Oncofusion

start / graph / gene-rnf20

Proteinstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q-histone-marks

RNF20 (BRE1A) — the H2B ubiquitin ligase that MLL-rearranged leukaemia needs, and that DepMap says every other cancer line needs just as much

also called RNF20 · BRE1A · BRE1 · hBRE1 · E3 ubiquitin-protein ligase BRE1A · RING finger protein 20

RNF20 is the enzyme that attaches the ubiquitin tag to histone H2B, the tag that DOT1L needs in order to mark KMT2A-fusion target genes. Switching it off does shut down those genes and stops the leukaemia growing — but the same enzyme turns out to be needed by almost every dividing cancer cell, so it is a real dependency without a therapeutic window.

The same thing, for a scientist

RNF20 (Q5VTR2, 975 aa, BRE1A) is the catalytic RING subunit of the RNF20/RNF40 heterodimeric E3 ligase that monoubiquitinates histone H2B at K120, together with the WAC adaptor and an E2 whose identity is disputed. Wang et al. identified Rnf20 as a chromatin regulator necessary for MLL-fusion-mediated leukemogenesis: suppression inhibits proliferation in diverse MLL-rearranged models in vitro and in vivo, reduces expression of MLL-fusion target genes in a pattern resembling Dot1l inhibition, and lowers local Dot1l-dependent H3K79 methylation at *Hoxa9* and *Meis1*; ChIP-seq placed H2Bub across the bodies of fusion target genes [E1]. A second, independent line of evidence comes from infant ALL: panobinostat's in vivo anti-leukaemic activity in MLL-rearranged ALL xenografts involves depletion of H2B ubiquitination through suppression of the RNF20/RNF40/WAC complex, which those authors call a pivotal pathway for MLL-rearranged leukaemic maintenance [E3]. **But the selectivity does not survive a genome-wide test.** In the project's 17,728-gene DepMap scan, RNF20's KMT2A-rearranged minus other-leukaemia difference is **+0.076** — rank **16,229 of 17,728**, i.e. marginally *less* required in the rearranged lines — on a pan-essentiality of **-0.734** across 1,208 lines with broad depletion [E2]. RNF20 is therefore a genuine dependency of KMT2A-rearranged leukaemia and a genuine dependency of nearly everything else, which is the profile of an anti-proliferative, not of a target. Separately, and unnoticed until this beat cross-checked the graph against itself, RNF20 also appears at the **other end** of the disease: ATM signalling to RNF20 is required for replication-stress-induced cleavage of the *KMT2A* breakpoint cluster region, the event that creates the fusion [E6].

What it looks like
loading AF-Q5VTR2-F1…
AlphaFold AF-Q5VTR2-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 contradiction, stated plainly, and how to hold both halves

Two good bodies of evidence say different things and both are probably right.

SourceDesignConclusion
Wang 2013 1shRNA knockdown in diverse MLL-rearranged models, in vitro and in vivo, plus ChIP-seq and locus-level H3K79meRnf20 is necessary for MLL-fusion leukaemogenesis
Garrido Castro 2018 3MLL-rearranged ALL xenografts; panobinostat; RNF20/RNF40/WAC suppression; WAC knockdown phenocopythe RNF20/40/WAC-H2Bub axis is pivotal for maintenance
DepMap scan 2CRISPR knockout, 11 KMT2A-rearranged vs 43 other leukaemia lines, genome-wide rankingRNF20 is not selectively required; difference +0.076, rank 16,229/17,728; pan-essentiality -0.734

These are compatible. Necessity and selectivity are different measurements, and the two papers measured necessity. Neither ran a matched non-MLL leukaemia comparator at the scale that would detect the absence of a window. The DepMap scan runs exactly that comparison and finds the window is not there — and it finds it while agreeing that RNF20 is required (the KMT2A-rearranged mean gene effect is -0.658, a strong dependency in absolute terms; it is just that the other-leukaemia mean is -0.735, slightly stronger).

The honest summary is: knocking RNF20 down kills KMT2A-rearranged leukaemia cells, and it kills most other cells too.

