Measure the pausing index at KMT2A::AFF1 target genes — the one number the whole pausing argument turns on, and it has never been reported
also called PRO-seq pausing index KMT2A::AFF1 · pause index experiment
Run a nascent-RNA sequencing experiment in an infant-leukaemia cell line carrying the KMT2A::AFF1 fusion, and count how much polymerase is stalled at the start of the fusion's target genes compared with other genes. One dataset either supports or kills the idea that these cells run with the transcriptional brake off.
The same thing, for a scientist
**Design.** PRO-seq (or equivalent nascent-transcription assay) in a KMT2A::AFF1-positive B-ALL line (for example SEM or RS4;11) with a degron-tagged fusion, plus a matched KMT2A-germline B-ALL comparator. Compute the **pausing index** (promoter-proximal read density divided by gene-body density) for (a) direct fusion targets defined by fusion ChIP or CUT&RUN in the same cells, (b) expression-matched non-target genes in the same cells, and (c) the same target set in the comparator line. Repeat after 2-4 h of acute fusion degradation. **Readout.** The difference in pausing index between target and matched non-target genes, and its change on fusion loss. **Controls.** Expression matching is the critical one, because pausing index correlates with expression level and an unmatched comparison will recover a difference that is about transcription rate and not about the fusion. Spike-in normalisation is required for the degron arm. **Why it is worth doing:** the field's model of the KMT2A fusion is an elongation hijack, and this number - the direct measurement of what the fusion does to the pause - is absent from the literature [E1][E2]. This graph's dependency data hint at an asymmetry [E3] and cannot test it.
What each outcome would mean
| Result | Interpretation |
|---|---|
| Target genes have a lower pausing index than matched non-targets, and it rises on fusion degradation | hyp-pause-axis-asymmetry-kmt2a-r supported; the fusion drives a constitutively pause-released state, which would explain why pause release is not rate-limiting 1 |
| No difference between target and matched non-target genes | The hypothesis is dead and, more usefully, the graph can stop treating pausing as an open question in this disease |
| Target genes have a higher pausing index | The fusion accumulates paused polymerase, which would be a genuinely new model and would make NELF and CDK9 far more interesting than this beat concluded |
| A difference that vanishes on expression matching | The commonest artefact; the experiment is designed around avoiding it |
All three informative outcomes are publishable and two of them change what this project believes.
Sequencing order and cost discipline
Do the free analysis first. Before any wet work, re-run the project's genome-wide selectivity scan for the six pausing-axis genes with permutation testing, lineage conditioning and line normalisation, using scripts this repository already has (analyses/depmap-null-distribution, analyses/depmap-lineage-conditioned, analyses/depmap-line-normalised-specificity). If the SUPT4H1 and NELFA/NELFB signals do not survive, the motivating pattern is an essentiality-floor artefact 3 and the wet experiment should not be run on that basis - though it would still be worth doing on the field's own terms, since the number is missing regardless.
Then the cheapest wet version. A single PRO-seq pair in one fusion-positive line with expression-matched internal controls answers the primary question without needing the degron. The degron arm is the confirmatory step and should follow only a positive primary result.
| design | PRO-seq (or NET-seq/TT-seq) in a KMT2A::AFF1-positive B-ALL line with a degron-tagged fusion, plus a KMT2A-germline B-ALL comparator. Fusion target genes defined by CUT&RUN or ChIP in the same cells. Primary comparison: pausing index at targets versus expression-matched non-targets in the same cells. Secondary: change in pausing index 2-4 h after acute fusion degradation, spike-in normalised. |
|---|---|
| readout | pausing index (promoter-proximal read density / gene-body read density) per gene; difference between target and matched non-target sets; change on fusion degradation |
| controls | expression matching of the non-target set (essential - pausing index correlates with expression); a KMT2A-germline comparator line; spike-in normalisation for the degron arm; a septin-partner KMT2A fusion line as a specificity control, since those fusions recruit no elongation machinery |
| cost tier | med |
| timeline | free re-analysis of the existing DepMap scan in days; primary PRO-seq comparison in 2-3 months; degron arm 3-6 months further |
| prerequisite analysis | re-run analyses/depmap-null-distribution, analyses/depmap-lineage-conditioned and analyses/depmap-line-normalised-specificity for SUPT4H1, SUPT5H, NELFA, NELFB, NELFCD and NELFE before committing to wet work |
| tests hypothesis | hyp-pause-axis-asymmetry-kmt2a-r |
| why not done | no published pausing-index measurement exists for any KMT2A fusion target gene set; the field's elongation-hijack model has been carried on recruitment evidence rather than on a direct pausing measurement |
- A negative result is likely and is worth having, but it will be hard to publish, so the experiment should be scoped as an internal decision-maker rather than as a paper.
- Pausing index is sensitive to normalisation choices and to how promoter-proximal windows are defined; a poorly specified analysis can produce either answer from the same reads.
