Biocatalytic Carbene Transfer Using Diazirines

Abstract

Biocatalytic carbene transfer from diazo compounds is a versatile strategy in asymmetric synthesis. However, the limited pool of stable diazo compounds constrains the variety of accessible products. To overcome this restriction, we have engineered variants of Aeropyrum pernix protoglobin (ApePgb) that use diazirines as carbene precursors. While the enhanced stability of diazirines relative to their diazo isomers enables access to a diverse array of carbenes, they have previously resisted catalytic activation. Our engineered ApePgb variants represent the first example of catalysts for selective carbene transfer from these species at room temperature. The structure of an ApePgb variant, determined by microcrystal electron diffraction (MicroED), reveals that evolution has enhanced access to the heme active site to facilitate this new-to-nature catalysis. Using readily prepared aryl diazirines as model substrates, we demonstrate the application of these highly-stable carbene precursors in biocatalytic cyclopropanation, N–H insertion, and Si–H insertion reactions.

Graphical Abstract

Over the past decades, biocatalysis has enabled the efficient and sustainable production of valuable compounds by drawing on the phenomenal activity and selectivity of enzymes.¹ Among the surge of novel biotransformations implemented in the past few years, hemoprotein-catalyzed carbene transfer reactions² have emerged as powerful new-to-nature transformations (Scheme 1a), including cyclopropanation,³ bicyclobutanation,⁴ and C–C bond formation.⁵ Though these enzymes yield many types of products, their synthetic potential has been limited by the small range of accessible carbenes. All efforts on these systems to date have used diazo compounds such as ethyl diazoacetate (EDA), which require the presence of an electron-with-drawing group (EWG) adjacent to the diazo moiety to be safely manipulated.⁶ In the absence of such substituents the diazo reagent is unstable and potentially explosive, making the corresponding carbenes unattractive for use at scale. More stable carbene precursors would allow access to a wider variety of iron carbenoid intermediates and promote preparative and industrial applications.

Scheme 1.

Background and summary

Diazirines are cyclic isomers of diazo compounds and are used as probes in photoaffinity labeling and chemical proteomics (Scheme 1b).^7–9 These well-studied carbene precursors are more stable than the corresponding diazo compounds and, as a result, do not require an EWG substituent to be handled safely. The tradeoff of increased stability is that more forcing conditions (high temperatures or UV photolysis, e.g.) are required for diazirine activation,¹⁰ resulting in low yields or poor selectivities in synthetic applications.^11,12 Main group Lewis acids have been shown to activate diazirines, albeit with limited synthetic utility.¹³ Meanwhile, in contrast to the well-explored transition metal-catalyzed carbene transfer from diazo compounds,¹⁴ the few examples of analogous diazirine-derived metallocarbenes known to form at room temperature have never demonstrated productive catalytic carbene transfer.¹⁵ This is unfortunate as diverse diazirines are readily made via numerous synthetic routes,^16–18 including the recent single-step conversion of α-amino acids to 3H-diazirines.¹⁹ The development of biocatalysts capable of using diazirines as carbene precursors could dramatically expand the scope of carbenes accessible in enzymatic transformations. In this work we introduce such a biocatalyst (Scheme 1c), demonstrating for the first time that stable diazirines can be biocatalytically activated to a carbene under mild conditions and productively transferred in multiple types of reactions.

Given the isomeric relationship between diazirines and diazo compounds, we began by assessing whether a hemoprotein might also accept a diazirine as a substrate. We first tested whether a Rhodothermus marinus cytochrome c variant that effects B–H insertion into borane 1 using diazo compound 2 as a carbene donor (Figure 1a)²⁰ could utilize the isomeric diazirine 4 to access the same putative iron-carbenoid intermediate and yield organoborane 3. Although the cytochrome c variant displayed no activity for this transformation, a control reaction containing free heme alone exhibited trace levels of product formation (Table S1). Emboldened by this observation, we screened a collection of engineered heme proteins for the same B–H insertion reaction using diazirine 4 as the carbene source. Gratifyingly, Aeropyrum pernix protoglobin variant Y60G F145Q (ApePgb GQ) exhibited formation of the organoborane product 3 in 0.9 % yield in a whole E. coli cell context, providing a reasonable starting point to assess reaction evolvability (Figure 1b). This enzyme was previously engineered for the cyclopropanation of unactivated alkenes using EDA.²¹ ApePgb is an attractive platform for enzyme engineering because its high thermostability²² allows it to accept activating mutations that are often destabilizing.²³ Additionally, its high expression level in Escherichia coli (>200 mg/L culture) means that protoglobin-based biocatalysts can be produced with ease and at low cost. Following validation of this activity with purified enzyme, amino acid positions near the enzyme active site were subjected to site-saturation mutagenesis (SSM) and the resultant libraries were screened for improved B–H insertion activity (Figure 1c). Iterative screening of these sites yielded a new variant, ApePgb W59L Y60V F145Q (ApePgb LVQ), that exhibited a nearly three-fold increase in organoborane yield. Notably, the mutations had a pronounced effect on enzyme activity only when using diazirine 4 as a carbene source, as yield of 3 with diazo compound 2 exhibited little change across these variants (Figure S1). This confirms that enzymatic activation of a diazirine can be enhanced by directed evolution.

