Oncofusion

start / graph / variant-h2a-z

Mechanismmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q59-marks2

H2A.Z - two genes of identical length, one KMT2A paper, and the measured quantity is a ratio nobody has repeated

also called H2A.Z · H2AZ · H2A.Z.1 · H2A.Z.2 · H2AZ1 · H2AZ2 · H2AFZ · H2AFV · H2A.V · histone variant H2A.Z

H2A.Z is a swap-in version of a core histone that sits at the nucleosomes flanking a gene's start, where it makes those nucleosomes easier to move. In KMT2A-fusion leukaemia one study found that the fusion protein brings an enzyme to the Hoxa9 gene and changes how much of the H2A.Z there is acetylated - and nobody has repeated it, in any other fusion, in any human cell.

The same thing, for a scientist

H2A.Z is the H2A variant enriched at the +1 and -1 nucleosomes of active and poised promoters. **It is a variant, not a mark, and it is encoded by two genes whose protein products are exactly the same length**: H2AZ1 (HGNC:4741, 4q23, previously H2AZ/H2AFZ, P0C0S5, 128 aa) and H2AZ2 (HGNC:20664, 7p13, previously H2AV/H2AFV, Q71UI9, 128 aa) [E4]. Deposition runs through the SRCAP and p400/TIP60 remodellers with VPS72/YL1 as the delivery chaperone; ANP32E is the eviction chaperone, preferentially binding H2A.Z-H2B over H2A-H2B and facilitating H2A.Z removal from the +1 nucleosome [E3]. **The entire KMT2A literature on this variant is one paper.** MLL-AF10 positively regulates the **ratio of acetylated H2A.Z to total H2A.Z** at the Hoxa9 locus; Tip60 (KAT5) is recruited by MLL-AF10 and MLL-ENL to that locus, acetylates H2A.Z there, and conditional Tip60 deletion prevents development of MLL-AF10 and MLL-ENL leukaemia [E1]. **That is the only measurement, it is in mouse, it is at one locus, it is in two non-index partners, and the analyte is an acetylation ratio rather than variant occupancy.** A bounded query over seven spellings returns that single record [E2], and a word-boundary sweep of nineteen KMT2A-rearranged chromatin full texts - including all six MLL-AF4 papers - finds the variant in **none** of them [E5]. The dependency data point the other way from the literature and this graph already records it under `gene-h2az1`: H2AZ1 is more required in KMT2A-rearranged lines while every component of its exchange machinery is less required.

Go deeper
The identifier trap, which is new and belongs in the registry

Two genes, two accessions, zero length difference. H2AZ1 is P0C0S5 at 128 aa and H2AZ2 is Q71UI9 at 128 aa 4. verify-accessions.py asserts accession plus length_aa and therefore cannot catch a swap between them at any tolerance - the fourth zero-spread case in this project after the RAS family, IMPDH1/2 and BMI1/CDK6.

And the protein-name defence fails too. Q71UI9's UniProt entry name is H2AV_HUMAN and its recommended protein name is "Histone H2A.V", with "H2A.Z variant histone 2" appearing only as an alternative name. A reader applying rule 65 - fetch the record and read the protein name back - asks "is this H2A.Z.2?", is told "Histone H2A.V", and may discard a correct record. That is the same failure shape already logged for DPF2 (returns REQU_HUMAN / Zinc finger protein ubi-d4) and for BICRA (protein name names BRD4).

The previous symbols are the searchable ones. H2AZ1's previous official symbols are H2AZ and H2AFZ; H2AZ2's are H2AV and H2AFV. Essentially all literature before the 2020 nomenclature change uses H2AFZ and H2AFV, so a query on current symbols alone under-counts - this project's retired-symbol rule, applied. The bounded query here ran all seven strings 2.

This is a correction owed to data/symbol-collisions.json, which this beat does not own. The proposed row is recorded in the memo with the fetched values.

