The AF9 acetylation arm: PCAF at K339, HDAC5 gated by CaMKII - a kinase-controlled switch the fusion throws away
also called AF9 K339 acetylation · HDAC5 S259 S498 · CaMKII AF9 · PCAF AF9 oligomerisation
A second control system on AF9 works by acetylation rather than phosphorylation, but a kinase still runs it: a kinase kicks the eraser enzyme out of the nucleus so the mark can build up. The spot it acts on sits in the part of AF9 that the leukaemia fusion discards, so this whole system is missing from the cancer protein.
The same thing, for a scientist
PCAF acetylates AF9 at K339. Acetyl-K339 monomerises AF9, reducing Poly-Ser-dependent oligomerisation and TFIID interaction, and a K339Q acetyl-mimic fails to restore target-gene expression, factor recruitment or paused-Pol II release in AF9-knockdown cells. The mark is erased by HDAC5, and the kinase enters here: **CaMKII-mediated phosphorylation of HDAC5 at S259 and S498 drives its nuclear export**, removing the eraser and allowing K339 acetylation to accumulate after ionising radiation; an HDAC5 S259A/S498A phospho-defective mutant blocks the IR-dependent change in AF9's TFIID and SEC interactions. **K339 lies at residue 339, N-terminal of the 376 breakpoint boundary, so this entire arm is lost in about 92% of KMT2A::MLLT3 patients** - as is the Poly-Ser domain it acts on. The entry exists to mark the contrast with S395: two kinase-gated switches on the same protein, one discarded and one kept.
The chain, and where the kinase sits
- PCAF acetylates AF9 at K339; a K339R mutant loses most PCAF-mediated acetylation 1.
- Acetyl-K339 reduces AF9 oligomerisation and TFIID binding; the K339Q mimic phenocopies this and fails to rescue AF9-knockdown cells 1.
- HDAC5 is the deacetylase for AF9 K339; HDAC5 knockdown raises endogenous K339 acetylation and reduces AF9 oligomerisation 2.
- CaMKII phosphorylates HDAC5 at S259 and S498, the residues that control its nuclear-cytoplasmic shuttling; the phospho-defective HDAC5(S259A,S498A) mutant blocks the IR-induced changes in AF9's TFIID and SEC interactions 2.
So the kinase does not touch AF9 at all. It controls the localisation of the eraser. That is a different topology from every other entry in this beat and from the CK2/LEDGF switch the previous beat found, and it is worth naming as a class: a kinase-gated contact where the kinase acts two steps away.
Lost, and what that costs the fusion
K339 is 37 residues N-terminal of the 376 boundary that about 92% of MLL-AF9 breakpoints fall at or after 3. The Poly-Ser domain the acetylation acts on (149-196) is lost with it.
That means KMT2A::MLLT3 has no CaMKII/HDAC5/PCAF input at all. A CaMKII inhibitor, an HDAC5-sparing or HDAC5-targeting HDAC inhibitor, or a PCAF inhibitor would act on wild-type AF9 in the leukaemic cell and on nothing in the fusion.
This is directly relevant to HDAC inhibitors, which are used and trialled in KMT2A-rearranged disease. Whatever they do there, they are not acting through this axis on the oncoprotein.
| kinase | CaMKII |
|---|---|
| kinase acts on | HDAC5 S259/S498, not on AF9 |
| topology | kinase-gated contact at two removes: kinase -> eraser localisation -> substrate acetylation -> protein-protein contact |
| writer | PCAF |
| eraser | HDAC5 (HDAC4 also active in vitro) |
| substrate | gene-mllt3 |
| uniprot | P42568 |
| site | K339 (acetylation, not phosphorylation) |
| gated contact | AF9 oligomerisation and TFIID binding |
| mapped in | wild-type AF9, HEK293T |
| mapped in fusion | false |
| retained in fusion | false |
| retained basis | K339 is N-terminal of residue 376; about 92% of MLL-AF9 breakpoints fall at or after 376 |
| system | HEK293T, ionising radiation 10 Gy |
| gene symbol | MLLT3 |
- HDAC inhibitors are in clinical use and in trials for KMT2A-rearranged disease. On this mechanism they would raise K339 acetylation on wild-type AF9 and monomerise it - an on-target effect on a non-target protein, in a cell where the oncoprotein does not carry the residue at all.
- A CaMKII inhibitor would act at two removes from AF9 through a shuttling mechanism shared by many nuclear proteins, so any effect attributed to the AF9 axis would be very hard to attribute at all.
