Lost from the KMT2A half: the MOF/KAT8 contact - required in the cell, but not shown to matter to the fusion
also called MLL1 MOF interaction · KAT8 H4K16ac · MLL C-terminal MOF site
KMT2A recruits a second enzyme that adds a different chemical mark, and the fusion cannot recruit it. That second enzyme really is needed for this leukemia - but when it was finally tested inside a fusion-driven leukemia, it turned out not to be running the fusion's own gene programme, and to be just as necessary in leukemias with no KMT2A break at all.
The same thing, for a scientist
MLL1 forms a stable complex with the H4K16 acetyltransferase MOF (KAT8); the interaction sites map to the **MLL1 C-terminal region** and the MOF zinc finger domain [E1]. Deleting *Mll1* selectively depletes H4K16ac at MLL1 target genes with reduced transcription, and SIRT1 inhibition prevents that loss - leading Mishra et al. to conclude that recruited MOF activity, not intrinsic MLL1 HMT activity, is central at these genes [E2]. Since the MOF contact is C-terminal, no der(11) fusion can make it. **Corrected cycle 31.** This entry previously called the MOF contact the functionally most important thing the KMT2A half loses, and said it had been assessed only on the wild-type protein. **Both statements are withdrawn.** Valerio et al. tested Mof directly in MLL-AF9: it is required - colony formation abolished, survival extended from 22-24 to 31 days, HAT activity indispensable - but **MOF loss does not clearly downregulate Meis1 or the Hoxa cluster**, the mechanism is global H4K16ac loss and DNA-damage-repair failure, and Mof is equally required for NUP98-fusion AML with MG149 killing three non-KMT2A-rearranged lines [E3]. The correct reading is therefore that **MOF is a dependency of the leukemic cell, probably a general one, and the fusion's inability to contact MOF has still never been shown to cost it anything.** That is the same class as the WIN/WRAD interface, not a special case.
CORRECTION (cycle 31): the 'best-evidenced loss' ranking is withdrawn
This entry ranked the MOF contact first among all KMT2A-side losses and described it as 'the output that actually mattered'. A direct test in a fusion-driven leukemia, published in 2017 and not cited here before, does not support that ranking 3.
What the direct test shows in MOF's favour. Mof came out of an unbiased chromatin-regulator RNAi screen in MLL-AF9 cells as one of only two genes with two independently scoring hairpins (the other was the already-known target). Conditional deletion abolishes colony formation, lowers tumour burden and extends median survival from 22-24 to 31 days. Rescue requires the HAT domain: a G327E point mutant and a CoA-site deletion both fail. So MOF is real, enzymatic and required, and this entry's previous statement that the module had been 'assessed only on the wild-type protein' was false.
What it shows against the ranking.
- The fusion's programme does not depend on MOF. The authors state that MOF loss does not lead to clear downregulation of MLL-AF9 target genes such as Meis1 and the Hoxa cluster. If the KMT2A half's loss of the MOF contact mattered by depriving the fusion's targets of H4K16ac, that is precisely the result that should not have appeared.
- It is not a KMT2A-fusion dependency. Mof is equally required for NUP98-fusion-driven AML, and MG149 kills three non-KMT2A-rearranged leukemia lines. The mechanism the authors reach is global H4K16ac depletion, downregulated DNA-damage-repair genes and increased gammaH2AX - general chromatin and genome-integrity biology.
The corrected claim. MOF is a dependency of the leukemic cell. Whether the fusion's missing MOF contact costs it anything is still untested, because no experiment has restored a MOF contact to a fusion. The entry's previous title clause - 'the output that actually mattered' - is withdrawn, and the loss class is now dependency-of-the-cell-not-of-the-fusion, the same class as the WIN/WRAD interface.
The authors themselves point at the wild-type allele: their suggestion is that MLL1 functions as a scaffold recruiting MOF to the targets of the oncogenic fusion. That is the beat's 'supplied-in-trans' column, reached independently.
Why this was called the most consequential loss, and why that ranking no longer stands1 deeper
This ranking is superseded - see the correction above. It is kept because the reasoning is instructive about how the error happened.
