Portrait of author

Oana Nicoleta Antonescu:
A massively parallel platform for engineering leave-one-out split fluorescent proteins

Date: 31-01-2020    Supervisor: Jakob Rahr Winther



A leave-one-out fluorescent protein (LOO-FP) is a split fluorescent protein where one of the secondary elements is removed, such as one of the 11 β-strands or the central α-helix. The two resulting broken protein fragments are not fluorescent on their own but can self-associate to reconstitute fluorescence. This property makes them extremely convenient and versatile biosensors. Engineering LOO-FPs for optimal complementation efficiency, spectral properties, solubility, and photostability is highly desirable. However, genetic methods, such as random mutagenesis, are laborious, they do not screen a large protein sequence space, and they often do not measure direct binding between the split fragments.

In this PhD project, a massively parallel in vitro method for assaying LOO-FP fragment function and binding has been developed. Hundreds of thousands of left-out LOO-FP peptide variants were synthesized on high-density peptide microarrays and examined for functional complementation of the partner fragment. The peptide microarray platform allowed us to map the substitutional tolerance of the left-out strand with high resolution, precision and robustness. A double substitution yielded a peptide with a 34-fold higher microarray fluorescence than the starting wild type peptide. For selected substitutions, the microarray fluorescence correlated with the affinity between the free LOO-FP fragments in solution and the spectral properties of the reconstituted proteins.

By labeling the partner LOO-FP fragment with a spectrally different fluorescent dye, the peptide microarray platform also allowed for studying direct binding with the left-out peptides, independent of functional complementation. The binding landscape between fragments was highly biased towards positive charges, but this non-specific effect was avoided by increasing the incubation buffer salt concentration. Indeed, the binding landscape in high ionic strength was strongly correlated (Pearson correlation coefficient = 0.94) with the functional interaction landscape. These results demonstrate that the platform can be extended to study the binding of other split protein fragments that do not have an internal reporter for function.

We have established and validated a novel platform for engineering the energetics and spectral properties of LOO-FPs in high throughput and with rapid turnover. We propose such a platform as a powerful alternative to biological synthesis, genotyping, and in cellulo analyses typically used in protein sequence-function mapping.