Progressive engineering of a homing endonuclease genome editing reagent for the murine X-linked immunodeficiency locus

LAGLIDADG homing endonucleases (LHEs) are compact endonucleases with 20–22 bp recognition sites, and thus are ideal scaffolds for engineering site-specific DNA cleavage enzymes for genome editing applications. Here, we describe a general approach to LHE engineering that combines rational design with directed evolution, using a yeast surface display high-throughput cleavage selection. This approach was employed to alter the binding and cleavage specificity of the I-Anil LHE to recognize a mutation in the mouse Bruton tyrosine kinase (Btk) gene causative for mouse X-linked immunodeficiency (XID)—a model of human X-linked agammaglobulinemia (XLA). The required re-targeting of I-AniI involved progressive resculpting of the DNA contact interface to accommodate nine base differences from the native cleavage sequence. The enzyme emerging from the progressive engineering process was specific for the XID mutant allele versus the wild-type (WT) allele, and exhibited activity equivalent to WT I-AniI in vitro and in cellulo reporter assays. Fusion of the enzyme to a site-specific DNA binding domain of transcription activator-like effector (TALE) resulted in a further enhancement of gene editing efficiency. These results illustrate the potential of LHE enzymes as specific and efficient tools for therapeutic genome engineering.

Equal amounts of WT-Ani enzyme were incubated with 20 nM alternative Alexa-647-conjugated ds-oligo substrates as indicated at 37°C for 1 hour. Oligos were separated by 10% non-denaturing polyacrylamide gel and quantified with an Odyssey infrared imaging system (Li-Cor Biosciences).
Due to tolerance of these two clusters, XID enzyme engineering was initiated with cluster +6T+7G and -6C-5C-4T. Figure S2. XID (+) half-site enzyme engineering. A. Based upon structural predictions, six AA residues (Y154, I164, S166, T189, K202 and T204) targeted to +6T+7G cluster were selected for randomization. To keep the theoretical library diversity within the range of yeast library size, randomization of T189 was limited to T, P, or R based on computational model prediction and previous selection. The remaining five residues were completely randomized with code NNK. The predicted diversity of such design is around 12.8 × 10 6 , which was well represented by the corresponding yeast library with the size of 30 million. B. Data showing three rounds of cleavage selection of +6T+7G randomization library. For each round of selection, around 0.3% potential active variants from reaction with Mg 2+ were selected for continued culture.

Supplementary
No obvious active population was observed until Round 3 selection (see small number of red events appearing just below the major population of APC/PE double positive events in blue). C.
Cleavage activity of +6T+7G-Ani variants selected from randomization library (lib) was significantly improved after random mutagenesis of this library followed by selection. D. Cleavage activity and specificity of five highly enriched variants from mutagenesis library. Variants exhibited similar mutations on DNA interface residues (Supplementary Table S1), and most of them (V2 to V5) only cut +6T+7G cluster, but not +6T or +7G single mismatch. E. Using five +6T+7G-Ani variants as seed template, four residues (L156, N157, D160 and Y162) targeted to +9C+10T cluster were randomized to generate a XID (+) half-site library. F. Cleavage activity and specificity of (+) half-site active variants selected from randomization library on yeast surface and in vitro. G. XID(+)-Ani variant 1 ((+) V1) not only had the WT enzyme comparable cleavage activity in vitro and on yeast surface, but showed significantly improved specificity at the +9+10 cluster, thus was selected as template to assemble XID full site enzyme (Supplementary Table S1). Figure S3. XID (-) half-site enzyme engineering. A. Schematic of Ani design (T22S, E31R and R70E) targeted to -6C. B. Data showing significantly improved specificity using this design which was selected as the seed for design extension. C. Using this seed design, five additional residues (Y18, S20, G33, R59, A68) targeted to -6C-5C-4T cluster were selected for randomization to generate a 6C-5C-4T cluster library. D-E. -6C-5C-4T-Ani active variants selected from randomization library had moderate cleavage activity on yeast surface and in vitro, that was significantly improved after random mutagenesis and selection. F. Cleavage activity and specificity of -6C-5C-4T-Ani variants (V1 to V3, Supplementary Table S1). All variants showed no activity towards WT and XID (-) full half site. G. Using -6C-5C-4T-Ani variants (V1 to V3) as seed template, I-Anil N-term loop domain (K24, G25, K26, Y27, L28, T29) targeted to the -10A-8T cluster were randomized to extend the design to the full (-) half site. However, this strategy did not generate any active variants (data not shown), suggesting structural shifts caused by multiple mutations at these two regions are not compatible and thus abolish enzyme activity. H. Utilizing an alternative, directed evolution method, XID (-) V1 selected from -6C-5C-4T random mutagenesis library gained cleavage activity towards XID (-) half site (Supplementary Table S1). I. Using XID (-) V1 as template, I-Anil N-term loop domain was re-designed for -10A-8T by randomization.

Supplementary
XID (-) V2 selected from N-term loop domain library exhibited increased activity and specificity towards the (-) half site (Supplementary Table S1).