Aim1: We set out to leverage the information collected during the directed evolution of SSRs by deep sequencing. The PacBio-based long-read protocol was fully implemented and more than 80 evolved SSR libraries were sequenced at a depth of at least 20.000 individual clones. This data allowed us to e.g. identify when specific mutations emerged in the library. We also observed that 157 positions in the 324 amino acid long recombinase coding sequence had been selected to be altered during at least one evolved enzyme library. More interestingly, certain positions were found to change to different amino acids, depending on the selected target site the recombinases were evolved to recombine. In some extreme cases, eight different amino acids were found to be selected from the different projects. Overall, we have obtained excellent insight into the evolution dynamics of evolved recombinases. Using this data, we could further optimize the SLiDE procedure. The obtained data was implemented into the computer algorithms and analyses helped to improve the directed evolution procedure and to rationalize the evolution dynamics and structure-function relationships. The data also revealed that certain amino acids repeatedly co-evolved during directed evolution of SSRs. Furthermore, the sequencing data was helpful to uncover that nearest-neighbour amino acids play an important role in enhancing SSR activity on foreign target sites. The deep sequencing data from additional designer-recombinase libraries were generated and integrated into the model to further optimize the evolution conditions. Based on these results, we have determined the best inflection point to reduce the arabinose concentration during the evolution cycles. Together with other measures the generation of new designer-recombinase was significantly accelerated. The sequencing data was also implemented into a first generative deep-learning algorithm to predict designer-recombinases with predefined specificities. Aim 2: To expand the sequences that can be addressed by designer-SSRs, we first demonstrated that the generation of heterodimer recombinases is feasible. Employing SLiDE, we developed a fused heterodimeric recombinase system (RecF8) that is able to correct the int1h inversion frequently found in patients with a severe form of haemophilia A. Importantly, we achieved 30% inversion of the target sequence in human tissue culture cells, indicating that RecF8 can efficiently function in the target cells. Transient RecF8 treatment of endothelial cells, differentiated from patient derived induced pluripotent stem cells (iPSCs) of a hemophilic donor, resulted in prominent correction of the inversion and restored Factor VIII mRNA expression. As the expression of 2 different designer-recombinases might increase potential off-target recombination, we developed an assay to screen for mutations that would render the monomers inactive, without compromising the dimer functionality. This screen identified single catalytic mutations in the individual monomers that fulfilled these criteria. We first explored the possibility of transplanting human liver organoids into TK-NOG mice in collaboration with Prof. Takanori Takebe (USA). However, due to the complexity and risks associated with these experiments, and after consultations with the German regulatory authorities, we decided to pursue an alternative approach. Aim 3: The principle of designer-recombinase induced gene replacement was demonstrated with loxLTR and loxBTR-specific recombinases by building a pSLIDE-4R system. Furthermore, we have generated a designer recombinase that catalyzes recombination of two genomic human sites to replace exon 8 of the F9 gene. Importantly, we could show that this recombinase can carry out the cassette exchange reaction in mammalian cells, when the cells were co-transfected with a donor and a receptor plasmid, plus a recombinase expression plasmid. Therefore, we achieved proof of concept for replacing a gene segment by DRiGR.
