Oncofusion

start / graph / concept-transcription-elongation

Explainerestablished confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.Q3-mechanism-core

Plain-English: what transcription elongation is, and why a traffic light explains this leukemia

also called promoter-proximal pausing · P-TEFb · elongation · what is RNA polymerase pausing

To use a gene, the cell makes an RNA copy of it with a machine called RNA polymerase. At most genes that machine starts, then stops about fifty letters in and waits for a green light — and the KMT2A fusion protein jams the green light permanently on at the wrong genes.

The same thing, for a scientist

At most metazoan genes RNA polymerase II initiates, transcribes 20-60 nucleotides and then enters a stable promoter-proximal paused state held by NELF and DSIF. The rate-limiting regulated step is release into productive elongation, executed by P-TEFb (CDK9/cyclin T1), which phosphorylates the Pol II CTD at Ser2 and phosphorylates NELF and DSIF. P-TEFb is delivered to genes by the super elongation complex (SEC), scaffolded by AFF1 or AFF4. KMT2A fusions retain KMT2A's chromatin-targeting module and acquire a constitutive connection to SEC, converting regulated pausing-release into permanent release at the fusion's target genes. This is why the disease is best measured by nascent-transcription assays rather than steady-state RNA.

Go deeper
The traffic-light picture, and why the pause exists at all2 deeper

A useful mental model: a gene is a road, RNA polymerase is a car, and there is a traffic light about fifty letters past the start. Most genes in a human cell already have a car sitting at that light with the engine running. Nothing is missing except permission to go.

Why build it that way? Because it is fast and reversible. A cell that needs to switch a gene on within seconds cannot wait to recruit the whole machine from scratch; it keeps the machine loaded and controls only the green light. Developmental genes — the ones a stem cell must be able to turn on and off decisively — are the ones most heavily regulated this way.

The green light is a kinase. P-TEFb, made of CDK9 and cyclin T1, phosphorylates the polymerase's tail and the two proteins holding it back (NELF and DSIF). Once phosphorylated, the polymerase leaves and transcribes the whole gene.

The green light is delivered by a courier. P-TEFb is brought to specific genes inside the super elongation complex, and the scaffold of that complex is AFF1 or AFF4 — the same AFF1 that is fused to KMT2A in t(4;11) infant leukemia 1.

What the fusion does, in one sentence

The fusion bolts the courier to the address label. KMT2A's front half knows how to find the developmental genes; AFF1 is part of the green-light delivery service. Join them and those genes get a permanent green light 12.

The experimental proof that this is the proximate function, not a side effect: when a degradable version of a KMT2A fusion protein is destroyed, transcriptional elongation at core target genes changes within 15 minutes, and the chromatin changes happen afterwards 3. Elongation first, chromatin second.

And there is a striking observation in the older literature that makes the same point from the phenotype side: with the fusion present, the polymerase stays in an actively elongating state and inhibitory marks never accumulate, so the HOX genes stay open and productive even when the cell is being pushed to differentiate 2. The cell is being told to grow up; the green light overrides it.

Why this changes which experiment you run

If the fusion's function is pausing-release, then standard RNA-seq is the wrong assay. Steady-state mRNA reflects transcription and RNA stability and hours of accumulated history. Pausing-release changes in minutes.

The right assays measure nascent transcription: PRO-seq, TT-seq, or 4sU labelling, reading out the pausing index (the ratio of polymerase at the promoter to polymerase in the gene body) at fusion target genes. A drug that works should shift that ratio back toward paused. A drug that merely kills cells will not.

This is a concrete recommendation for the lab: score candidate interventions on pausing index at fusion targets, not on target-gene mRNA and not on viability alone.