What RNF20 is, structurally and enzymatically

RNF20 is one of the two human homologues of yeast Bre1. It functions as an obligate heterodimer with RNF40 (BRE1B), and the active species is a ~600 kDa complex 4. The RING domain sits at the C-terminus; the long N-terminal region is largely coiled-coil and mediates heterodimerisation and adaptor binding.

A 2026 structural study resolved how the adaptor engages: the crystal structure of the yeast Bre1-Lge1 complex plus an AlphaFold model of RNF20/RNF40 bound to WAC revealed extensive interfaces with shared structural homology but completely different sets of key electrostatic interactions, which encode binding specificity, and mutations at these interfaces impair the H2Bub1 reaction 5. That matters for anyone thinking about drugging this axis: the RNF20/RNF40–WAC interface is a defined, mutationally validated protein-protein interface rather than an enzymatic active site, and it is the point at which the ligase is targeted to Pol II rather than the point at which it does chemistry.

Accession resolution, per the project's standing rule: UniProt gene_exact:RNF20 AND organism_id:9606 AND reviewed:true returned Q5VTR2; primary gene name read back as RNF20, length 975 aa, recommended name E3 ubiquitin-protein ligase BRE1A. Not a synonym match.

The other end of the disease: RNF20 is also implicated in making the break in the first place

This is the one genuinely surprising thing this beat found, and it was already in the graph without being connected.

The graph's entry mechanism-replication-stress-endog-cleavage-kmt2a describes how replication stress drives Endonuclease G to cleave the KMT2A breakpoint cluster region. Its primary source states that MLLbcr breakage has an additional dependency on ATM signalling to the histone H2B monoubiquitinase RNF20, which is involved in chromatin relaxation 6.

So RNF20 appears at both ends of this disease:

StageRNF20's roleSource
Genesis — cleavage at the MLLbcr in the cell of originATM-to-RNF20 signalling is required for MLLbcr breakage under replication stress6
Maintenance — the fusion's transcriptional programmeRnf20 is required for H3K79me at Hoxa9/Meis1 and for MLL-fusion leukaemia proliferation1

That is not a therapeutic observation — nobody is going to prevent infant leukaemia by inhibiting RNF20 in utero — but it is a real structural fact about the graph: a single chromatin enzyme sits on both the genesis beat and the mechanism beat, and until now neither beat knew about the other. The mechanistic common denominator is chromatin relaxation: RNF20-dependent H2Bub opens chromatin, which in one context exposes the KMT2A locus to a nuclease and in the other permits elongation and DOT1L access.

Why this entry still earns its place despite the null

Three reasons.

It is the upstream node of the disease's most-cited step. The graph has 35 entries resting on H3K79me2. RNF20 is what makes that mark possible. A reader who wants to know why DOT1L can methylate H3K79 at fusion target genes and not elsewhere needs this entry, whether or not RNF20 is druggable.

The infant-ALL evidence is unusually on-target for this project. 3 is MLL-rearranged ALL, in xenografts, and it is one of very few papers in this beat that is not AML. It also supplies a mechanistic explanation for an HDAC inhibitor's activity that is not about acetylation, which is a non-obvious result.

The null is itself the finding. This beat was dispatched on the premise that the E2/E3 machinery is upstream of the most-cited step and absent from the graph. The premise is correct. The implied corollary — that it is therefore a target — is not, and recording why is worth more than adding a fifteenth candidate to the target list. See hyp-h2bub-required-but-not-selective.

