Enzymes that catalyze carbon–silicon bond formation are unknown in nature, despite the natural abundance of both elements. Such enzymes would expand the catalytic repertoire of biology, enabling living systems to access chemical space previously only open to synthetic chemistry. We have discovered that heme proteins catalyze the formation of organosilicon compounds under physiological conditions via carbene insertion into silicon–hydrogen bonds. The reaction proceeds both in vitro and in vivo, accommodating a broad range of substrates with high chemo- and enantioselectivity. Using directed evolution, we enhanced the catalytic function of cytochrome c from Rhodothermus marinus to achieve more than 15-fold higher turnover than state-of-the-art synthetic catalysts. This carbon–silicon bond-forming biocatalyst offers an environmentally friendly and highly efficient route to producing enantiopure organosilicon molecules.

To reduce experimental effort associated with directed protein evolution and to explore the sequence space encoded by mutating multiple positions simultaneously, we incorporate machine learning into the directed evolution workflow. Combinatorial sequence space can be quite expensive to sample experimentally, but machine-learning models trained on tested variants provide a fast method for testing sequence space computationally. We validated this approach on a large published empirical fitness landscape for human GB1 binding protein, demonstrating that machine learning-guided directed evolution finds variants with higher fitness than those found by other directed evolution approaches. We then provide an example application in evolving an enzyme to produce each of the two possible product enantiomers (i.e., stereodivergence) of a new-to-nature carbene Si–H insertion reaction. The approach predicted libraries enriched in functional enzymes and fixed seven mutations in two rounds of evolution to identify variants for selective catalysis with 93% and 79% ee (enantiomeric excess). By greatly increasing throughput with in silico modeling, machine learning enhances the quality and diversity of sequence solutions for a protein engineering problem.

Recently, heme proteins have been discovered and engineered by directed evolution to catalyze chemical transformations that are biochemically unprecedented. Many of these nonnatural enzyme-catalyzed reactions are assumed to proceed through a catalytic iron porphyrin carbene (IPC) intermediate, although this intermediate has never been observed in a protein. Using crystallographic, spectroscopic, and computational methods, we have captured and studied a catalytic IPC intermediate in the active site of an enzyme derived from thermostable () cytochrome High-resolution crystal structures and computational methods reveal how directed evolution created an active site for carbene transfer in an electron transfer protein and how the laboratory-evolved enzyme achieves perfect carbene transfer stereoselectivity by holding the catalytic IPC in a single orientation. We also discovered that the IPC in cytochrome has a singlet ground electronic state and that the protein environment uses geometrical constraints and noncovalent interactions to influence different IPC electronic states. This information helps us to understand the impressive reactivity and selectivity of carbene transfer enzymes and offers insights that will guide and inspire future engineering efforts.

Developing catalysts that produce each stereoisomer of a desired product selectively is a longstanding synthetic challenge. Biochemists have addressed this challenge by screening nature’s diversity to discover enzymes that catalyze the formation of complementary stereoisomers. We show here that the same approach can be applied to a new-to-nature enzymatic reaction, alkene cyclopropanation via carbene transfer. By screening diverse native and engineered heme proteins, we identified globins and serine-ligated “P411” variants of cytochromes P450 with promiscuous activity for cyclopropanation of unactivated alkene substrates. We then enhanced their activities and stereoselectivities by directed evolution: just 1–3 rounds of site-saturation mutagenesis and screening generated enzymes that transform unactivated alkenes and electron-deficient alkenes into each of the four stereoisomeric cyclopropanes with up to 5,400 total turnovers and 98% enantiomeric excess. These fully genetically encoded biocatalysts function in whole Escherichia coli cells in mild, aqueous conditions and provide the first example of enantioselective, intermolecular iron-catalyzed cyclopropanation of unactivated alkenes.

Following the recent discovery that heme proteins can catalyze the cyclopropanation of styrenyl olefins with high efficiency and selectivity, interest in developing new enzymes for a variety of non-natural carbene transfer reactions has burgeoned. The fact that diazo compounds and other carbene precursors are known mechanism-based inhibitors of P450s, however, led us to investigate if they also interfere with this new enzyme function. We present evidence for two inactivation pathways that are operative during cytochrome P450-catalyzed cyclopropanation. Using a combination of UV-Vis, mass spectrometry, and proteomic analyses, we show that the heme cofactor and several nucleophilic side chains undergo covalent modification by ethyl diazoacetate (EDA). Substitution of two of the affected residues with less-nucleophilic amino acids led to a more than two-fold improvement in cyclopropanation performance (total TTN). Elucidating the inactivation pathways of heme protein-based carbene transfer catalysts should aid in the optimization of this new biocatalytic function.