The original argument ranked the lost KMT2A modules by how much evidence there was that each carries functional load at KMT2A target genes, and put the MOF contact first on the strength of Mll1-deletion experiments in normal haematopoiesis 2. The inference was: deleting Mll1 costs H4K16ac at its targets, therefore a protein that cannot recruit MOF has lost something central.
The missing step is the one this beat exists to insist on. Mll1 deletion removes the enzyme's scaffold from the cell. It does not model a fusion that cannot bind MOF while a wild-type allele still can. When the experiment that actually addresses the leukemia was done, MOF turned out to be required for the leukemia but not for the fusion's target programme, and required just as much in a leukemia with no KMT2A break 3.
The corrected ordering of KMT2A-side losses by evidence that the fusion's loss costs it something is, after the cycle-31 pass: PHD3 (one unreplicated in-cis add-back); then the degradation determinants - the taspase1 sites, FYRN, PHD2 - which have measured consequences in the right direction but only on wild-type protein; then everything else, including MOF, at zero.
The coordinate this entry deliberately does not give
Dou et al. map the MOF interaction to 'MLL1 C-terminal' without a residue range in the abstract, and I did not read the full text 1. This entry therefore records no start and end for the MOF contact rather than inventing one.
That is a deliberate choice under this project's join rules: a domain entry with a fabricated or null coordinate is either wrong or invisible, and 'C-terminal, position unresolved' is the honest state of the literature as far as I checked it. A future agent reading the Dou 2005 full text should add the range.
The strategic consequence: a lost module can still be a target
As with the WIN/WRAD interface, the leukemic cell still owns a wild-type KMT2A allele that can recruit MOF. So the question 'is the MOF axis targetable in KMT2A-rearranged leukemia?' is not answered by 'the fusion lost it'.
Two levers follow from 2 and neither has been pursued in KMT2A-rearranged disease specifically:
- KAT8/MOF itself. In DepMap, KAT8 is depleted in ~99.5% of lines - pan-essential, so no window.
- SIRT1. Mishra et al. show SIRT1 inhibition prevents the loss of H4K16ac at MLL1 targets, i.e. SIRT1 opposes the MLL1-MOF axis. That makes SIRT1 a lever on the wild-type allele's output, in the opposite direction, and SIRT1 modulators exist. Whether raising H4K16ac at MLL1 targets would help or harm in a leukemia where those same targets are the oncogenic programme is genuinely unclear - it could as easily feed the disease.
Update, cycle 31. The 'no window' reading of KAT8's DepMap pan-essentiality is partly contradicted by the in-vivo data: conditional Mof deletion in established MLL-AF9 leukemia extended survival rather than killing the animals, and MG149 has a measurable in-vitro window 3. But the same data close the selectivity argument: MG149's IC50s in MLL-rearranged and non-MLL-rearranged lines are in the same tens-of-micromolar range, and Mof is required for NUP98-fusion AML too. A MOF inhibitor, if one were made properly, would be a general leukemia agent, not a KMT2A-directed one - which is a different and much harder development proposition. The SIRT1 lever 2 is untouched by any of this and remains the more interesting and more uncertain of the two.
| module | MOF/KAT8 recruitment contact |
|---|---|
| uniprot | Q03164 |
| coordinates | not assigned - primary source states 'MLL1 C-terminal' without a residue range in the abstract; deliberately left unspecified rather than guessed |
| fate in fusion | lost |
| partner lost | KAT8 / MOF |
| downstream mark | H4K16 acetylation at KMT2A target genes |
| opposing enzyme | SIRT1 |
| demonstrated functional consequence of loss | none. MOF is required for MLL-AF9 leukemia maintenance (PMID 28202522) but not via Hoxa/Meis1 and equally in NUP98-fusion AML, so no experiment yet shows that the FUSION's inability to contact MOF costs it anything. |
| loss class | dependency-of-the-cell-not-of-the-fusion (withdrawn: 'lost, and the best-evidenced functional loss on the KMT2A side') |
| tested in fusion context | true |
| tested in fusion context result | Mof is required for MLL-AF9 leukemia (median survival 22-24 to 31 days; HAT activity indispensable), but the mechanism is global H4K16ac loss and DNA-damage-repair failure, not the fusion's target programme (PMID 28202522, full text) |
- KAT8/MOF is pan-essential in DepMap (~99.5% of lines depleted), so direct MOF inhibition has no therapeutic window regardless of its mechanistic centrality.
