Oncofusion

start / graph / genome-aff1-breakpoint-region-architecture

Mechanismmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q32-break-machinery

The other half of the break: what is actually known about the AFF1 breakpoint region, and the negative-control gene nobody cites

also called AF4 breakpoint cluster region · 4q21 breakpoint architecture · partner-side breakpoint structure t(4;11)

A translocation needs two breaks, and almost everything written about this one describes only the KMT2A side. The AFF1 side has been mapped, and the study that did it also examined a near-identical gene that never translocates - which is the comparison the KMT2A literature has been missing for thirty years.

The same thing, for a scientist

This entry closes a gap that `breakpoint-kmt2a-bcr-architecture` names as an open question: 'what is the corresponding structural map of the AFF1 breakpoint region? The KMT2A side is well characterised and the AFF1 side is not.' Four primary results define it. The AFF1 gene structure was assembled from genomic phage contigs into 21 exons with an intron phase map, and the **translocation breakpoint cluster region is flanked by exons 3 and 6** [E1]. Halo mapping and scaffold reassociation found nearly the entire AFF1 breakpoint cluster region to be nuclear-matrix attached, with t(4;11) breakpoints preferentially clustered OUTSIDE high-affinity matrix attachment regions on both the KMT2A and AFF1 sides [E2]. Strick et al. mapped chromatin structural elements across BCR, ABL and AFF1 and found topoisomerase II and DNase I cleavage sites associated with all breakpoint cluster regions and scaffold attachment regions associated with ABL and AFF1 but not BCR; crucially, **no close breakpoint clustering with the TOP2/DNase I sites was observed**, only a statistically significant 5' or 3' displacement of patient breakpoints relative to them, which the authors read as implicating DNA repair and exonucleases [E3]. And fine-structure analysis of both derivative chromosomes in three t(4;11) cell lines and two primary patients found multiple double- or single-stranded breaks near the breakpoints on BOTH participating chromosomes, with intervening fragments deleted, inverted or duplicated, and no evidence for target sequences or immune-system recombinases [E4].

Go deeper
The negative control the field has and does not use

Strick et al. included MLL2 - described by them as a homolog of MLL that is not associated with chromosome translocations - as an internal comparator. It was expressed in every cell line tested, and apart from a single DNase I site in its promoter, no other chromatin structural elements were found in it 3.

That is the shape of the control every feature-association claim about the KMT2A BCR needs and almost none of them have. The standard argument runs: the KMT2A BCR contains TOP2 sites, DNase hypersensitivity and scaffold attachment, therefore it breaks. The question that argument never answers is how many non-breaking genes contain the same features. Here is one that does not, which converts the association from a description into something with a contrast behind it.

Two honest limits. First, it is a single comparator gene, and one control is an anecdote about background rather than a background distribution - this is precisely why ana-repeat-feature-enrichment-kmt2a-breakpoints specifies matched decoy genes rather than one. Second, the modern symbol MLL2 is contested: the name has been applied both to KMT2B and to KMT2D across the literature, and this entry does not assert which gene the 2006 paper assayed. It reports the comparison as the authors framed it, as a non-translocating MLL homolog.

The result that undercuts the simple TOP2 story on BOTH sides1 deeper

Strick et al.'s central negative is easy to miss: no close breakpoint clustering with the TOP2/DNase I sites was observed across BCR, ABL and AFF1. What was observed was a significant 5'-or-3' displacement of patient breakpoints relative to those sites, which the authors interpret as the signature of DNA repair and exonuclease processing rather than of cleavage exactly where the enzyme sits 3.

Read against the KMT2A side, this is consistent rather than contradictory. Cowell et al. reported that TOP2 occupancy and measured etoposide-induced cleavage do not explain where KMT2A breakpoints cluster. Two independent groups, on two sides of the same translocation, found that the enzyme's position does not predict the breakpoint's position. The recurring lesson is that the breakpoint marks where the ends were joined after processing, not where the cut was made, and this is exactly why junction structure (ana-junction-microhomology-signature-mining) carries information that breakpoint position does not.

Both chromosomes break more than once

Reichel et al. analysed breakpoints on both derivative chromosomes in three t(4;11) cell lines and two primary patients and concluded that multiple double- or single-stranded breaks must have occurred near the breakpoints on both participating chromosomes, with fragments between them deleted, inverted or duplicated 4. They found no evidence for specific target sequences or immune-system recombinases, and reported the same features on re-interpretation of published t(6;11) and t(9;22) data. Their conclusion - that these translocations are reciprocal but not balanced at the fine-structure level - is the founding statement of the DNA damage-repair model for this fusion and it applies symmetrically to the AFF1 side.