Figure 1.

(a) Previous biocatalytic B–H insertion reaction catalyzed by BOR^P*, a variant of Rhodothermus marinus cytochrome c. (b) Reaction targeted in enzyme discovery substitutes the diazo with the equivalent diazirine carbene source. c) Amino acid positions targeted in evolution (orange) shown on an ApePgb GQ homology model (blue) alongside evolved improvements in B–H insertion yield in whole cell reactions (OD₆₀₀ = 60).

Having demonstrated that carbene transfer from diazirine 4 was both enzyme-catalyzed and amenable to evolution, we then set out to pursue the primary objective: selective enzymatic transfer of a carbene bearing no EWG substituent. We decided to pursue 3-phenyl-3H-diazirine 6a due to its similarity to 4 and its ease of synthesis. Notably, the corresponding diazo isomer, phenyldiazomethane, is known to be unstable and explosive.⁶ Successful carbene transfer from 6a would enable safe biocatalytic access to the phenyl carbene. However, the use of a mono-substituted diazirine removed stereochemistry in the B–H insertion product. Thus, to assess the selectivity conferred by the enzyme, we shifted focus to a carbene transfer activity yielding a chiral product. Using 6a as a carbene source, ApePgb LVQ catalyzed the cyclopropanation of acrylate 5 to give chiral cyclopropane 7a at <1% yield with 2:1 cis:trans diastereoselectivity. Free heme also catalyzed this transformation, albeit at a much lower level, and heavily favored production of trans-7a. Thus, we targeted production of the higher energy diastereomer cis-7a for directed evolution of the enzyme.

ApePgb variant libraries made by SSM and random mutagenesis were screened for enhanced production of the desired cis-cyclopropane. Five sequential rounds of directed evolution resulted in variant ApePgb GLAVRSQLL bearing mutations C45G, W59L, Y60A, G61V, V63R, C102S, F145Q, I149L, and F175L relative to wild-type ApePgb. This variant exhibited >150-fold improvement in cis-7a formation in whole E. coli cells relative to ApePgb LVQ, producing 7a in 28% yield with 6:1 cis:trans diastereomeric ratio (dr) and 86:14 enantiomeric ratio (er) for the cis- isomer (Figure 2). Using optimized conditions developed with purified enzyme resulted in 53% yield with 6.9:1 cis:trans dr and 80:20 er (Figure 3a).

Figure 2.

Cyclopropanation reaction targeted in this work using E. coli cells harboring ApePgb variants (relative product stereochemistry shown) and yields for each sequentially evolved variant, showing a >150-fold improvement overall.

Figure 3.

(a) Enzyme-catalyzed cyclopropanation of benzyl acrylate with aryldiazirines. Reactions were performed anaerobically at room temperature using purified ApePgb GLAVRSQLL in 50 mM potassium phosphate (pH = 8.0) and 150 mM NaCl. Yields and diastereomeric ratios were quantified by GC-FID. Enantioselectivities were quantified by HPLC (7a) or SFC (7b & 7c) on a chiral stationary phase (details in SI). (b) Alternative carbene transfer reactions carried out using purified enzyme under identical conditions to those shown in (a), only substituting the shown substrate for benzyl acrylate. All yields and dr measurements were quantified by GC-FID.

Having reached good activity for the cyclopropanation reaction, we were curious how substitutions on the aryl ring of the diazirine are accommodated by the enzyme. To this end, p-fluoro- (6b), p-chloro- (6c), and p-methoxyphenyl-3H-diazirine (6d) were synthesized and used in activity assays. Diazirine 6b gave 69% yield of cyclopropane 7b with 8.5:1 cis:trans dr and 88:12 er, and 6c gave 64% yield of cyclopropane 7c with 9.6:1 cis:trans dr and 81:19 er. (Figure 3a). No product formation was detected by GC-MS using 6d, suggesting that the presence of either electron-donating or bulky substituents on the ring impedes binding or catalysis. Additionally, 3-benzyl-3H-diazirine 6e was tested to assess activity with an alkyl diazirine. No product was observed by GC-MS, indicating that deeper understanding of enzymatic diazirine activation may be required to access carbenes from such diazirines.