- SIRT1 modulation acts on a deacetylase with hundreds of substrates and both tumour-suppressive and tumour-promoting reports; using it to tune one histone mark at one gene set is not a realistic selectivity proposition.
- If the leukemic cell depends on the wild-type KMT2A allele to supply MOF recruitment at shared targets, then therapies that damage wild-type KMT2A would be effective for a reason that also predicts developmental toxicity in infants.
- This entry is a worked example of the error it was meant to catch. It ranked a loss as most important on evidence about the enzyme in normal cells, and the ranking survived a cycle because nobody checked whether anyone had tested the enzyme in a fusion-driven leukemia. Somebody had, in 2017.
- MOF inhibition looks like a pan-leukemia agent, not a KMT2A-directed one. Developing it on a KMT2A-rearranged rationale would mean running the selectivity argument backwards from a compound that shows none.
- Does the fusion recruit any acetyltransferase in place of MOF, via the partner? Still open, and now the only question in this entry that could restore the MOF contact to significance.
- Does SIRT1 inhibition help or harm in KMT2A-rearranged leukemia? It raises H4K16ac at KMT2A target genes - which are the oncogenic programme - so the sign is genuinely unknown and this remains a cheap in-vitro question, untouched by the 2017 MOF work.
- If MOF's requirement in leukemia runs through DNA-damage repair and global H4K16ac rather than through HOX genes, is MOF inhibition synergistic with DNA-damaging or PARP-directed agents in KMT2A-rearranged disease? Nobody has asked, and the mechanism practically proposes the combination.
- Where exactly does MOF bind MLL1? Still unresolved from the 2005 abstract, and now a lower-priority question than it looked.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A stable complex containing MLL1 and MOF was immunoaffinity purified and confirmed by reciprocal immunoprecipitation, cosedimentation and cotransfection; the interaction sites were mapped to MLL1 C-terminal and MOF zinc finger domains; and both the MLL1-mediated H3K4 methyltransferase activity and the MOF-mediated H4K16 acetyltransferase activity are required for optimal transcriptional activation on a chromatin template in vitro and at the endogenous MLL1 target gene HOXA9 in vivo.
Physical association and coordinate function of the H3 K4 methyltransferase MLL1 and the H4 K16 acetyltransferase MOF · Dou Y, Milne TA, Tackett AJ, Smith ER, Fukuda A, Wysocka J, Allis CD, Chait BT, Hess JL, Roeder RG · Cell 121(6):873-885 2005What this evidence cannot show
ABSTRACT ONLY. The abstract gives no residue range for the MLL1 C-terminal MOF-binding region, so this entry assigns no coordinates. HeLa-cell biochemistry and in-vitro chromatin templates; 'in vivo' means in cultured cells. Note the internal tension: this paper reports MLL1's own HMT activity as required for optimal activation, which E2 later contradicts in mice.
Evidence that argues against this (1)
- A 2005 immunoaffinity purification with reciprocal IP and cotransfection establishes association under those conditions. Stoichiometry, occupancy and whether the contact is constitutive or regulated are not addressed, and a stable complex in a preparation is compatible with a small fraction of either protein being engaged in a cell. Cell 2005
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Upon Mll1 deletion, histone H4 lysine 16 acetylation is selectively depleted at MLL1 target genes in conjunction with reduced transcription; inhibition of SIRT1 is sufficient to prevent the loss of H4K16 acetylation and the reduction in MLL1 target gene expression; and recruited MOF activity, not the intrinsic histone methyltransferase activity of MLL1, is central to maintaining HSC target genes - the SET domain itself being dispensable for haematopoiesis and for MLL-AF9-driven leukemogenesis.
The histone methyltransferase activity of MLL1 is dispensable for hematopoiesis and leukemogenesis · Mishra BP, Zaffuto KM, Artinger EL, Org T, Mikkola HKA, Cheng C, Djabali M, Ernst P · Cell Reports 7(4):1239-1247 2014What this evidence cannot show
ABSTRACT ONLY. Mouse genetics in normal HSCs and an MLL-AF9 model; the H4K16ac and SIRT1 results concern wild-type MLL1 target genes, not fusion target genes, and the paper does not claim that the fusion's inability to recruit MOF is consequential. The inference that the MOF contact is the highest-evidence loss on the KMT2A side is this Queen's ranking, not a claim in either paper.