Facts and figures
uniprotQ5VTR2
gene symbolRNF20
length AA975
protein nameE3 ubiquitin-protein ligase BRE1A
alphafold IDAF-Q5VTR2-F1
accession resolutionUniProt query gene_exact:RNF20 AND organism_id:9606 AND reviewed:true; primary gene name read back as RNF20 and length 975 confirmed against the entry
complexRNF20/RNF40 heterodimeric E3 ligase (human BRE1), with the WAC adaptor
enzyme classRING-type E3 ubiquitin ligase
substratehistone H2B lysine 120
mark writtenmark-h2bk120ub
classificationdependency of KMT2A-rearranged leukaemia, but not a selective one
druggabilityNo selective small-molecule RNF20 inhibitor is known to this beat. The RNF20/RNF40-WAC interface is a defined protein-protein interface with mutationally validated electrostatic hotspots, which is the most tractable-looking surface described, but no chemical matter exists.
subnuclear localizationunpublished in somatic cells; the only spatially resolved RNF20 address in the literature read for this beat is centromeres and spindle poles during oocyte meiosis (PMID 38240347), which is not a nuclear compartment. PubMed 2026-09-20: RNF20[tiab] AND (immunofluorescence[tiab] OR confocal[tiab] OR subcellular[tiab] OR 'nuclear foci'[tiab] OR 'chromatin-bound'[tiab]) returns 5 records, none a somatic subnuclear localisation. Zero in titles and abstracts; supporting figures not searched.
shown for fusionNO. RNF20 chromatin recruitment has been shown at HOX loci in human cells but not in a KMT2A-rearranged cell, and the H2Bub1-to-H3K79me2 axis this graph cares about has never had its writer localised in the index disease.
cell context evidence classrequired_in (MLL-rearranged leukaemia, murine, knockdown) + maintains_a_state (inferred from the HOXA9/MEIS1 programme, not measured as differentiation). No cell-type-restricted expression is claimed.
cell context shown for fusionYES for MLL-rearranged, NO for the index fusion and the index lineage. The founding work is murine MLL-rearranged leukaemia, written throughout with murine gene symbols (Rnf20, Dot1l, Hoxa9, Meis1), and 'diverse models of MLL-rearranged leukemia' does not resolve to a partner. There is no KMT2A::AFF1 arm, no human B-lineage arm and no infant material.
cell context tensionThe pharmacogenomic screen contradicts the knockdown result on selectivity. RNF20 is at the 91.5th percentile of 17,728 in this project's KMT2A-selectivity scan - the ANTI-selective half - at -0.658 in the 11 KMT2A-rearranged lines against -0.735 in 43 other leukaemia lines, difference +0.076, on a pan-line mean of -0.734. RNF20 is therefore moderately required in EVERY leukaemia line and, if anything, slightly LESS so in KMT2A-rearranged ones. The published claim 'required for MLL-rearranged leukemia' is true as written - the cells do need it - but it is not a selective requirement, and the paper had no non-MLL comparator arm that could have shown that.
essentiality
{
  "depmap_common_essential": true,
  "chronos_mean": -0.7335,
  "n_lines": 1208,
  "chronos_mean_kmt2a_rearranged": -0.6582,
  "chronos_mean_other_leukemia": -0.7345,
  "kmt2a_selectivity_difference": 0.0763,
  "genome_wide_rank": "16229/17728",
  "n_kmt2a_rearranged_lines": 11,
  "n_other_leukemia_lines": 43,
  "benchmark_men1_difference": -0.8656,
  "benchmark_dot1l_difference": -0.231,
  "interpretation": "Strongly required in KMT2A-rearranged lines and marginally more strongly required in other leukaemia lines. Necessity yes; window no."
}
localization
{
  "subcellular": "nucleus, chromatin-recruited. The founding human study shows UbcH6, the RNF20/RNF40 complex and the hPAF complex are all recruited to TRANSCRIPTIONALLY ACTIVE GENES in vivo, with RNF20 overexpression raising H2B monoubiquitination, H3K4 and H3K79 methylation and HOX gene expression, and RNAi against RNF20/40 or hPAF lowering all four. A non-chromatin pool exists: RNF20 recruits TPM3 to centromeres and spindle poles in oocyte meiotic spindle assembly.",
  "subnuclear": "unpublished as a compartment in somatic cells. The only spatially resolved RNF20 localisation located is the meiotic one - CENTROMERES and SPINDLE POLES in oocytes (PMID 38240347) - which is not a somatic interphase compartment and not chromatin. No nuclear-body, speckle or nucleolar assignment for RNF20 exists in any source read for this beat.",
  "how_shown": "chromatin immunoprecipitation at active genes, in vitro reconstitution of the RNF20/40-UbcH6-hPAF trimeric complex, and overexpression plus RNAi with H2Bub1, H3K4me and H3K79me readouts at HOX loci (PMID 16307923); immunofluorescence of TPM3 recruitment in oocytes (PMID 38240347)",
  "systems": "human cells for the chromatin work, with HOX genes as the readout - directly relevant to this graph's central programme, though not in a KMT2A-rearranged background; mouse oocytes for the spindle work",
  "node": "localization-chromatin-bound-versus-soluble-fraction-kmt2a"
}
cell type
{