What is still missing on this side

Everything modern. There is no CTCF or cohesin map of the AFF1 breakpoint region in a haematopoietic progenitor, no replication-timing measurement, no repeat-enrichment test, and no DSB map. The structural work is 1997-2006 and predates every genome-wide chromatin assay now standard. The consequence is that hyp-kmt2a-aff1-shared-loop-anchor-hub is half-testable at best: its KMT2A-side prediction can be checked against public data and its AFF1-side prediction has no prior literature to check against at all.

AFF1 spans 206,871 bp on GRCh38 (chr4:86,934,203-87,141,073, plus strand), more than twice the 90,377 bp of KMT2A 5, and its breakpoint cluster region occupies a small part of it. Any enrichment test on this side has more sequence to control for and less prior annotation to lean on.

Facts and figures
geneAFF1 (AF-4), 4q21
grch38 spanchr4:86934203-87141073 (+), 206871 bp
bcr location 1997 numberingflanked by exons 3 and 6
newest primary structural work2006
features reported
[
  {
    "feature": "nuclear matrix attachment",
    "finding": "nearly the entire AFF1 BCR matrix-attached; breakpoints cluster OUTSIDE high-affinity MARs",
    "refs": [
      "E2"
    ]
  },
  {
    "feature": "scaffold attachment regions",
    "finding": "present at AFF1 and ABL, absent at BCR",
    "refs": [
      "E3"
    ]
  },
  {
    "feature": "topoisomerase II and DNase I cleavage sites",
    "finding": "present, but patient breakpoints do NOT cluster at them; significant 5' or 3' displacement instead",
    "refs": [
      "E3"
    ]
  },
  {
    "feature": "multiple breaks per event",
    "finding": "multiple DSB/SSB near the breakpoint on BOTH chromosomes with deleted, inverted or duplicated intervening fragments",
    "refs": [
      "E4"
    ]
  },
  {
    "feature": "recombination signal sequences",
    "finding": "none found",
    "refs": [
      "E4"
    ]
  }
]
internal negative control
{
  "gene_as_named_by_authors": "MLL2",
  "finding": "expressed in all cell lines tested; apart from one promoter DNase I site, no chromatin structural elements found",
  "caution": "the symbol MLL2 has been applied to more than one gene across the literature; this entry does not assert which",
  "refs": [
    "E3"
  ]
}
missing entirely
[
  "CTCF or cohesin occupancy at the AFF1 BCR in any haematopoietic cell",
  "replication timing",
  "repeat-class enrichment test",
  "any genome-wide DSB map covering this locus"
]
How this could bite us back
  • Nothing here is a target. Describing the partner-side architecture constrains mechanism and offers no intervention, and the entry exists to stop one-sided reasoning rather than to open a therapeutic route.
  • The single negative-control gene is the most quotable result in this entry and is the one most likely to be over-cited. One non-breaking gene lacking these features is not a background distribution, and quoting it as proof that the features cause breakage would repeat the error it was included to expose.
Still open
  1. Does the AFF1 breakpoint region overlap a CTCF/cohesin loop anchor in a haematopoietic progenitor? No measurement of any kind exists, which makes the shared-loop-anchor hypothesis half-testable.
  2. Which gene did the 2006 study assay as 'MLL2'? Resolving the symbol would turn its negative-control result from a suggestive comparison into a usable one, and it is a question for the authors or the paper's figures rather than for a database.
  3. Why do patient breakpoints sit 5' or 3' of the TOP2/DNase I cleavage sites rather than at them, on three different loci? The authors implicate exonucleolytic processing; a junction-structure analysis could test that directly by asking whether the displacement distance matches the resection lengths seen at TOP2-induced breaks.
  4. Is there any structural work on the AFF1 breakpoint region after 2006? Searches found none, which means the partner side of the commonest infant leukaemia fusion has not been looked at with any post-genomic assay.
Every source behind this entry (5)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The human AFF1 (AF-4) gene at 4q21 was assembled from genomic phage contigs into 21 exons with an intron phase map, two alternative first exons each carrying a translation initiation codon, and a translocation breakpoint cluster region flanked by exons 3 and 6.

    PMID 9233580DOI
    What this evidence cannot show

    1997 phage-contig assembly; exon numbering in this paper is not automatically the numbering used by modern diagnostic reports or by the 2024 KMT2A consensus-structure conventions, and exon numbers for AFF1 must be re-mapped before being compared across sources.

