Oncofusion

start / graph / gene-tet2

Proteinstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q79-cxxc-readers

TET2 is NOT a CXXC protein - its CXXC domain is a separate gene, and it is DEPLETED in KMT2A-rearranged AML rather than enriched

also called TET2 · ten-eleven translocation 2 · KIAA1546 · Methylcytosine dioxygenase TET2

TET2 is the gene people reach for when they want to argue that a leukaemia's DNA-methylation landscape has been altered by a second mutation. But TET2 does not actually carry the CG-reading module that the KMT2A fusion uses - that module was split off onto a different gene long ago - and in two large adult studies TET2 mutations are LESS common in KMT2A-rearranged leukaemia than in other kinds.

The same thing, for a scientist

**TET2 (Q6N021, 2,002 aa) carries no CXXC domain.** It is absent from the reviewed human PF02008 set, from InterPro IPR002857 and from the UniProt `CXXC-type` zinc-finger feature query - all three of which return the same twelve proteins, and TET2 is in none of them [E1]. The reason is evolutionary: the ancestral *TET2* gene underwent a chromosomal inversion that separated the CXXC domain from the catalytic domain, and the orphan CXXC is now **CXXC4/IDAX**, which binds unmethylated CpG, localises to CpG islands and interacts directly with the TET2 catalytic domain in trans [E2]. **This entry exists to correct a premise, including one in the dispatch brief for this beat**: TET2 is in the CpG pathway, it is not in the CXXC family, and it does not compete for the KMT2A address as a reader. The co-mutation argument was then tested on its own terms and does not survive either. In the index disease TET2 does not appear at all: **0 occurrences of the string TET2 in 71,489 characters of the full text** of the whole-genome study of infant KMT2A-rearranged ALL (22 WGS discovery cases), which is 0/22 and bounds the rate below **13.6%** by the rule of three [E3]. In adult KMT2A-rearranged AML two independent 2026 cohorts agree closely on the point estimate and one supplies the comparator: **50/490 = 10.2% (Wilson 7.8-13.2)** in the ECLA series [E4], and **10.1% of 521 KMT2A-altered cases against 16.4% of 3,342 KMT2A-wild-type cases, p = 0.0002** in a 3,863-case FoundationOne Heme series - that is **significant DEPLETION, odds ratio 0.58** [E5]. In DepMap the mean gene effect is **positive** (+0.111 across 1,208 lines; +0.129 in the KMT2A-rearranged arm), i.e. knocking TET2 out very slightly helps, consistent with a tumour suppressor and inconsistent with a dependency [E6].

What it looks like
loading AF-Q6N021-F1…
AlphaFold AF-Q6N021-F1 ↗
colourshape

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

very high (90+)confident (70–90)low (50–70)very low (<50)
I have never read one of these pictures before

A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.

Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.

Anything in yellow sticks is a bound small molecule, which is usually the drug.

Go deeper
The correction, stated plainly

A dozen human proteins read unmethylated CpG through a CXXC zinc finger. TET2 is not one of them, and no amount of pathway proximity makes it one. The three orthogonal database queries are in ana-cxxc-domain-family-census-human; all three return CXXC1, CXXC4, CXXC5, DNMT1, FBXL19, KDM2A, KDM2B, KMT2A, KMT2B, MBD1, TET1 and TET3 1.

TET2 still reaches CpG islands - through CXXC4/IDAX and through CXXC5, in trans 27. That is a real mechanism and it is written up in gene-cxxc4 and gene-cxxc5. It is a different claim from 'TET2 is a CXXC reader', and the difference matters for a selectivity argument: a molecule that blocks a CXXC-DNA interface cannot hit TET2, because TET2 has no such interface.