The one KMT2A measurement, stated at the resolution it was actually made

PMID 33967269 is the whole of it 1. What its figure titles state, read from the publisher page through this project's route-5 DOI fetch:

figureclaim
Fig. 1MLL-AF10 positively regulates the ratio of acetylated H2A.Z to total H2A.Z on the Hoxa9 locus
Fig. 2Tip60 acetylates H2A.Z
Fig. 3Tip60 colocalizes with MLL-AF10 and MLL-ENL at the Hoxa9 locus
Fig. 4Tip60 is essential for MLL-AF10 and MLL-ENL-transformed cells in vitro
Fig. 5Tip60 is required for development of MLL-AF10 and MLL-ENL leukaemia in vivo
Fig. 6Tip60 globally regulates MLL-AF10 target genes

Four things this graph should take from it exactly:

  1. The analyte is a ratio, not an occupancy. Acetylated H2A.Z over total H2A.Z means total variant occupancy was the denominator - so the fusion changes the acetylation state of H2A.Z at Hoxa9, and the paper's own framing does not claim it changes how much H2A.Z is there.
  2. It is locus-anchored, at one locus. Hoxa9. Figure 6's 'globally' refers to target-gene expression, not to a genome-wide H2A.Z map.
  3. The partners are MLL-AF10 and MLL-ENL - fusion-kmt2a-mllt10 and fusion-kmt2a-mllt1. Not KMT2A::AFF1, and not human.
  4. The enzyme is KAT5/Tip60, which this graph already carries as gene-kat5, and it is a MYST-family acetyltransferase - the same family as KAT8/MOF in mark-h4k16ac, and the same family the non-selective inhibitor MG149 hits.
The exchange machinery is well defined and points away from the disease

ANP32E is a higher-eukaryote-specific H2A.Z chaperone that preferentially associates with H2A.Z-H2B dimers over H2A-H2B, dissociates non-nucleosomal DNA-H2A-H2B aggregates, co-occupies the genome with H2A.Z, and on over-expression causes strong global H2A.Z loss at +1 nucleosomes while depletion causes a moderate increase 3. Deposition runs the other way through SRCAP and p400/TIP60 with VPS72/YL1 delivering the dimer.

This graph already holds the dependency asymmetry in gene-h2az1: H2AZ1 has a KMT2A-selective DepMap difference of -0.184 (more required in the rearranged lines) while ANP32E is +0.093, SRCAP +0.118, EP400 +0.121 and VPS72 -0.051 - the variant matters more and the machines that move it matter less. That entry states it and this one does not restate it; the point here is the consequence for design. If the variant is required but its exchange machinery is not, then the requirement is more likely to be for H2A.Z's presence in nucleosomes than for its turnover, and a therapeutic strategy aimed at the remodellers is aimed at the part of the system the data say is dispensable.

The bounded absence, and where it is bounded

(MLL-rearranged OR KMT2A-rearranged OR MLL-AF9 OR MLL-AF4 OR KMT2A OR "MLL fusion")[tiab] AND (H2A.Z OR H2AZ OR H2AFZ OR H2AZ1 OR H2AZ2 OR "H2A.Z.1" OR "H2A.Z.2" OR H2AFV)[tiab] returns 1 record - 1 - on 2026-09-20. The operands are 1,233 and 3,209; (leukemia OR leukaemia)[tiab] against the variant name list returns 11. The instrument works and both operands are large 2.

Then the full-text bound. Nineteen full texts of the core KMT2A-rearranged chromatin literature were swept with word boundaries: Guenther 2008, Kerry 2017, Prange 2017, Godfrey 2017, Godfrey 2019, Godfrey 2021, Harman 2021, Crump 2023, Chen 2015, Ropa 2018, Ropa 2020, Chu 2020, Valerio 2017, Mishra 2014, Sakamoto 2014, the paediatric mutation landscape, and three menin-inhibitor chromatin papers. H2A.Z, H2AZ1, H2AZ2, H2AFZ and H2AFV occur zero times in all nineteen 5. Unlike H3K27ac, H3K9me3, H4K16ac and H3.3, rule 80 did not rescue this variant - the full-text sweep confirmed the abstract-level absence.

So the correct statement is: H2A.Z has been measured once in a KMT2A-rearranged system, at one locus, in mouse, as an acetylation ratio, for two partners that are not the index fusion - and it has never been measured in KMT2A::AFF1, in B-lineage disease, in a human cell, in infant material, or under either of this disease's two targeted drugs.

