Synthetic shuffling of positively selected mutations identifies Ll.LtrB variants with enhanced retrohoming into plasmid and chromosomal target sites in human cells.

<p>Optimal combinations of mutations were identified by the recombination-based technique of synthetic shuffling. A synthetic shuffling library of Ll.LtrB variants was constructed in which mutations at 18 positions that were under positive selection (>80% of one nucleotide type in >5% of the population in selection rounds 8 or 12 (subsets of the nucleotides indicated by green triangles or green or black arrows in <a href="http://www.plosgenetics.org/article/info:doi/10.1371/journal.pgen.1005422#pgen.1005422.g007" target="_blank">Fig 7</a>) were doped fifty percent against the wild-type nucleotide and position 642 was randomized. Sanger sequencing showed that the starting frequency of each nucleotide in the initial library was as expected for the degree of doping or randomization. The introns were selected in HEK-293 cells for four cycles of retrohoming (<a href="http://www.plosgenetics.org/article/info:doi/10.1371/journal.pgen.1005422#pgen.1005422.s007" target="_blank">S7 Fig</a>) with either 80 or 40 mM MgCl₂ added to the culture medium. Pacific Biosciences RS circular consensus sequencing was used to identify the mutations present in the populations with 4,770 and 4,768 reads obtained for the selections at 80 and 40 mM MgCl₂, respectively. (A) Sequence logos depicting nucleotide percentages in the population at the indicated positions after four rounds of selection at 80 and 40 mM Mg²⁺, and (B) sequence logos depicting the nucleotide percentages in variants that appear >3 times in the PacBio sequencing in the same selections. The size of the nucleotide indicates its frequency in the population. The wild-type (WT) nucleotide is indicated above the logo, and nucleotide position number is indicated beneath the logo. A green dot indicates that the mutant nucleotide increased in frequency to at least 60% of the population. (C) and (D) Retrohoming frequencies measured by Taqman qPCR for variants identified by synthetic shuffling. The bar graphs in (C) show retrohoming frequencies for four variants from the 80 mM MgCl₂ selection and two from the 40 mM MgCl₂ (sequences shown below) into a plasmid target site in HEK-293 cells with 80 mM MgCl₂ added to the culture medium. The wild-type intron and the best variant from <a href="http://www.plosgenetics.org/article/info:doi/10.1371/journal.pgen.1005422#pgen.1005422.g008" target="_blank">Fig 8</a>, which contained only the prevalent EBS1 and DIVb mutations (G282A, U642A, G651A, U652C), were assayed in parallel. The bar graphs in (D) show retrohoming frequencies into the genomic HEK-293 cell target site with 80 mM MgCl₂ added to the culture medium for the best four variants from panel (C) compared to wild-type and the best variant from <a href="http://www.plosgenetics.org/article/info:doi/10.1371/journal.pgen.1005422#pgen.1005422.g008" target="_blank">Fig 8</a>. The values are the average for three experiments, with the error bars indicating the SD. The retrohoming frequency of the wild-type intron ranged from 0.034–0.050% in the assays of panel (C) and from 0.017–0.026% in the assays of panel (D). (E) The most prevalent variants identified by PacBio sequencing from synthetic shuffling after four rounds of selection in HEK-293 cells with 80 and 40 mM MgCl₂ added to the culture medium. The nucleotide position in the Ll.LtrB intron is indicated at the top, and the table shows the nucleotide sequence at that position in the variants. Upper case letters indicate the wild-type nucleotide, and lower case letters indicate mutant nucleotides.