Alu repeats and nuclear-matrix anchors in the KMT2A breakpoint region — the genome-feature side of the break

The most defensible statement is an indirect one. When Echlin-Bell and colleagues sequenced the full 8.3-kb KMT2A BCR in 87 individuals (22 topoisomerase-II-inhibitor-treated patients who developed t-AML, 37 treated patients who did not, 8 infants, 20 normal controls), every one of the four polymorphic sites they found lay inside an Alu repetitive element — three altering poly-A tract length and one a (GAA)n trinucleotide repeat 1. You cannot find four polymorphisms and have all four fall in Alu unless Alu occupies a large fraction of the region.

That same study is also a useful negative: the three poly-A polymorphisms occurred at equal frequency in leukemic patients and controls, so they are not risk predictors. The (GAA)4/5 heterozygote excess in adult t-AML (83% vs 55% of normal individuals) was not statistically significant 1. Alu content is a property of the locus, not a demonstrated susceptibility allele.

Alu sequences are annotated at the t(4;11) breakpoint junctions themselves in the Marschalek group's breakpoint-isolation work 2, but that paper's purpose was PCR primer design for minimal-residual-disease monitoring, not repeat-enrichment statistics, so it supports presence rather than causal enrichment.

This is the most honest thing in the node, and it is a disagreement that has never been resolved.

Broeker and colleagues mapped breakpoints in 31 de novo acute leukemias and 8 t-AML cases across the 8.3-kb BamHI BCR and identified a weak-affinity SAR just centromeric to the BCR and a high-affinity SAR inside the telomeric half. 74% of de novo breakpoints fell in the centromeric half (between the two SARs), whereas 75% of t-AML breakpoints fell in the telomeric half containing the high-affinity SAR and six topoisomerase-II consensus sites; the de novo distribution difference gave p = 0.02 4. Their reading: chromatin structure partitions the breakpoints, and de novo and therapy-related leukemia break by different mechanisms.

Hensel and colleagues did halo mapping and scaffold-reassociation with Distamycin-A competition on both MLL and AF4, found roughly 20 kb of MLL and nearly the whole AF4 BCR matrix-attached, and concluded that DNA with strong MAR affinity contained fewer breakpoints while weak or absent MAR affinity marked higher breakpoint density 5.

These are not trivially reconcilable. They used different assays (in-vitro SAR binding versus halo mapping plus competition), different patient mixes (Broeker's telomeric enrichment is driven by t-AML, Hensel's cohort is t(4;11) ALL), and they disagree on the sign of the correlation. Anyone citing "the MLL breakpoint region is a matrix attachment region, therefore it breaks" should be asked which of these two papers they mean.

All of the primary work above is from 1996-2003 and predates RepeatMasker-era genome builds, ENCODE chromatin maps and large breakpoint catalogues. The Meyer/Marschalek recombinome now holds thousands of sequenced KMT2A breakpoints. Nobody has intersected that breakpoint catalogue with current repeat annotation, G-quadruplex predictions, replication-timing maps and ENCODE candidate cis-regulatory elements and asked whether repeat class explains breakpoint position. That is a purely computational analysis the lab could run this month; it is written up as ana-repeat-feature-enrichment-kmt2a-breakpoints.

On the specific PI question about transcriptionally active repeats and torsional stress: we found no study measuring torsional stress, supercoiling or R-loop formation over repeats inside the KMT2A BCR. The transcription-associated-break argument for this locus rests on the transcriptional state of KMT2A itself, not on its repeats. Treat "Alu transcription creates torsional stress that breaks KMT2A" as an untested hypothesis, recorded as hyp-r-loop-landscape-kmt2a-r.