Oncofusion

start / graph / gene-zeb2

Proteinmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q-scan-hits

ZEB2 — an EMT transcription factor that turns out to be a leukemia dependency and a guardian of lineage fidelity

also called ZEB2 · SIP1 · ZFHX1B · Smad-interacting protein 1 · O60315

ZEB2 is best known for helping solid-tumour cells change shape and spread, but in blood it does something else: it stops immature blood cells from switching on the programs of the mature cell types they have not become yet. A scan run by this project found KMT2A-fusion leukemia cells need it, and independent laboratories had already shown that leukemia cells of several kinds do.

The same thing, for a scientist

ZEB2 (UniProt O60315, 1,214 aa) is a zinc-finger E-box-binding homeobox transcriptional repressor. It entered this graph through the coordinator's unbiased DepMap scan at rank 11 of 4,589 for KMT2A-rearranged selectivity (difference -0.464; -1.205 in 11 rearranged lines vs -0.740 in 43 others; pan-essentiality -0.358). The literature independently supports a leukemia dependency and, importantly, does so **with an MLL-AF9 model in the validating experiment**: Li et al. identified ZEB2 from genome-scale shRNA screens in 17 AML lines and validated it in a syngeneic murine MLL-AF9 AML model, showing that ZEB2 depletion impairs proliferation and causes aberrant differentiation by de-repressing myeloid differentiation genes [E2]. Wang et al. showed ZEB2 acts predominantly as a repressor **restraining mature haematopoietic lineage programs from being expressed too early in HSPCs**, and that Zeb2 inactivation prolongs survival in secondary MLL-AF9 AML models [E3]. Where the scan and the literature diverge is selectivity: the published work frames ZEB2 as a broad AML dependency, also implicated downstream of AML1-ETO and PML-RARα [E4], which the scan's KMT2A-selective framing does not predict.

What it looks like
loading AF-O60315-F1…
AlphaFold AF-O60315-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
How this gene got into the graph, and what that does and does not license

ZEB2 was surfaced by a scan, not by reading. The coordinator ranked 4,589 protein-coding genes by the difference in mean DepMap Chronos gene effect between 11 KMT2A-rearranged and 43 other leukemia lines, with no candidate list; ZEB2 came eleventh 1.

That licenses a literature search and nothing more. In particular it does not license the claim that ZEB2 is a KMT2A-selective dependency, and on this point the literature actively disagrees with the scan — see 'Where the scan and the literature part company' below.

What the scan did get right is that there is something here. ZEB2 was not a guess: two independent groups had already made it a validated leukemia dependency, and one of them validated it in an MLL-AF9 model.

What ZEB2 does in blood1 deeper

The haematopoietic function is the interesting part and it is not the EMT function.

ZEB2 is a repressor of premature lineage programs. Wang et al., using conditional loss-of-function and bone-marrow reconstitution, report that ZEB2 acts predominantly as a transcriptional repressor 'involved in restraining mature hematopoietic lineage gene expression programs from being expressed too early in HSPCs', with ZEB1 fine-tuning the same repression to ensure lineage fidelity. Inducible inactivation of both Zeb1 and Zeb2 causes acute bone-marrow failure at the short-term-HSC-to-MPP1 transition, with increased stem-cell numbers and decreased differentiation into every lineage 3.

In leukemia that repression is what the cells are using. Li et al. found ZEB2 depletion impaired proliferation of both human and murine AML cells and caused aberrant differentiation, and showed mechanistically that ZEB2 transcriptionally represses genes regulating myeloid differentiation, including cell adhesion and migration genes. They also found epigenetic silencing of the miR-200 family affects ZEB2 expression 2.

So ZEB2 is a differentiation brake. That is the same category of function as the KMT2A fusion's own central effect, and it is why a ZEB2 dependency in these cells is mechanistically plausible rather than merely statistical.

ZEB2 can also be an oncogene in its own right

Zeb2 overexpression in mice causes spontaneous immature T-cell lymphoblastic leukemia; pre-leukemic Zeb2-overexpressing thymocytes are delayed at beta-selection through aberrant IL-7-receptor signalling and, unlike Lmo2-overexpressing thymocytes, acquire no self-renewal properties, so ZEB2 and LMO2 transform immature T-cell progenitors by distinct mechanisms 5.

