KMT2A-CLOCK: one paper, six IntAct records, and the bait/prey grade splits it down the middle - CLOCK falls in the discard bin and SIRT1 does not, from the same experiments
also called MLL1-CLOCK · KMT2A-CLOCK · circadian H3K4me3 · MLL1 K1130 K1133 acetylation
A 2015 paper showed that the body clock controls the MLL1 enzyme by switching an acetyl group on and off, which makes the activating H3K4 mark rise and fall through the day. Our filing system kept one partner from that paper and threw the other away, purely because of which protein the antibody was pointed at.
The same thing, for a scientist
**Verdict: ADMITTED as a lead, and filed mainly as a worked demonstration that the bait/prey grade is a property of assay direction, not of evidence.** IntAct holds **six** records from PMID 25751424 involving KMT2A [E1]. Four are anti-tag co-IP with **KMT2A as bait and SIRT1 as prey** (miscore 0.51, human, plus one mouse Sirt1 row at 0.40); one is a two-hybrid with KMT2A prey and SIRT1 bait; and one is a two-hybrid with **CLOCK as bait and KMT2A as prey** (miscore 0.37). **`SIRT1` is therefore graded BAIT-SUPPORTED and `CLOCK` PREY-ONLY, from the same paper, the same authors and the same figure set.** The paper is Aguilar-Arnal, Katada, Orozco-Solis and Sassone-Corsi, *Nat Struct Mol Biol* 2015: *NAD(+)-SIRT1 control of H3K4 trimethylation through circadian deacetylation of MLL1* [E2]. Its abstract states that MLL1 coordinates recruitment of CLOCK-BMAL1 activator complexes to chromatin; that mammalian MLL1 is acetylated at two conserved residues **K1130 and K1133**; that the acetylation is cyclic and controlled by the clock and by SIRT1; that it affects MLL1 methyltransferase activity; and that H3K4 methylation at clock-controlled-gene promoters responds to pharmacological or genetic SIRT1 inactivation [E2]. **`confidence_overall` rates the KMT2A-CLOCK contact, which is one two-hybrid row. The surrounding biology is much stronger than the row.**
The six records, and the split
| IntAct AC | partners | method | our role | host | miscore | grade this project gives the partner |
|---|---|---|---|---|---|---|
| EBI-16148380 | KMT2A - SIRT1 (Q96EB6) | anti tag coip | bait / prey | human | 0.51 | bait-supported |
| EBI-16148422 | KMT2A - SIRT1 | anti tag coip | bait / prey | human | 0.51 | bait-supported |
| EBI-16148445 | KMT2A - SIRT1 | anti tag coip | bait / prey | human | 0.51 | bait-supported |
| EBI-16148469 | KMT2A - SIRT1 | anti tag coip | bait / prey | human | 0.51 | bait-supported |
| EBI-16148510 | KMT2A - SIRT1 | 2 hybrid | prey / bait | human | 0.51 | (already bait-supported above) |
| EBI-16148361 | KMT2A - Sirt1 (Q923E4) | anti tag coip | bait / prey | mouse | 0.40 | excluded, non-human |
| EBI-16148403 | KMT2A - CLOCK (O15516) | 2 hybrid | prey / bait | human | 0.37 | PREY-ONLY |
Nothing distinguishes the CLOCK row from the SIRT1 rows except which protein carried the tag. The corpus already has gene-sirt1; it does not have CLOCK. The coverage figure that says the directed-evidence interactome is fully written and the prey-only interactome is not is, for this paper, a statement about antibody choice.
This is a different failure from the AKAP8 one in the sibling entry. There the method had no bait; here the method has a bait and it happens to be the other protein. Both land in the same discard bin.
The biology, which is the part worth keeping
From the abstract of PMID 25751424, read in full 2:
- MLL1 coordinates recruitment of CLOCK-BMAL1 activator complexes to chromatin, an event associated with cyclic H3K4 trimethylation at circadian promoters.
- In mouse liver, circadian H3K4 trimethylation is modulated by SIRT1, the NAD(+)-dependent deacetylase.
