DNA Ligases Discriminate Between Natural and Non-Natural Base Pairs

Abstract

Artificially Expanded Genetic Information Systems (AEGIS) increase the information content of nucleic acids by including new nucleobase pairings that are orthogonal to those of canonical Watson-Crick nucleobases. DNA ligases do not form direct interactions with the nucleobases during catalytic turnover, suggesting that these enzymes should efficiently and faithfully join double-stranded AEGIS substrates. Here we report the systematic investigation into the validity of this hypothesis for structurally-diverse DNA ligases employing substrates built from the eight nucleotide hachimoji genetic alphabet, where orthogonality is achieved by rearranging the hydrogen bonding patterns seen in canonical Watson-Crick pairs. We find that single, or multiple, non-canonical bases are well tolerated at the 5-end of the nick. However, tracts of consecutive non-canonical bases at the 3-end of the break significantly decrease ligation efficiency or abolish it altogether. Possible reasons for this apparent bias against non-canonical nucleobases could include incompatibility in electrostatic interactions between the ligase active site and the non-canonical substrates or altered conformational preferences and/or dynamics in key catalytic intermediates. We also observe single hachimoji mismatches are ligated more frequently than mis paired canonical bases, potentially due to promiscuous pairing of tautomeric forms of the non-canonical bases.