Oncofusion

start / graph / gene-mapk7

Geneweak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.Q75-dot1l-partners

MAPK7 (ERK5) - a DOT1L partner on one 1:365 spoke of one atlas pulldown, and a kinase with real leukaemia literature that has never been connected to DOT1L

also called ERK5 · BMK1 · BMK-1 · PRKM7 · Big MAP kinase 1 · Extracellular signal-regulated kinase 5 · Mitogen-activated protein kinase 7

ERK5 is a signalling enzyme that carries growth signals from the outside of a cell into its nucleus, and one large screen saw a tiny amount of it stuck to DOT1L. The amount was about one ERK5 for every 365 DOT1L molecules, and no other study has ever reported the two together.

The same thing, for a scientist

MAPK7/ERK5 (Q13164, 816 aa) is an atypical MAP kinase that shuttles to the nucleus on activation and has a substantial cancer literature, including 59 PubMed records in leukaemia [E4]. Its only recorded contact with anything in this graph is a single spoke of the Hein et al. GFP-DOT1L pulldown in HeLa [E1]: interactor confidence class **B-**, interaction stoichiometry **0.0027 (1 MAPK7 per 365 DOT1L)**, prey-bait profile correlation **0.033**, and `core.interactor` not set [E2]. Its abundance stoichiometry is the one genuinely interesting number - 1.009, i.e. **MAPK7 and DOT1L are present at essentially identical copy number in HeLa** (3,263 against 3,779) [E2] - so the substoichiometry is not an abundance artefact in the usual direction and cannot be explained away by the prey being scarce. The two proteins do share a compartment: UniProt gives MAPK7 Cytoplasm, Nucleus and Nucleus/PML body, translocating to the nucleus on activation [E3]. **`DOT1L[tiab] AND (MAPK7 OR ERK5 OR BMK1)[tiab]` returns 0 records and `(MLL OR KMT2A)[tiab] AND (MAPK7 OR ERK5 OR BMK1)[tiab]` returns 1, a bovine oocyte paper [E4].** Written as a graded, bounded single-source observation, not as an interactor of the fusion.

Go deeper
Why this one was worth checking outside the screen, and what came back

ERK5 is not an obscure protein. It is a druggable kinase with published inhibitors, a MEF2-directed transcriptional output, and an active oncology literature - (ERK5 OR MAPK7)[tiab] AND (leukemia OR leukaemia)[tiab] returns 59 records 4. A DOT1L-ERK5 link would be mechanistically live: ERK5 activates MEF2 family transcription factors, and MEF2C is a central node in this graph's KMT2A-rearranged programme.

The check found nothing. Zero records pair DOT1L with any ERK5 name, and the single (MLL OR KMT2A) hit is Theriogenology 2026 on bovine cumulus cells (PMID 42105620), where MLL matches something else entirely 4. The operand controls pass - DOT1L[tiab] = 710, (MAPK7 OR ERK5)[tiab] = 1,379 - so this is a real zero and not a non-composing query (brief rule 67).

So the attractive hypothesis has no support, and the graph should not manufacture one. The honest statement is that a large, well-studied kinase and a well-studied methyltransferase, both nuclear, both abundant in the same cell at the same level, have never been reported together except as one B- row of one atlas.

The abundance argument does not rescue this spoke, and that is worth saying plainly

The usual dismissal of a substoichiometric AP-MS hit is that the prey is abundant background. That is not available here. MAPK7 is at 3,263 copies per HeLa cell and DOT1L at 3,779 - an abundance stoichiometry of 1.009, the closest to unity of any prey in the pulldown 2. It is not a sticky abundant protein and it is not a ribosomal or nucleolar passenger.

What argues against it instead is the profile: an enrichment of 2.66 SD against the controls, a prey-bait intensity correlation of 0.033 where the three A+ DotCom subunits sit at 0.275-0.420, and an interaction stoichiometry of 1:365 2. A correlation near zero means MAPK7's intensity did not track DOT1L's across the replicate pulldowns - which is what the class B- encodes.

One reading that keeps the observation alive: ERK5's nuclear pool is conditional. UniProt records that it translocates to the nucleus on activation 3, and HeLa cells in standard culture are not stimulated. A transient, signal-dependent kinase-substrate encounter is exactly the kind of interaction that would appear at 1:365 with a poor correlation and be real. That is a hypothesis, not a finding, and the experiment that would settle it is named in open_questions.

