Back

De Novo Design of Flavoproteins with Controlled Excited State Dynamics

Anderson, R.; Bungay, E.; Underhill, J.; Liu, M.; Stuart, A.; Back, C.; Berrones-Reyes, J.; Dubiel, P.; Birrell, J.; G. Rao, A.; Williams, C.; Barringer, R.; Hinchliffe, P.; Lichtenstein, B.; Curchod, B.; Butt, J.; Oliver, T.; Mulholland, A. J.

2026-10-10 biophysics
10.64898/2026.10.08.757584 bioRxiv
Show abstract

Designing photochemical properties in proteins is a frontier challenge in biomolecular design, which, when realised, will open new avenues in catalysis and biomolecular electronics. Here we demonstrate an integrated computational design framework, supported by ultrafast spectroscopy, combining protein design, quantum mechanical/molecular mechanics molecular dynamics (QM/MM MD) and electronic structure calculations, to control flavin excited states in de novo proteins. Iterative design yields FC-4D2, a flavocytochrome with covalently bound flavin and non-covalently bound heme B and Fl-4D2, where the heme-binding site is removed. Both proteins are highly thermostable, and access all physiologically relevant flavin redox and protonation states, while high-resolution X-ray crystallography confirms the near-[A]ngstrom-level accuracy in design. Incorporating QM/MM MD and linear-response time-dependent density functional theory (LR-TDDFT) into the design workflow enables engineering of a flavin-tryptophan charge-transfer variant, Fl-4D2CT, that modifies the flavin absorption spectrum, substantially alters its redox properties and suppresses fluorescence. Ultrafast studies reveal how protein design reshapes the excited state relaxation timescale and mechanism. Our study establishes a modular strategy for programming flavin photochemistry in designed proteins.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.