We next assessed the ability of ApePgb GLAVRSQLL to catalyze other carbene transfer reactions (Figure 3b). Using 6a as a carbene source, the enzyme cyclopropanated styrene 8, affording 1,2-diphenylcyclopropane 9 in 61% yield with 12:1 cis:trans dr. We also tested for enzymatic C–X bond formation, using aniline 10, silane 12, and anisole 14 as substrates to assess N–H, Si–H, and C–H insertion activities, respectively. ApePgb GLAVRSQLL produced the N–H insertion product 11 in 15% yield and Si–H insertion product 13 in 5% yield. However, none of the corresponding C–H insertion product was observed by GC-MS. The formation of 11 and 13 by ApePgb GLAVRSQLL illustrates that this enzyme can serve as a good starting point for the evolution of variants catalyzing such X–H insertion reactions.

Taken together, the accommodation of different diazirines and substrates, including those for unoptimized activities, highlights the functional flexibility of the ApePgb enzyme scaffold. To interrogate the functional changes introduced through evolution, we determined the first structure of an ApePgb variant, GLVRSQL, by microcrystal electron diffraction (MicroED). In general, the protoglobin fold is maintained with an overall r.m.s.d. of <0.8 Å across backbone atoms in comparison to structures of M. acetivorans protoglobin (76% sequence similarity), the only homolog whose structure has been reported.^24,25 Additionally, the porphyrin ring of the heme cofactor exhibits the ruffled distortion present in other protoglobin structures.²⁶ However, residues 60–70, which adopt a rigid helical conformation in the homologous protein, appear to be restructured as a loop in the MicroED structure of ApePgb GLVRSQL (Figure 4). This region contains the mutation V63R, the first mutation introduced during evolution for cyclopropanation chemistry, which resulted in a 14-fold boost in product yield. This boost reasonably stems from the unwinding of the rigid helix into a more flexible/dynamic loop, which creates a larger cavity around the enzyme active site and likely facilitates access to the heme iron. Two mutations subsequently introduced into this region, V60A and G61V, contributed to an additional 4-fold boost in cyclopropane yield. This structure-function analysis, further elaborated in the Supplementary Information, suggests that rearrangement of the polypeptide backbone around the newly formed active site of ApePgb enhances enzymatic diazirine activation.

Figure 4.

Superposition of Y61A M. acetovorans protoglobin (gray; PDB 3ZJI) with the four monomers observed in the MicroED structure of ApePgb GLVRSQL (blue) with positions 60–70 in orange, showing the reconfigured conformation of these residues relative to the helical conformation in the M. acetovorans protoglobin (76 % sequence similarity). Dashed lines represent regions disordered in subunits B, C, and D of the unit cell in comparison with the resolved loop in subunit A.

Given the mild conditions required for this enzymatic activity, we sought to further interrogate the chemical details underlying this catalysis. During condition optimization with purified protein, control experiments demonstrated that ambient light is not required for diazirine activation as reactions run under strictly dark conditions exhibited comparable yield (Figure S2). We hypothesized that the enzyme may isomerize the diazirine to the corresponding diazo in situ, then perform carbene transfer in a manner similar to that previously demonstrated by other heme enzymes. To test this hypothesis, we used dibenzocyclooctyne amine S1 (DBCO-NH₂) as a probe for the diazo intermediate. In the presence of a diazo compound, the DBCO-NH₂ alkyne undergoes a rapid [3 + 2] cycloaddition to afford the pyrazole product S2 (Figure S3a).²⁷ Intriguingly, when DBCO-NH₂ was added to a sample of ApePgb GLAVRSQLL in the presence of 6a, the pyrazole product was observed at a higher level than for either the same concentration of free heme or heat-denatured enzyme (Figure S3b). Since the concentration of pyrazole product S2 is directly related to the amount of diazo compound in solution,²⁷ this is consistent with enzymatic isomerization of diazirine 6a to the corresponding diazo isomer followed by capture by DBCO-NH₂. This suggests that ring opening is a potential mechanism for diazirine activation by ApePgb GLAVRSQLL.

In conclusion, this work presents the first example of catalytic diazirine activation and subsequent selective carbene transfer; this new-to-nature chemistry is accomplished enzymatically under mild aqueous conditions and requires no exogenous heat or light. We believe that the enzymatic platform will open a whole new set of transformations accessible by biocatalytic carbene transfer. For example, access to arylcarbenes, as shown here, would otherwise require hazardous diazo compounds. Encouraged by these results, we envision that further engineering of protoglobin variants could provide access to even more challenging species, such as alkylcarbenes, and provide access to more X–H insertion reactions based on the nascent activities presented here. The potential synthetic scope achievable with diazirines is both exciting and promising for the advancement of biocatalysis.