Computational reprogramming of homing endonuclease specificity at multiple adjacent base pairs

Abstract

Site-specific homing endonucleases are capable of inducing gene conversion via homologous recombination. Reprogramming their cleavage specificities allows the targeting of specific biological sites for gene correction or conversion. We used computational protein design to alter the cleavage specificity of I-MsoI for three contiguous base pair substitutions, resulting in an endonuclease whose activity and specificity for its new site rival that of wild-type I-MsoI for the original site. Concerted design for all simultaneous substitutions was more successful than a modular approach against individual substitutions, highlighting the importance of context-dependent redesign and optimization of protein-DNA interactions. We then used computational design based on the crystal structure of the designed complex, which revealed significant unanticipated shifts in DNA conformation, to create an endonuclease that specifically cleaves a site with four contiguous base pair substitutions. Our results demonstrate that specificity switches for multiple concerted base pair substitutions can be computationally designed, and that iteration between design and structure determination provides a route to large scale reprogramming of specificity.

Figure 1.

Amino acid base interactions in wild-type and designed complexes. The interactions between amino acid residues 28, 30, 43, 70, 83, 85 and DNA bases −8, −7, −6 are shown. Blue spheres are crystallographic water molecules. Dashed lines depict selected hydrogen-bonding interactions. (a) Wild-type I-MsoI interactions observed in the original crystal structure (pdb: 1M5X). (b) Predicted model of computationally designed interactions between novel amino acids and DNA bases for the I-MsoI ‘GCG’ design.

Figure 2.

Complete switch of activity and specificity for three novel adjacent base pairs by computational design of I-MsoI. The cleavage of either the wild-type site (blue) or the designed ‘gcg’ site (red) is plotted as a function of the endonuclease concentrations of wild-type I-MsoI (a) and the I-MsoI ‘GCG’ design (b). Data are densitometric measurements of ethidium bromide-stained agarose-electrophoresed DNA cleavage products. The data were fit to determine the endonuclease concentrations that correspond to half-maximal cleavage (EC₅₀, gray lines). In (b), the best fit to the wild-type data in (a) is shown in dashed lines for comparison.

Figure 3.

Comparison of designed and crystallographically observed interactions. (a–c) the crystal structure of the triple-base pair I-MsoI ‘GCG’ design (cyan) is aligned with the designed model (green) and with the crystal structures and designed models of each single-base pair design: (a) I-MsoI ‘−8G’ (X-ray: yellow, model: orange), (b) I-MsoI ‘−7C’ (X-ray: white, model: pink), (c) I-MsoI ‘−6G’ (X-ray: purple, model: beige). (d) A conformational shift in the DNA backbone is observed near Trp85 in the I-MsoI ‘−7C’ crystal structure (colored by increasing B-factor from light blue to red), compared to the designed model (dark blue).

Figure 4.

Designed specific cleavage activity for an asymmetric four-base pair cluster. In vitro cleavage of wild-type (blue) and asymmetric ‘tgcg’ (red) DNA sites by monomerized I-MsoI (mMsoI) endonuclease designs. (a) wild-type mMsoI endonuclease, (b) N-terminal mMsoI ‘GCG’ design, (c) N-terminal mMsoI ‘TGCG’ design. Dashed lines in (b and c) represent the mMsoI trace from (a). Data are densitometric measurements of ethidium bromide-stained agarose-electrophoresed DNA cleavage products (Supplementary Figure S5).