Regulated complex assembly safeguards the fidelity of Sleeping Beauty transposition

Parts used

Key findings

• Replacing only the PAI recognition motif inside each of the two inner 12DRs of the Sleeping Beauty (SB) transposon, while leaving the rest of every direct repeat wild type, raises transposition to 175% of the wild-type transposon in a colony-based transposition assay in HeLa cells. The substituted sequence is derived from CAST-5, a high-affinity SB transposase binding site isolated by CASTing, and this optimized construct is named pT4. The same CAST-5/wild-type hybrid motif placed at the outer 14DRs, or at the inner and outer positions together, lowers transposition instead, so the gain depends on the inner 12DR position specifically.
• Raising transposase binding affinity across the Sleeping Beauty inverted repeats by swapping in whole high-affinity CAST-5 motifs suppresses rather than improves transposition. Relative to the wild-type transposon set at 100%, a colony-based transposition assay gives 65% when CAST-5 replaces the inner 12DRs, 3% when it replaces the outer 14DRs and 2.2% when all four DRs are replaced. CAST-5 sequences are predicted to be sub-optimal for RED interaction, so the loss is attributed both to excess affinity and to compromised inner-versus-outer discrimination.
• Six cycles of CASTing on a 35-bp random oligonucleotide library bound by recombinant Sleeping Beauty transposase yield sequences that the PAIRED-like DNA-binding domain N123 binds up to eight-fold more strongly than the wild-type 14DR of the SB left inverted repeat in electromobility shift assay. The selected high-affinity sites resemble the PAI recognition motif far more than the RED motif (70% versus 20% identity) and show no preference for 12DR over 14DR, indicating that PAI specifies primary substrate recognition while RED contributes little to first contact and does not distinguish inner from outer sites at that stage.
• The two helix-turn-helix subdomains of the Sleeping Beauty transposase DNA-binding domain split substrate recognition by position. PAI (amino acids 1-57) binds the inner 12DR and the outer 14DR equally well in electromobility shift assay, whereas RED (amino acids 58-123) binds the 12DR and gives no detectable complex with the 14DR. Filling a 12DR oligonucleotide with two extra nucleotides so that it matches the length of 14DR abolishes specific RED binding but leaves PAI binding unaffected, showing that RED discriminates inner from outer direct repeats by length rather than by sequence.
• The RED subdomain (amino acids 58-123) of the Sleeping Beauty transposase carries a protein-protein interaction function in addition to DNA binding. The dominant RED-12DR complex in electromobility shift assay migrates consistently with two RED molecules bound, and a monomeric RED-12DR complex appears only at a 20-fold lower RED concentration, while chemical crosslinking of the RED peptide followed by western blotting resolves dimeric, tetrameric and higher-order multimers both with and without DNA substrate. PAI binds the 12DR and 14DR as a monomer, so of the two subdomains only RED dimerizes upon DNA binding.
• Second-end capture in Sleeping Beauty transposition follows a 12/12DR rule. In staged electromobility shift assay, a preformed RED-12DR complex captures a second 12DR molecule to give a lower-mobility 12DR-RED-12DR complex, 14DR offered as the second substrate is rejected, and no comparable complex forms when assembly is initiated on 14DR instead. Human HMGB1 increases the low-mobility RED-12DR complex at RED concentrations of 0.2-0.6 pmol and does so without Mg2+, yet HMGB1 is not stably incorporated because no supershift is detectable.
• Cleavage of the Sleeping Beauty transposon is blocked at the inner 12DR even when the canonical TA target dinucleotide is supplied. A transposon whose right inverted repeat is truncated at the internal direct repeat and modified to end in TA (12DR-TA) yields almost no product in a PCR-based excision assay and no detectable transposition in HeLa cells, whereas the matching construct truncated to a 14DR with its canonical TA (14DR-TA) is cleaved and transposes at roughly 26% of wild-type frequency. The residual activity of 14DR-TA also shows that the 12/12DR rule is not absolute.
• A pre-cleaved Sleeping Beauty transposon is not accepted as a transposition substrate. Linear transposon DNA released from the pTBsaXI donor by BsaXI, a restriction enzyme that removes three nucleotides from the transposon end just as the SB transposase does, produces no detectable colonies in a transposition assay in HeLa cells, indicating that the ordered assembly process cannot resume from an already processed transposon end.
