Oncofusion

start / graph / gene-cebpa

Proteinmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.king

CEBPA — ranks 25th on the KMT2A scan and sits at the median of random genes

also called CEBPA

C/EBP-alpha is another master switch of myeloid identity. On one way of measuring it looked like the 25th most important gene for this leukaemia. Split the cell lines by blood-cell type and the effect is entirely in the myeloid ones, and against random genes it scores no better than average.

The same thing, for a scientist

CEBPA ranks 25 of 17,728 in the genome-wide KMT2A-selectivity scan and is the starkest example of the panel confound in the graph. It scores **-2.79 SD** below the other-leukaemia arm in KMT2A-rearranged MYELOID lines and **+0.15 SD** in KMT2A-rearranged B-lineage lines; on a line-normalised lineage-consistency statistic it falls at the **48.53rd percentile** of random genes — the median. Its myeloid half alone would look impressive; the statistic is capped by the weaker half, which is the point. Like PU.1, C/EBP-alpha-dependent enhancers are part of the mechanism by which LSD1 inhibition acts.

Go deeper
The clearest case of the panel confound

CEBPA is the example worth remembering because both numbers are extreme in opposite directions. A -2.79 SD myeloid half is a strong dependency by any standard; a +0.15 SD B-lineage half is nothing at all. Averaged across a myeloid-enriched index arm, those produce a top-25 genome-wide rank.

Against a null that takes the weaker half, it lands at the median of random genes. A gene can be both strongly required in most of the panel and completely unremarkable as a KMT2A dependency, and CEBPA is the proof.

Localization: the informative answer is isoform-scoped and nucleolar (Q87, 2026-09-21)

Subcellular. Nucleus - but UniProt's evidence for it is a functional AML-mutation paper with no localisation experiment in its abstract (ECO:0000269 PubMed:11242107; rule 91). HPA v25.1 IF, reliability Approved: nucleoplasm main, vesicles additional.

Subnuclear: THE NUCLEOLUS, for the N-terminally extended isoform. A basic motif in its extended N-terminus is required for nucleolar retention and for binding nucleophosmin; there it occupies the rDNA promoter with UBF-1 and SL1 and stimulates rRNA synthesis and cell size (PMID 20075868). UniProt scopes this correctly as 'Isoform 4'.

Why the scoping is the point. CEBPA's AML biology is already an isoform story - the classic N-terminal mutations spare the downstream-initiated 30 kDa product. This gene has one isoform axis deciding phenotype and another deciding compartment, and a pan-CEBPA antibody sums them.

Cell type and state. Granulocyte precursors and myeloid blasts; the nucleolar work is in HL-60 during differentiation - genuinely haematopoietic. Two state axes: differentiation block, and ribosome output.

Shown for fusion. No. All three fusion-scoped hits were retrieved and rejected; one matched only because BAALC's full name contains the word 'cytoplasmic'.