Evidence that argues against this (1)
- Tested directly in a fusion-driven leukemia, MOF's requirement does not run through the fusion's target programme: Mof loss in MLL-AF9 cells does not clearly downregulate Meis1 or the Hoxa cluster, and the mechanism the authors reach is global H4K16ac loss and DNA-damage-repair failure. Mof is also required for NUP98-fusion AML. So the MLL1-MOF axis cannot be read as the fusion's lost activating output. Valerio DG et al. 2017
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A chromatin-regulator RNAi screen in murine MLL-AF9 leukemia cells identified Kat8/Mof as one of only two genes with two independently depleted shRNAs; conditional Mof deletion abolishes colony formation by MLL-AF9-transformed cells, lowers tumour burden and extends median survival from 22-24 days to 31 days in vivo, and rescue requires an intact MOF HAT domain (both a G327E point mutant and a CoA-site deletion fail). Mechanistically MOF loss depletes GLOBAL H4K16 acetylation and downregulates DNA-damage-repair genes with increased gammaH2AX foci - and the authors state explicitly that MOF loss does NOT lead to clear downregulation of MLL-AF9 target genes such as Meis1 and the Hoxa cluster. Mof is equally required for NUP98-fusion-driven AML, and the MYST HAT inhibitor MG149 kills three non-KMT2A-rearranged leukemia lines as well as MLL-rearranged lines.
Histone Acetyltransferase Activity of MOF Is Required for MLL-AF9 Leukemogenesis · Valerio DG, Xu H, Chen CW, Hoshii T, Eisold ME, Delaney C, Cusan M, Deshpande AJ, Huang CH, Lujambio A, Zheng YG, Zuber J, Pandita TK, Lowe SW, Armstrong SA · Cancer Research 77(7):1753-1762 2017What this evidence cannot show
FULL TEXT read via PMC5501293. This measures MOF the enzyme inside a fusion-driven leukemia; no arm restores a MOF contact to a fusion, so it still does not test what the fusion's missing contact costs. MG149 is a weak, non-selective first-generation tool (IC50 19.6-50.9 microM across the lines tested). The authors' own reading is that MLL1 may act merely as a scaffold recruiting MOF to the fusion's targets - i.e. the wild-type allele in trans.
Evidence that argues against this (1)
- Two independently depleted shRNAs in a screen is a low bar for a hit, and the in vivo magnitude is modest - median survival 22-24 days to 31. The entry should carry the animal number rather than the screen, since a chromatin-regulator RNAi screen nominating Kat8 is a starting point and not the demonstration. Cell 2005
Prerequisite and potential cheap kill: whether the KMT2A-MOF contact survives the KMT2A::AFF1 breakpoint is examined in this subtraction entry, and if the contact maps to a discarded region the hypothesis dies without any bench work
- ← related toMSL2 — the second H2B ubiquitin ligase, sitting inside the MOF complex the KMT2A fragment contacts
if the retained KMT2A fragment contacts MOF it is in contact with an H2B ubiquitin ligase
THE CONTACT THAT WRITES IT IS DISCARDED BY THE FUSION, and already tested in the fusion context: Mof is required for MLL-AF9 leukaemia but through global H4K16ac loss and DNA-damage repair, not the fusion's target programme.
the MLL1-MOF contact that makes the requirement look fusion-directed
[contradicts] MOF is a dependency of the leukemic cell, not a demonstrated cost of the fusion's missing contact - it does not run through Hoxa/Meis1
MLL1 C-terminal region binds the MOF zinc finger; contact absent from every fusion
- part of →KMT2A / MLL1 — the gene that breaks
C-terminal MOF-recruitment contact, coordinates unresolved
explains why SET loss is tolerated: the load at KMT2A targets is carried by recruited MOF, not by SET catalysis
SIRT1 opposes the MLL1-MOF axis; its inhibition restores H4K16ac at KMT2A targets
row: MOF contact, the one module with no resolvable coordinates
the best-evidenced KMT2A-side loss is an activating output, which is what breaks the strong restraint reading