V(D)J recombination: the immune system deliberately cuts its own chromosomes, using a domesticated transposon
also called RAG1/RAG2 recombination · programmed DSB · antigen receptor assembly
Every developing lymphocyte deliberately cuts its own DNA in several places and glues the pieces back in a new order, which is how the body builds millions of different antibodies from a handful of gene segments. The scissors that do this are a tamed jumping gene that a vertebrate ancestor captured and put to work.
The same thing, for a scientist
The answer to 'is fusion intended in biology sometimes' begins here. RAG1/RAG2 introduce paired, site-specific double-strand breaks at recombination signal sequences flanking V, D and J gene segments, generating blunt signal ends and covalently sealed hairpin coding ends. Artemis in complex with DNA-PKcs opens the hairpins, and classical NHEJ joins them [E3]. The machinery is literally transposon-derived: RAG1/RAG2 catalyse coupled transposition in vitro, which Gellert's group proposed as a source of oncogenic translocations [E1], and cryo-EM of the amphioxus ProtoRAG transposase identifies the specific jawed-vertebrate adaptations (a RAG1 arginine, a RAG2 acidic region) that suppress transposition more than a thousand-fold [E2]. Deliberate genome breakage is therefore not an exotic exception; it is a load-bearing part of adaptive immunity, tolerated because the risk was engineered down rather than eliminated. That risk is visible in two directions: losing the repair arm causes human SCID [E4][E5], and mis-targeting the recombinase drives oncogenic rearrangement in childhood ALL [E6].
Why this is in a KMT2A knowledge graph1 deeper
Two reasons, and neither is that RAG causes KMT2A fusions — it almost certainly does not (mechanism-rag-off-target-recombination). First, existence: the cell already possesses a programmable, sequence-directed machine that makes paired DSBs and hands them to NHEJ. Machinery that exists to be used is machinery that can be hijacked, and in ETV6::RUNX1 ALL it demonstrably is 6. Second, borrowing: the same machinery is what makes a fusion-junction-directed genome-editing strategy conceivable at all (target-crispr-correction-fusion-junction), and the fact that vertebrates spent 500 million years engineering RAG's transposition activity down by three orders of magnitude 2 is a sober prior on how hard it is to make a programmable nuclease safe in a haematopoietic stem cell.
The engineered-down risk, quantified
The ProtoRAG work is the useful number here: evolution reduced RAG transposition by >1,000-fold and still the residual off-target activity is enough to drive recurrent oncogenic deletions in a common childhood leukaemia 26. Any engineered nuclease deployed in fetal or infant haematopoiesis inherits that arithmetic.
What the failure modes look like in humans
Null mutations in RAG1 or RAG2 abolish V(D)J recombination and cause T-negative B-negative SCID, while hypomorphic missense alleles with partial activity cause Omenn syndrome 5. Mutation of Artemis (DCLRE1C), which opens the coding-end hairpins, causes radiosensitive SCID 4. These are the clearest demonstrations that the programmed-break pathway is both essential and finely dose-dependent: too little activity is immunodeficiency, and mis-directed activity is leukaemia.
- The precedent that a genome-cutting machine can be hijacked is the strongest single argument for caution about introducing an engineered nuclease into infant haematopoiesis.
- Interventions that deplete or arrest early B-cell development to reduce RAG exposure would also produce immunodeficiency, because the same activity builds the antibody repertoire [E4][E5]. There is no separating the function from the hazard.
- Fetal liver B-lymphoid progenitors, the likely KMT2A::AFF1 target compartment, are largely CD10-negative and pre-RAG. Is the target cell for the KMT2A translocation defined partly by being in the window BEFORE RAG expression starts? If so, RAG's absence is itself a property of the cell of origin and belongs in the cell-type entries.
- Does RAG expression status at the moment of the KMT2A break leave a signature at the junction that has never been looked for, such as a bias against cryptic RSS proximity relative to chance?
- Vertebrates suppressed RAG transposition by >1,000-fold [E2]. What is the corresponding suppression factor a therapeutic nuclease would need in a fetal or infant HSC, and is any current editor within that range?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
RAG1 and RAG2 catalyse DNA transposition in vitro, biochemically linking the V(D)J recombinase to transposable elements and to a possible origin of oncogenic translocations.