Facts and figures
gene symbolCEBPA
depmap scan rank of 1772825th
depmap scan difference-0.4377
lineage consistency percentile48.53
verdictmyeloid-identity dependency, not KMT2A-specific
uniprotP49715
length AA358
gene primaryCEBPA
accession addedverified against live UniProt 2026-09-20; rules 66/68/69 require accession AND length_aa so verify-accessions.py can fail a swap
subnuclear localizationISOFORM-DEPENDENT: the NUCLEOLUS for the N-terminally extended translational isoform of C/EBPalpha, whose N-terminal basic motif is required for nucleolar retention and for binding nucleophosmin, and which occupies the rDNA promoter with UBF-1 and SL1 to stimulate rRNA synthesis (PMID 20075868; UniProt P49715 records it molecule-scoped as 'Isoform 4: Nucleus, nucleolus'). The full-length protein's own 'Nucleus' term rests on a functional AML-mutation paper with no localisation experiment in its abstract, so the better-evidenced subnuclear statement for this gene is the isoform one.
cell stateGRANULOCYTIC DIFFERENTIATION versus DIFFERENTIATION BLOCK, plus a size/growth state. The differentiation axis is the classical one: dominant-negative CEBPA mutations fail to induce granulocytic differentiation and are proposed to induce the differentiation block found in AML (PMID 11242107). The second, less-quoted axis is anabolic: the nucleolar isoform stimulates rDNA transcription and CELL SIZE (PMID 20075868), so C/EBPalpha sets ribosome output as well as lineage identity.
shown for fusionNO - nobody has looked in any fusion. Measured 2026-09-21: (CEBPA[tiab] OR "C/EBPalpha"[tiab] OR "CEBPalpha"[tiab]) AND the eight-term fusion clause AND the ten-term localisation clause returns 3 records, ALL retrieved and read and all off topic: PMID 41021397 (cytomorphology and immunophenotype of lymphoblastic lymphoma in serous effusion), PMID 24413067 (BAALC expression as a prognostic marker - where the query's 'cytoplasmic' comes from BAALC's own name, 'brain and acute leukemia, CYTOPLASMIC') and PMID 18342811 (a review of normal-karyotype AML). Screened count on topic: 0. Note separately that this entry's own DepMap section records CEBPA's KMT2A-scan rank as a LINEAGE ARTEFACT, so there is no selectivity claim here for a localisation result to support.
z vs other leukaemia
{
  "kmt2a_myeloid": -2.79,
  "kmt2a_b_lineage": 0.15
}
uniprot synonyms
[
  "CEBP"
]
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.8273,
  "chronos_mean_other_leukaemia": -0.3896,
  "difference": -0.4377,
  "pan_essentiality_all_lines": -0.1691,
  "selectivity_rank": 25,
  "rank_denominator": 17728,
  "provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
  "how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}
localization
{
  "subcellular": "NUCLEUS, and the database's evidence for it is weaker than it looks. UniProt P49715 carries 'Nucleus' for the full-length protein on ECO:0000269 PubMed:11242107 - which is the Nature Genetics paper reporting dominant-negative CEBPA mutations in AML, a paper about DNA binding and transactivation whose abstract describes no localisation experiment. That is the rule-91 pattern, and it is exactly the trap this beat was warned about for CEBPA: a functional paper standing in for a localisation record. HPA v25.1 IF, reliability Approved, gives Nucleoplasm main with Vesicles additional, and is the better-founded nuclear call of the two.",
  "subnuclear": "THE NUCLEOLUS, for one specific translational isoform, and this is the real localisation result in the CEBPA literature. The intronless CEBPA mRNA is translated into several isoforms from consecutive initiation sites; an N-TERMINALLY EXTENDED isoform, initiated at a non-AUG codon, carries a basic amino-acid motif in its N-terminus that is REQUIRED for nucleolar retention and for interaction with nucleophosmin (NPM). In the nucleolus, extended-C/EBPalpha occupies the ribosomal DNA promoter and associates with the Pol I factors UBF-1 and SL1 to stimulate rRNA synthesis, and the phenomenon was followed during differentiation of HL-60 cells (PMID 20075868). UniProt records this as a separate molecule-scoped location comment - 'Isoform 4: Nucleus, nucleolus', ECO:0000269 PubMed:20075868.",
  "why_the_isoform_scoping_matters_here": "CEBPA's biology in AML is ALREADY an isoform story - the classic N-terminal CEBPA mutations truncate the full-length protein while sparing the 30 kDa isoform initiated further downstream (PMID 11242107). So this gene has a documented mechanism in which which-isoform decides the phenotype, AND a documented mechanism in which which-isoform decides the subnuclear compartment. A bare 'CEBPA is nuclear' loses both.",
  "how_shown": "deletion and point mutation of the N-terminal basic motif with immunofluorescence for nucleolar retention, co-immunoprecipitation with NPM, ChIP at the rDNA promoter, and Pol I factor association, followed through HL-60 differentiation (PMID 20075868); HPA antibody IF",
  "tissue_of_evidence": "HL-60 human myeloid leukaemia cells for the nucleolar isoform work - genuinely haematopoietic, which is unusual in this tranche.",
  "node": "localization-nucleolus-kmt2a-fusion"
}
cell type
{
  "haematopoietic_answer": "GRANULOCYTE PRECURSORS AND MYELOID BLASTS. C/EBPalpha is crucial for granulocyte differentiation - conditional expression triggers neutrophilic differentiation and Cebpa-mutant mice have no mature granulocytes - and the nucleolar-isoform work was done in HL-60 human myeloid leukaemia cells during differentiation. This is a genuinely haematopoietic answer for both the cell type and the compartment.",
  "expression_atlas": "HPA v25.1: RNA tissue specificity 'Group enriched'; IF reliability Approved."
}
How this could bite us back
  • The rank of 25 in this project's own published scan is wrong as evidence about KMT2A, and that scan's ranking has been quoted in several entries. This entry exists partly to make the correction findable.
  • Calling CEBPA a confound risks discarding it: it is a strong dependency in myeloid KMT2A-rearranged disease, which is a large share of patients. The claim is narrow.
Still open
  1. CEBPA's myeloid half is -2.79 SD, stronger than MEN1's -2.29. Is there anything to be made of that within KMT2A-rearranged AML specifically, as distinct from AML generally?
  2. The full genome-wide scan has never been re-run with a lineage-balanced index arm. How many of its top 50 are CEBPA-like?
Every source behind this entry (5)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.analysis