Facts and figures
uniprotQ13164
length AA816
gene symbolMAPK7
protein nameMitogen-activated protein kinase 7 (ERK5, Big MAP kinase 1)
uniprot resolutionaccession:Q13164 AND organism_id:9606 AND reviewed:true; one reviewed hit; primary gene name read back as MAPK7; length 816 aa checked
shown for fusionNO - nobody has looked in any fusion. Zero PubMed records pair DOT1L with any MAPK7/ERK5 name, and the only (MLL OR KMT2A) hit is a bovine oocyte paper.
interaction with dot1l
{
  "source_pmid": "26496610",
  "source_class": "proteome-scale AP-MS atlas",
  "intact_interaction_ac": "EBI-11060394",
  "imex_id": "IM-24272-464",
  "detection_method": "psi-mi:MI:0007 anti tag coimmunoprecipitation",
  "interaction_type": "psi-mi:MI:0914 association",
  "expansion": "psi-mi:MI:1060 spoke expansion - binary pair inferred, not observed",
  "intact_miscore": 0.35,
  "author_confidence_class": "B-",
  "core_interactor": false,
  "interaction_stoichiometry_prey_per_bait": 0.00274,
  "interaction_stoichiometry_as_ratio": "1 MAPK7 per 365 DOT1L",
  "log10_prey_bait_ratio": -2.562,
  "abundance_stoichiometry_prey_per_bait": 1.009,
  "log10_prey_bait_expression_ratio": 0.003908126,
  "enrichment_sd": 2.662,
  "neg_log10_p": 3.812598,
  "prey_bait_correlation": 0.033,
  "valid_values": 3,
  "cell_system": "HeLa, LAP-tagged DOT1L at 3,734 copies (0.99 of endogenous)",
  "independent_sources": 0,
  "reciprocal_tested": false
}
localization
{
  "subcellular": "Cytoplasm, Nucleus, and Nucleus/PML body; UniProt notes translocation to the nucleus upon activation. Compartment overlap with DOT1L exists, but the nuclear pool is conditional on signalling and HeLa cells in standard culture are unstimulated.",
  "source": "UniProt Q13164, fetched 2026-09-21; the location comment carries no ECO code in the returned fields, so it is a curated statement rather than a traced measurement (brief rule 91)"
}
absence claim
{
  "queries_run": [
    "DOT1L[tiab] AND (MAPK7 OR ERK5 OR BMK1)[tiab]",
    "(MLL OR KMT2A)[tiab] AND (MAPK7 OR ERK5 OR BMK1)[tiab]",
    "(ERK5 OR MAPK7)[tiab] AND (leukemia OR leukaemia)[tiab]"
  ],
  "records": [
    0,
    1,
    59
  ],
  "on_topic": [
    0,
    0,
    "not title-checked"
  ],
  "operand_controls": {
    "DOT1L[tiab]": 710,
    "(MAPK7 OR ERK5)[tiab]": 1379
  },
  "date": "2026-09-21",
  "tool": "PubMed E-utilities esearch",
  "bounded_by": "titles and abstracts only; supporting figures not searched"
}
How this could bite us back
  • If a later cycle reads this entry as 'DOT1L binds ERK5' and builds a kinase-inhibitor rationale on it, the whole chain rests on one B- spoke of one 2015 screen with a prey-bait correlation of 0.033. The title of this entry says so deliberately.
  • ERK5 inhibitors have a documented off-target problem: several early compounds inhibit BRD4 as well as the kinase, so any phenotype obtained with them in a KMT2A-rearranged cell would be uninterpretable in exactly the pathway this graph cares about, and that confound must be designed out before any such experiment is read.
  • The abundance argument that usually dismisses a substoichiometric hit is unavailable here - MAPK7 and DOT1L are at the same copy number - so dismissing this spoke requires the profile correlation, which is a softer number.
Still open
  1. Does activated ERK5 contact DOT1L? The decisive experiment is a GFP-DOT1L pulldown in stimulated versus unstimulated cells - EGF or sorbitol drives ERK5 into the nucleus, and a conditional interaction would appear as a stoichiometry shift rather than a new hit.
  2. ERK5 activates MEF2 family transcription factors and MEF2C is central to this graph's KMT2A-rearranged programme. Is there an ERK5-MEF2C-DOT1L axis, and has any ERK5 inhibitor ever been profiled for H3K79 methylation?
  3. Is MAPK7 expressed in t(4;11) blasts at all? Everything here is HeLa, and the entry would change character if the kinase is absent from the disease cell.
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    The entire recorded interaction evidence for MAPK7 with DOT1L is one row of one IntAct interaction record - EBI-11060394 / IM-24272-464, spoke-expanded (MI:1060) from a single anti-tag co-immunoprecipitation of LAP-tagged DOT1L in HeLa, interaction type 'association' (MI:0914), intact-miscore 0.35, author-confidence B- - from PMID 26496610; MAPK7 appears exactly once across all 49 IntAct rows involving DOT1L from seven publications.

