P-TEFb phosphorylates AFF4 and rearranges the complex the fusion hijacks
also called CDK9 AFF4 phosphorylation · SEC assembly phospho-control · CDK9 T186 autophosphorylation
The super elongation complex is the machine KMT2A fusions hijack. Its own kinase, CDK9, phosphorylates the complex's scaffold protein, which changes how the parts fit together and where the complex goes on DNA.
The same thing, for a scientist
Three findings put the SEC's assembly under phosphorylation control by its own catalytic subunit. **P-TEFb phosphorylates AFF4 and thereby modulates SEC assembly, AFF1/AFF4 dimerisation and recruitment to a promoter** [E1]. Separately, monomeric dephosphorylated CDK9 released from 7SK snRNP **autophosphorylates T186 on binding Brd4 or SEC**, so the scaffold's engagement is itself the trigger for kinase activation [E2]. And the AFF1 scaffold specifically associates with **HEXIM1-containing inactive CDK9** to suppress a set of paused genes, with the switch to active CDK9 driving differentiation [E3]. **AFF4 carries 53 observed phosphosites and AFF1 22**, none of them site-mapped to a kinase in this literature. For a beat asking which contacts on a KMT2A fusion are phospho-gated, this is the largest unmapped surface in the system: the fusion's principal effector complex is assembled under phospho-control and not one site is assigned.
The three phospho-controls, and what is missing from each
| finding | what is shown | what is missing |
|---|---|---|
| P-TEFb phosphorylates AFF4, modulating SEC assembly, AFF1/AFF4 dimerisation and promoter recruitment 1 | a phospho-dependent assembly step | no residue; the system is HIV transcription, not leukaemia |
| CDK9 T186 autophosphorylation is induced by binding Brd4 or SEC 2 | scaffold binding activates the kinase | whether a KMT2A fusion changes this, in either direction |
| AFF1-SEC associates with HEXIM1-inactive CDK9 to repress paused genes; PKC activates CDK9 3 | an AFF1-specific, not AFF4, activity switch | epidermal progenitors, not blasts |
Every one of these is a phosphorylation-dependent contact within the complex the KMT2A fusion recruits, and none of them has been tested on a fusion.
Where the site map would come from
The raw material exists. AFF4 (Q9UHB7) has 53 observed phosphosites in the PRIDE re-analysis and AFF1 (P51825) has 22, including AFF1 S588, which is the one residue in the whole AFF family with a named binding partner (14-3-3 theta). AFF1's S-P-containing sites are candidate CDK substrates by consensus alone.
What nobody has done is the obvious pairing: take the SEC phospho-assembly result, which has no residue, and the AFF-family phosphosite list, which has no function, and cross them. A CDK9-inhibitor phosphoproteomics experiment in a KMT2A-rearranged line would produce the site map directly, and the graph already holds a CDK9 inhibition dossier that would be the natural home for it.
| kinase | CDK9 / P-TEFb |
|---|---|
| sites | unmapped on AFF4; T186 on CDK9 |
| gated contact | SEC assembly, AFF1/AFF4 dimerisation, promoter recruitment |
| tested on fusion | false |
| fate in fusion | AFF1 phosphosites C-terminal to the junction are retained in KMT2A::AFF1; AFF4 is a recruited partner and is not part of any fusion chain |
substrates
[ "AFF4", "CDK9 T186 (autophosphorylation)" ]
phosphosites available
{
"AFF4_Q9UHB7": 53,
"AFF1_P51825": 22
}- CDK9 inhibition collapses transcription globally; any selectivity argument for KMT2A-rearranged disease has to survive the observation that MYC and MCL1 loss explains much of what CDK9 inhibitors do in every leukaemia.
- If SEC binding activates CDK9 rather than the reverse, then a drug that disrupts the AFF1/AFF4 scaffold would lower kinase activity indirectly, and a drug that inhibits the kinase would leave the scaffold assembled. Two interventions described as hitting the same axis would leave very different complexes behind.
- The AFF1-repressive result means SEC inhibition can induce a differentiation programme in a non-leukaemic tissue within hours. Whatever that predicts for efficacy, it also predicts on-target effects in skin and other progenitor compartments.
- Which AFF4 residues does CDK9 phosphorylate? Not one is named in the literature this beat searched, despite 53 observed sites being available to cross-reference.
- Does a KMT2A fusion alter CDK9 T186 autophosphorylation at its target loci? If scaffold binding activates the kinase, a constitutive scaffold should.
