needed to access active enzymes, we explored the possibility of bypassing HPLC purification of the primary amino acid sequence following synthesis by AFPS as recently demonstrated by our laboratories.⁴⁵ This simplification of the production workflow would accelerate the experimental evaluation of synthetic enzymes, akin to using whole-cell lysates in the screening of enzymes produced by directed evolution.⁴⁶ Following AFPS synthesis of the 132 AA sequence of BsMb^Nle and cleavage of the polypeptide from resin, we directly subjected the resulting material to dialysis refolding conditions previously optimized for synthetic BsMb^Nle. After reconstitution with the heme cofactor, crude BsMb^Nle (termed CR-BsMb^Nle) was fully characterized and displayed a Soret band red-shift in the UV–vis spectrum consistent with successful incorporation of heme (Figure 3C). When deployed in the cyclopropanation reaction between 1 and 2, CR-BsMb^Nle provided (R,R)-3 in 66% yield and slightly reduced 90:10 e.r._trans compared to the HPLC-purified material (Figure 3A, entry 4). Despite the low purity of CR-BsMb^Nle (estimated 19% based on absorbance, Figures S74 and S75), relatively high enantioselectivity was still achieved, demonstrating the utility of this strategy in bypassing the significant time bottleneck associated with HPLC purification. Hence, synthetic enzymes can be used to perform stereoselective catalysis without purification, an outcome which we anticipate could accelerate the discovery and development of synthetic biocatalysts.

Additional Reactions Catalyzed by Synthetic Heme Enzymes.

We sought to demonstrate the ability of BsMb^Nle to catalyze cyclopropanation reactions on additional substrates. Enzymatic cyclopropanation reactions of 1,1-disubstituted styrenes are relatively rare,^33,47–51 and we found that BsMb^Nle promoted highly enantioselective conversion of α-methylstyrene 4 to cyclopropanated product 5 (70% yield, 98:2 e.r._trans) possessing a quaternary center using low catalyst loading (0.04 mol %, Figure 3D). We further evaluated the performance of BsMb^Nle on a larger scale (0.1 mmol with respect to 2) in the cyclopropanation reaction between 2 and 4. The reaction was found to provide (R,R)-5 in high isolated yield (84%, 17 mg) and excellent enantioselectivity (96:4 e.r._trans) using a 10-fold reduction in catalyst loading (0.02 mol %).

In addition to enantioselective cyclopropanation reactions, we investigated the chemistry catalyzed by another heme-dependent enzyme, sperm whale myoglobin (SwMb). Synthetic SwMb H64 V V68A was found to be active toward N–H insertion reactions using ethyl diazoacetate, as reported by Fasan and co-workers.⁵² Incubating 4-chloroaniline 6 and ethyl diazoacetate 2 with SwMb H64 V V68A at 0.2 mol % loading led to the formation of product 7 in 82% yield (Figure 3E). These results compare well with the reported performance of recombinant SwMb in this reaction, and further demonstrate that AFPS can be used generally to produce active heme-based enzymes for biocatalysis.⁵²

Enantiodivergent Catalysis Is Achieved with Synthetic D-Heme Enzymes.

While directed evolution campaigns have excelled at delivering enzymes for the production of a specific enantiomer of a product, evolving catalysts from the same starting sequence that generate the opposite enantiomer of product is often challenging and not generalizable.^53–58 In practice, such an undertaking essentially involves performing directed evolution workflows twice, and though the sense of asymmetric induction can be inverted, enantioselectivity values are generally not equal and opposite.⁵⁹ Computational tools have made important progress in predicting enantiocomplementary mutants, but have not yet been demonstrated to be widely applicable.^60,61 In principle, synthetic mirror-image D-enzymes^62,63 could predictably deliver the opposite enantiomer of product with unaltered efficiency and selectivity, bypassing multiple rounds of evolution.

Hence, we aimed to demonstrate that mirror-image D-BsMb^Nle could be accessed with AFPS and applied to the cyclopropanation of 1 to generate the (S,S) enantiomer of 3 (Figure 4A). Notably, recombinant methods to express D-proteins have not yet been developed, necessitating total chemical synthesis as the only current means to access mirror-image proteins.^26,62,64 Using AFPS technology, D-BsMb^Nle was synthesized in ~6.5 h starting from commercially available D-amino acids. The homogeneity and purity of the primary amino acid sequence of synthetic D-BsMb^Nle were confirmed by HPLC and LC-MS methods (see Section 3.3 in the Supporting Information). Following refolding and reconstitution, D-BsMb^Nle was characterized by CD to display opposite spectral features compared to its L-analog (Figure 4B) and by UV–vis to effectively bind heme (Figure 4C). In the model cyclopropanation reaction between styrene 1 and ethyl diazoacetate 2, D-BsMb^Nle provided the cyclopropyl product of opposite configuration (S,S)-3 in identical and reversed enantioselectivity (1:99 e.r._trans) without erosion of yield (73%; Figure 4D). The synthetic D-BsMb^Nle variant is readily available as an enantiocomplementary mutant, requiring no additional engineering efforts for its generation. To the best of our knowledge, this is the first reported application of a full-length D-enzyme in small-molecule biocatalysis.