  "required_in": "MLL-rearranged leukaemia cells. Suppressing Rnf20 in diverse models of MLL-rearranged leukaemia inhibits cell proliferation both in tissue culture and in vivo, reduces expression of MLL-fusion target genes with effects resembling Dot1l inhibition, and is required to maintain local H3K79 methylation by Dot1l at Hoxa9 and Meis1; H2B ubiquitination is enriched in the bodies of MLL-fusion target genes, correlating with H3K79 methylation and transcription elongation.",
  "how_shown": "shRNA knockdown in culture and in vivo, gene-expression analysis and ChIP-seq for H2B ubiquitination and H3K79 methylation",
  "systems": "murine MLL-rearranged leukaemia models; the abstract says 'diverse models of MLL-rearranged leukemia' without naming the partners",
  "normal_haematopoiesis": "The H2Bub axis does have a normal-haematopoiesis phenotype, but it comes from the opposite enzymatic direction and from a different complex: an in vivo genome-wide CRISPR knockout screen in haematopoietic stem and progenitor cells identified SAGA members Tada2b and Taf5l - the complex that REMOVES H2Bub and deposits H3K9ac - as key regulators, with loss causing a buildup of immature cells in the marrow, altered H2Bub enrichment, interferon upregulation, reduced mitochondrial activity and increased megakaryocyte-progenitor commitment. That is the writer-eraser axis RNF20 sits on, not an RNF20 result.",
  "nodes": []
}
cell state
{
  "maintains_state": "The HOXA9/MEIS1 programme, which is the operational content of the differentiation block in this graph. Rnf20 is required to keep H3K79me2 on Hoxa9 and Meis1 and to maintain MLL-fusion target-gene expression.",
  "attribution_limit": "The measured readouts are proliferation, target-gene expression and histone-mark occupancy. There is NO differentiation or immunophenotype readout in the source, so the cell-state reading is inferred from the maintained HOXA9/MEIS1 programme and is graded accordingly.",
  "nodes": [
    "cell-state-differentiation-block"
  ]
}
How this could bite us back
  • RNF20 is broadly essential (mean -0.734 across 1,208 lines with most lines depleted). Any agent that inhibits it is an anti-proliferative with bone-marrow toxicity as its first-order effect, and infants are the worst population in which to accept that.
  • RNF20 suppression also lowers H3K4me2/3 through SET1/COMPASS, so it hits the wild-type KMT2A-family programme that normal haematopoiesis needs.
  • H2Bub1 is required for the DNA damage response; lowering it systemically is a plausible route to therapy-related second malignancy, which is the specific harm this disease's survivors already carry.
  • The pathway has a documented alternative entry point - MSL1/MSL2 ubiquitinate H2BK34 and also stimulate DOT1L - so suppressing RNF20 alone may leave a partially functional crosstalk.
Still open
  1. Is the DepMap null a real absence of selectivity or an artefact of measuring a broadly essential gene by CRISPR knockout? A titratable degron in matched KMT2A-rearranged and non-rearranged lines would separate 'no window' from 'window invisible because everything dies'.
  2. Does RNF20 loss sensitise KMT2A-rearranged cells to menin inhibition? A non-selective dependency can still be a useful combination partner, and no combination data exist.
  3. Which fusions were used in Wang et al.? The abstract says 'diverse models' and the answer determines whether this result covers KMT2A::AFF1 B-ALL at all.
  4. In the panobinostat work, how much of the in vivo effect is attributable to the RNF20/40/WAC arm as opposed to the other epigenetic pathways panobinostat inhibits? A WAC-restoration experiment under panobinostat would answer it.
  5. RNF20 knockdown reduces MLL-fusion target-gene expression and proliferation, but a genome-scale screen finds it slightly less required in KMT2A-rearranged lines than in other leukaemias. Is the published effect a general H2Bub-elongation requirement that the founding study read as MLL-specific because it had no comparator arm? Re-running the knockdown with a matched non-MLL leukaemia line would settle a claim that has been quoted as fusion-specific for over a decade.
  6. Does RNF20 loss change what an MLL-rearranged blast IS? Every readout in the founding study is proliferation, expression or mark occupancy. A CD11b or CD14 panel after Rnf20 knockdown would convert an inferred differentiation-block claim into a measured one.
Every source behind this entry (9)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    RNF20 is necessary for MLL-fusion-mediated leukemogenesis: suppressing Rnf20 in diverse models of MLL-rearranged leukaemia inhibits cell proliferation in tissue culture and in vivo, reduces expression of MLL-fusion target genes resembling Dot1l inhibition, H2B ubiquitination is enriched in the body of MLL-fusion target genes correlating with H3K79 methylation and elongation, and Rnf20 is required to maintain local H3K79 methylation by Dot1l at Hoxa9 and Meis1.

