Now, Andy Hsien-Wei Yeh, David Baker and co-workers report de novo designed luciferases that selectively oxidize diphenylterazine (DTZ) (pictured), which possesses a high quantum yield, red-shifted emission and favourable in vivo pharmacokinetics compared to natural substrates. By docking DTZ into 4,000 native small-molecule-binding proteins, the nuclear transport factor 2 (NTF2)-like superfamily was identified as a suitable scaffold. However, the stability of NTF2 structures is not ideal, resulting from the presence of long loops. Using a deep-learning approach, which the researchers refer to as hallucination, they searched for non-natural sequences that are predicted by a neural network to have the desired properties to substitute the loop and variable regions (pictured) of NTF2 structures, while structure-guided sequence optimization was performed for the core regions.
Moreover, the protein active site was computationally designed to contain an arginine residue to stabilize the anionic state of DTZ required for the chemiluminescent reaction and to optimize the pocket polarity for the single-electron transfer process with triplet molecular oxygen. Thereafter, a library of designed luciferases was screened for the targeted activity with DTZ. A variant was identified that generated luminescence with an emission peak at around 480 nm. This enzyme is small, highly expressed in Escherichia coli, monomeric and thermostable. Subsequently, its activity was significantly enhanced by active-site mutagenesis reaching a catalytic efficiency (kcat/Km) of 106 M−1 s−1, which is in the range of native luciferases. Finally, the applicability of the developed enzyme for cell imaging and multiplexed bioassays was demonstrated in live mammalian cells, showing the highly specific nature for DTZ and promising potential for biomedical research.
This is a preview of subscription content, access via your institution