This matters for interpreting a dependency: a gene that is oncogenic when overexpressed and required when present is behaving like a lineage-determining transcription factor, which is a class that dependency screens hit often and drugs reach rarely.

Where the scan and the literature part company

The scan's claim is selectivity: ZEB2 is more required in KMT2A-rearranged lines than in other leukemia lines. The literature's claim is breadth: ZEB2 is an AML dependency found by screening 17 AML lines against 199 lines of other cancer types 2, and Zeb2 knockdown in vivo also delays leukemia onset in t(8;21) and t(15;17) models, where Zeb2 is upregulated by the AML1-ETO and PML-RARα oncogenes 4.

These are not formally contradictory — a gene can be required across AML and more required in one subtype — but the published evidence gives no reason to expect KMT2A specificity, and the two other fusion oncoproteins that upregulate ZEB2 are precisely the ones this graph holds up as the curable comparators (fusion-runx1-runx1t1, fusion-pml-rara). Read plainly, ZEB2 looks like a general myeloid-leukemia differentiation brake that the scan happened to catch in a rearranged panel, not a fusion-specific dependency. The scan is a weaker piece of evidence than the papers, so on this point the papers win and the entry says so.

Why a lineage-fidelity factor scoring highly is worth taking seriously anyway

This graph already holds that lineage switch is the characteristic escape from CD19-directed therapy in KMT2A-rearranged leukemia (risk-lineage-switch-cd19-directed-therapy, mechanism-lineage-switch-kmt2a). A gene whose documented job is to enforce lineage fidelity in haematopoietic stem and progenitor cells 3, surfacing near the top of an unbiased dependency scan in exactly this disease, is a coincidence worth a hypothesis rather than a shrug.

The hypothesis is written separately as hyp-zeb2-sets-lineage-switch-permissiveness, with its falsifiers. It is speculative and is labelled as such. Nothing read in this cycle connects ZEB2 to a documented lineage switch in a patient, and the absence of that link is the main reason the hypothesis stays speculative.

Druggability: poor, and honestly so

ZEB2 is a 1,214-residue zinc-finger/homeodomain transcriptional repressor. No ligand, pocket, structure or chemical probe is asserted here because none was verified in this cycle. The protein class has the worst record in oncology for direct small-molecule targeting, which the scan's own author flagged before this entry was written.

The indirect handles that exist in the literature are the miR-200 family, whose epigenetic silencing raises ZEB2 2, and the ZEB1/ZEB2 interplay 3. Neither is a therapeutic strategy today. The honest position is that ZEB2 is a good piece of biology and a bad target, and this entry should not be used to argue otherwise.