    In the genome-wide DepMap scan contrasting KMT2A-rearranged against other leukaemia lines, CEBPA ranks 25th of 17,728 with a difference of -0.4377. Splitting the index arm by lineage dissolves it: CEBPA scores -2.79 SD below the other-leukaemia arm in KMT2A-rearranged MYELOID lines and 0.15 SD in KMT2A-rearranged B-lineage lines, i.e. no dependency at all in KMT2A-rearranged B-ALL.

    Lineage-split test of myeloid transcription-factor dependencies in KMT2A-rearranged leukaemia · Oncofusion knowledge graph (internal analysis); DepMap Chronos via Open Targets 2026
    What this evidence cannot show

    Computed by the coordinator on 2026-09-19. Five B-lineage and eight myeloid lines, so each half is small and one line moves a mean. Lineage is assigned from the DepMap disease label rather than an expression signature, and KMT2A-rearranged disease includes genuinely ambiguous-lineage cases. Chronos measures knockout.

    Evidence that argues against this (2)
    • A genome-wide rank from that scan is not evidence about KMT2A, because the scan's index arm is myeloid-enriched (8 AML lines against 5 B-ALL). A myeloid-identity gene rises in that ranking for reasons that have nothing to do with the fusion, which is precisely what the lineage split demonstrates for this gene. The rank is retained here as the observation that needed explaining, not as support. Oncofusion knowledge graph (internal analysis); DepMap Chronos via Open Targets 2026
    • The B-lineage half is five lines and two of them are the field's most heavily passaged KMT2A::AFF1 models, so the absence of dependency rests on a small and culture-adapted set. A dependency that is real but weak in B-lineage disease would be hard to distinguish from zero here. Oncofusion knowledge graph (internal analysis); DepMap Chronos via Open Targets 2026
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.analysis

    Scored against 647 random protein-coding genes on a line-normalised statistic that takes the WEAKER lineage half - max(z_myeloid, z_lymphoid), so a one-sided effect cannot rank - CEBPA falls at the 48.53th percentile. MEN1 and PSIP1 fall at the 0.00th and the pan-essential control ACTB at the 43.59th.

    Lineage-consistency null test for KMT2A dependencies: line-normalised, random-gene percentile · Oncofusion knowledge graph (internal analysis); DepMap Chronos via Open Targets 2026
    What this evidence cannot show

    The null is random protein-coding genes, most not expressed in leukaemia, so percentiles are flattered for genuinely leukaemia-relevant genes and this one still does not rank. The statistic is capped by a four-line B-lineage half by construction.

    Evidence that argues against this (1)
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    C/EBP-alpha-dependent enhancers are, with PU.1-dependent enhancers, the route through which LSD1 inhibition exerts its antileukemic effect in AML.

    PMID 29453291DOI
    What this evidence cannot show

    AML models. The work does not address KMT2A-rearranged B-lymphoid disease, where CEBPA shows no DepMap dependency.

