Complete, Programmable Decoding of Oxidized 5-Methylcytosine Nucleobases in DNA by Chemoselective Blockage of Universal Transcription-Activator-Like Effector Repeats.

5-Methylcytosine (5mC) and its oxidized derivatives are regulatory elements of mammalian genomes involved in development and disease. These nucleobases do not selectively modulate Watson-Crick pairing, preventing their programmable targeting and analysis by traditional hybridization probes. Transcription-activator-like effectors (TALEs) can be engineered for use as programmable probes with epigenetic nucleobase selectivity. However, only partial selectivities for oxidized 5mC have been achieved so far, preventing unambiguous target binding. We overcome this limitation by destroying and re-inducing nucleobase selectivity in TALEs via protein engineering and chemoselective nucleobase blocking. We engineer cavities in TALE repeats and identify a cavity that accommodates all eight human DNA nucleobases. We then introduce substituents with varying size, flexibility, and branching degree at each oxidized 5mC. Depending on the nucleobase, substituents with distinct properties effectively block TALE-binding and induce full nucleobase selectivity in the universal repeat. Successful transfer to affinity enrichment in a human genome background indicates that this approach enables the fully selective detection of each oxidized 5mC in complex DNA by programmable probes.