Removing the fusion is a transcription event first and a chromatin event later, and the gap is enormous: nascent transcription changes within 15 minutes, while genome-wide RNA Pol II peak calls are 99.6% retained at 3 hours and not one active histone mark loses a peak call by 24 hours - the fusion's immediate job is licensing elongation, not building a chromatin landscape
also called transcription first, chromatin later · 15 minutes versus 24 hours · Pol II at 99.6% while transcription collapses · the fusion licenses elongation
When the cancer fusion protein is destroyed the genes it controls stop being read within fifteen minutes, yet the reading machinery is still sitting on the DNA three hours later and the chemical marks on DNA packaging have not budged a day later. So the fusion's immediate job is not to assemble machinery or write marks - it is to let machinery that is already there actually move.
The same thing, for a scientist
**Two clocks, measured in the same cells with the same degron, running an order of magnitude apart.** | layer | measurement | when it moves | |---|---|---| | **nascent transcription** | SLAM-seq at 15/30/60/120 min; PRO-seq at 15/30/180 min; pausing index [E1] | **within 15 minutes** | | **the fusion itself** | HA ChIP | **by 3 hours** (2.5-7.1% retained) [E2] | | **Pol II occupancy, genome-wide peak calls** | total 99.6%, pS2 98.3%, pS5 86.8% retained at 3 h [E2] | **not at 3 hours** | | **SEC occupancy, genome-wide** | AFF4 92.7%, CDK9 83.6% retained at 3 h [E2] | **not at 3 hours** | | **active histone marks, genome-wide** | H3K79me2 99.7%, H3K9ac 100.0% at 3 h AND 24 h [E2] | **not by 24 hours** | **The mechanistic reading.** The elongation machinery is **already present and stays present**. What the fusion supplies is not recruitment but **licence to elongate** — and when it is removed, transcription stops while the apparatus remains in place. The authors state the conclusion directly: *"MLL::AF9 promotes transcription by maintaining productive RNA Pol II elongation"*, with the transcriptional changes preceding the chromatin ones [E1]. **Where this analysis adds something rather than restating the paper.** The authors report Pol II, pS2, AFF4 and CDK9 loss **at MLL::AF9 target gene bodies and promoters** by normalised read counts. This analysis shows that **genome-wide, essentially none of that machinery loses a peak call** — 99.6% for total Pol II. Together the two say something sharper than either alone: **the change is confined to the fusion's own loci and is quantitative rather than presence/absence.** The machinery does not leave the genome; it leaves, partially, a small set of genes. **And the therapeutic corollary follows immediately.** If the fusion's proximate function is licensing elongation at a defined gene set, then the quantity that reports on a drug working is **nascent transcription at those genes** — not occupancy of anything, and certainly not a bulk histone mark. This graph's eviction thread is large and its elongation thread is small; this entry argues that ratio is inverted relative to where the causal action is.
Why the chromatin layer is so slow, and what that costs
H3K79me2 has no known demethylase and is diluted principally by replication, so it cannot fall quickly however fast its writer leaves — and DOT1L is the most-displaced protein in the panel (68.7% retained) while its mark is the least-moved (99.7% at both 3 h and 24 h). Writer displacement and mark persistence are separable and this deposit separates them within one experiment.
The cost is practical. A mark whose removal requires cell division cannot report on acute target engagement, and a therapeutic strategy that aims to reset the epigenome is aiming at the layer with the longest latency and the least causal proximity to the transcriptional failure that actually kills the cell.
The limits of the synthesis
The two clocks are measured by different instruments, and this beat's contribution is entirely the genome-wide presence layer. The 15-minute figure is the authors' SLAM-seq and PRO-seq; the 99.6% is peak presence. Constraint U2 binds — a peak call is a threshold, so 'Pol II did not move' means 'Pol II did not lose its calls', and the authors' own read-count measure shows it did lose signal at target genes.
HCB1 is KMT2A::MLLT3, not the index fusion. In KMT2A::AFF1 the 3' partner is itself an elongation-complex scaffold, so if anything the elongation-licensing reading should be stronger there — but that is a prediction, not a result, and this project's rule is that ordering transfers between these lines while magnitude does not.
The authors' pS2 antibody showed cross-reactivity by their own account (their Figure S4G), which weakens the elongation-specific arm of the occupancy data on both sides.
- If the causal action is transcriptional and fast while the measurable chromatin changes are slow and confined, then most of this project's chromatin-occupancy effort is measuring the echo rather than the event. That is uncomfortable for a knowledge graph whose eviction thread is one of its largest.