Facts and figures
uniprotO60315
gene symbolZEB2
length AA1214
alphafold IDAF-O60315-F1
cytoband2q22.3
accession resolutionResolved by fetching https://rest.uniprot.org/uniprotkb/O60315.json on 2026-09-18 and reading back the primary gene name ZEB2 (synonyms listed: KIAA0569, SIP1, ZFHX1B, ZFX1B); UniProtKB ID ZEB2_HUMAN; sequence length 1,214 aa, matching the value stored here. Not resolved by a gene_exact: search, which matches synonyms.
domains noteNo domain coordinates asserted. ZEB2's zinc-finger clusters, homeodomain-like region and SMAD-binding domain were not resolved from a primary source in this cycle, and the graph's rule is to leave a boundary out rather than guess it.
rolezinc-finger E-box-binding transcriptional repressor; in haematopoiesis it restrains premature expression of mature lineage programs in HSPCs and enforces lineage fidelity; in AML it represses myeloid differentiation genes
published dependency scopeAML broadly (shRNA screens across 17 AML lines vs 199 other-cancer lines, PMID 27756750), with in vivo validation in MLL-AF9 and separate in vivo evidence in t(8;21) and t(15;17) models (PMID 33327558). The published scope is wider than the scan's KMT2A-selective framing, and where they differ the literature is the stronger evidence.
druggabilityPoor. Large zinc-finger transcriptional repressor; no ligand, pocket, structure or chemical probe verified in this cycle, and none is asserted. Indirect handles in the literature are the miR-200 family and the ZEB1/ZEB2 interplay, neither of which is a therapeutic strategy today.
entered graph viaana-depmap-kmt2a-selective-scan (cycle 25), rank 11 of 4,589; no prior entry in this graph
subnuclear localizationNucleoli and nucleoplasm by antibody immunofluorescence (HPA v25.1, Supported). No nucleolar experiment or function published; database-grade only. Explicitly NOT relocalised by Pc2-mediated SUMOylation (PMID 16061479).
shown for fusionNO - nobody has localised ZEB2 in any KMT2A-rearranged cell, in any lineage; the bounded query returned 0. The graph's KMT2A-selective dependency call for ZEB2 (ana-depmap-kmt2a-selective-scan) is a fitness measurement and says nothing about where the protein is.
domains
[]
essentiality
{
  "depmap_common_essential": false,
  "chronos_mean_all_lines": -0.3581,
  "chronos_mean_kmt2a_rearranged": -1.2046,
  "chronos_mean_other_leukemia": -0.7404,
  "kmt2a_selectivity_difference": -0.4642,
  "scan_rank": 11,
  "scan_n_genes": 4589,
  "scan_percentile": 0.24,
  "empirical_p_vs_scan": 0.002615,
  "n_kmt2a_rearranged_lines": 11,
  "n_other_leukemia_lines": 43,
  "n_all_lines": 1208,
  "mouse_ko": "Conditional Zeb1/Zeb2 double inactivation causes acute bone-marrow failure at the ST-HSC to MPP1 transition with decreased differentiation into all lineages (PMID 34550965); single-gene germline knockout phenotype not established in this entry"
}
localization
{
  "subcellular": "Nucleus (UniProt O60315, ECO:0000269 PubMed:16061479, 9853615) and Chromosome (ECO:0000269 PubMed:20516212). HPA immunofluorescence gives main locations Nucleoplasm AND NUCLEOLI with Plasma membrane additional, reliability Supported.",
  "subnuclear": "NUCLEOLI plus nucleoplasm, on antibody immunofluorescence (HPA v25.1, main locations Nucleoplasm and Nucleoli, reliability Supported). NO nucleolar function or nucleolar experiment exists in the primary literature for ZEB2 - the bounded subnuclear query returned 14 records, none of which is a ZEB2 nucleolar study. This is a database observation awaiting an experiment, and it is recorded at that grade.",
  "how_shown": "antibody immunofluorescence (HPA); yeast nuclear-transportation-trap selection plus COS-7 immunofluorescence (PMID 9853615); chromatin association (PMID 20516212)",
  "rule_91_note": "UniProt's nuclear call for ZEB2 rests in part on PMID 9853615, which is a METHODS paper - a yeast selection system for genes encoding nuclear-targeted proteins, in which 75% of the unknown proteins examined localised to the nucleus in COS-7 cells. SIP1/ZEB2 was one clone recovered by that trap. It is real evidence but it is a screen readout in monkey kidney cells, not a dedicated ZEB2 localisation study, and nothing in the UniProt row says so.",
  "an_explicit_negative_worth_keeping": "PMID 16061479 shows SIP1/ZEB2 is SUMOylated by the polycomb protein Pc2 at Lys391 and Lys866 - and states plainly that sumoylation DOES NOT AFFECT its subcellular localisation, only its transcriptional activity. The obvious inference, that a Pc2 substrate is drawn into polycomb bodies, was tested and is false. Recorded so nobody re-derives it.",
  "tissue_of_evidence": "COS-7 and epithelial EMT systems for the localisation work; HPA's non-haematopoietic immunofluorescence panel. Nothing haematopoietic has been imaged.",
  "node": "localization-nucleolus-kmt2a-fusion"
}
cell type
{
  "haematopoietic_answer": "Broad across haematopoiesis, and now with a B-LINEAGE LEUKAEMIC anchor: in 992 Philadelphia-negative adult B-ALL patients, ZEB2 p.H1038R defines subtypes - one co-occurring with CEBP alterations (n=18) and one isolated ZEB2alt subtype (n=15) (PMID 42306066). HPA Blood Atlas: low immune cell specificity, detected in many.",