    Evidence that argues against this (2)
    • As with PU.1, this places CEBPA on a drug's mechanism rather than nominating it as a target: the enhancers are RECOMMISSIONED, meaning the therapy restores their function rather than removing it. A high dependency rank and this mechanism point in opposite therapeutic directions. Blood 2018
    • CEBPA is also a recurrently mutated AML driver in its own right, independently of any KMT2A rearrangement, so its myeloid dependency is expected from general AML biology and carries no information about fusion leukaemia specifically. Nature Genetics 2015
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    An N-terminally extended translational isoform of C/EBPalpha, initiated at an alternative non-AUG codon, is retained in the NUCLEOLUS. A basic amino-acid motif within its N-terminus is required both for that nucleolar retention and for interaction with nucleophosmin; in the nucleolus the extended isoform occupies the ribosomal DNA promoter and associates with the Pol I-specific factors UBF-1 and SL1 to stimulate rRNA synthesis and cell size. The phenomenon was followed during differentiation of HL-60 cells.

    Nucleolar retention of a translational C/EBPalpha isoform stimulates rDNA transcription and cell size. · Müller C, Bremer A, Schreiber S, Eichwald S, Calkhoven CF · EMBO J 2010
    PMID 20075868DOI
    What this evidence cannot show

    The compartment belongs to ONE isoform, so this is not a statement about C/EBPalpha in general, and UniProt correctly scopes it that way ('Isoform 4: Nucleus, nucleolus'). Non-AUG-initiated isoforms are intrinsically hard to detect at endogenous levels and the mapping work uses constructs. HL-60 is a human myeloid leukaemia line and not a KMT2A-rearranged one; nothing here is fusion-scoped.

    Evidence that argues against this (1)
    • This entry should not let the nucleolar isoform become 'where CEBPA is', because the isoform question for this gene is already load-bearing in the opposite direction. The classic N-terminal CEBPA mutations in AML truncate the full-length protein while leaving a 30 kDa isoform initiated further downstream intact, and the leukaemic phenotype is attributed to that surviving short isoform acting dominant-negatively. So CEBPA has at least three translational products with different N-termini, at least one confers a nucleolar address, and at least one confers a dominant-negative phenotype - and they are not the same one. Any experiment reading 'C/EBPalpha localisation' with a pan-CEBPA antibody is summing species that do different things and sit in different places. Pabst T et al. 2001
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    UniProt P49715's 'Nucleus' term for full-length C/EBPalpha rests on ECO:0000269 PubMed:11242107, a Nature Genetics report of dominant-negative CEBPA mutations in AML whose abstract describes DNA binding, transactivation and granulocytic differentiation and no localisation experiment. The nuclear call is certainly correct for a transcription factor, but its cited evidence is a functional paper, and the independent support is HPA v25.1 immunofluorescence (Nucleoplasm main, Vesicles additional, reliability Approved) rather than the database's own citation.

    Dominant-negative mutations of CEBPA, encoding CCAAT/enhancer binding protein-alpha (C/EBPalpha), in acute myeloid leukemia. · Pabst T, Mueller BU, Zhang P, Radomska HS, Narravula S, Schnittger S, et al. · Nat Genet 2001
    PMID 11242107DOI
    What this evidence cannot show

    This is an assessment of a database annotation made from the cited paper's abstract, not from its figures, so the paper may contain a localisation panel that is simply not mentioned there (rule 80). The narrow claim is that the CITED EVIDENCE does not establish the compartment, not that the compartment is wrong. HPA's 'Vesicles' additional location is unexplained and is not corroborated by UniProt.

    Evidence that argues against this (1)
    • The counter-argument is that demanding a dedicated localisation paper for a sequence-specific transcription factor is the wrong standard, and the CEBPA literature shows why: the same paper cited here demonstrates that mutant C/EBPalpha proteins BLOCK wild-type DNA binding and transactivation in a dominant-negative manner, which is an assay that only works if both proteins reach the same DNA - a functional demonstration of nuclear residence that is stronger than an immunofluorescence image would be. The reason to flag the annotation anyway is not that it is unreliable but that it is uninformative about the question this tranche is asking: it cannot distinguish the isoforms, and the isoform is where the subnuclear answer lives. Müller C et al. 2010
Connected to