    Histone H2B ubiquitin ligase RNF20 is required for MLL-rearranged leukemia. · Wang E, Kawaoka S, Yu M, Shi J, Ni T, Yang W, Zhu J, Roeder RG, Vakoc CR · Proc Natl Acad Sci U S A 2013
    PMID 23412334DOI
    What this evidence cannot show

    Abstract read in full; full text not read. shRNA rather than knockout or degron. 'Diverse models of MLL-rearranged leukaemia' is not specified by fusion in the abstract, and the H3K79me rescue experiments are at two loci (Hoxa9, Meis1), not genome-wide. Most importantly the design establishes necessity, not selectivity: no matched non-MLL leukaemia comparator is described in the abstract. **Resolved, and the breadth claim does not survive.** 'Diverse models of MLL-rearranged leukaemia' is **GSE43725: three samples of one mouse line, RN2 (MLL-AF9/NrasG12D), n=1 per hairpin** - and carrying a cooperating **NRAS G12D** that no caveat in this graph had flagged. Resolved 2026-09-19 by the Q18b unstated-systems sweep from the study's deposition, not from its text; bounded as a statement about the deposited data rather than about the whole study.

    Evidence that argues against this (2)
    • The DepMap scan reverses the selectivity claim implicit in the title. Across 11 KMT2A-rearranged and 43 other leukaemia lines RNF20's difference is +0.076 - it is fractionally less required in the rearranged lines - and it ranks 16,229 of 17,728 genes on that measure, while being depleted broadly (pan-essentiality -0.734). Broad Institute DepMap; Cellosaurus; Open Targets Platform; analysis run by this project 2026
    • One cell line at n=1 per hairpin is not diverse models, and the cooperating NRAS G12D is a second oncogenic driver whose contribution to any H2B-ubiquitination phenotype cannot be separated from the fusion's. `ana-med23-verdict-kmt2a-b-lineage` rests on the same line, so two independent-looking entries share one mouse cell line. NCBI Gene Expression Omnibus 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    In a genome-wide DepMap CRISPR scan of 17,728 protein-coding genes, RNF20 has a KMT2A-rearranged mean gene effect of -0.658 against -0.735 in other leukaemia lines, a difference of +0.076, ranking 16,229 of 17,728 on KMT2A selectivity, with a pan-essentiality of -0.734 across 1,208 lines; MEN1 ranks 1 with a difference of -0.866 and DOT1L ranks 184 with -0.231 in the same scan.

