Structure-guided reprogramming of serine recombinase DNA sequence specificity

Abstract

Routine manipulation of cellular genomes is contingent upon the development of proteins and enzymes with programmable DNA sequence specificity. Here we describe the structure-guided reprogramming of the DNA sequence specificity of the invertase Gin from bacteriophage Mu and Tn3 resolvase from Escherichia coli. Structure-guided and comparative sequence analyses were used to predict a network of amino acid residues that mediate resolvase and invertase DNA sequence specificity. Using saturation mutagenesis and iterative rounds of positive antibiotic selection, we identified extensively redesigned and highly convergent resolvase and invertase populations in the context of engineered zinc-finger recombinase (ZFR) fusion proteins. Reprogrammed variants selectively catalyzed recombination of nonnative DNA sequences > 10,000-fold more effectively than their parental enzymes. Alanine-scanning mutagenesis revealed the molecular basis of resolvase and invertase DNA sequence specificity. When used as rationally designed ZFR heterodimers, the reprogrammed enzyme variants site-specifically modified unnatural and asymmetric DNA sequences. Early studies on the directed evolution of serine recombinase DNA sequence specificity produced enzymes with relaxed substrate specificity as a result of randomly incorporated mutations. In the current study, we focused our mutagenesis exclusively on DNA determinants, leading to redesigned enzymes that remained highly specific and directed transgene integration into the human genome with > 80% accuracy. These results demonstrate that unique resolvase and invertase derivatives can be developed to site-specifically modify the human genome in the context of zinc-finger recombinase fusion proteins.

Fig. 1.

Structure-guided analysis of serine resolvase and invertase DNA sequence specificity. (A) γδ resolvase dimer (green and blue) in complex with substrate DNA (gray). The variable arm region residues targeted for saturation mutagenesis, Asn 127, Arg 130, Met 134, Phe 140, Lys 143, and Arg 144, are represented as green balls and sticks (inset, with DNA) (PDB ID: 1GDT). (B) Multiple sequence alignment of the representative resolvase/invertase family members—Gin, Hin, γδ, Tn3, Sin, and Beta—and alignments of DNA sequences recognized by these enzymes. The secondary structures observed in the γδ resolvase crystal structure are denoted above the multiple sequence alignment. Conserved residues are highlighted red and conservative amino acid substitutions are highlighted yellow. Arm region amino acid residues targeted for saturation mutagenesis are highlighted green.

Fig. 2.

Serine resolvase and invertase DNA sequence specificity can be reprogrammed for recombination against nonnative DNA sequences. (A) Schematic representation of the selection strategy used for isolating unique ZFR variants. GFPuv flanked by ZFR target sites (white) was inserted into the gene encoding TEM-1 β-lactamase (black). ZFR-mediated modification resulted in restoration of the β-lactamase gene. Active ZFR variants were enriched by carbenicillin selection and isolated following restriction digestion. (B–C) Substrate specificity shift analysis of isolated (B) Gin invertase and (C) Tn3 resolvase variants. Resolvase and invertase derivatives were isolated for analysis following four rounds of positive antibiotic selection. The recombination efficiency of an isolated enzyme variant against the nonnative target core sequence and the native core sequence was determined by measuring the fraction of carbenicillin-resistant transformants following enzyme-mediated reconstitution of the gene encoding TEM-1 β-lactamase. The specificity shift of an enzyme variant was calculated as the quotient of nonnative target core sequence recombination efficiency divided by native core sequence recombination efficiency. (D–E) Recombination efficiency of parental and reprogrammed (D) Gin invertase and (E) Tn3 resolvase variants against gix (gray) and resI (black) DNA sequences. Specificity shift and recombination efficiency are presented on a logarithmic scale. Error bars indicate the standard deviation of three independent replicates.

Fig. 3.

Functional analysis of the resolvase/invertase amino acid residues implicated in mediating DNA sequence specificity. Recombination efficiency of alanine-substituted parental (Gin and Tn3) and reprogrammed (GinT and Tn3G) resolvase and invertase variants against (A) gix and (B) resI DNA sequences. Recombination efficiency is presented on a logarithmic scale. Error bars indicate the standard deviation of three independent replicates.

Fig. 4.

ZFR heterodimers catalyze recombination against unnatural and asymmetric DNA sequences. (A) γδ resolvase dimer (green and blue) in complex with substrate DNA (gray). Arm region amino acid residues implicated in mediating substrate recognition (red spheres) are sequestered from the resolvase dimer interface (PDB ID: 1GDT). (B) Schematic representation of ZFR-mediated modification of asymmetric DNA sequences. The H1/P2.T/G DNA sequence is depicted. (C–D) Recombination efficiencies of (C) Gin invertase-based and (D) Tn3 resolvase-based ZFR heterodimers against the asymmetric ZFR target site H1/P2.T/G. “-” indicates no ZFR_L added. Recombination efficiency is presented on a logarithmic scale. Error bars indicate the standard deviation of three independent replicates.

Fig. 5.

Engineered ZFRs accurately and efficiently direct plasmid integration into the human genome. (A) Schematic representation of ZFR-mediated plasmid integration into the human genome. Donor plasmid containing ZFR target-site, CMV promoter (black arrow), and constitutively expressed puromycin-resistance gene were cotransfected with ZFR expression vector into human cells that contain a single genomic ZFR target site. ZFR-mediated integration resulted in the formation of puromycin-resistant cells and two phenotypically distinct integration products. (B) Efficiency of ZFR-mediated plasmid integration into the human genome. Following cotransfection of ZFR expression vector and donor plasmid, puromycin selection was used to assess total integration efficiency. Efficiency of ZFR-mediated integration is represented as fold-enhancement relative to donor plasmid only. (C) Specificity of ZFR-mediated plasmid integration into the human genome. Following puromycin selection and clonal expansion, specificity of ZFR-mediated integration was determined by flow cytometry. The efficiency and specificity of ZFR-mediated integration was analyzed with various ZFR (Gin_C4, Tn3_C4, GinT_C4, and Tn3G_C4), donor (C4.20G, C4.20T) and genomic target (C4.20G, C4.20T) combinations. Error bars indicate the standard deviation of three independent replicates. “ND” indicates not detectable.