- MLL1 is acetylated at two conserved residues, K1130 and K1133. The acetylation is cyclic, is controlled by the clock and by SIRT1, and affects the methyltransferase activity of MLL1.
- H3K4 methylation at clock-controlled-gene promoters is influenced by pharmacological or genetic inactivation of SIRT1.
Why a leukaemia graph should care about a residue pair. K1130 and K1133 sit in the N-terminal portion of KMT2A. Every KMT2A fusion in this disease retains the N-terminus and loses the SET domain, so a regulatory acetylation site at 1130/1133 is in the part of the protein that the oncoprotein keeps. Whether the fusion is acetylated there, whether SIRT1 deacetylates it, and whether that changes anything for a protein with no SET domain are three questions nobody appears to have asked. The mechanism described is a modulation of methyltransferase activity, which the fusion does not have - so the naive prediction is that the axis is irrelevant to the fusion, and the interesting version of the question is whether the same acetylation does something else on a protein that cannot methylate.
Cell type and cell state. The circadian work is mouse liver and the cell state is time of day - the most explicitly non-steady-state cell state in this graph. No haematopoietic cell and no fusion.
What is weak here, stated plainly
The KMT2A-CLOCK row itself is a single two-hybrid, miscore 0.37, the lowest of the six. Two-hybrid in this context is a heterologous yeast or mammalian reconstitution; it does not establish that the contact occurs in a mammalian nucleus at endogenous stoichiometry, and CLOCK and MLL1 are both large multi-domain proteins where a fragment-level interaction may not survive in the full-length context.
The functional statement that MLL1 recruits CLOCK-BMAL1 is not the same claim as a direct MLL1-CLOCK contact, and this entry does not conflate them. Recruitment can be indirect through BMAL1, through chromatin, or through a third subunit.
And the honest limit of this entry's own reading: the abstract was read in full; the body and figures were not. A PMC identifier was guessed during this beat and resolved to an unrelated schizophrenia consortium paper, which is exactly why every statement above is scoped to support_location: abstract and why no figure panel is cited.
| partner a | gene-kmt2a |
|---|---|
| partner b | CLOCK |
| partner b uniprot | O15516 |
| partner b length AA | 846 |
| partner a uniprot | Q03164 |
| partner a length AA | 3969 |
| verdict | ADMITTED AS A LEAD |
| adjudicated by | q160, cycle 146, 2026-09-24 |
| druggability | Not assessable as an interface. The druggable object in the surrounding biology is SIRT1, for which pharmacological inhibitors and activators exist and were used in the source study; nothing in this record set maps a KMT2A-CLOCK surface. |
| n methods | 1 |
| n platforms | 1 |
| n publications | 1 |
| endogenous side | no - the CLOCK row is a two-hybrid |
| intact miscore | 0.37 |
| cell state | time of day - a circadian oscillation, the most explicitly non-steady-state cell state in this corpus |
| epigenetics | Cyclic H3K4 trimethylation at circadian promoters; MLL1 acetylation at K1130 and K1133 is clock- and SIRT1-controlled and modulates MLL1 methyltransferase activity. |
| why this matters for the fusion | K1130 and K1133 lie in the N-terminal portion of KMT2A that every fusion in this disease retains. The described mechanism acts on methyltransferase activity, which the fusion lacks, so the naive prediction is irrelevance - which is exactly what makes the measurement informative. |
| absence class | never-measured |
| absence class note | No experiment has tested whether a KMT2A fusion protein is acetylated at the residues corresponding to K1130/K1133, whether SIRT1 deacetylates it, or whether the CLOCK-BMAL1 axis operates in a KMT2A-rearranged blast. Mass spectrometry of an immunoprecipitated fusion protein for acetyl-K would answer the first in one experiment. |
interface residues a
[]
structures
[]
ligands
[]
records
[ "EBI-16148403, 2 hybrid, physical association, human, CLOCK bait / KMT2A prey, miscore 0.37, PMID 25751424" ]
sibling records same paper
[ "EBI-16148380, -16148422, -16148445, -16148469: anti tag coip, KMT2A BAIT / SIRT1 prey, human, miscore 0.51", "EBI-16148510: 2 hybrid, SIRT1 bait / KMT2A prey, human, miscore 0.51", "EBI-16148361: anti tag coip, KMT2A BAIT / mouse Sirt1 prey, Mus musculus, miscore 0.40" ]
cell systems
[ "mouse liver (circadian H3K4me3)", "cultured cells, unspecified in the abstract" ]
localization
{
"compartment": "nucleus",
"subnuclear": [],
"shown_for_fusion": "NO - nobody has looked in any KMT2A fusion. The MLL1 evidence is wild-type MLL1 in mouse liver and cultured cells."