- This entry's main practical value is negative: it removes three candidate handles from the KMT2A::MLLT3 target list. A programme that did not check retention could spend a year on a residue the oncoprotein does not contain.
- Is there an equivalent acetylation site on the retained portion of AF9? The graph has no acetylation map of the AHD.
- ENL retains everything AF9 loses. Does ENL have a K339-equivalent residue, and is KMT2A::MLLT1 therefore subject to the PCAF/HDAC5 arm that KMT2A::MLLT3 escapes?
- HDAC inhibitors are trialled in KMT2A-rearranged leukaemia. Do they act on wild-type AF9 oligomerisation in those cells, and does that matter when the fusion cannot oligomerise through AF9 anyway?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
PCAF acetylates AF9 at K339; a K339R mutant loses most PCAF-mediated acetylation, acetyl-K339 reduces AF9 oligomerisation and TFIID interaction, and a K339Q acetyl-mimic fails to restore target-gene expression, factor recruitment and paused-Pol II release when re-expressed in AF9-knockdown cells.
Post-translational modification-dependent oligomerization switch in regulation of global transcription and DNA damage repair during genotoxic stress. · Talukdar P, et al. · Nat Commun 2024What this evidence cannot show
Full text read via Europe PMC fullTextXML (PMC11096357). HEK293T throughout; the K339Ac-specific antibody validation panel is stated as n=1 replicate and several downstream panels are n=1 or n=2. Mapping used a multi-lysine mutant series in which K339R only showed a dramatic effect in combination before being confirmed alone.
Evidence that argues against this (1)
- The authors themselves record that the K339R mutant showed ENHANCED rather than merely preserved TFIID interaction after IR, and note that AF9 is retained on chromatin after IR by a mechanism the acetylation model does not explain - so the acetylation arm is not a complete account even in the system where it was defined. Kabra A et al. 2022
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
HDAC5 deacetylates AF9 at K339 and CaMKII-mediated phosphorylation of HDAC5 at S259 and S498, which regulates its nuclear-cytoplasmic shuttling, is required for genotoxic-stress-dependent modulation of AF9 acetylation; an HDAC5(S259A,S498A) phospho-defective mutant blocks the IR-induced change in AF9 interaction with TFIID and SEC components.
Post-translational modification-dependent oligomerization switch in regulation of global transcription and DNA damage repair during genotoxic stress. · Talukdar P, et al. · Nat Commun 2024What this evidence cannot show
Same paper and same system. This arm is the most thinly replicated part of it: the in vitro deacetylation by purified HDAC5, the PCAF/HDAC5 dynamic interaction panel, the HDAC5 knockdown interaction time-course and the nascent-transcription readout are each stated as n=1 replicate, and the authors state of one nascent-RNA experiment that it was done once. HDAC4 also deacetylates AF9 in the same assay, so HDAC5 is not shown to be the only eraser.
Evidence that argues against this (1)
- HDAC5 S259/S498 phosphorylation and 14-3-3-dependent nuclear export is a general, long-established mechanism acting on many nuclear substrates, so demonstrating that blocking it changes AF9 acetylation does not establish an AF9-specific regulatory relationship; the same manipulation changes the acetylation state of much of the nucleus. Etchegaray JP et al. 2019
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
K339 lies N-terminal of AF9 residue 376 and is therefore absent from the fusion protein in approximately 92% of KMT2A::MLLT3 patients, as is the Poly-Ser domain at 149-196 whose oligomerisation the acetylation controls.
The YEATS domain epigenetic reader proteins ENL and AF9 and their therapeutic value in leukemia. · Hu H, et al. · Exp Hematol 2023What this evidence cannot show
The breakpoint figure is a pooled distribution from a review, not a residue-resolved primary series. A patient breaking N-terminal of 339 would retain K339.
Evidence that argues against this (1)
- Retention arguments of this shape have been wrong in this graph before in the opposite direction: the dispatch premise that MLLT1 discards its YEATS domain was refuted by the actual breakpoint distribution. A pooled percentage is a weak basis for asserting what any individual fusion protein contains. Hu H et al. 2023
HDAC5 is the eraser named in that mechanism's own id
PCAF is the writer named in that mechanism's own id
PCAF/HDAC5 acetylation switch at K339
the lost arm in the four-element map
- related to →KMT2A::MLLT3 (MLL-AF9), t(9;11)(p21.3;q23.3)
K339 and Poly-Ser both lost from the fusion
relation 5, the only lost one