DNA transposition by the RAG1 and RAG2 proteins: a possible source of oncogenic translocations · Hiom K, Melek M, Gellert M · Cell 1998What this evidence cannot show
In vitro reconstitution with purified core proteins; transposition is strongly suppressed in vivo in jawed vertebrates, so the in vitro activity overstates the in vivo hazard.
Evidence that argues against this (1)
- Twenty years of looking found the in vivo product is a DIFFERENT reaction. Next-generation sequencing of rearrangements in primary murine B cells showed that RAG1/2 causes aberrant insertions by releasing cleaved antibody-gene fragments that then reintegrate into breaks induced independently on a heterologous chromosome, and the same signature was found in human lymphoma and leukaemia. The authors state only a few putative transposition events have ever been documented in vivo and describe this pathway as DISTINCT from DNA transposition and from trans-V(D)J recombination. The in vitro transposition reaction is real chemistry; the oncogenic route it was proposed to explain turns out to be end-donation into pre-existing breaks. Rommel PC et al. 2017
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
RAG derives from a domesticated transposase, and specific vertebrate adaptations suppress its transposition activity by more than a thousand-fold.
Transposon molecular domestication and the evolution of the RAG recombinase · Zhang Y, Cheng TC, Huang G, Xu A, Xiong Y, Schatz DG, et al. · Nature 2019What this evidence cannot show
Structural and biochemical work on the amphioxus ProtoRAG; the evolutionary narrative is inference from comparative structure. An alternative co-evolution model for the origin of V(D)J recombination has been proposed and the single-transposon account should not be presented as uncontested.
Evidence that argues against this (1)
- A published alternative hypothesis argues for co-evolution of two distinct elements rather than a single Transib transposon splitting a primordial receptor gene. Yakovenko I et al. 2021
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Artemis in complex with DNA-PKcs opens the hairpin coding ends generated by RAG, coupling programmed breakage to non-homologous end joining.
Hairpin opening and overhang processing by an Artemis/DNA-dependent protein kinase complex in nonhomologous end joining and V(D)J recombination · Ma Y, Pannicke U, Schwarz K, Lieber MR · Cell 2002What this evidence cannot show
In vitro biochemistry with purified components; the in vivo requirement is established separately by the human genetics in E4.
Evidence that argues against this (1)
- Hairpin opening is not absolutely Artemis-dependent in a living animal. Artemis-deficient mice show a LEAKY SCID: coding-joint formation is defective but not abolished, so another activity opens some hairpins. The same study separates Artemis from DNA-PKcs - signal-joint fidelity is reduced in DNA-PKcs-deficient but not Artemis-deficient fibroblasts - showing the purified two-component reaction does not map one-to-one onto the in vivo requirement. Rooney S et al. 2002
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Mutation of Artemis (DCLRE1C) causes human radiosensitive T-negative B-negative severe combined immunodeficiency.
Artemis, a novel DNA double-strand break repair/V(D)J recombination protein, is mutated in human severe combined immune deficiency · Moshous D, Callebaut I, de Chasseval R, Fischer A, de Villartay JP, et al. · Cell 2001What this evidence cannot show
Human genetics; the gene was identified by positional cloning in a founder population, so allelic spectrum in other populations was established later.
Evidence that argues against this (1)
- The genotype-phenotype statement is the null end of a graded allelic series, and the series reaches into the general population. Functional assay of 41 naturally occurring DCLRE1C mutations in Dclre1c-/- pro-B cells found low or absent recombination activity only for mutations causing typical SCID; most leaky-SCID patients were compound heterozygous for one loss-of-function and one HYPOMORPHIC allele with significant residual recombination and repair activity, and C-terminal-disrupting deletions produce truncated but partially functional protein. The authors note hypomorphic variants already reported in the general population can be disease-causing in trans with a null. 'Artemis mutation causes radiosensitive T-B-SCID' therefore describes one end of a spectrum. Felgentreff K et al. 2015
- E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Complete loss of RAG activity causes T-negative B-negative SCID, while partial (hypomorphic) V(D)J recombination activity causes Omenn syndrome.