    Evidence that argues against this (1)
    • Exon-numbering drift across decades is not hypothetical in this gene family: the same group later had to publish a consensus gene structure for KMT2A explicitly because research and diagnostic numbering had diverged. An AFF1 breakpoint described as 'intron 3' in 1997 and one described as 'intron 3' in a 2014 case report are not guaranteed to be the same interval without lift-over. Meyer C et al. 2024
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Halo mapping and scaffold reassociation with Distamycin-A competition found about 20 kb of KMT2A and nearly the entire AFF1 breakpoint cluster region to be nuclear-matrix attached, with t(4;11) breakpoints preferentially clustered outside regions of high matrix-attachment affinity.

    PMID 11527155DOI
    What this evidence cannot show

    In-vitro scaffold reassociation is an assay whose relationship to in-vivo chromatin anchoring is indirect, and Distamycin-A competition defines affinity classes operationally. t(4;11) ALL cohort only, so the therapy-related case is not represented.

    Evidence that argues against this (1)
    • The other primary mapping of the KMT2A side reached the opposite sign on the same variable, placing a high-affinity scaffold attachment region inside the telomeric half of the BCR alongside six TOP2 consensus sites with therapy-related breakpoints concentrated there. The matrix-attachment literature at this locus pair disagrees about whether anchoring attracts or repels breakpoints, and neither result has been repeated with a modern assay. Broeker PL et al. 1996
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Topoisomerase II and DNase I cleavage sites associate with the BCR, ABL and AFF1 breakpoint cluster regions and scaffold attachment regions associate with ABL and AFF1 but not BCR; no close clustering of patient breakpoints with the TOP2/DNase I sites was observed, only a statistically significant 5' or 3' displacement implicating DNA repair and exonucleases; and MLL2, an expressed MLL homolog not associated with translocations, contained no such structural elements apart from one promoter DNase I site.

    PMID 16572268DOI
    What this evidence cannot show

    2006 gel-based cleavage-site mapping in cell lines and primary cells, not genome-wide sequencing. One comparator gene is a single control, not a background distribution. The symbol MLL2 has been applied to different genes across the literature and this entry does not assert which gene was assayed; it reports the comparison as the authors framed it.

    Evidence that argues against this (1)
    • Genome-wide DSB mapping has since reported the opposite emphasis at the class level: TOP2-mediated breaks ARE enriched at specific structural positions, namely CTCF/cohesin-bound chromatin loop anchors, and leukaemia-driving translocating genes are enriched there. A 2006 gel-based survey finding no clustering with TOP2 sites and a 2017-2019 sequencing survey finding strong positional enrichment are not measuring the same thing at the same resolution, and the older negative should not be read as refuting the newer positive. Canela A et al. 2017
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Fine-structure analysis of breakpoints on both derivative chromosomes in three t(4;11) cell lines and two primary patients showed that multiple double- or single-stranded DNA breaks must have occurred near the breakpoints on BOTH participating chromosomes, with intervening fragments deleted, inverted or duplicated, and no evidence for specific target sequences or immune-system recombinases; the translocations are reciprocal but not balanced at the fine-structure level.

    PMID 9881706DOI
    What this evidence cannot show

    Three cell lines and two patients, Sanger-era cloning. Cell-line breakpoints can acquire structure in culture. The re-interpretation of published t(6;11) and t(9;22) data is secondary analysis of others' figures.

    Evidence that argues against this (1)
    • Complex, unbalanced junctions are the generic signature of error-prone end joining rather than evidence of a KMT2A-specific multi-break mechanism. Junctions from patients with a different prenatal fusion show the same repertoire of microhomologies with small deletions and duplications and no consistent recombination motifs, and were read by their authors as products of ordinary imprecise repair. Establishing that ends were processed before ligation is not the same as establishing that multiple independent strand breaks occurred on both chromosomes. Wiemels JL et al. 1999
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    On GRCh38 AFF1 spans chr4:86,934,203-87,141,073 on the plus strand, 206,871 bp, against KMT2A's 90,377 bp at chr11:118,436,456-118,526,832.

    What this evidence cannot show

    Canonical gene span on the primary assembly; alternative transcripts and the historic restriction-fragment coordinate systems used in the 1990s breakpoint literature do not map onto these numbers without explicit lift-over.

    Evidence that argues against this (1)
    • A gene span is a scope-limited fact: the breakpoint-competent interval is a small part of it, flanked by exons 3 and 6 in the 1997 numbering, so using the full 206,871 bp as a permutation domain would over-represent sequence in which no break has ever been reported. Nilson I et al. 1997
Connected to