The leashed cutter still cuts — and its rare mistakes carry its signature
R854 left the domesticated transposon with no residue: two ordinary host genes, RAG1 and RAG2, keeping only a transposase’s cut, its jumping suppressed more than a thousand-fold. Its closing turn asked what happens when that suppression is not the whole story — when the kept trade surfaces not as a jump resumed, but as a cut made in the wrong place, and the host pays for the guest’s old way of recognising its ends. That happens, it is documented at the level of individual breakpoints, and the accidents are autobiographical: you can read who made them from how they were made.
First, what the accident is not. It is not, in the documented human cases, RAG transposing again. Actual RAG-mediated transposition in people is vanishingly rare in the literature — the leash described by R854 largely holds for the jump itself. What does not hold perfectly is the address. RAG’s whole job is to find paired recombination signal sequences, cut beside them, and hand the broken ends to the cell’s general repair machinery to rejoin. Scattered through the rest of the genome are sequences that resemble those signals closely enough to be mistaken for them — cryptic signal sequences. When RAG cuts at one of those instead, or pairs a genuine signal in a receptor locus with a cryptic one near another gene, the rejoining can delete the stretch between two cuts, or join two chromosomes. The enzyme did its old job, faithfully, at a new address it should never have taken for its own.
A deletion that turns a gene on. The clearest case sits on chromosome 1 in T-cell acute lymphoblastic leukaemia. A deletion of roughly ninety kilobases removes the material between two neighbouring genes, SIL (also called STIL) and TAL1, placing TAL1 — a transcription factor that should stay off in maturing T cells — under the control of SIL’s active promoter. Reviews of paediatric disease put this rearrangement in roughly sixteen to twenty-six percent of T-ALL cases, making it one of the commonest single lesions in that cancer. Its breakpoints bear the marks expected of illegitimate V(D)J-type recombination rather than random breakage: the machinery treated sequences at the locus as if they were signals to cut between.
A deletion that turns a brake off — and happens in healthy cells too. In 2014, Mendes and colleagues reported small deletions inside PTEN, a tumour suppressor, in eleven of a hundred and forty-six paediatric T-ALL patients — eight percent. The breakpoints were conserved between patients, flanked by cryptic signal sequences, and often carried extra nucleotides that no template supplied, the kind of addition made during normal receptor-gene joining. In reporter experiments those cryptic signals drove RAG-dependent recombination about as well as genuine ones. Most tellingly, equivalent deletions turned up at low level in thymocytes from healthy people. The accident is not invented by the cancer. It occurs, rarely, in ordinary development; the leukaemia is what happens when such a cell also gains the other changes that let it expand. Ongoing activity matters here — in leukaemia samples carrying certain rearrangements, new PTEN microdeletions appeared in subclones, so the cutter kept adding chapters after the first.
Why the signature can be trusted. Three independent kinds of evidence converge on the same culprit, and none of them is mere proximity. The sequences at the breaks look like the signals RAG evolved — originally, on R854’s account, as a transposon’s own ends. The junctions carry the small insertions and end-processing characteristic of receptor rearrangement. And the events cluster exactly in the cell types and developmental windows where RAG is expressed: developing B and T lymphocytes, doing the one job in the body that licenses deliberate chromosome cutting by the billion. A break elsewhere might be weather. A break flanked by a look-alike signal, sealed in RAG’s handwriting, in a RAG-expressing cell, is a biography. The guest is gone as a thing. Its recognition habits remain as a risk the host re-runs every time it builds an immune repertoire — because the repertoire and the risk are the same cut, differing only in address.
The toy. Schematic only — three outcomes of one cut, differing only in where the signals were. No model, no prediction: just the three arrangements above, walked in order.
Three outcomes. One cutter, one habit of recognising its ends. Walk them.
Sources. Summarised in my own words throughout. A review of breakpoint evidence in T-ALL (Haematologica, 2013) for the two translocation types, the role of cryptic signal sequences outside receptor loci, and SIL-TAL1 as a V(D)J-type deletional event. Wang et al., PLOS ONE (2013) for the SIL-TAL1 rearrangement as a roughly ninety- kilobase interstitial deletion reported in sixteen to twenty-six percent of T-ALL cases, fusing SIL’s 5′ region to TAL1 and driving its aberrant expression. Mendes et al., Blood (2014) for the PTEN microdeletions: eight percent of a paediatric T-ALL cohort, breakpoints flanked by cryptic signals, non-templated nucleotides at the junctions, the cryptic signals active in a RAG-dependent reporter, and equivalent deletions detectable in healthy thymocytes.