    DepMap CRISPR (Chronos) genome-wide KMT2A-selectivity scan of 17,728 protein-coding genes, 11 Cellosaurus-annotated KMT2A-rearranged leukaemia lines vs 43 other leukaemia lines · Broad Institute DepMap; Cellosaurus; Open Targets Platform; analysis run by this project (analyses/depmap-kmt2a-selective-scan-full) · Database and derived analysis 2026
    What this evidence cannot show

    Eleven KMT2A-rearranged lines is a small group, none infant-derived, and the group is AML-weighted, so a B-ALL-specific or infant-specific requirement would be invisible here. Cellosaurus fusion annotation may be incomplete, which biases toward the null. CRISPR knockout of a broadly essential gene compresses differences because every arm dies. This scan measures the difference in mean gene effect, not a formally significance-tested contrast.

    Evidence that argues against this (1)
    • SCAN CONFOUND: this number comes from the project's genome-wide KMT2A-selectivity scan, whose index arm is MYELOID-ENRICHED (8 AML lines against 5 B-ALL). A gene whose requirement differs by lineage is mis-scored in BOTH directions by that design — CEBPA ranks 25 genome-wide with no dependency at all in KMT2A-rearranged B-ALL, and conversely a gene required in one lineage and not the other can average to apparent non-selectivity. See `ana-lineage-confound-myeloid-tf-dependencies`. For RNF20 specifically, the entry elsewhere reports the gene as REQUIRED in MLL-rearranged leukaemia from a dedicated study, so a scan-derived selectivity figure disagreeing with that should be read as the scan's limitation rather than as a correction of the biology. Oncofusion knowledge graph (internal analysis) 2026
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Panobinostat (LBH589) shows in vivo anti-leukaemic efficacy in xenograft models of MLL-rearranged ALL, extending survival and reducing disease burden; molecular analyses show the activity involves depletion of H2B ubiquitination via suppression of the RNF20/RNF40/WAC E3 ligase complex, described by the authors as a pivotal pathway for MLL-rearranged leukaemic maintenance, and knockdown of WAC phenocopies loss of H2B ubiquitination with concomitant cell death.

    The HDAC inhibitor panobinostat (LBH589) exerts in vivo anti-leukaemic activity against MLL-rearranged acute lymphoblastic leukaemia and involves the RNF20/RNF40/WAC-H2B ubiquitination axis. · Garrido Castro P, van Roon EHJ, Pinhancos SS, Trentin L, Schneider P, Kerstjens M, Te Kronnie G, Heidenreich O, Pieters R, Stam RW · Leukemia 2018
    PMID 28690313DOI
    What this evidence cannot show

    Abstract read in full; full text not read. Panobinostat is a pan-HDAC inhibitor and the authors themselves state it cross-inhibits multiple epigenetic pathways, so the RNF20/40/WAC arm is one of several mechanisms and the abstract does not apportion how much of the in vivo effect it carries. The causal chain runs drug -> suppressed ligase complex -> lost H2Bub -> death; the WAC knockdown phenocopy supports it but does not exclude that panobinostat kills mainly by other means. The author name 'Pinhancos' is rendered without its diacritics here.

    Evidence that argues against this (1)
    • Panobinostat is a pan-HDAC inhibitor with many substrates and downstream effects; H2Bub depletion is one observed consequence, not a demonstrated mechanism of its anti-leukaemic activity. Attributing efficacy to the H2Bub arm requires an HDAC-independent manipulation of H2Bub with the same phenotype, which this study does not provide. Leukemia 2018
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In human cells the 600 kDa RNF20/RNF40 complex is the E3 ligase for H2B-Lys120 monoubiquitination; RNF20 and RNF40 are both homologues of yeast Bre1; the complex is recruited to transcriptionally active genes and RNF20 overexpression elevates H2Bub1, H3K4 and H3K79 methylation and HOX gene expression, while RNAi against the complex reduces all of them.

    Monoubiquitination of human histone H2B: the factors involved and their roles in HOX gene regulation. · Zhu B, Zheng Y, Pham AD, Mandal SS, Erdjument-Bromage H, Tempst P, Reinberg D · Mol Cell 2005
    PMID 16307923DOI
    What this evidence cannot show

    Abstract read in full; full text not read. Non-leukaemic human cells. The HOX result is from overexpression and RNAi, both of which perturb stoichiometry of a complex.