Facts and figures
objecthistone variant H2A.Z (NOT a post-translational mark)
length spread AA0
length defence usablefalse
protein name defence usablefalse
protein name trapQ71UI9's recommended protein name is 'Histone H2A.V' and its entry name is H2AV_HUMAN, so reading the name back for H2AZ2 looks like a failed lookup - same class as DPF2/REQU_HUMAN and BICRA
acetyltransferase in this diseasegene-kat5 (Tip60), recruited by MLL-AF10 and MLL-ENL to Hoxa9
measured in kmt2a rtrue
measured in kmt2a aff1false
measured in human cells in this diseasefalse
measured in infant materialfalse
measurement scopeONE LOCUS - the ratio of acetylated H2A.Z to total H2A.Z at Hoxa9 in murine MLL-AF10 cells (PMID 33967269). No genome-wide H2A.Z map exists in any KMT2A-rearranged system.
analyte is a ratio not an occupancytrue
menin inhibitor effectnever measured
dot1l inhibitor effectnever measured
abstract query count 2026 09 201
full texts swept19
full texts containing the variant0
rule 80 changed the answerfalse
named in entries when written12
mentions when written122
count methodregex /H2A\.Z/i over all 1,701 node JSON files, run 2026-09-20
genes
[
  "H2AZ1",
  "H2AZ2"
]
hgnc IDS
[
  "HGNC:4741",
  "HGNC:20664"
]
previous symbols
{
  "H2AZ1": [
    "H2AZ",
    "H2AFZ"
  ],
  "H2AZ2": [
    "H2AV",
    "H2AFV"
  ]
}
cytobands
{
  "H2AZ1": "4q23",
  "H2AZ2": "7p13"
}
ensembl
{
  "H2AZ1": "ENSG00000164032",
  "H2AZ2": "ENSG00000105968"
}
subunits
{
  "H2AZ1": {
    "accession": "P0C0S5",
    "length_aa": 128,
    "entry_name": "H2AZ_HUMAN"
  },
  "H2AZ2": {
    "accession": "Q71UI9",
    "length_aa": 128,
    "entry_name": "H2AV_HUMAN"
  }
}
deposition machinery
[
  "gene-srcap",
  "gene-ep400",
  "gene-kat5",
  "VPS72/YL1 (no entry in this graph)"
]
eviction chaperone
[
  "gene-anp32e"
]
partners with data
[
  "fusion-kmt2a-mllt10",
  "fusion-kmt2a-mllt1"
]
abstract query name list
[
  "H2A.Z",
  "H2AZ",
  "H2AFZ",
  "H2AZ1",
  "H2AZ2",
  "H2A.Z.1",
  "H2A.Z.2",
  "H2AFV"
]
reading routes tried for pmid 33967269
[
  "Europe PMC core (no pmcid)",
  "efetch db=pubmed ArticleId type pmc (absent)",
  "route 5: curl -sL https://doi.org/10.1038/s41375-021-01244-y with a browser User-Agent - HTTP 200, 19,501 stripped characters, abstract and figure titles present, article body absent"
]
How this could bite us back
  • KAT5/Tip60 is a MYST-family acetyltransferase with substrates spanning ATM, p53 and H4; the only chemical tool used against this family in KMT2A-rearranged leukaemia, MG149, inhibits KAT5, KAT6A, KAT7 and KAT8 together, so an 'H2A.Z acetylation' strategy is in practice a pan-MYST strategy and cannot be attributed to the variant.
  • H2A.Z is required for normal development and is enriched at the promoters of essentially every poised gene; a therapy that removed or de-acetylated it would act on the transcriptional start site of most of the genome, and the fact that its exchange machinery scores as LESS required in KMT2A-rearranged lines removes the obvious selective handle.
  • Because the only published measurement is an acetylation RATIO, a pharmacodynamic assay reading total H2A.Z occupancy would return no change from an intervention that worked, and one reading bulk acetyl-H2A.Z would be dominated by the genome-wide pool rather than by the fusion's targets.
  • The two paralogues are indistinguishable by length and their UniProt protein names do not name the variant consistently, so a resistance or dependency claim assigned to the wrong gene would pass every automated check this project runs.
Still open
  1. Is H2A.Z present at KMT2A::AFF1-bound loci at all? No genome-wide H2A.Z map exists in any KMT2A-rearranged cell, so the variant that DepMap says is selectively required in these lines has never been located in their genomes.
  2. Which of H2AZ1 and H2AZ2 carries the requirement? The two are identical in length, differ at three residues, and every source touching this disease writes only 'H2A.Z' - so the selective dependency recorded in gene-h2az1 is assigned to a gene by the screen's reagents and not by any experiment on the protein.
  3. Does KMT2A::AFF1 change the acetylated-to-total H2A.Z ratio at its targets the way MLL-AF10 does at Hoxa9? The experiment is a ChIP with two antibodies at named loci and it has never been done for the index fusion.
  4. Is Tip60's requirement in MLL-AF10 and MLL-ENL leukaemia an H2A.Z-acetylation requirement? Conditional deletion removes the whole enzyme; an acetylation-dead H2A.Z knock-in, or a Tip60 HAT-dead rescue scored for H2A.Z acetylation at Hoxa9, would separate them.
Every source behind this entry (5)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In murine MLL-fusion leukaemia, MLL-AF10 positively regulates the ratio of acetylated H2A.Z to total H2A.Z at the Hoxa9 locus; the MYST acetyltransferase Tip60 (KAT5) is required for HoxA cluster and Meis1 expression, is recruited by MLL-AF10 and MLL-ENL to the Hoxa9 locus where it acetylates H2A.Z and promotes Hoxa9 expression, colocalises with both fusions at that locus, and conditional Tip60 deletion prevents the development of MLL-AF10 and MLL-ENL leukaemia.

