Trans-splicing: healthy human cells make cancer-associated fusion transcripts without any chromosome rearrangement
also called RNA trans-splicing · chimeric RNA without translocation · JAZF1-SUZ12
Some fusion proteins that we think of as cancer-causing are also made in perfectly healthy tissue, not by joining chromosomes but by joining two different RNA messages together after they are copied. The DNA is entirely normal; only the message is hybrid.
The same thing, for a scientist
Trans-splicing joins exons from two separately transcribed pre-mRNAs into one mature chimeric transcript, producing a fusion RNA and protein with no underlying genomic rearrangement. The canonical case is JAZF1-SUZ12 (JJAZ1), the product normally created by t(7;17) in endometrial stromal sarcoma, which Li et al. detected as a trans-spliced chimeric RNA and protein in normal endometrial stromal cells lacking the translocation [E1]. The same laboratory later reported PAX3-FOXO1 RNA and protein in normally differentiating myoblasts and fetal muscle in the absence of t(2;13) [E2], and a CML-associated CLEC12A chimera was shown to be a trans-splicing product present in normal as well as transformed cells [E3]. Chimeric immunoglobulin heavy-chain transcripts were described in normal human B cells without the corresponding DNA-level class switch as early as 1996 [E4]. The operational consequence for this project is blunt: **detecting a fusion transcript is not the same as detecting a rearranged genome**, and any screening or diagnostic strategy built on RT-PCR inherits this false-positive class.
What this does and does not say about KMT2A1 deeper
There is no published report of a trans-spliced KMT2A::AFF1 transcript in normal tissue, and this entry does not claim one. What it establishes is a prior: for at least three different cancer fusions, the chimeric transcript exists physiologically. The KMT2A field has never systematically looked, and the assays in routine use (RT-PCR for the fusion transcript, concept-molecular-confirmation-fusion-transcript) are exactly the assays that would not distinguish. The correct confirmatory test is genomic — FISH or a DNA-level breakpoint assay — and the graph should be explicit that RNA-level confirmation alone is weaker than it looks.
A physiological role, not just noise
In the endometrial and myogenic cases the chimeric transcripts are described as arising during normal, regulated differentiation rather than as random mis-splicing 12. If a chimeric RNA has a physiological function in a developmental programme, then the oncogenic version is a quantitative or contextual perversion of a normal process rather than a wholly foreign object, which is exactly the framing the PI's 'is fusion intended in biology' question was reaching for.
- Any diagnostic or screening assay in this project that confirms a fusion by RT-PCR alone can be fooled by this biology; genomic confirmation should be mandatory wherever a positive result would change management.
- If a fusion transcript has a physiological role in normal differentiation, a therapeutic RNA-targeting agent aimed at the fusion junction could hit a normal transcript in a normal tissue, which is a toxicity route that small-molecule interface disruption does not have.
- Has anyone deep-sequenced normal fetal liver and fetal bone marrow for a trans-spliced KMT2A::AFF1 transcript? If such a transcript exists physiologically in the exact tissue where the leukaemia arises, it would reframe the disease as the genomic fixation of a normally transient RNA event. Nobody has looked, and the tissue is available through fetal tissue banks.
- Read the Science 2015 erratum to E1 and record what was corrected; until that is done the JAZF1-SUZ12 claim should not be used as load-bearing evidence anywhere in this graph.
- Are trans-spliced chimeras translated at functionally meaningful levels, or are they transcript-level curiosities? Ribosome profiling in the normal tissues where these chimeras are reported would settle whether 'the fusion protein exists in healthy cells' is a real statement.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The JAZF1-SUZ12 chimeric RNA and protein, normally produced by t(7;17) in endometrial stromal sarcoma, are present in normal endometrial stromal cells without the genomic translocation, arising by trans-splicing.
A neoplastic gene fusion mimics trans-splicing of RNAs in normal human cells · Li H, Wang J, Mor G, Sklar J · Science 2008What this evidence cannot show
A published erratum exists for this Report (Science 2015;350:aad3463, PMID 26430125) and its content was not read by this beat; the claim should be re-checked against the erratum before being relied on for a decision. Confidence is held at moderate for that reason despite the paper's prominence.
Evidence that argues against this (1)
- Accession resolved, as the entry's open question asked. The erratum is PMID 26430125, DOI 10.1126/science.aad3463, Science 2015;350(6256):aad3463, indexed publication type 'Published Erratum', authors Li H, Wang J, Mor G, Sklar J, titled 'Erratum for the Report "A neoplastic gene fusion mimics trans-splicing of RNAs in normal human cells"'. PubMed carries no abstract for it and this beat could not obtain its text, so WHAT was corrected is still unread and the entry's caution should stand. The identifier is now verified as real and correctly attached to the right Report, which was not established before. Read: PubMed record only, no text. Li H et al. 2015
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
PAX3-FOXO1 RNA and protein, the hallmark fusion of alveolar rhabdomyosarcoma, are detectable in normal differentiating myoblasts and fetal muscle without t(2;13).