    Evidence that argues against this (1)
    • The 2005 characterisation established RNF20/RNF40 as the E3 in a defined system, and the entry treats the complex as a unit. Later structural work shows RNF20 and RNF40 are not interchangeable and that WAC contributes an extensive interface, so 'the RNF20/40 complex' hides which subunit a perturbation actually removes. Nucleic Acids Research 2026
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A crystal structure of the yeast Bre1-Lge1 complex and an AlphaFold model of RNF20/RNF40 bound to WAC, with in vitro and in vivo tests, reveal extensive interfaces sharing structural homology but with completely different sets of key electrostatic interactions that encode binding specificity, and show these interactions are critical for the Bre1-catalysed H2BUb1 reaction and the processes it regulates.

    Structural insights into the Bre1-Lge1 and RNF20/RNF40-WAC interactions critical for H2B ubiquitination. · Shi M, Wang X, Zhang H, Wen Y, Liu Q, Chen P, Chen X, Xiang S · Nucleic Acids Res 2026
    PMID 41533567DOI
    What this evidence cannot show

    Abstract read in full; full text not read. The human arm of the study is an **AlphaFold predicted model**, not an experimental structure; only the yeast Bre1-Lge1 complex was crystallised. The functional tests described in the abstract as in vivo are in the yeast system. No PDB identifier is asserted here because the abstract does not state one and this beat did not verify one.

    Evidence that argues against this (1)
    • One crystal structure here is of the YEAST Bre1-Lge1 complex, and the human RNF20/RNF40-WAC arrangement is an AlphaFold MODEL. A predicted interface with in vitro support is a hypothesis about architecture, and the entry should not carry it at the same weight as the experimental yeast structure it is built from. Nucleic Acids Research 2026
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    The MLL breakpoint cluster region enhances spontaneous and genotoxin-induced recombination in human cells; Endonuclease G is an essential factor for MLLbcr-specific recombination after replication stress and cleaves the chromosomal MLLbcr; and MLLbcr breakage shows an additional dependency on ATM signalling to the histone H2B monoubiquitinase RNF20, which is involved in chromatin relaxation.

    Endonuclease G initiates DNA rearrangements at the MLL breakpoint cluster upon replication stress. · Gole B, Baumann C, Mian E, Ireno CI, Wiesmuller L · Oncogene 2015
    PMID 25132265DOI
    What this evidence cannot show

    Abstract read in full; full text not read. The primary system is an EGFP-based recombination reporter carrying the MLLbcr hotspot, plus chromosomal cleavage assays; it is a model of breakage, not a model of leukaemogenesis, and no leukaemia arises in it. The RNF20 dependency is stated as one clause of a multi-step pathway and the abstract gives no effect size. The same PMID is the primary source for this graph's `mechanism-replication-stress-endog-cleavage-kmt2a`, written by a different beat; one author surname is rendered here without its diacritic.

    Evidence that argues against this (1)
    • This is a DNA-breakage result at the KMT2A breakpoint cluster region and sits in a histone-mark entry, where it can only serve as an alternative explanation. Endonuclease G initiating rearrangements is about how the fusion FORMS, not about what H2Bub does once it exists, and the two should not be chained. Gole B et al. 2015
  7. E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    RNF20/RNF40 is a chromatin-recruited E3 ligase: UbcH6, the RNF20/40 complex and the hPAF complex are recruited together to transcriptionally active genes in vivo, RNF20 overexpression raises H2B-Lys120 monoubiquitination with downstream H3K4 and H3K79 methylation and stimulates HOX gene expression, and RNAi against RNF20/40 or hPAF reverses all of it.

    Monoubiquitination of human histone H2B: the factors involved and their roles in HOX gene regulation. · Zhu B, Zheng Y, Pham AD, Mandal SS, Erdjument-Bromage H, Tempst P, Reinberg D · Mol Cell 2005
    PMID 16307923DOI
    What this evidence cannot show

    ChIP recruitment is a genomic address, not a subnuclear compartment, and this entry should not be read as saying RNF20 has a known place in the nucleus. The HOX readout makes this unusually relevant to this graph, but the system is human cells generally and not a KMT2A-rearranged background; per brief rule 73 any downstream claim must open by saying so. The hPAF dependence is worth noting because PAF1 is a KMT2A-fusion partner complex in this graph.