    PMID 33967269DOI
    What this evidence cannot show

    **Read at abstract level plus figure titles only.** The paper has no PMCID; routes 1-4 are structurally unavailable and the route-5 DOI fetch returned the publisher landing page (19,501 stripped characters) carrying the abstract, the six figure titles and the reference list but not the article body. No quantification, antibody, ChIP protocol or statistics was read. **Murine MLL-AF10 and MLL-ENL** - two partners, neither of which is the index KMT2A::AFF1 fusion - and a single locus, Hoxa9. The measured quantity is an acetylation ratio, so this is evidence about H2A.Z acetylation state and not about variant deposition or occupancy. Figure 6's 'globally regulates MLL-AF10 target genes' refers to expression, not to a genome-wide H2A.Z map.

    Evidence that argues against this (1)
    • Tip60/KAT5 acetylates many substrates including H4, H2A, ATM and p53, and the leukaemogenesis phenotype comes from conditional deletion of the whole enzyme rather than from an H2A.Z-acetylation-dead allele - so the requirement for Tip60 in MLL-AF10/MLL-ENL leukaemia is not demonstrated to run through H2A.Z. The parallel MYST case in this disease, MOF/KAT8, resolves to DNA damage repair rather than to transcription at fusion targets. Valerio DG; Xu H; Chen CW; Hoshii T; Eisold ME; Delaney C; et al. 2017
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    As of 2026-09-20 PubMed returns 1 record for the KMT2A term set intersected with (H2A.Z OR H2AZ OR H2AFZ OR H2AZ1 OR H2AZ2 OR 'H2A.Z.1' OR 'H2A.Z.2' OR H2AFV)[tiab], and that record is PMID 33967269; the variant name list alone returns 1,233, the KMT2A term set alone returns 3,209, and (leukemia OR leukaemia)[tiab] against the variant name list returns 11.

    What this evidence cannot show

    **One in titles and abstracts; supporting figures not searched by this query.** The name list deliberately includes both current symbols and both sets of previous official symbols (H2AFZ, H2AFV) per this project's retired-symbol rule, and both dotted and undotted spellings. Free-text phrases such as 'histone variant H2A' and 'H2A variant' were not run. The querytranslation was read back and contained no [UID] term. A positive control through the identical query shape - the same KMT2A term set against H3K27ac - returns 14.

    Evidence that argues against this (1)
    • H2A.Z's leukaemia literature is not empty - 11 records intersect leukemia/leukaemia - and the GAS41/YEATS2 YEATS domain reads histone acetylation to promote H2A.Z deposition in other cancers, which is mechanistically adjacent to the YEATS biology at the centre of this graph. The KMT2A intersection is empty; the relevance is not. Yamagata K; Shino M; Aikawa Y; Fujita S; Kitabayashi I 2021
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    ANP32E is a higher-eukaryote-specific histone chaperone that preferentially associates with H2A.Z-H2B dimers rather than H2A-H2B in vitro and in vivo, dissociates non-nucleosomal DNA-H2A-H2B aggregates, contacts the extended H2A.Z alphaC helix through residues 214-224 absent from other ANP32 family members, and co-occupies the genome with H2A.Z; ANP32E over-expression causes strong global H2A.Z loss at +1 nucleosomes and ANP32E depletion a moderate global increase.

