Chemical fluorophores offer tremendous size and photophysical advantages over fluorescent proteins but are much more challenging to target to specific cellular proteins. Here, we used Rosetta-based computation to design a fluorophore ligase that accepts the red dye resorufin, starting from Escherichia coli lipoic acid ligase. X-ray crystallography showed that the design closely matched the experimental structure. Resorufin ligase catalyzed the site-specific and covalent attachment of resorufin to various cellular proteins genetically fused to a 13-aa recognition peptide in multiple mammalian cell lines and in primary cultured neurons. We used resorufin ligase to perform superresolution imaging of the intermediate filament protein vimentin by stimulated emission depletion and electron microscopies. This work illustrates the power of Rosetta for major redesign of enzyme specificity and introduces a tool for minimally invasive, highly specific imaging of cellular proteins by both conventional and superresolution microscopies.fluorescence microscopy | enzyme redesign | LplA | PRIME | chemical probe targeting F luorescent proteins are used ubiquitously in imaging, but their dim fluorescence, rapid photobleaching, and large size limit their utility. At ∼27 kDa (∼240 aa), fluorescent proteins can disrupt protein folding and trafficking or impair protein function (1, 2). Chemical fluorophores, in comparison, are typically less than 1 kDa in size, and are brighter and more photostable. These properties allow chemical fluorophores to perform better than fluorescent proteins in advanced imaging modalities such as singlemolecule tracking and superresolution microscopies (3, 4).Site-specific labeling of proteins with chemical fluorophores inside living cells is challenging because these fluorophores are not genetically encodable and therefore must be posttranslationally targeted inside a complex cellular milieu. Existing methods to achieve this targeting either require large fusion tags [such as HaloTag (5), the SNAP tag (6), and the DHFR tag (7)] or have insufficient specificity [such as biarsenical dye targeting (8) and amber codon suppression (9)]. To achieve a labeling specificity comparable to fluorescent proteins, we developed PRIME (PRobe Incorporation Mediated by Enzymes), which uses Escherichia coli lipoic acid ligase to attach small molecules to a 13-aa peptide tag (Fig. 1A) (10). To make PRIME more useful for cellular protein imaging, we sought to engineer the system for the targeting of bright chemical fluorophores. The challenge, though, is that lipoic acid ligase (LplA) has a small and fully enclosed substrate-binding pocket that even with extensive structure-guided mutagenesis has not until this point been able to accommodate large chemical structures.
ResultsSynthesis of Resorufin Derivatives for PRIME. We considered fluorophores for PRIME based on the steric constraints of LplA. Far-red emitters such as Cy5 and Atto 647N, although photophysically desirable, are so bulky that binding inside LplA would require...