  "why_this_matters_here": "The entry's existing haematopoietic framing is HSPC lineage fidelity and AML dependency. The B-ALL subtype result puts a recurrent ZEB2 POINT MUTATION in the index disease's lineage, which is a different and stronger kind of cell-type evidence than a dependency rank.",
  "other_lineages_with_dedicated_results": "ZEB2 is a conserved regulator in monocyte/macrophage and conventional dendritic cell specification and in CD8 effector versus memory differentiation; those literatures are large and were not enumerated here.",
  "non_haematopoietic_baseline": "HPA RNA tissue-enhanced in brain (176.9 nTPM), which is the highest single tissue - a reminder that ZEB2's largest expression domain is not blood."
}
cell state
{
  "state": "Lineage-fidelity restraint in HSPCs and repression of myeloid differentiation genes in AML, both already recorded in data.role. The new state-adjacent fact is that ZEB2 p.H1038R is a RECURRENT SOMATIC ALTERATION defining B-ALL subtypes rather than a state the cell passes through (PMID 42306066).",
  "bounded": "No ZEB2 cell-state experiment in KMT2A::AFF1 specifically."
}
localization bounded absence
{
  "date": "2026-09-21",
  "tool": "NCBI E-utilities esearch, db=pubmed",
  "shape": "gene clause AND localisation clause AND KMT2A/MLL-fusion clause, one [tiab] tag per term (brief rule 94), every cytogenetic term quoted (brief rule 95)",
  "querytranslation_fusion_clause": "\"KMT2A\"[Title/Abstract] OR \"MLL\"[Title/Abstract] OR \"MLL1\"[Title/Abstract] OR \"MLL-AF4\"[Title/Abstract] OR \"MLL-AF9\"[Title/Abstract] OR \"MLL-ENL\"[Title/Abstract] OR \"KMT2A-rearranged\"[Title/Abstract] OR \"MLL-rearranged\"[Title/Abstract] OR \"t 4 11\"[Title/Abstract] OR \"t 9 11\"[Title/Abstract] OR \"MLL fusion\"[Title/Abstract]",
  "count": 0,
  "controls": {
    "fusion_clause_alone": 8743,
    "gene_clause_alone": 2105
  },
  "reading": "Counted in TITLES AND ABSTRACTS only; supporting figures were not searched (brief rule 80)."
}
How this could bite us back
  • The essential-function problem is severe and documented: combined Zeb1/Zeb2 loss causes acute bone-marrow failure with loss of differentiation into every lineage [E3]. An agent that suppressed this axis would reproduce a marrow-failure phenotype in a patient whose marrow is already failing.
  • ZEB2 is a large zinc-finger transcriptional repressor with no known pocket. A selective dependency in a protein class with essentially no direct-ligand precedent is a finding, not a plan.
  • ZEB2 is oncogenic when overexpressed [E5] and required when present, so a partial knockdown has no obvious safe direction. Dose-response for such a factor is not monotonic in the way a kinase inhibitor's is.
  • If ZEB2 is what enforces lineage fidelity, then suppressing it could in principle *promote* the lineage plasticity that already defeats CD19-directed therapy in this disease. The therapeutic direction is not obviously the one the dependency implies.
  • The dependency is shared with AML1-ETO and PML-RARA leukemias [E4], both of which are curable by other means. A target common to the curable and the incurable is unlikely to be what makes the incurable one incurable.
Still open
  1. Is ZEB2 dependence actually greater in KMT2A-rearranged cells than in other AML, or did the scan simply rediscover a general AML dependency in a panel that happens to be rearranged? A myeloid-matched comparison would settle it and this project already has the lineage-conditioning machinery.
  2. Does ZEB2 expression or dependence change across a lineage switch? Paired diagnosis/switch samples from KMT2A-rearranged patients who relapsed as myeloid after CD19-directed therapy would be the direct test, and `exp-diagnostic-subclone-archaeology-lineage-switch` already contemplates that sample set.
  3. Is ZEB2 required in KMT2A::AFF1 B-lineage cells, or only in myeloid ones? Every published functional experiment is myeloid or T-lineage, and infant KMT2A-rearranged disease is overwhelmingly B-lineage. This is the largest single gap in the entry.
  4. Does menin inhibition alter ZEB2 expression? If the fusion program sustains ZEB2, the differentiation induced by a menin inhibitor and the differentiation induced by ZEB2 loss may be the same event reached two ways.
  5. HPA scores ZEB2 in NUCLEOLI at Supported reliability and no paper has ever asked why. Is there a nucleolar pool of ZEB2, and does the recurrent B-ALL p.H1038R allele change it? This is the cheapest unexplored subnuclear lead in this tranche.
Every source behind this entry (9)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    In this project's unbiased scan of 4,589 protein-coding genes, ZEB2 ranks 11 by KMT2A-rearranged selectivity (difference -0.464; mean DepMap Chronos -1.205 across 11 Cellosaurus-annotated KMT2A-rearranged leukemia lines versus -0.740 across 43 other leukemia lines) with a pan-essentiality of -0.358 across 1,208 lines, and had no entry in this graph before this cycle.