A chimeric RNA characteristic of rhabdomyosarcoma in normal myogenesis process · Yuan H, Qin F, Movassagh M, Park H, Golden W, Xie Z, Zhang P, Sklar J, Li H · Cancer Discovery 2013What this evidence cannot show
From the same laboratory as E1, so the two are not fully independent; detection is at low abundance and depends on sensitive RT-PCR, the very method whose specificity is in question.
Evidence that argues against this (1)
- A low-abundance chimeric transcript detected by sensitive RT-PCR in normal tissue is the exact signature of the most common library artefact, and the discriminating control is not in the published design. Yan et al. estimated from a cross-species spike-in that approximately 1% of RNA-seq reads are artifactually chimeric through random reverse-transcriptase template switching and alignment error, and showed that the property separating real events from artefacts is reproducibility of the same chimeric fragment across independent biological isolates. PAX3-FOXO1 detected at low abundance in differentiating myoblasts is compatible with trans-splicing and with template switching; the protein-level detection in that paper is the part that artefact cannot explain, so the claim is not refuted, but the RNA evidence alone does not carry it. Read: abstract only. Yan B et al. 2021
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
A CML-associated CLEC12A chimeric transcript arises from trans-splicing and is expressed in normal as well as transformed human cells.
The Fusion of CLEC12A and MIR223HG Arises from a trans-Splicing Event in Normal and Transformed Human Cells · Dhungel BP, Monteuuis G, Giardina C, Bailey CG, Schmitz U, Rasko JEJ, et al. · International Journal of Molecular Sciences 2021What this evidence cannot show
A haematopoietic example, which makes it the most relevant here, but published in a lower-impact venue and not independently replicated.
Evidence that argues against this (1)
- Same objection, applied to the haematopoietic example that matters most here. The CLEC12A-MIR223HG chimera was reported from RNA-level evidence in normal and transformed cells and has not been independently replicated; Yan et al.'s measured approximately 1% artefactual-chimera background from RT template switching means that a single-laboratory RNA-level report of a normal-tissue chimera needs cross-isolate reproducibility or an enrichment control before it is load-bearing. This is the entry's most relevant evidence item and also its least protected one. Read: abstract only. Yan B et al. 2021
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Chimeric germline immunoglobulin heavy-chain transcripts occur in normal human B cells in the absence of DNA-level class switch recombination, consistent with trans-splicing of Ig RNA.
Multiple types of chimeric germ-line Ig heavy chain transcripts in human B cells: evidence for trans-splicing of human Ig RNA · Fujieda S, Lin YQ, Saxon A, Zhang K · Journal of Immunology 1996What this evidence cannot show
Pre-genomic-era methods; no DOI indexed. Important mainly as an early demonstration that chimeric transcripts can mimic a DNA-level recombination product.
Evidence that argues against this (1)
- Cis-splicing of ADJACENT genes is an unexcluded alternative here, and it is the alternative that has since overturned the best-known trans-splicing claim. The immunoglobulin heavy-chain constant-region genes are tandem neighbours on one chromosome, so a readthrough transcript spliced in cis would produce exactly the chimeric germline transcripts reported, with no trans-splicing and no DNA-level recombination. That is precisely what happened to SLC45A3-ELK4, a chimera long attributed to trans-splicing and later shown to arise by cis-splicing of adjacent genes. Pre-genomic-era methods cannot distinguish the two, so this observation bounds the DNA-recombination interpretation without establishing trans-splicing. Zhang Y et al. 2012
Counter-evidence (Yamamoto 1998, PMID 9737688): no trans-spliced MLL/AF-4 chimera was found in 65 normal cord bloods, but 31% carried an in-frame AFF1 exon 11 to exon 4/5 aberrant transcript -- the RNA-assay confounder for this fusion sits on the AFF1 side, not in a genuine chimera
Counter-evidence (Yamamoto 1998, PMID 9737688): no trans-spliced MLL/AF-4 chimera in 65 normal cord bloods, but an in-frame AFF1 exon 11 to exon 4/5 transcript in 31% of them -- a junction gapmer is safe on this evidence, an AFF1-side reagent is not
Shared counter-evidence (Yan 2021, PMID 33980601): a cross-species spike-in measures approximately 1% of RNA-seq reads as artifactually chimeric from random reverse-transcriptase template switching, so a third category sits underneath both read-through and trans-splicing -- the chimeric read that corresponds to no molecule; cross-isolate reproducibility is the discriminating control and neither entry's sources applied it
chimeric transcript without rearrangement, mechanism 4 of 5
RNA-level confirmation alone cannot distinguish a rearrangement from a trans-spliced chimera
junction specificity must be verified against normal chimeric transcripts, not assumed
Counter-evidence (Zhang 2012, PMID 22719019): cis-splicing of ADJACENT genes overturned the best-known trans-splicing assignment, and the IGH constant-region genes are tandem neighbours -- so chimeric germline Ig transcripts are explained by readthrough cis-splicing without invoking trans-splicing