    Anp32e, a higher eukaryotic histone chaperone directs preferential recognition for H2A.Z. · Mao Z; Pan L; Wang W; Sun J; Shan S; Dong Q; et al. · Cell Res 2014
    PMID 24613878DOI
    What this evidence cannot show

    Crystallography and genome-wide profiling in non-haematopoietic human cells; no leukaemia, no fusion, no KMT2A. The deposition side (SRCAP, p400/TIP60, VPS72/YL1) is established elsewhere and is not demonstrated in this paper. The over-expression and depletion effects are described by the authors as strong and moderate respectively, i.e. asymmetric, so ANP32E is a facilitator of removal rather than the sole determinant of H2A.Z occupancy.

    Evidence that argues against this (1)
    • This graph's own DepMap reading, recorded in gene-h2az1, finds ANP32E LESS required in KMT2A-rearranged lines (+0.093) while H2AZ1 is MORE required (-0.184) - so in this disease the removal chaperone and the variant move in opposite directions, and a mechanism built on H2A.Z exchange is not supported by the dependency data. Yamagata K; Shino M; Aikawa Y; Fujita S; Kitabayashi I 2021
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    H2A.Z is encoded by two human genes - H2AZ1 (HGNC:4741, 4q23, previous official symbols H2AZ and H2AFZ, UniProt P0C0S5, 128 aa, entry name H2AZ_HUMAN, recommended protein name 'Histone H2A.Z') and H2AZ2 (HGNC:20664, 7p13, previous official symbols H2AV and H2AFV, UniProt Q71UI9, 128 aa, entry name H2AV_HUMAN, recommended protein name 'Histone H2A.V' with 'H2A.Z variant histone 2' only as an alternative name) - whose protein products are of identical length.

    What this evidence cannot show

    Fetched live on 2026-09-20 from rest.genenames.org and rest.uniprot.org, with gene_primary read back on every returned row per this project's accession rules. **The identical 128 aa length means verify-accessions.py cannot detect a swap between the two paralogues**, and the recommended protein name of the second makes the standard read-back check return something that looks like a failed lookup. Both failure modes are properties of the registry records, not of any claim made here.

    Evidence that argues against this (1)
    • Identical length does not mean identical protein: H2A.Z.1 and H2A.Z.2 differ at three residues and have been reported to have distinct functions and distinct turnover, so treating the two as interchangeable - which every KMT2A-adjacent source does, including the one paper in this entry, which says only 'H2A.Z' - discards a real distinction. Mao Z; Pan L; Wang W; Sun J; Shan S; Dong Q; et al. 2014
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    A word-boundary sweep for H2A.Z, H2AZ1, H2AZ2, H2AFZ and H2AFV across nineteen full texts of the core KMT2A-rearranged chromatin literature - including all six Milne-laboratory MLL-AF4 papers, Guenther 2008, Prange 2017, Chen 2015, Ropa 2018 and 2020, Chu 2020, Valerio 2017, Mishra 2014, Sakamoto 2014 and three menin-inhibitor chromatin papers - returns zero occurrences in every document.

    What this evidence cannot show

    **Nineteen full texts is a purposive sample, not a census**, biased towards papers this graph already cites and towards open-access content reachable through PMC. Supplementary files and figure source data outside the retrieved article body were not searched. Text was stripped tags-first and entities-second, and document lengths were checked to exclude body-less records. The same sweep did find H3.3 in two documents, so it is capable of returning a positive.

    Evidence that argues against this (1)
    • The same sweep's H3.3 result shows that a variant can appear in a KMT2A paper as an engineered tool without being measured; an equivalent H2A.Z reagent use - for instance an H2A.Z-depletion or ANP32E-overexpression control - would be equally invisible to an abstract query and could have been missed in the papers outside this nineteen. Sparbier CE; Gillespie A; Gomez J; Kumari N; Motazedian A; Chan KL; et al. 2023
Connected to