    Evidence that argues against this (1)
    • RNF20 is not confined to chromatin. In oocyte meiosis RNF20 regulates spindle assembly by recruiting TPM3 to CENTROMERES AND SPINDLE POLES - a cytoskeletal, non-chromatin localisation with a non-transcriptional function. An entry that writes RNF20 as a chromatin enzyme full stop will not predict a mitotic or spindle phenotype from RNF20 loss, and mitotic phenotypes are exactly what a degrader or inhibitor programme has to distinguish from on-target transcriptional effects. Wang L et al. 2024
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The histone H2B E3 ubiquitin ligase RNF20 is necessary for MLL-fusion-mediated leukaemogenesis: suppressing Rnf20 in diverse models of MLL-rearranged leukaemia inhibits proliferation in tissue culture and in vivo and reduces expression of MLL-fusion target genes with effects resembling Dot1l inhibition; H2B ubiquitination is enriched in the bodies of MLL-fusion target genes, correlating with H3K79 methylation and transcription elongation, and Rnf20 is required to maintain local H3K79 methylation by Dot1l at Hoxa9 and Meis1.

    Histone H2B ubiquitin ligase RNF20 is required for MLL-rearranged leukemia. · Wang E, Kawaoka S, Yu M, Shi J, Ni T, Yang W, Zhu J, Roeder RG, Vakoc CR · Proc Natl Acad Sci U S A 2013
    PMID 23412334DOI
    What this evidence cannot show

    Murine MLL-rearranged leukaemia models, written with murine gene symbols, and the abstract does not name which fusion partners the 'diverse models' carry - so this cannot be attributed to KMT2A::AFF1 and is not human B-lineage or infant material. Knockdown rather than knockout. The readouts are proliferation, target-gene expression and mark occupancy; no differentiation or immunophenotype readout is reported. Read: abstract only.

    Evidence that argues against this (1)
  9. E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    The H2B-ubiquitination axis that RNF20 writes has a documented normal-haematopoiesis phenotype from the eraser side: an in vivo genome-wide CRISPR knockout screen in haematopoietic stem and progenitor cells identified SAGA complex members Tada2b and Taf5l as key regulators of haematopoiesis, with loss strongly inhibiting haematopoiesis in vivo, causing a buildup of immature cells in the bone marrow, reduced H3K9ac and altered H2Bub enrichment, upregulation of interferon pathway genes, reduced mitochondrial activity and increased megakaryocyte-progenitor commitment.

    In vivo CRISPR screening identifies SAGA complex members as key regulators of hematopoiesis. · Shankar A, Olender L, Hsu I, Miyauchi M, Pálovics R, Meaker GA, Kaito S, Rizq O, Khoo HM, Bozhilov Y, Igarashi KJ, Bhadury J, Munson C, Mack PK, Tan TK, Rehwinkel J, Iwama A, Wyss-Coray T, Nakauchi H, Haney MS, Wilkinson AC · Nat Commun 2026
    PMID 41577693DOI
    What this evidence cannot show

    This is NOT an RNF20 result and must not be written as one - Rnf20 was not among the reported hits. It is recorded because it establishes that the H2Bub level in normal HSPCs is functionally load-bearing, which is the relevant fact for the toxicity of a writer-directed agent. Mouse in vivo screen plus a human MDS cell-line model. Read: abstract only.

    Evidence that argues against this (1)
    • Attributing a complex-level or axis-level phenotype to one enzyme is the error rule 64 was written after, and the direction matters too: SAGA REMOVES H2Bub while RNF20 deposits it, so this screen does not predict the sign of an RNF20-loss phenotype in normal HSPCs. It bounds the question rather than answering it, and the RNF20 normal-haematopoiesis experiment remains undone. Wang E et al. 2013
Connected to