• Sleeping Beauty does not favour internally deleted donors. SB transposons carrying internal deletions of 977 or 1654 bp are both excised less frequently than the full-length element in a GFP-restoration excision assay read by FACS, and the MITE-mimicking Mutant#2, which retains almost nothing beyond its two 230-bp inverted repeats, is mobilized at only about 10% of full-length efficiency. This substrate preference is the opposite of the Hsmar1 transposase, which prefers its MADE1 MITE derivatives over full-length elements, and it explains why MITE-like copies have not accumulated for IR/DR-type elements.
• Single-ended substrates are used far less efficiently by the IR/DR-type Sleeping Beauty transposon than by the simple-IR mariner element Hsmar1. Transposons lacking one of their two inverted repeats integrate at 6.4% of the respective wild-type frequency for Hsmar1 but only 0.54% for SB in HeLa cell transposition assays scored after discarding transposase-independent genomic integration events, with integration counted only where a precise left-inverted-repeat junction flanked by TA was recovered by linker-mediated PCR.
• Alanine-scanning mutagenesis of residues 236-285 in the catalytic domain of the Sleeping Beauty transposase identifies M243A, K252A, V254A, D260A, S270A and P277A as hyperactive substitutions, raising transposition of the neomycin-marked transposon pT/nori in HeLa cells to as much as 180% of wild-type SB10, while most other single substitutions across this stretch reduce or abolish activity. Excision and transposition track together for these mutants, with K248A the exception.
• The single substitution K248A in the catalytic domain of the Sleeping Beauty transposase uncouples excision from integration: K248A still excises the transposon but reintegrates it at only 8% of wild-type frequency, an excision-positive, integration-negative phenotype. Circular extrachromosomal transposon molecules recovered from K248A reactions with the RescueSB reporter carry no large deletions but lack a few nucleotides, mostly from the right inverted repeat, and their junction sequences match repair after hairpin resolution. K248A therefore nicks the transposon ends without completing second-strand hydrolysis, letting the 3-OH of the nicked strand attack the opposing strand in a transesterification reaction that liberates the transposon through a hairpin intermediate and diversifies the inverted repeat ends.
• The IR/DR-type transposon Tdr1 of zebrafish and the simple-IR mariner element Hsmar1 of the human genome leave different footprints of internal deletion. Hsmar1 has spawned MITEs consisting only of its inverted repeats, including about 300 copies of the MADE1 variant that are 100% identical to the Repbase consensus and are built from two short inverted repeats joined by a 6-bp intervening sequence, and solo Hsmar1 inverted repeats outnumber full-length copies 12.5-fold (about 2500 versus about 200). No Tdr1-derived MITE sequence can be identified in the zebrafish genome, internally deleted Tdr1 copies always retain more than 400 bp between their IR/DRs, and solo Tdr1 inverted repeats exceed full-length copies only 5.6-fold (660 versus 118). Both families also carry mutational damage at the very ends of their inverted repeats, and for Tdr1 these self-inflicted wounds occur more symmetrically at the two ends than they do for Hsmar1.

Abstract

The functional relevance of the inverted repeat structure (IR/DR) in a subgroup of the Tc1/mariner superfamily of transposons has been enigmatic. In contrast to mariner transposition, where a topological filter suppresses single-ended reactions, the IR/DR orchestrates a regulatory mechanism to enforce synapsis of the transposon ends before cleavage by the transposase occurs. This ordered assembly process shepherds primary transposase binding to the inner 12DRs (where cleavage does not occur), followed by capture of the 12DR of the other transposon end. This extra layer of regulation suppresses aberrant, potentially genotoxic recombination activities, and the mobilization of internally deleted copies in the IR/DR subgroup, including Sleeping Beauty (SB). In contrast, internally deleted sequences (MITEs) are preferred substrates of mariner transposition, and this process is associated with the emergence of Hsmar1-derived miRNA genes in the human genome. Translating IR/DR regulation to in vitro evolution yielded an SB transposon version with optimized substrate recognition (pT4). The ends of SB transposons excised by a K248A excision+/integration- transposase variant are processed by hairpin resolution, representing a link between phylogenetically, and mechanistically different recombination reactions, such as V(D)J recombination and transposition. Such variants generated by random mutation might stabilize transposon-host interactions or prepare the transposon for a horizontal transfer.