Partial V(D)J recombination activity leads to Omenn syndrome · Villa A, Santagata S, Bozzi F, Vezzoni P, Spanopoulou E, et al. · Cell 1998What this evidence cannot show
Genotype-phenotype correlation from a limited patient series; subsequent work showed the RAG allelic spectrum produces a continuum rather than two discrete syndromes.
Evidence that argues against this (1)
- The Omenn phenotype is not a specific readout of partial V(D)J recombinase activity. Pathogenic variants in RMRP - the untranslated RNA component of the RNase MRP endoribonuclease, which has no role in V(D)J recombination - also produce Omenn syndrome, and a newborn homozygous for RMRP n.35C>A presented with the cartilage-hair-hypoplasia skeletal phenotype together with neonatal erythroderma and immunodeficiency. Reading back from an Omenn presentation to 'hypomorphic recombinase' is therefore not safe, and the two-syndrome mapping in this claim omits the genetic heterogeneity of the phenotype. Insalaco A et al. 2026
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Off-target activity of the legitimate V(D)J recombinase is the predominant driver of oncogenic rearrangement in ETV6::RUNX1 childhood ALL.
RAG-mediated recombination is the predominant driver of oncogenic rearrangement in ETV6-RUNX1 acute lymphoblastic leukemia · Papaemmanuil E, Rapado I, Li Y, Greaves M, Campbell PJ, et al. · Nature Genetics 2014What this evidence cannot show
Applies to secondary deletions in ETV6::RUNX1 ALL, and explicitly not to KMT2A fusions, whose junction architecture argues against recombinase involvement (see mechanism-rag-off-target-recombination).
Evidence that argues against this (1)
- This bounds the claim to SECONDARY lesions: the founding translocation itself carries no recombinase signature. Long-distance inverse PCR of TEL-AML1 genomic fusions from nine patients placed breakpoints scattered across 14 kb of TEL intron 5-6 and two cluster regions in AML1 intron 1, with junctions showing characteristic non-homologous end joining - microhomologies, small deletions and duplications - and explicitly NO consistent V(D)J recombinase signal sequences, no topoisomerase II consensus sites and no other recombination motifs. The nearby polymorphic TEL repeat formed paranemic structures in nuclease-sensitivity assays. So the recombinase drives the deletions that accumulate around the fusion, not the fusion; this is the same correlation-of-sequence-features trap the KMT2A intron-11 literature fell into. Wiemels JL et al. 1999
umbrella concept over the V(D)J case
Counter-evidence: Clappier 2014 (PMID 24064621) shows the intragenic ERG deletion is V(D)J-recombinase-mediated and defines the entity in 29/897 children on its own, an origin that does not require DUX4
programmed DNA breakage, mechanism 1 of 5
the physiological programme whose off-target activity that entry evaluates and excludes for KMT2A
- expressed in →Pro-B cell — the stage at which KMT2A::AFF1 leukaemia is arrested, and the phenotype the diagnosis is made on
V(D)J recombination runs in developing B-lineage progenitors
the KMT2A target compartment may sit in the window before RAG activity
second programmed-break programme in the same lineage
RAG's engineered-down transposition sets the safety bar for a therapeutic nuclease in haematopoiesis
programmed DNA breakage and rejoining: machinery that exists to be used can be hijacked
Counter-evidence (Rommel 2017, PMID 28179379): the in vivo RAG hazard is end-donation of cleaved antibody-gene fragments into independently generated breaks, a pathway the authors call distinct from transposition and from trans-V(D)J recombination
Counter-evidence (Wiemels 1999, PMID 10463610): the founding ETV6-RUNX1 junctions show NHEJ signatures and explicitly NO V(D)J signal sequences and NO topoisomerase II consensus sites -- the recombinase drives the secondary deletions, not the founding fusion, the same correlation-of-sequence-features trap as the KMT2A intron-11 account
Counter-evidence (Rommel 2017, PMID 28179379): the residual hazard of a lymphoid-active programmable nuclease is not transposition but migration of cut ends into whatever other breaks exist -- a property of making a break, not of enzyme cleanliness