- If the answer is that the fusion does control the pause, the therapeutic implication points at CDK9 - a pan-essential kinase whose inhibitors have a long history of dose-limiting toxicity - so a positive result creates a target problem rather than solving one.
- Which KMT2A::AFF1 line has the cleanest degron-tagged fusion available, and is a matched KMT2A-germline B-ALL comparator with similar proliferation rate available?
- Is expression matching sufficient, or does pausing index in B-lineage cells also require matching on CpG-island status and gene length?
- Should the septin-partner specificity control be run in the first pass rather than held back? It would cost one more library and would make a negative result far more interpretable.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Okuda and Yokoyama state that it is hypothesised that relieving promoter-proximal pausing of RNA polymerase II is the rate-limiting step of MLL fusion-dependent transcription, and report the contrary: the key activity responsible for the oncogenic property of MLL-AEP fusion proteins is TBP loading through SL1, not mediator recruitment or transcriptional elongation activities.
TBP loading by AF4 through SL1 is the major rate-limiting step in MLL fusion-dependent transcription · Okuda H, Takahashi S, Takaori-Kondo A, Yokoyama A · Cell Cycle 15(20):2712-22 2016What this evidence cannot show
Abstract fetched and read on 2026-09-19; full text not read. Murine myeloid transformation assays with engineered MLL-AEP constructs; the conclusion is about what limits transformation by those constructs, not about a measured pausing index at any gene. Notably, the paper does not report a pausing index either - which is the gap this experiment fills.
Evidence that argues against this (1)
- The pause-release model has real support from a different direction: ENL, an MLL fusion partner, initiates multivalent phase separation of SEC that concentrates P-TEFb and controls pause release, and MLL-ENL fusion is reported to boost SEC droplet formation. If the fusion increases pause-release capacity, the experiment could find lower pausing at target genes for a reason that has nothing to do with this beat's dependency asymmetry. Guo C et al. 2020
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Acute perturbation of AQR/XAB2-dependent functions of the ECAS complex, an AFF4-ELL2-EAF1 elongation module joined to a U2 snRNP-associated intron binding complex that occupies promoter-proximal chromatin, impairs productive gene-body elongation without altering pause release, increases intron retention and preferentially affects long genes; Aqr perturbation impairs MLL-rearranged leukaemia initiation, maintenance and leukaemia stem cell function.
A dual-module ECAS complex couples transcription elongation and RNA processing · Xiang Y, Wang H, Yin Z, Xiao R, Song Z, Lu R, Dai R, Ye J, Li C, Cui Y, Wang M, Lv J, Li Q, Wang F, Li Z, Zhang K, Zhang Y, Wu D, Fang P, Liang K · Molecular Cell, online ahead of print 16 September 2026 2026What this evidence cannot show
Abstract fetched and read on 2026-09-19; full text not read. The 'without altering pause release' observation is for AQR perturbation in human cells and is the closest published measurement to the one this experiment proposes - but it is a perturbation readout, not a baseline comparison of fusion targets against non-targets, which is what is missing.
Evidence that argues against this (1)
- This project's own scan does not support AQR as a KMT2A-selective dependency (rank 17,654, difference +0.244), so the MLL-relevance of the ECAS mechanism is itself contested within this graph. Using it to motivate the experiment is therefore weaker than using Okuda and Yokoyama. Broad Institute DepMap / Open Targets Platform; analyses/depmap-kmt2a-selective-scan-full 2026
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
In the project's genome-wide KMT2A-rearranged selectivity scan, the two DSIF subunits score on the more-required side (SUPT4H1 -0.2642 rank 138; SUPT5H -0.1215 rank 767) while two of four NELF subunits score at the anti-dependency extreme (NELFB +0.2650 rank 17,671; NELFA +0.2403 rank 17,650), with NELFE +0.0583 and NELFCD -0.0208 unremarkable.
Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap Chronos via Open Targets) · Broad Institute DepMap / Open Targets Platform; analyses/depmap-kmt2a-selective-scan-full 2026What this evidence cannot show
Rows read out of analyses/depmap-kmt2a-selective-scan-full/results.json on 2026-09-19. Eleven KMT2A-rearranged lines, AML-heavy, none infant-derived; the proposed experiment is in an infant B-ALL line, so the motivating data and the experimental system do not match in lineage.
Evidence that argues against this (1)
- Within NELF the selectivity ordering tracks pan-essentiality depth exactly, which is the signature of a floor artefact rather than of biology. This is why the experiment's first step is the free re-analysis of the existing scan rather than any wet work. Vos SM et al. 2018
pausing index at fusion targets versus expression-matched non-targets; a single dataset that supports or kills the hypothesis
measures directly what the fusion does to promoter-proximal pausing at its own target genes - never published for any KMT2A fusion