    DepMap CRISPR (Chronos) gene effect via the Open Targets Platform GraphQL API, 4,589 HGNC protein-coding genes split by Cellosaurus KMT2A gene-fusion annotation; analyses/depmap-kmt2a-selective-scan/ · Broad Institute DepMap; Cellosaurus (SIB); HGNC; Open Targets Platform; analysis run by this project · Databases and derived analysis 2026
    What this evidence cannot show

    This project's own unreviewed computation, read from the raw results file. 11 lines, none infant-derived; 24% of the protein-coding genome; the comparison does not condition on lineage or on myeloid-versus-lymphoid composition, which matters because the published ZEB2 dependency is a myeloid one. Not methodologically independent of ana-depmap-kmt2a-rearranged-specificity.

    Evidence that argues against this (1)
    • The published evidence frames ZEB2 as a dependency of AML broadly, found by comparing 17 AML lines against 199 lines of other cancer types — a comparison that says nothing about KMT2A selectivity and gives no reason to expect it. Li H et al. 2017
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Genome-scale shRNA screens in 17 AML cell lines analysed against parallel screens in 199 cell lines of other cancer types identified 353 genes specifically required for AML proliferation; a secondary screen in a syngeneic murine AML model driven by the MLL-AF9 fusion, integrated with expression data, identified ZEB2 as a novel AML dependency; ZEB2 depletion impaired proliferation of human and mouse AML cells and caused aberrant differentiation of human AML cells; ZEB2 transcriptionally represses genes regulating myeloid differentiation including cell adhesion and migration genes; and epigenetic silencing of the miR-200 family affects ZEB2 expression.

    The EMT regulator ZEB2 is a novel dependency of human and murine acute myeloid leukemia · Li H, Mar BG, Zhang H, Puram RV, Vazquez F, Weir BA, Hahn WC, Ebert B, Pellman D · Blood 129:497-508 2017
    PMID 27756750DOI
    What this evidence cannot show

    Read as abstract only. The MLL-AF9 model is used as the in vivo validation system for an AML-wide hit, not to claim KMT2A specificity — an important distinction this entry must not blur. shRNA rather than CRISPR, with the off-target liabilities of that era. AML only; no lymphoid or infant data. The screen predates the DepMap CRISPR data the scan used, so the two are genuinely independent.

    Evidence that argues against this (1)
    • ZEB2 dependence is demonstrated in other fusion-driven AML too: Zeb2 knockdown in vivo delays leukemia onset in t(8;21) and t(15;17) models, where Zeb2 is upregulated by AML1-ETO and PML-RARA. A dependency shared with the two fusions this project cites as curable comparators is not a KMT2A story. De Conti G et al. 2020
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    ZEB2 acts predominantly as a transcriptional repressor restraining mature haematopoietic lineage gene expression programs from being expressed too early in haematopoietic stem and progenitor cells, with ZEB1 fine-tuning that repression to ensure haematopoietic lineage fidelity; inducible inactivation of both Zeb1 and Zeb2 causes acute bone-marrow failure at the short-term-HSC-to-MPP1 transition with increased stem-cell numbers and decreased differentiation into all lineages; and inactivation of Zeb2 alone or of Zeb1 and Zeb2 together enhances survival in secondary MLL-AF9 acute myeloid leukemia models.

