Molecular evolution of a novel hyperactive Sleeping Beauty transposase enables robust stable gene transfer in vertebrates

Individual mutations imported into the Sleeping Beauty (SB) transposase from related Tc1-family transposases, or previously reported in the literature, each increase transpositional activity by no more than fourfold on their own. Combining them produces much larger, non-additive gains: a combinatorial DNA-shuffling library averaging 2.2 mutations per clone yields variants up to 25-fold more active than the original SB transposase from an initial screen of about 2,000 clones, and only 25 of 41 individually hyperactive mutations recur in these higher-activity combinations ("friendly" mutations), while the remaining 16 are never recovered together with others ("unfriendly" mutations). SB100X, the most hyperactive variant identified, combines six specific hyperactivating mutations out of nearly 4.5 million possible six-mutation combinations and shows roughly 100-fold higher transpositional activity than the original SB transposase.

Hyperactive mutations in the SB transposase map throughout the protein except within three regions — amino acids 148–186, 218–242 and 317–340 — which fall within the conserved DDE catalytic domain and the C terminus, indicating that positions tolerant of hyperactivating substitutions lie outside the core catalytic machinery. At amino acid position 243, replacing the native methionine of SB with either glutamine or histidine, both of which occur at the equivalent position in other Tc1-family transposases, increases transpositional activity.

SB100X achieves up to 35% stable transgenesis frequency in HeLa cell colony-forming assays, compared with 8% for the original SB transposase at the same high transposon dosage, and it retains high transgenic frequencies even when transposon DNA is limiting. SB100X is more than 100-fold more potent than SB at remobilizing a single-copy transposon already integrated into a chromosome, a property directly relevant to genetic screens, and it generates three- to five-fold more integrated transposon copies per cell than SB independent of DNA dosage — showing that its gain in copy number is not proportional to its gain in transpositional activity.

SB100X does not differ from the original SB transposase in protein stability, overproduction-inhibition profile or affinity for the transposon's inverted terminal repeats (ITRs), but it shows altered subcellular localization after heat shock, suggesting its hyperactivating mutations act on protein folding and on interactions between the DNA-binding and catalytic domains rather than on these other biochemical properties directly.

SB100X outperforms a codon-optimized piggyBac transposase in side-by-side comparisons: it is approximately sevenfold more active than piggyBac in HeLa colony-forming transposition assays, and in human CD34+ cells it produces 33–35% GFP-positive colonies in erythroid (CFU-E) and granulocyte/monocyte/macrophage (CFU-GM) assays, compared with 7–9% for piggyBac.

Nucleofection of human CD34+ hematopoietic stem/progenitor cells with a GFP-marked SB transposon and each of several SB transposase variants — SB, SB11, SB16X, SB32X and SB100X — gives statistically indistinguishable transfection efficiencies (50–60% GFP-positive cells at two days), so downstream differences in stable colony formation reflect differences in transposition efficiency rather than delivery efficiency. SB100X yields the highest stable GFP-positive colony-forming unit (CFU) frequencies of any variant tested across all three lineages assayed — 34 ± 4% of erythroid (CFU-E), 45 ± 17% of megakaryocytic (CFU-MK) and 33 ± 7% of granulocyte/monocyte/macrophage (CFU-GM) colonies — corresponding to 70–90% of transfected cells having undergone stable transposition once corrected for transfection efficiency. This significantly exceeds the second-generation SB11 transposase (P < 0.002) and vastly exceeds the original SB transposase, which gives only about 1% CFU-E and 0.2% CFU-GM colonies under the same conditions.

Transplantation of SB100X-transfected human CD34+ cells into NOD-SCID γc-null mice produces long-term (4–5 month) engraftment of human CD45+ cells, reaching 44–73% of peripheral blood or bone marrow cells in one cohort and averaging 33 ± 10% in a second cohort, with 0.4–1.0 transposon copies per cell detected by quantitative PCR in engrafted animals. Gene marking persists in both the CD20+ lymphoid and CD33+ myeloid compartments, and identical transposon integration sites recovered from both lineages confirm that transposition occurred in a common, multipotent hematopoietic stem or progenitor cell. This was the first demonstration of robust, stable, multilineage gene transfer into human CD34+ cells and their in vivo progeny using a nonviral, transposon-based delivery method.

Hydrodynamic co-injection of a human factor IX (hFIX)-expressing transposon (pT2/Apo/AAT-FIX) with SB100X into adult mouse liver sustains plasma hFIX at 2,000–3,000 ng/ml for more than one year, significantly exceeding expression driven by the second-generation SB11 transposase (P < 0.0001), which produces only marginal, near-background hFIX levels. Consistent with this difference, transposon excision efficiency in the liver at seven days post-transfection is roughly eightfold higher with SB100X than with SB11, tracking the 9- to 14-fold difference in hFIX expression between the two transposases. Following partial hepatectomy, which forces hepatocyte proliferation and clears episomal DNA, hFIX expression driven by SB100X remains stable while expression in control animals declines to background — confirming that the sustained hFIX expression reflects stable genomic integration rather than persistence of episomal plasmid.

Lowering the amount of SB100X transposase plasmid co-injected with a fixed transposon dose in hydrodynamic liver delivery (2 μg versus 25 μg, at 50 μg transposon DNA) increases plasma hFIX into the supraphysiological range (>8,000 ng/ml, about 160% of normal levels) rather than decreasing it — consistent with overproduction inhibition, in which an excess of transposase relative to transposon substrate reduces net transposition efficiency.

Co-injecting SB100X transposase mRNA (5 ng/μl) with a Venus-marked transposon plasmid (pT2/CAGGS-Venus, 0.4 ng/μl) into fertilized mouse oocytes reproducibly yields Venus-positive fluorescence in an average of 45% of surviving embryos at day 7 and produces transgenic offspring at a rate of 37% (7 of 19 newborns in one experiment), with transgenic animals typically carrying 1–2 integrated transposon copies per genome and no evidence of mosaicism. This exceeds the roughly 10% transgenesis frequency typical of classical pronuclear microinjection and avoids the concatemeric transgene insertions produced by classical methods, which are more prone to silencing and recombination.

Transposon integration sites recovered from SB100X-treated mouse liver show no bias toward genic regions: approximately 70% of 60 recovered sites are intergenic and 30% fall within genes, matching the integration pattern reported for earlier-generation SB transposases and contrasting with retroviral, lentiviral and piggyBac vectors, which preferentially integrate into active genes.