- Is AFF1's repressive, HEXIM1-associated state relevant in blasts, and does the fusion convert it?
- Would CDK9-inhibitor phosphoproteomics in a t(4;11) line produce the AFF-family site map in one experiment? It appears nobody has run it.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
P-TEFb phosphorylates AFF4 and modulates super elongation complex assembly, AFF1/AFF4 dimerisation and recruitment to the promoter; AFF1 and AFF4 are differentially recruited and differ in their association with the PAF complex.
Super elongation complex promotes early HIV transcription and its function is modulated by P-TEFb. · Kuzmina A, et al. · Transcription 2017What this evidence cannot show
Abstract read 2026-09-19; full text not retrieved. The entire system is HIV-1 transcription in cell lines, with Tat present or absent, and no KMT2A fusion anywhere in the study. No phosphorylation site on AFF4 is named. Whether the same phospho-control operates at fusion target genes is untested.
Evidence that argues against this (1)
- SEC composition is not a single entity, and a later study reports that AFF4-ELL2-EAF1 also occurs in an assembly molecularly distinct from canonical P-TEFb-containing SEC. A phospho-assembly result obtained on one configuration does not necessarily describe the assembly a KMT2A fusion recruits. Nandy A et al. 2026
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Cdk9 and cyclin T1 released from 7SK snRNP dissociate into monomers under stress; Brd4 or SEC then recruits the monomers to reassemble core P-TEFb, and binding of monomeric dephosphorylated Cdk9 to either Brd4 or SEC induces autophosphorylation of Cdk9 T186.
Disrupting the Cdk9/Cyclin T1 heterodimer of 7SK snRNP for the Brd4 and AFF1/4 guided reconstitution of active P-TEFb. · Zhou K, et al. · Nucleic Acids Res 2022What this evidence cannot show
Abstract read 2026-09-19. A biochemical and cell-biological study of P-TEFb regulation with no leukaemia model and no KMT2A fusion; the nocodazole-release cell-cycle arm is the closest thing to a physiological context. The direction of causation between scaffold binding and T186 autophosphorylation rests on the authors' interpretation of their reconstitution.
Evidence that argues against this (1)
- If SEC binding is what activates CDK9, then a KMT2A fusion that constitutively recruits SEC would be expected to raise CDK9 activity at its targets - but a phosphoproteomic study of KMT2A-rearranged AML found the discriminating kinase signal in CDK1 activity and DOT1L phosphorylation rather than in a global CDK9 signature, which is not what the simple prediction implies. Casado P et al. 2023
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In epidermal progenitors, SEC suppresses a group of rapid-response genes in a manner dependent on the AFF1 scaffold but not AFF4, with AFF1-SEC associating with HEXIM1-containing inactive CDK9; SEC peptidomimetic inhibitors or PKC signalling activate CDK9 and rapidly induce differentiation transcription factors within hours.
CDK9 activity switch associated with AFF1 and HEXIM1 controls differentiation initiation from epidermal progenitors. · Lloyd SM, et al. · Nat Commun 2022What this evidence cannot show
Abstract read 2026-09-19. Human keratinocytes, not haematopoietic cells, and the readout is differentiation of epidermal progenitors. The AFF1-versus-AFF4 distinction is the most transferable part; the specific gene programme is not.
Evidence that argues against this (1)
- That SEC can be repressive through AFF1 and inactive CDK9 runs against this graph's dominant account, in which the KMT2A fusion's recruitment of SEC is an activating hijack. If the AFF1 scaffold's default state is repressive, then what the fusion does to it is a switch rather than a recruitment, and the two models predict different consequences for CDK9 inhibition. Yang Z et al. 2024
where every CDK9 claim currently lives
phospho-S395 AF9 loses CDK9/SEC binding; CDK9 is the SEC kinase (CDK9 lives in protein-ptefb-cdk9-ccnt1)
P-TEFb phosphorylates AFF4, modulating SEC assembly and AFF1/AFF4 dimerisation
CDK9 T186 autophosphorylation is induced by SEC or BRD4 binding
the phospho-assembly layer a CDK9 inhibitor would also hit
AFF1 S588 is a proline-directed site in a P-TEFb-containing complex; CDK9 is the untested candidate kinase
- regulates →Step 3 — The hijack: the fusion drags the super elongation complex to paused RNA polymerase II
phospho-control of the complex the fusion hijacks