}kmt2a residues of interest
[ 1130, 1133 ]
measurements
[
{
"quantity": "count",
"subject": "IntAct records from PMID 25751424 involving KMT2A",
"value": 7,
"units": "records",
"n": 7,
"universe": "IntAct REST findInteractions on Q03164, 2026-09-24, filtered to publicationPubmedIdentifier 25751424",
"null": null,
"note": "6 SIRT1 (5 human, 1 mouse) plus 1 CLOCK. The grade splits the paper: SIRT1 bait-supported, CLOCK prey-only."
},
{
"quantity": "intact_miscore",
"subject": "KMT2A|CLOCK vs KMT2A|SIRT1 from the same publication",
"value": 0.37,
"units": null,
"n": 1,
"universe": "IntAct REST, PMID 25751424; the SIRT1 rows score 0.51 (human) and 0.40 (mouse)",
"null": null,
"note": "The CLOCK row is the weakest of the seven on IntAct's own score as well as on role."
}
]- SIRT1 modulators are in clinical development and NAD(+) precursors are sold as supplements. If SIRT1 activity sets MLL1 methyltransferase output, an NAD(+)-raising supplement is an unmonitored modifier of the pathway a menin inhibitor is aimed at, in patients who may be taking it without telling anyone.
- If the H3K4me3 landscape at clock-controlled promoters oscillates, then the time of day a marrow sample is drawn is an uncontrolled variable in every epigenomic dataset this project reasons from. That is a cheap thing to record and nobody records it.
- Is KMT2A::AFF1 acetylated at the positions corresponding to K1130 and K1133? Those residues are retained in the fusion, the modification is reported to regulate methyltransferase activity, and the fusion has no methyltransferase activity - so either the modification does something else there or it is a vestigial site, and both answers are informative.
- Does anything in KMT2A-rearranged leukaemia oscillate? Every ChIP, CUT&RUN and RNA-seq dataset in this corpus is a single time point. A circadian component to H3K4me3 at HOXA would make every one of them a sample from an unrecorded phase.
- How many other partner pairs in this graph are split across the bait-supported / prey-only line by a single paper? This is the first one identified; the census file has the material to count them and it has not been counted.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
IntAct REST returns six records involving Q03164 from PMID 25751424. Four are anti-tag co-immunoprecipitation with KMT2A as the recorded bait and human SIRT1 (Q96EB6) as prey, intact-miscore 0.51 (ACs EBI-16148380, -16148422, -16148445, -16148469); one is a two-hybrid with SIRT1 as bait and KMT2A as prey (EBI-16148510, 0.51); one is an anti-tag co-IP with KMT2A as bait and mouse Sirt1 (Q923E4) as prey in Mus musculus (EBI-16148361, 0.40); and one, EBI-16148403, is a two-hybrid with CLOCK (O15516) as BAIT and KMT2A as PREY, human, intact-miscore 0.37. SIRT1 therefore qualifies as bait-supported and CLOCK as prey-only from the same publication.
IntAct molecular interaction database, REST findInteractions on Q03164, retrieved 2026-09-24 by q160 · IntAct (EMBL-EBI) 2026What this evidence cannot show
Read from the live REST response; the IMEx records themselves (IM-24440-1 through -8) were not opened at the curator level, so the mapping of each AC to a specific figure panel in the paper is unknown. Four co-IP rows with identical partners and identical miscore are very likely one experiment annotated four ways (different constructs or conditions), not four experiments - the same row-versus-experiment trap the corpus records for the AFF4 two-hybrid set.