The co-mutation test, with its null

The hypothesis worth testing was: if TET2 loss changes the methylome the KMT2A CXXC reads, then TET2 mutation should be enriched alongside KMT2A rearrangement. The null is the mutation rate of TET2 in the comparator leukaemia population, which is 16.4% (548/3,342) in KMT2A-wild-type AML 5. Against that null:

populationTET2 mutatedrate95% Wilson
infant KMT2A-rearranged ALL, WGS0/220%0-14.9; rule of three <13.6%
adult KMT2A-rearranged AML, ECLA50/49010.2%7.8-13.2
adult KMT2A-altered AML, FoundationOne Heme53/52110.2%7.9-13.1
adult KMT2A-wild-type AML, same assay548/3,34216.4%15.2-17.7

The two independent adult KMT2A-rearranged estimates are 10.2% and 10.2%, and both sit below the wild-type rate; the same-assay comparison is significant in the direction of depletion (p = 0.0002, odds ratio 0.58) 5. The prediction is not merely unsupported, the sign is wrong. If anything, a KMT2A rearrangement makes a TET2 mutation less likely to be found alongside it - which is what one expects when two lesions are partly redundant routes to the same phenotype, and which is a mild point in favour of the fusion supplying its own address rather than depending on a co-mutated methylome.

The honest limits: the ECLA count of 50 is back-calculated from a reported 10.2% of 490 and is therefore +/- 1 case; the FoundationOne assay is a targeted panel, so both arms are ascertained the same way but neither is whole-genome; and the infant arm is a different disease, a different lineage and n = 22.

What the infant full-text search actually covered

The index whole-genome study of infant MLL-rearranged ALL was retrieved as full text and searched. fullTextXML returned HTTP 500; efetch db=pmc on PMC4553269 returned a well-formed document with a body, 71,489 characters after tag-stripping-then-unescaping 3. Term counts in that text:

termhits
TET20
TET10
TET31
DNMT1, DNMT3A, KDM2B, CXXC4, CXXC5, IDAX0 each
methylation0

The single TET3 hit is in the non-infant arm: somatic mutations were found in eleven epigenetic regulatory genes - CHD4, SETD2, CREBBP, L3MBTL3, ATR, KAT6A, KDM6A, NSD1, PARP8, SUPT3H and TET3 - in 9/20 (45%, Wilson 25.8-65.8) non-infant MLL-R leukaemias against 3/22 (14%, Wilson 4.7-33.3) of infant MLL-R ALLs, P = 0.04 3. So exactly one CXXC-family gene has ever been found somatically mutated in a KMT2A-rearranged leukaemia cohort, and it is TET3, and it is in the older children.

Bounding, per rule 80: this is a full-text search of one paper, not of the field. Supplementary tables 42 and 44, where the per-case gene assignments live, were not parsed. A gene named only in a supplementary table would be invisible to this count, and the epigenetic-regulator analysis in that paper explicitly runs against a 633-gene list held in a supplementary table.

What would change this conclusion

Three results would reopen it:

  1. A methylome, not a mutation list. Every figure above counts mutations. A KMT2A-rearranged leukaemia could have an altered 5mC landscape with no TET2 mutation at all - through TET2 expression, through IDH1/IDH2 (which this graph does not hold either), through age, or through the cell of origin. This graph already carries mechanism-kmt2a-r-dna-methylation-signature; the co-mutation null says nothing about it.
  2. An infant cohort with TET2 sequenced and reported. 0/22 bounds the rate below 13.6% and no lower. A 200-case infant series would settle it.
  3. A functional test. Nobody has depleted TET2 in a KMT2A-rearranged line and measured fusion occupancy. The depletion finding above is epidemiological and could reflect lineage, age or assay rather than biology.
Facts and figures
uniprotQ6N021
gene symbolTET2
protein nameMethylcytosine dioxygenase TET2
length AA2002
alphafold IDAF-Q6N021-F1
accession resolutionUniProt REST gene_exact:TET2 AND organism_id:9606 AND reviewed:true, run 2026-09-21; one row; primary gene name read back as TET2 and length 2002 from the same fetched record.
carries cxxc domainfalse
why no cxxcThe ancestral TET2 gene underwent a chromosomal inversion separating its CXXC domain from its catalytic domain; the orphan CXXC is now the separate gene CXXC4/IDAX, which binds unmethylated CpG and interacts with the TET2 catalytic domain in trans.
domains noteDeliberately no residue coordinates. The project's join rule is that a null start makes a domain invisible; the honest alternative is to state that the boundary is unknown to this entry, which is what is done here.
classificationNOT a CXXC-family protein; a 5mC dioxygenase that reaches CpG islands in trans via CXXC4/IDAX and CXXC5
shown for fusionNO for any mechanism; for co-mutation the answer is a measured DEPLETION in adult KMT2A-rearranged AML (10.2% vs 16.4%, p = 0.0002) and 0/22 in the infant KMT2A-rearranged ALL whole-genome series
cell type evidenceadult AML cohorts (two, 2026); infant KMT2A-rearranged ALL whole genomes (n = 22, zero); DepMap leukaemia lines. No functional experiment in any KMT2A-rearranged system.
subnuclear localizationNot established. UniProt records Nucleus and Chromosome (both ECO:0000269) with no subnuclear term and no isoform-scoped location. Never imaged in any KMT2A-fusion cell.
cell typeAdult AML patients (two 2026 cohorts, n = 490 and n = 3,863); infant KMT2A-rearranged ALL whole genomes (n = 22, zero TET2 events); DepMap leukaemia cell lines. No functional experiment in any KMT2A-rearranged cell of any type.
domains
[
  {
    "name": "Tet_JBP (2OG-Fe(II) dioxygenase)",
    "pfam": "PF12851",
    "boundary_note": "Pfam cross-reference only. UniProt's feature table gives NO Domain feature for TET2, so no residue range is asserted and no viewer lane is created - an integer would be invented rather than read."
  }
]
comutation
{
  "infant_kmt2a_r_all_wgs": {
    "k": 0,
    "n": 22,
    "pct": 0,
    "wilson_95": [
      0,
      14.9
    ],
    "rule_of_three_upper_pct": 13.6,
    "source_pmid": "25730765"
  },
  "adult_kmt2a_r_aml_ecla": {
    "k": 50,
    "n": 490,
    "pct": 10.2,
    "wilson_95": [
      7.8,
      13.2
    ],
    "k_note": "back-calculated from a reported 10.2%",
    "source_pmid": "42187108"
  },
  "adult_kmt2a_altered_aml_f1h": {
    "k": 53,
    "n": 521,
    "pct": 10.1,
    "wilson_95": [
      7.9,
      13.1
    ],
    "source_pmid": "41514668"
  },
  "adult_kmt2a_wt_aml_f1h_NULL": {
    "k": 548,
    "n": 3342,
    "pct": 16.4,
    "wilson_95": [
      15.2,
      17.7
    ],
    "source_pmid": "41514668"
  },
  "test": "Fisher exact on the same-assay pair: odds ratio 0.577, p = 1.6e-4, DEPLETED in KMT2A-altered AML"
}
localization
{
  "subcellular": "nucleus and chromosome, both ECO:0000269 and both UNSCOPED - TET2 has no isoform-scoped location comment, which is the opposite of TET1",
  "subnuclear": "not established in UniProt",
  "localization_shown_for_fusion": "NO - nobody has looked in any KMT2A-fusion system"
}
essentiality
{
  "depmap_scan": "analyses/depmap-kmt2a-selective-scan-full",
  "difference_kmt2a_r_minus_other": 0.0233,
  "mean_kmt2a_rearranged": 0.1287,
  "mean_other_leukemia": 0.1054,
  "pan_essentiality_all_lines": 0.1114,
  "n_all_lines": 1208,
  "rank_of_17728": 12109,
  "percentile": 68.304,
  "release_note": "Open Targets dataVersion 26.06, read 2026-09-21"
}
How this could bite us back
  • A therapeutic argument that runs 'TET2 co-mutation reshapes the methylome and therefore the fusion's address is not fixed' has no support in either of the two cohorts that can test it, and the sign runs the other way. Do not build a stratification hypothesis on it without first re-running the analysis on IDH2.
  • Because TET2 has no CXXC domain, a CXXC-directed molecule cannot reach it - which is a rare piece of good news for the selectivity gate in the phase-2 CXXC dossier, and it should be recorded there.
Still open
  1. IDH2 is significantly ENRICHED in KMT2A-rearranged AML (16.0% vs 10.4%, p = 0.0004 (recomputed; the paper's ABSTRACT says p < 0.0001, its Table 2 says 0.0004, and an independent Fisher exact on 83/521 vs 348/3,342 gives OR 1.630, p = 4.1e-4 - corrected cycle 45)) in the same series that finds TET2 depleted, and IDH2 mutation inhibits TET2 catalytically through 2-hydroxyglutarate. This graph holds no IDH1 or IDH2 entry at all. Is the methylome-co-mutation argument actually an IDH argument?
  2. Is TET2 mutation depleted in KMT2A-rearranged AML because the two lesions are redundant, because of lineage, or because of age? The FoundationOne series has a median age of 62 and TET2 mutation is strongly age-associated through clonal haematopoiesis.
  3. Has anyone depleted TET2 in a KMT2A-rearranged line and measured fusion occupancy or 5mC at fusion target CpG islands? Not found in this beat's searches.
  4. Does CXXC4 or CXXC5 expression, rather than TET2 mutation, vary with KMT2A rearrangement? That is the in-trans version of the same question and it is answerable from existing expression data.
Every source behind this entry (7)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    TET2 (Q6N021, 2,002 aa) carries no CXXC-type zinc finger: it is absent from the reviewed human Pfam PF02008 set, from InterPro IPR002857 and from the UniProt CXXC-type zinc-finger feature query, all three of which return the same twelve proteins.