Figure 4.

Artificial D-BsMb^Nle is active for enantiodivergent cyclopropanation. (A) A schematic comparison of the enantiomeric forms of BsMb^Nle. Both proteins were obtained as C-terminal primary amides. (B) CD spectrum of apo-D-BsMb^Nle and apo-L-BsMb^Nle obtained in 10 mM TRIS·HCl, pH 8.0 at 25 °C. (C) UV–vis spectrum of reconstituted D-BsMb^Nle confirms successful incorporation of the heme cofactor. (D) Mirror-image enantioselective cyclopropanation catalyzed by D-BsMb^Nle. Reaction conditions: 20 μM of D-BsMb^Nle, 30 mM of 1, 10 mM of 2, 10 mM of Na₂S₂O₄, 5% MeOH, 2 h, r.t., 250 μL scale, reaction performed under N₂ atmosphere (left). Stacked chiral GC traces obtained from crude reaction mixtures with Rh₂(AcO)₄ yielding a racemic mixture, L-BsMb^Nle, and D-BsMb^Nle (right). All yields, d.r., and e.r. were evaluated based on GC analysis using dodecane as an internal standard, relative to 2. All reactions were performed in technical triplicate.

BsMb Mutants Containing Noncanonical Amino

Acids Offer Mechanistic Insights.

The limited structural diversity of canonical AAs can restrict the breadth of mechanistic insight achievable through mutagenesis studies. We considered that ncAAs could enable focused examination of specific noncovalent interactions by precise tuning of electronic and/or steric effects. Historically, mechanistic enzymology has benefitted from studies involving mutants possessing ncAA in place of native residues.^14,65,66 We envisioned that AFPS could deliver ncAA-bearing BsMb analogs required for precise structure–activity characterization, an advance that could offer unprecedented insight into enzymatic carbene transfer behavior.

Guided by crystallographic information available for wild-type BsMb (PDB 1UX8), we identified several aromatic amino acids located in proximity to the Fe-heme cofactor that could be involved in enantioinduction (Figure 5A). Using AFPS, we incorporated a panel of ncAAs at F38, H76, and W89 of BsMb^Nle to elucidate the origins of enantioselectivity in the model cyclopropanation reaction for the formation of (R,R)-3 (Figure 5B). In all cases, mutant enzymes were characterized by HPLC and LC-MS analysis (see Sections 3.5 to 3.13 in the Supporting Information) and were found to exist as soluble monomers (Figures S72 and S73) with each producing a significant red-shift in the UV–vis Soret band upon reconstitution, consistent with productive heme binding (Figure S79).

Figure 5.

Noncanonical amino acids inform the origins of enantioselectivity in BsMb-catalyzed cyclopropanation. (A) Three-dimensional (3D) model of BsMb^Nle with the iron carbene intermediate (in green) modeled in the active site (adapted from crystal structure PDB 1UX8) (left). Enlarged depiction of the active site region with the side chains selected for engineering with ncAAs (right). (B) Enantioselective cyclopropanation catalyzed by BsMb F38, H76, and W89 mutants. Reaction conditions for model reaction: 20 μM of Mb mutant, 30 mM of 1, 10 mM of 2, 10 mM of Na₂S₂O₄, 5% MeOH, 2 h, r.t., 250 μL scale, reaction performed under N₂ atmosphere. All yields, d.r. and e.e. were evaluated based on GC analysis using dodecane internal standard, relative to 2. All reactions were performed in technical triplicates (e.e. = enantiomeric excess). (C) Proposed roles of F38, H76, and W89 during the myoglobin catalyzed carbene insertion reaction.