    What this evidence cannot show

    One database on one day, and IntAct's coverage of AP-MS supplementary tables is incomplete by construction. Spoke expansion means the binary pair was inferred from co-purification by a curator and never observed as a binary interaction. No leukaemia cell, no KMT2A fusion, no primary sample anywhere in this evidence - HeLa cervical carcinoma only (brief rule 73).

    Evidence that argues against this (1)
    • The same pulldown recovered MLLT3, MLLT6 and MLLT10, the genuine DotCom core, so the experiment demonstrably works and its failure to find corroboration for MAPK7 elsewhere may reflect nothing more than that nobody has looked at a kinase-methyltransferase pair. Hein MY et al. 2015
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    In the source study's own interactions table the DOT1L-MAPK7 spoke is class B- with interaction stoichiometry 0.00274 (1 MAPK7 per 365 DOT1L, log10 prey:bait -2.562), abundance stoichiometry 1.009 (log10 0.0039), enrichment 2.662 SD, -log10 p 3.813 and prey-bait profile correlation 0.033, with core.interactor not set; in the same paper's HeLa proteome MAPK7 is at 3,263 copies per cell against DOT1L's 3,779.

    What this evidence cannot show

    Downloaded 2026-09-21 from the Elsevier static CDN with magic bytes checked; the article narrative itself is unreadable (no PMCID, DOI resolves to a 13-character Elsevier stub), so the exact thresholds behind the class labels could not be read and the class is used as an ordinal grade published by the authors. Stoichiometries are medians of three replicates with no dispersion reported in the table. The prey protein group is Q13164;Q13164-3;C9JUK9;J3KT61, so the row cannot distinguish MAPK7 isoforms.

    Evidence that argues against this (1)
    • Interaction stoichiometry does not discriminate here in the direction one would expect: MLLT3/AF9, an A+ hit with solved DOT1L co-structures, sits at 1:172 in the same pulldown - twice as substoichiometric as several of the B- rows and only two-fold better than MAPK7. A stoichiometry cut would be a bad filter and this number should not be used as one. Cell 163(3), supplementary material 2015
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB Q13164 (MK07_HUMAN) is human MAPK7, 816 aa, protein name 'Mitogen-activated protein kinase 7 / Big MAP kinase 1 / Extracellular signal-regulated kinase 5', gene synonyms BMK1, ERK5 and PRKM7, with subcellular location Cytoplasm, Nucleus and Nucleus/PML body, noting that it translocates to the nucleus upon activation.

    What this evidence cannot show

    Resolved 2026-09-21 by accession with the primary gene name read back as MAPK7 and the length checked (brief rules 62 and 65). The location comment carries no ECO code in the fields returned, so by brief rule 91 this is a curated statement and not a measurement I have traced to an experiment; the conditional wording ('translocates upon activation') is the part that matters here and it is unsourced in the summary field.

    Evidence that argues against this (1)
    • This graph has measured that baseline compartment annotations are frequently miscalled - localization-partner-baseline-compartment-is-often-miscalled - so a shared UniProt compartment is at most a weak prior in favour and its absence would have been at most a weak prior against. UniProt Consortium 2026
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    Counted absence, PubMed E-utilities esearch run 2026-09-21: DOT1L[tiab] AND (MAPK7 OR ERK5 OR BMK1)[tiab] returns 0 records; (MLL OR KMT2A)[tiab] AND (MAPK7 OR ERK5 OR BMK1)[tiab] returns 1 record, PMID 42105620, a Theriogenology paper on bovine cumulus cells, off topic; the operand controls return 710 for DOT1L[tiab] and 1,379 for (MAPK7 OR ERK5)[tiab], and (ERK5 OR MAPK7)[tiab] AND (leukemia OR leukaemia)[tiab] returns 59 records, so an ERK5 leukaemia literature exists and none of it touches DOT1L or KMT2A.

    What this evidence cannot show

    Bounded by titles and abstracts (brief rule 80): a co-IP in a supporting figure is unreachable. Three name spellings were run (MAPK7, ERK5, BMK1); the hyphenated BMK-1 was not run separately and PubMed tokenises hyphens, so that spelling is not covered. The 59 leukaemia records were counted, not title-checked, so that figure is a raw count and should not be quoted as 59 on-topic papers.

    Evidence that argues against this (1)
    • A zero here is exactly what one expects for any prey of a 2015 proteome-scale atlas, true or false, because nobody publishes a title about one row of somebody else's screen. This absence therefore discriminates weakly, and the entry's confidence rests on the confidence class and the profile correlation, not on the literature silence. NCBI PubMed (queries run 2026-09-21) 2026
Connected to