    What this evidence cannot show

    Fetched 2026-09-21; accession resolved by gene_exact:TET2 AND organism_id:9606 AND reviewed:true with gene_primary read back as TET2 and length 2002 from the same record. A domain census measures annotation, so a divergent CXXC too weak for the Pfam HMM would be missed - but the positive explanation (the domain is on a different gene) removes that worry here.

    Evidence that argues against this (1)
    • TET2 does reach unmethylated CpG islands, through direct interaction of its catalytic domain with the CXXC protein IDAX/CXXC4 - so 'TET2 has no CXXC domain' and 'TET2 has no CpG-island address' are different statements and only the first is true. Ko M et al. 2013
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    IDAX/CXXC4 was originally encoded within an ancestral TET2 gene that underwent a chromosomal inversion separating the CXXC domain from the catalytic domain; IDAX binds unmethylated CpG, localises to promoters and CpG islands, and interacts directly with the TET2 catalytic domain.

    PMID 23563267DOI
    What this evidence cannot show

    Mouse embryonic stem cells and the human monocytic line U937. The inversion is a synteny inference. The functional consequence reported is TET2 protein DOWN-regulation via caspase activation, which is an unusual mechanism and has not been widely replicated.

    Evidence that argues against this (1)
    • Liu et al. report that Tet3 has an alternative transcript incorporating the adjacent Cxxc10-1 CXXC in cis, so 'CXXC in trans' is a TET2-specific arrangement rather than a TET-family rule. Liu N et al. 2013
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The string TET2 occurs zero times in 71,489 characters of the full text of the whole-genome study of infant MLL-rearranged ALL (22 WGS discovery cases), and the only CXXC-family gene named anywhere in it is TET3, in the non-infant arm, among eleven epigenetic regulatory genes mutated in 9/20 non-infant versus 3/22 infant cases (P = 0.04).

    PMID 25730765DOI
    What this evidence cannot show

    A full-text search of one paper bounds a claim about that paper, not about the field. Europe PMC fullTextXML returned HTTP 500; the text came from efetch db=pmc on PMC4553269 and was tag-stripped before entity-unescaping. Supplementary tables, including the 633-gene epigenetic-regulator list and the per-case assignments, were NOT parsed, so a gene named only in a supplement is invisible to this count. n = 22 for the infant arm; 0/22 bounds the TET2 rate below 13.6% by the rule of three and no lower.

    Evidence that argues against this (1)
    • TET2 mutations are found in about one in ten adult KMT2A-rearranged AML cases, so the infant zero is a statement about infant B-lineage disease and not about KMT2A-rearranged leukaemia generally. Pan J et al. 2026
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In 490 patients with KMT2A-rearranged AML from the East China Leukemia Alliance, TET2 was the third most frequent co-occurring gene mutation at 10.2%, after KRAS (20.1%) and NRAS (19.0%).

