When one infant's KMT2A::AFF1 B-ALL became an AML, EBF1 lost fusion occupancy and PAX5 kept it - the B-lineage transcription-factor programme is not vacated as a block
In the same child, before and after the leukaemia switched from lymphoid to myeloid, the fusion came off one master B-cell gene but stayed on another. The B-cell programme is not simply switched off wholesale.
The same thing, for a scientist
**Per-promoter values, fold over each sample's own mean promoter coverage, with the N/C ratio and the fusion-target call from that sample's own KMT2A tracks** [E1]: | gene | p279 (B-ALL) | p318 (AML, same patient) | p454 (B-ALL, other patient) | |---|---|---|---| | **EBF1** | N 7.84, C 2.23, N/C **3.09**, ENL 9.1, **fusion target** | N 2.54, C 2.50, N/C **1.02**, ENL 3.0, **not a target** | N 8.95, C 1.96, N/C **3.68**, ENL 23.1, **fusion target** | | **PAX5** | N 20.69, N/C **7.14**, ENL 91.8, **fusion target** | N 19.37, N/C **8.47**, ENL 44.4, **fusion target** | N 20.85, N/C **10.41**, ENL 86.1, **fusion target** | | **CD19** | N 0.61, N/C 0.83, H3K4me3 2.15, not a target | N 0.11, N/C 1.27, **H3K4me3 0.06**, not a target | N 0.74, N/C 1.29, H3K4me3 2.18, not a target | **Three separable things happen to three B-lineage genes in the same switch:** 1. **EBF1** loses fusion occupancy outright - N falls 3.1-fold and the N/C ratio falls to 1.02, i.e. what remains is wild-type KMT2A. 2. **PAX5** keeps it entirely - the N/C ratio actually rises to 8.47, the highest of the three states. 3. **CD19** was never a fusion target in any state, and simply goes silent - H3K4me3 falls from 2.15x to 0.06x its own sample's mean promoter. **So CD19 loss at lineage switch is not a loss of fusion occupancy at CD19.** The fusion was not on the CD19 promoter in the B-ALL either. Whatever removes CD19 acts upstream, and EBF1 - which the fusion does leave - is a candidate for that upstream step, since EBF1 is required for CD19 expression in normal B-cell development. **This is a hypothesis generated by the data, not a result: no causal step was tested here.** **The number of patients is one.** This is a single lineage-switching patient profiled twice, with a second unrelated B-ALL as the only contemporaneous comparison.
Why PAX5 keeping the fusion matters for target lists
PAX5 is among the most strongly fusion-occupied promoters in all three states - ENL at 44-92x mean promoter, N/C 7-10. A target list built from the B-ALL and a target list built from the AML of the same patient would both contain PAX5, and only the first would contain EBF1. The genes that stay are as informative as the genes that go, and this corpus has mostly recorded the ones that go.
| script | analyses/q142-aff1-in-primary-infants/analyze.py |
|---|---|
| status | run |
| result summary | EBF1 N/C 3.09 to 1.02 across the switch; PAX5 7.14 to 8.47; CD19 never a target and loses H3K4me3 35-fold. |
- CD19 was not a fusion target in the B-ALL state, so a model in which CD19-directed therapy 'selects for cells that switch off a fusion target' is not supported by these data; the CD19 loss is downstream of something else.
- If PAX5 stays occupied through a lineage switch, a PAX5-anchored fusion-occupancy biomarker would not flag the switch that changes everything else.
- Does EBF1 lose fusion occupancy before or after CD19 goes silent? A timecourse does not exist in any KMT2A::AFF1 patient.
- Is EBF1 occupancy retained in the other two lineage-switching pairs in this deposit (KMT2A::ENL and KMT2A::EPS15)?
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
In one infant's KMT2A::AFF1 B-ALL the EBF1 promoter carries KMT2A-N at 7.84x mean promoter with N/C 3.09 and is called a fusion target; in the AML that the same patient's disease became, N falls to 2.54x with N/C 1.02 and it is not called; PAX5 remains a fusion target in both with N/C 7.14 rising to 8.47; CD19 is a fusion target in neither and loses H3K4me3 from 2.15x to 0.06x.
GSE252378 - AutoCUT&RUN and AutoCUT&Tag profiles of 36 MLL-rearranged leukemia samples (171 GSMs, hg38) · Janssens DH, Henikoff S (Fred Hutchinson Cancer Center / Van Andel Institute) · NCBI Gene Expression Omnibus 2024What this evidence cannot show
One patient, two timepoints. Single 2 kb promoter windows at one RefSeq TSS per gene - a promoter-window value is the least robust statistic in this beat and a distal element would be missed. Within-sample shares; no spike-in. p318 is a post-chemotherapy relapse, so therapy, clonal selection and lineage change are confounded.
Evidence that argues against this (1)
- The source paper reports that oncoprotein levels are globally reduced in lineage-switching samples, so EBF1's fall could be the generic loss of the weakest sites rather than a lineage-specific eviction - PAX5's retention is consistent with that too, since PAX5 is a much stronger site to begin with. A level-matched comparison would be needed to separate the two and was not run. Janssens DH et al. 2024
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
p318 is the AML arising from the lineage switch of the infant B-ALL p279 under ALL-directed chemotherapy.
MLL oncoprotein levels influence leukemia lineage identities · Janssens DH, Duran M, Otto DJ, Wu W, Xu Y, Kirkey D, Mullighan CG, Yi JS, Meshinchi S, Sarthy JF, Ahmad K, Henikoff S · Nature Communications 15:9341 2024What this evidence cannot show
From the paper and the GEO descriptions.
Evidence that argues against this (1)
- The pairing is asserted only by the depositing group; no independent clonality evidence is presented in the deposit. Janssens DH et al. 2024
a worked locus inside the same comparison
PAX5 retains fusion occupancy through the switch
- related to →CD19 - the molecule this disease is treated by, named in 114 entries and written in none of them until now
CD19 is not a fusion target in any state and loses H3K4me3 at the switch
the retained/lost split