- It also means a drug could be fully effective transcriptionally while every occupancy and histone-mark assay reads negative, and conversely that a drug producing dramatic occupancy changes may not have stopped transcription. Both failure modes have been observed in this deposit.
- If the fusion licenses elongation at a defined gene set rather than recruiting machinery, what is the licensing step molecularly - P-TEFb activation, release from the 7SK snRNP reservoir, or something at the pause site? This graph holds a ptefb-reservoir folder and this is the question that connects it to the degron data.
- Does the elongation-first ordering hold for KMT2A::AFF1, whose 3' partner is itself an elongation-complex scaffold? The prediction is that it holds more strongly, and no KMT2A::AFF1 degron exists to test it.
- Is there a mark or a protein that DOES lose peak calls genome-wide within 24 h of fusion degradation? Across eight proteins and four active marks, nothing did. A repressive mark was never profiled, and this project has repeatedly found the polycomb side carrying signal the active side does not.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
MLL::AF9 degradation changes nascent transcription within 15 minutes by SLAM-seq and PRO-seq, alters the pausing index, and the authors conclude that MLL::AF9 promotes transcription by maintaining productive RNA Pol II elongation, with chromatin changes following subsequently.
MLL::AF9 degradation induces rapid changes in transcriptional elongation and subsequent loss of an active chromatin landscape. · Olsen SN, Godfrey L, Healy JP, Choi YA, Kai Y, Hatton C, Perner F, Haarer EL, Nabet B, Yuan GC, Armstrong SA · Molecular Cell 2022What this evidence cannot show
Read in full text at PMC9044330. The 15-minute figure comes from SLAM-seq (15/30/60/120 min) and PRO-seq (15/30/180 min) volcano plots at stated thresholds (SLAM-seq log2 FC 1, p 0.05; PRO-seq FC 1.5, p 0.0001), in HCB1 only. The authors note their pSer2 RNA Pol II antibody showed cross-reactivity, which weakens the elongation-specific occupancy arm. HCB1 is KMT2A::MLLT3.
Evidence that argues against this (1)
- A transcriptional change within 15 minutes of degrading a chromatin-bound factor does not by itself establish that elongation licensing is the proximate mechanism rather than, say, rapid loss of a labile co-activator downstream. The pausing-index and metagene analyses are what carry that claim, and those were not independently reproduced here. Olsen SN et al. 2022
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
Genome-wide, RNA Pol II retains 99.6%, pS2 98.3%, pS5 86.8%, AFF4 92.7% and CDK9 83.6% of their strongest peak calls 3 h after fusion degradation, and H3K79me2 and H3K9ac retain 99.7% and 100.0% at both 3 h and 24 h.
GSE173599, the seven protein and four histone-mark DMSO/dTAG-VHL pairs (GSM5271643-GSM5271709) · Olsen SN, Armstrong SA, et al. · NCBI Gene Expression Omnibus 2022What this evidence cannot show
n = 1 per arm, no replicates, no spike-in, no noise floor. Genome-wide peak presence is a threshold measure and is close to blind to a target-gene-restricted quantitative change, which is exactly what the authors report - so the high retention figures corroborate the 'machinery stays on the genome' reading and say nothing about the fusion's own loci. Arms span several sequencing batches and genome builds, so cross-channel ranking is suggestive only.
Evidence that argues against this (1)
- The authors report loss of RNA Pol II and pSer2 throughout the gene body of MLL::AF9 target genes, and specific AFF4/CDK9 loss at target promoters, at the same 3 h timepoint. If read as a locus-level statement the genome-wide retention figures would contradict the paper; read as a genome-scale statement they do not. Olsen SN et al. 2022
Supplies the alternative readout: if the fusion's proximate job is licensing elongation and transcription changes within 15 minutes, nascent transcription at target genes is both faster and closer to the biology than any occupancy measure.
- supports →Plain-English: what transcription elongation is, and why a traffic light explains this leukemia
The degron's sharpest result for the elongation explainer: the machinery is already present and stays present (Pol II 99.6% of peak calls retained at 3 h) while nascent transcription falls within 15 minutes. The fusion licenses elongation rather than recruiting the apparatus.
The chromatin half of the two clocks: no active mark loses a genome-wide peak call by 24 h, against nascent transcription moving at 15 minutes - a latency gap of roughly two orders of magnitude.
- regulates →KMT2A::MLLT3 (MLL-AF9), t(9;11)(p21.3;q23.3)
KMT2A::MLLT3's proximate function, read from its acute removal: maintaining productive RNA Pol II elongation at a defined target-gene set, not assembling the elongation apparatus or building the active chromatin landscape.