    PMID 42187108DOI
    What this evidence cannot show

    Adult and older-adult AML, single regional alliance, 2013-2025, retrospective. The count of 50 is back-calculated from the reported percentage and is accurate to about one case. No KMT2A-wild-type comparator arm is reported, so this cohort alone cannot say whether 10.2% is high or low. Wilson 95% 7.8-13.2%.

    Evidence that argues against this (1)
    • A same-assay comparison in a larger series puts TET2 at 16.4% in KMT2A-wild-type AML, which makes the 10.2% a depletion rather than an enrichment. Batayneh O et al. 2026
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 3,863 AML cases profiled by one hybrid-capture assay, TET2 alteration was significantly LESS frequent in KMT2A-rearranged cases than in KMT2A-wild-type cases: 10.1% of 521 versus 16.4% of the remainder, p = 0.0002.

    PMID 41514668DOI
    What this evidence cannot show

    A commercial targeted panel on peripheral blood, so both arms share the same ascertainment but neither is whole-genome and germline filtering is inferred. Median age 62 - this is adult AML and says nothing directly about infant ALL. 43.1% of the KMT2A-altered cases are duplications rather than fusions, which dilutes the rearranged arm with a biologically different lesion. Recomputed here from the published percentages: 53/521 versus 548/3,342, Fisher odds ratio 0.577, p = 1.6e-4, matching the reported p.

    Evidence that argues against this (1)
    • The same assay finds IDH2 significantly ENRICHED in KMT2A-rearranged AML (16.0% versus 10.4%, p = 0.0004 (recomputed; the paper's ABSTRACT says p < 0.0001, its Table 2 says 0.0004, and an independent Fisher exact on 83/521 vs 348/3,342 gives OR 1.630, p = 4.1e-4 - corrected cycle 45)), and IDH2 mutation inhibits TET2 through 2-hydroxyglutarate - so a methylome-altering co-mutation IS enriched, just not the one predicted, and this graph holds no IDH1 or IDH2 entry. Batayneh O et al. 2026
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    TET2 shows no dependency in KMT2A-rearranged cell lines and its mean gene effect is positive: +0.129 in the KMT2A-rearranged arm, +0.105 in other leukaemia lines, +0.111 across 1,208 lines, rank 12,109 of 17,728.

    Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap CRISPR Chronos via the Open Targets Platform GraphQL API) · Oncofusion knowledge-graph project · analyses/depmap-kmt2a-selective-scan-full/results.json, generated 2026-09-19; Open Targets dataVersion year 26 month 06, re-read from the API on 2026-09-21 2026
    What this evidence cannot show

    The scan compares 11 Cellosaurus-annotated KMT2A-rearranged lines against 43 other leukaemia lines; none of the 11 is infant-derived and the arm is myeloid-enriched, so a difference mixes fusion status with lineage. Both the difference and the pan-essentiality across all 1,208 lines are quoted, because a gene required everywhere is not a selective dependency. Open Targets dataVersion 26.06; DepMap's own release identifier is not returned by that endpoint.

    Evidence that argues against this (1)
    • A positive gene effect in established cell lines is weak evidence about a tumour suppressor whose loss acts at initiation in a stem cell, and none of the 11 lines is infant-derived. Oncofusion knowledge-graph project 2026
  7. E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    CXXC5 recruits TET2 to maintain hypomethylation of a subset of CpG islands, so TET2 reaches unmethylated CpG through more than one CXXC partner.

    PMID 28416650DOI
    What this evidence cannot show

    Murine plasmacytoid dendritic cells and viral infection models; the locus analysed in detail is the Irf7 CpG island. Nothing haematopoietic-malignant and nothing in a KMT2A-rearranged system.

    Evidence that argues against this (1)
    • Rinf/CXXC5 in embryonic stem cells is reported to regulate Tet1 and Tet2 at the level of TRANSCRIPTION, binding their promoters and enhancers, which is a different mechanism from escorting TET2 to a target CpG island. Ravichandran M et al. 2019
Connected to