Facts and figures
audiencelayman entry point
key terms
{
  "RNA polymerase II": "the machine that copies a gene into RNA",
  "promoter-proximal pausing": "the machine stops 20-60 letters in and waits",
  "NELF and DSIF": "the two proteins holding the paused machine back",
  "P-TEFb (CDK9 + cyclin T1)": "the kinase that gives the green light",
  "SEC / super elongation complex": "the courier that delivers P-TEFb, scaffolded by AFF1 or AFF4",
  "pausing index": "polymerase at the promoter divided by polymerase in the gene body — the number to measure"
}
recommended assays
[
  "PRO-seq",
  "TT-seq",
  "4sU metabolic labelling"
]
links to
[
  "mechanism-sec-elongation-hijack",
  "mechanism-fusion-dependency-reversibility-kinetics"
]
How this could bite us back
  • Pausing release depends on CDK9, which is essential in every cell; the mechanism is shared with all of normal transcription, so it is explanatory rather than targetable.
  • A drug that reduces elongation without restoring the paused state may leave a primed locus that restarts when drug pressure lifts.
Still open
  1. Has anyone measured the pausing index at KMT2A::AFF1 target genes in primary infant ALL blasts, before and after menin inhibition? If not, the field's best mechanistic readout has never been applied to the disease's actual cells.
  2. Does a clinically effective menin inhibitor restore pausing, or does it simply reduce the number of engaged polymerases? These predict different combination partners.
  3. Why are only some genes pause-regulated? If the fusion can only hijack pause-regulated genes, the pausing landscape of the cell of origin would predict the fusion's target set — a testable link to the HOXA-high/low split.
Every source behind this entry (3)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Multiple commonly occurring KMT2A chimeras all co-purify the super elongation complex containing ELL, P-TEFb and AFF4; AFF4 is required for SEC stability and for proper transcription by poised RNA polymerase II, and AFF4 knockdown reduces KMT2A-chimera target gene expression.

    AFF4, a component of the ELL/P-TEFb elongation complex and a shared subunit of MLL chimeras, can link transcription elongation to leukemia · Lin C, Smith ER, Takahashi H, et al. (Shilatifard A, senior author) · Molecular Cell 37(3):429-437 2010
    PMID 20159561DOI
    What this evidence cannot show

    Biochemical purification from overexpressing cell lines plus RNAi; the pausing framework described in this explainer is standard transcription biology rather than a finding of this paper.

    Evidence that argues against this (1)
    • The elongation account also holds that the SEC is a defined stoichiometric complex. Guo et al. report instead that SEC components compartmentalise P-TEFb by liquid-liquid phase separation, that ENL or its IDR alone initiates SEC droplet formation, that AFF4 fluidises those droplets, and that fusing ENL to MLL boosts SEC phase separation. Whether the courier is a complex or a condensate is an open dispute inside the elongation model, not a settled part of it. Guo C et al. 2020
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    KMT2A fusions capture an elongation complex integrating H3K79 methylation, Pol II CTD phosphorylation and the fusion partner; the fusion keeps Pol II in an actively elongating state, prevents accumulation of inhibitory histone methylation, and keeps Hox loci open and productive even under forced differentiation, with KMT2A-transformed cells particularly sensitive to inhibition of Pol II phosphorylation.

    Misguided transcriptional elongation causes mixed lineage leukemia · Mueller D, Garcia-Cuellar MP, Bach C, Buhl S, Maethner E, Slany RK · PLoS Biology 7(11):e1000249 2009
    PMID 19956800DOI
    What this evidence cannot show

    Engineered conditional fusion in a myeloid cell system; the Pol II-phosphorylation inhibitor used is flavopiridol-class and not CDK9-selective in cells.

    Evidence that argues against this (1)
    • The therapeutic prediction of an elongation-centric model has underperformed. Pinometostat, targeting the elongation-linked mark H3K79me2, reduced H3K79 methylation as intended in 51 adults with acute leukaemia and produced two complete remissions, both in t(11;19); efficacy as a single agent was described by the authors as modest. Correctly identifying the elongation step has not, so far, produced correspondingly large clinical effects. Stein EM et al. 2018
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Degradation of degron-tagged MLL::AF9 changes transcriptional elongation at a core subset of target genes within 15 minutes, with loss of the active chromatin landscape occurring subsequently — establishing elongation control as the fusion's proximate function.

    MLL::AF9 degradation induces rapid changes in transcriptional elongation and subsequent loss of an active chromatin landscape · Olsen SN, Godfrey L, Healy JP, et al. (Armstrong SA, senior author) · Molecular Cell 82(6):1140-1155.e11 2022
    PMID 35245435DOI
    What this evidence cannot show

    MLL::AF9 in cell lines; no equivalent KMT2A::AFF1 degron exists, so the 15-minute figure is transferred across fusion partners.

    Evidence that argues against this (1)
    • The 15-minute elongation-first kinetics were measured on MLL::AF9 in a myeloid system, and the downstream biology is lineage-dependent: Adriaanse et al. found KMT2A-r ALL dies rapidly under menin inhibition whereas KMT2A-r AML differentiates slowly. Transferring a degron time-course from an AML model to KMT2A::AFF1 ALL is an assumption that the entry's own caveat already flags, and the ALL/AML response difference gives a concrete reason to doubt it. Adriaanse FRS et al. 2024
Connected to