Fasan and Zhang have provided computational evidence to support the presence of stabilizing aromatic interactions between the styrene substrate and a key phenylalanine residue within the active site of a related myoglobin that catalyzes enantioselective styrene cyclopropanation.⁶⁷ BsMb possesses an analogous phenylalanine residue (F38) in proximity to the heme binding site that could play a similar role. To test this hypothesis, we used AFPS to synthesize a mutant in which F38 was replaced with a ncAA bearing a cyclohexyl substituent—which cannot engage in such a π-interaction—in place of the corresponding phenyl ring (Figure 5A). When used in the model reaction, BsMb^Nle F38a led to a significant decrease in the yield of 3 (30 vs 74%) while maintaining high levels of enantioselectivity (98:2 e.r._trans). These results are consistent with the proposal that F38 plays an insignificant role in asymmetric induction but is involved in effective catalysis, possibly by preorganization of the styrene nucleophile via attractive aromatic interactions. Additionally, given the known impact of the axial ligand on the chemistry catalyzed by hemedependent cytochrome enzymes, we investigated mutations at H76.^32,68–72 Electronic tuning of the axial ligand on Fe was achieved through replacement of the imidazole of H76 with oxazole (BsMb^Nle H76b), thiazole (BsMb^Nle H76c), and N-methylimidazole (BsMb^Nle H76d) (Figure 5B). All mutants proved to be catalytically active toward cyclopropanation to provide 3 in similar yields relative to BsMb^Nle and with varying levels of enantiomeric purity.

Finally, we probed the role of W89 in asymmetric induction using a series of ncAAs, ultimately demonstrating the importance of the heterocyclic N–H bond. Incorporating an N-methylated tryptophan analog in BsMb^Nle W89e enabled precise removal of the indole N–H bond. When used in the model reaction, the W89e mutant delivered 3 in racemic form and only moderately suppressed yield (56%). Similarly, replacement of the indole group with benzofuran (BsMb^Nle W89f), benzothiophene (BsMb^Nle W89g), and 1-naphthyl (BsMb^Nle W89h) analogs lacking the indole N–H bond also led to the formation of 3 with moderate yields and in essentially racemic form (Figure 5B). In light of the negligible racemic background reaction involving free heme (Figure 3A, entry 5), these results strongly implicate the indole N–H bond of W89 in enantiodiscrimination in this reaction.¹² Indeed, high enantioselectivity was recovered when using BsMb^Nle W89i, which possesses a 6-chloroindole substituent with an unaltered heterocyclic N–H bond. To gain further insight in the geometry of the iron carbene intermediate and its interactions with active site residues, we performed density functional theory (DFT) calculations (see Section 7 in the Supporting Information). The optimized geometry of the carbene intermediate displayed a clear hydrogen bonding contact between the W89 indole N–H bond and the carbonyl oxygen of the iron carbene (N···O distance of 1.9 Å, Figure S81). In contrast, a separate calculation employing the same methods, but with W89 mutated to benzofuran (as in BsMb^Nle W89f), revealed a markedly different carbene orientation and significantly increased distance between benzofuran and carbonyl oxygen atoms (O···O distance of 3.6 Å, Figure S82). Taken together, these computational and experimental structure–activity data are consistent with an anchoring hydrogen-bonding interaction between the carbonyl of the metal-carbene intermediate and W89 as being the primary driver of enantioselectivity (Figure 5C). Corroborating this hypothesis, decreasing the coordinating ability of the axial ligand on Fe in mutants BsMb^Nle H76b and BsMb^Nle H76c resulted in diminished enantioselectivity, presumably by lowering the electron density on the carbonyl oxygen and rendering it a weaker hydrogen-bond acceptor. Overall, these studies further support the utility of AFPS as an enabling tool for mechanistic enzymology.

CONCLUSIONS

This work demonstrates that artificial heme enzymes obtained through total chemical synthesis using AFPS technology exhibit catalytic activity on-par with traditionally produced recombinant enzymes. In a streamlined workflow consisting of polypeptide synthesis, purification, refolding, and reconstitution with heme, fully synthetic enzymes were accessed in up to milligram quantities. Synthetic enzymes displayed excellent performance in model enantioselective cyclopropanation and N–H bond insertion reactions, yielding highly enantiomerically enriched products. Unpurified synthetic enzymes could also be used efficaciously as catalysts, highlighting AFPS total synthesis as a viable method for the rapid evaluation of enzyme mutants. Mirror-image D-BsMb^Nle was shown to catalyze the formation of the opposite enantiomer of cyclopropane product and straightforward incorporation of ncAA at programmed sites in BsMb^Nle was leveraged to provide mechanistic insight into the origins of asymmetric induction, revealing the role of an active site hydrogen-bond. AFPS technologies represent a complementary method to recombinant or cell free expression techniques. Further optimizations are expected to decrease solvent and reagent usage, and enhance coupling efficiencies to access longer peptide sequences beyond current limits.²⁵ Heme-dependent myoglobin enzymes are important biocatalysts known to catalyze a wide range of nonbiological reactions, and our current efforts are directed toward exploring new chemistry carried out by these enzymes. Beyond the application to heme enzymes described in this report, we anticipate that the strategies disclosed herein will provide a useful complementary method for the field of biocatalysis in general by delivering fully synthetic designer enzymes.