    Interplay between the EMT transcription factors ZEB1 and ZEB2 regulates hematopoietic stem and progenitor cell differentiation and hematopoietic lineage fidelity · Wang J, Farkas C, Benyoucef A, Carmichael C, Haigh K, Wong N, Huylebroeck D, Stemmler MP, Brabletz S, Brabletz T, Nefzger CM, Goossens S, Berx G, Polo JM, Haigh JJ · PLoS Biology 19:e3001394 2021
    PMID 34550965DOI
    What this evidence cannot show

    Read as abstract only. Mouse genetics throughout; 'lineage fidelity' is a normal-haematopoiesis claim, and the extension to leukemic lineage switch in humans is an inference this entry makes explicit and does not assert. The MLL-AF9 survival result is in secondary transplant models, a stringent but artificial setting. The acute bone-marrow failure on combined Zeb1/Zeb2 loss is also the clearest toxicity signal against targeting this axis.

    Evidence that argues against this (1)
    • The same paper shows Zeb1/Zeb2 double inactivation causes acute bone-marrow failure with loss of differentiation into every lineage. Anything that suppressed this axis therapeutically would be doing to normal marrow what the disease already does, which is a severe essential-function problem rather than an incidental toxicity. Wang J et al. 2021
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Inducible knockdown of Zeb2 in transgenic t(8;21) and t(15;17) driven murine acute myeloid leukemia, using a dual-fluorescence in vivo platform, produced a significant delay in leukemia onset; Zeb2 is upregulated by the AML1-ETO and PML-RARA oncogenes in the pre-leukemic haematopoietic stem cell compartment.

    A Novel Platform to Test In Vivo Single Gene Dependencies in t(8,21) and t(15,17) AML Confirms Zeb2 as Leukemia Target · De Conti G, Gruszka AM, Valli D, Cammarata AU, Righi M, Mazza M, Pelicci PG · Cancers (Basel) 12:3768 2020
    PMID 33327558DOI
    What this evidence cannot show

    Read as abstract only. The paper's primary purpose is to present a methods platform, and the Zeb2 result is used as a proof of concept for that platform, which is a weaker framing than a dedicated dependency study. Mouse models of two fusions this project does not work on. A delay in onset is not a treatment effect in established disease. Its value here is as the cleanest available evidence that ZEB2 dependence is not specific to KMT2A rearrangement.

    Evidence that argues against this (1)
    • Against reading this as 'therefore not KMT2A-relevant': a gene can be required across several fusion-driven leukemias and still be more required in one. The scan's difference statistic and these in vivo onset experiments are not measuring the same quantity and cannot directly contradict each other. analysis run by this project 2026
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Mice overexpressing Zeb2 spontaneously develop immature T-cell lymphoblastic leukemia; pre-leukemic Zeb2-overexpressing thymocytes show a differentiation delay at beta-selection due to aberrant activation of interleukin-7 receptor signalling and, unlike Lmo2-overexpressing thymocytes, display no acquired self-renewal properties; Zeb2 activation in more differentiated T-cell precursors also drives malignant T-cell development, so ZEB2 and LMO2 transform immature T-cell progenitors by distinct mechanisms.

    ZEB2 and LMO2 drive immature T-cell lymphoblastic leukemia via distinct oncogenic mechanisms · Goossens S, Wang J, Tremblay CS, De Medts J, T'Sas S, Nguyen T, Saw J, Haigh K, Curtis DJ, Van Vlierberghe P, Berx G, Taghon T, Haigh JJ · Haematologica 104:1608-1616 2019
    PMID 30679322DOI
    What this evidence cannot show

    Read as abstract only. T-lineage, mouse, and a gain-of-function model, so it is background on what kind of factor ZEB2 is rather than evidence about KMT2A-rearranged disease. Useful mainly because it establishes that ZEB2 sits upstream of differentiation timing in a second lineage, which is consistent with the lineage-fidelity role.

    Evidence that argues against this (1)
    • A gene that is transforming when overexpressed and required when present cannot be read as a simple dependency: the DepMap knockout phenotype may reflect loss of a lineage-determining factor that the cell needs for identity rather than a selective vulnerability created by the fusion. Wang J et al. 2021
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    Human Protein Atlas v25.1 immunofluorescence assigns ZEB2 main subcellular locations Nucleoplasm AND Nucleoli, with Plasma membrane additional, at reliability Supported.