Evidence that argues against this (1)
- The split is not evidence that the bait/prey grade is useless. SIRT1 has four co-IP rows and CLOCK has one two-hybrid, so on any reasonable reading SIRT1 IS better supported in this paper than CLOCK. The claim here is narrower: the grade is assigned by role rather than by strength, and the two happen to agree in this case while the AKAP8 case shows they need not. IntAct (EMBL-EBI) 2026
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
PMID 25751424 (Aguilar-Arnal L, Katada S, Orozco-Solis R, Sassone-Corsi P, Nat Struct Mol Biol 2015) reports that MLL1 coordinates recruitment of CLOCK-BMAL1 activator complexes to chromatin, an event associated with cyclic trimethylation of H3K4 at circadian promoters; that in mouse liver circadian H3K4 trimethylation is modulated by SIRT1; that mammalian MLL1 is acetylated at two conserved residues, K1130 and K1133; that MLL1 acetylation is cyclic, controlled by the clock and by SIRT1, and affects the methyltransferase activity of MLL1; and that H3K4 methylation at clock-controlled-gene promoters is influenced by pharmacological or genetic inactivation of SIRT1.
NAD(+)-SIRT1 control of H3K4 trimethylation through circadian deacetylation of MLL1. · Aguilar-Arnal L, Katada S, Orozco-Solis R, Sassone-Corsi P · Nat Struct Mol Biol 2015What this evidence cannot show
ABSTRACT ONLY. The body and figures were not read, and an attempt to fetch full text via a guessed PMC identifier returned an unrelated paper, so nothing here is sourced to a figure. The system is mouse liver and cultured cells; no haematopoietic cell, no KMT2A fusion, and no leukaemia appears in the abstract - `expected_absent_terms` records that as part of the claim rather than as an oversight. The acetylation-site numbering K1130/K1133 is quoted from the abstract and was not checked against a KMT2A sequence record.
Evidence that argues against this (1)
- This graph already records, as rule 72, that a bounded absence about H3K27me3 under menin inhibition survived only in its narrow pharmacological form. The analogous risk here is reading 'MLL1 coordinates recruitment of CLOCK-BMAL1' as a direct MLL1-CLOCK contact. The abstract asserts coordination of recruitment, which is compatible with an entirely indirect relationship through BMAL1 or through chromatin. Aguilar-Arnal L et al. 2015
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
UniProtKB O15516 (CLOCK_HUMAN) is 846 aa, with three experimentally coded (ECO:0000269) entries for Nucleus and one for Cytoplasm, cytosol, plus Cytoplasm by similarity. Tissue specificity records expression in all tissues examined, explicitly including leukocytes, spleen and thymus.
UniProtKB entry O15516 (CLOCK_HUMAN), subcellular location and tissue specificity, retrieved 2026-09-24 · UniProt Consortium 2026What this evidence cannot show
Rule 91: the ECO:0000269 codes mean a curator read an experiment, not that the cited paper is about circadian chromatin. The leukocyte expression line removes the compartment- and expression-mismatch objection that killed several other prey-only partners in this corpus, but it does not supply positive evidence in a leukaemic blast.
Evidence that argues against this (1)
- Ubiquitous expression is a weak argument for relevance: it removes an objection rather than supplying support. CLOCK being present in leukocytes says nothing about whether it contacts KMT2A there, and the only record that speaks to the contact is a single two-hybrid in a heterologous system. IntAct (EMBL-EBI) 2026
verifies the residue placement that entry asserts, from the UniProt sequence
the same source paper, bounded by its citation network
single two-hybrid, CLOCK bait / KMT2A prey, miscore 0.37
same paper; SIRT1 is graded bait-supported and CLOCK prey-only from the same figure set
cyclic H3K4me3 at circadian promoters, SIRT1-modulated via MLL1 K1130/K1133 acetylation
worked case: the grade tracks assay direction, not evidence strength
the second mechanism: one paper split across the grade by assay direction