    What this evidence cannot show

    Antibody immunofluorescence in non-haematopoietic lines. A nucleolar call with no corroborating functional experiment anywhere in the ZEB2 literature should be treated as a lead, not a fact; nucleolar staining is a common antibody artefact. The Plasma membrane additional call for a zinc-finger transcriptional repressor is itself a reason for caution about this antibody.

    Evidence that argues against this (1)
    • No orthogonal source supports the nucleolar call and one property of the same antibody argues against trusting it: the identical HPA run also scores Plasma membrane for a zinc-finger transcriptional repressor, which is not credible. UniProt's coded record contains Nucleus and Chromosome and NO nucleolar line at all. A subnuclear assignment supported by one antibody, whose same run produces an implausible second call, is a lead and should not be counted as a subnuclear record in any coverage figure that implies measurement. UniProt Consortium 2026
  7. E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    UniProt O60315's nuclear call for ZEB2 rests partly on PMID 9853615, a methods paper describing a yeast nuclear-transportation-trap selection in which 75% of recovered unknown proteins localised to the nucleus in COS-7 cells - a screen readout rather than a dedicated ZEB2 localisation study.

    PMID 9853615
    What this evidence cannot show

    Brief rule 91 in its purest form: the ECO:0000269 code tells you a curator saw an experiment, not that the paper is about ZEB2's location. COS-7 is monkey kidney. Read at abstract level.

    Evidence that argues against this (1)
    • The rule-91 criticism impugns the record and not the fact, and pushing it further would be wrong. A yeast transport-trap screen is weak provenance, but an independent antibody immunofluorescence at 'Supported' reliability and a separate chromatin-association citation reach the same nuclear conclusion by different routes. The correct conclusion is that ZEB2 is nuclear on adequate evidence and that UniProt's chosen citation is a poor pointer to it - not that the nuclear call is in doubt. Human Protein Atlas 2026
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    SIP1/ZEB2 is SUMOylated by the polycomb protein Pc2 at Lys391 and Lys866, and that modification does NOT change its subcellular localisation - it changes only its transcriptional activity.

    PMID 16061479
    What this evidence cannot show

    An explicit negative in an epithelial EMT system, read at abstract level. It excludes Pc2-driven relocalisation; it does not exclude polycomb-body association by another route, which nobody has tested.

    Evidence that argues against this (1)
    • An explicit negative in one system does not close the question, and this one was not designed to find a compartment. The experiment asked whether SUMOylation changes ZEB2's gross subcellular distribution in epithelial cells; it did not look for a subnuclear pool, and HPA's independent nucleolar call suggests there is one to look for. What the negative establishes is narrow - Pc2-driven relocalisation does not happen - and it leaves both polycomb proximity by other routes and the nucleolar pool untested. Human Protein Atlas 2026
  9. E9strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 992 Philadelphia-negative adult B-cell acute lymphoblastic leukaemia patients, ZEB2 p.H1038R defines distinct subtypes - co-occurring with CEBP alterations in 18 cases and isolated in 15 - placing a recurrent ZEB2 alteration in the index disease's lineage.

    PMID 42306066DOI
    What this evidence cannot show

    ADULT B-ALL, Philadelphia-negative, not infant and not KMT2A-rearranged; the paper's subtypes are defined against CEBP enhancer hijacking and this cohort's KMT2A-rearranged fraction is not stated in the abstract. Abstract-level reading. A recurrent mutation is a genetic fact about the cell type, not a localisation or state measurement.

    Evidence that argues against this (1)
    • A recurrent mutation locates the gene in a lineage, and this one locates it in the wrong part of it for this project: the cohort is ADULT, Philadelphia-negative B-ALL, while the index disease is infant KMT2A-rearranged. ZEB2's best-defined haematopoietic role is not B-lineage at all but dendritic and myeloid - ZEB2 sits downstream of a pleiotropic enhancer governing cDC1 specification - and this entry's own published dependency scope is AML with in vivo validation in MLL-AF9, t(8;21) and t(15;17). Treating p.H1038R as 'ZEB2 is a B-lineage protein' would invert the weight of evidence. Nat Immunol 2026
Connected to