Oncofusion

start / graph / gene-rnf2

Proteinmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q5-interactome4

RNF2 (RING1B) — the one catalytic subunit shared by every PRC1 flavour, which is why the genetics point two ways at once

also called RNF2 · RING1B · RING2 · DING · HIPI3 · BAP-1 · E3 ubiquitin-protein ligase RING2

RNF2 is the enzyme that attaches a small tag (ubiquitin) to histone H2A, the chemical mark that Polycomb complexes use to keep genes quiet. Because the same enzyme sits inside every version of Polycomb complex 1, deleting it removes several different machines at once — which is why experiments that delete it and experiments that delete only its PRC1.1 partners give opposite-looking answers.

The same thing, for a scientist

RNF2/RING1B (Q99496, 336 aa) is the RING-type E3 ubiquitin ligase that, with RING1A, deposits **H2AK119ub**. It is the invariant catalytic core of all six PCGF-defined PRC1 complexes, PRC1.1 included [E1]. Genetic depletion of Ring1a and Ring1b in mouse leukaemia showed that **PRC1 activity and H2AK119Ub deposition are required for the leukemogenic potential of both Hox-activating fusions (MLL-AF9) and non-Hox-activating fusions**, preventing differentiation independently of the Cdkn2a locus [E2]. BCOR loss-of-function reduces RING1B **protein** levels and lowers H2AK119ub upstream of up-regulated HoxA genes, so PRC1.1 integrity and RING1B abundance are coupled [E3]. A physical link to an MLL fusion partner exists but is to the *canonical* complex: ATM-phosphorylated ENL (MLLT1) binds the PRC1 E3 module via BMI1 [E4]. **RNF2 is not a KMT2A-selective dependency in DepMap** — rank 16,819 of 17,728, difference **+0.096**, pan-essentiality only -0.047 [E5]. The two datasets are not contradictory so much as measuring different things, and this entry argues that explicitly.

What it looks like
E3 ubiquitin-protein ligase RING2336 aa
RING (RING-type zinc finger)RING (RING-type zinc finger) · residues 51–91Nucleosome acidic-patch binding region · residues 93–980100200300
loading AF-Q99496-F1…
AlphaFold AF-Q99496-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
Why RNF2 is the awkward node in the PRC1.1 story

The beat's PRC1.1 entries all point one way: PCGF1, BCOR, BCORL1 loss confers drug resistance, and all three sit on the anti-dependency side of the genome-wide scan 5. RNF2 sits on the same side of the scan (+0.096, 94.9th percentile) — but the one clean in vivo genetic experiment says PRC1 catalytic activity is required for MLL-AF9 leukaemia 2.

Three things reconcile this without waving either result away:

  1. RNF2 is not a PRC1.1 gene. It is in all six PRC1 flavours 1. Deleting Ring1a/Ring1b removes canonical PRC1 (BMI1/CBX/PHC) at the same time, and the MLL-AF9 requirement reported by Rossi et al. is a requirement for PRC1 activity in general, not for PRC1.1. Deleting only the PCGF1 arm leaves the canonical arm intact.
  2. Paralogue redundancy. RING1A (RING1) covers for RING1B, which is why Rossi et al. had to deplete both. A single-gene CRISPR screen such as DepMap knocks out RNF2 alone and will under-read the requirement by exactly that amount.
  3. Assay. Rossi's readout is leukaemogenic potential and differentiation in mice; DepMap's is bulk proliferation of established human lines.

So the honest statement is: PRC1 catalysis is required for MLL-AF9 leukaemia; RNF2 as a single gene is not a selective vulnerability of KMT2A-rearranged cell lines. Both can be true and the graph should carry both.

What RNF2 does, mechanically

RING1B's RING domain (UniProt aa 51-91) heterodimerises with a PCGF RING domain, and a short region at aa 93-98 is annotated as the contact with the nucleosome acidic patch on H2A/H2B — the geometry that positions the catalytic module over H2A Lys119 6. RYBP association stimulates RING1B activity toward H2AK119ub1 and simultaneously excludes the canonical CBX/PHC/SCM subunits, which is the biochemical basis for calling RYBP-containing assemblies 'variant' PRC1 1.

The coupling to PRC1.1 integrity runs both ways: mutating Bcor lowers RING1B protein and reduces H2AK119ub upstream of up-regulated HoxA genes in murine myeloid cells, with enhanced proliferation and differentiation 3. That is a direct mechanistic bridge between this entry, gene-bcor, and the HOXA programme the fusion drives.

Cell type, cell state, localisation and epigenetic readout
AxisDocumentedWhere
Cell typeMouse MLL-AF9 leukaemia and non-Hox-fusion leukaemia; murine myeloid bone marrow cells23
Cell statePRC1 activity prevents differentiation and maintains the undifferentiated identity of the leukaemic cell; BCOR loss (and consequent RING1B loss) increases proliferation and differentiation23
SubcellularNucleus, chromatin-bound1
SubnuclearNot established in a KMT2A-rearranged cell. RING1B is classically associated with Polycomb bodies in other systems, but no such assignment is published for a KMT2A-fusion leukaemia and none is asserted here.—
Epigenetic readoutH2AK119ub1, catalysed directly; loss removes the mark globally12

No B-lineage or KMT2A::AFF1 result exists for RNF2. Everything above is myeloid.

Facts and figures
uniprotQ99496
gene symbolRNF2
protein nameE3 ubiquitin-protein ligase RING2
length AA336
alphafold IDAF-Q99496-F1
accession resolutionUniProt REST query gene_exact:RNF2 AND organism_id:9606 AND reviewed:true, run 2026-09-19; one reviewed hit; primary gene name read back as RNF2 and length 336 confirmed from the same fetched record.
complexcatalytic subunit of ALL PRC1 complexes, canonical and variant, PRC1.1 included
paralogueRING1 (RING1A) - functionally redundant; Rossi et al. deleted both to reveal the phenotype
catalytic activityE3 ubiquitin ligase depositing H2AK119ub1 (with RING1A); stimulated by RYBP/YAF2
classificationshared catalytic core of PRC1; NOT a demonstrated interactor of KMT2A::AFF1
classification reasonno published experiment places RING1B on a KMT2A fusion protein; the nearest physical link is ENL-BMI1 in the DNA damage response, using wild-type ENL
interaction evidence classconditional mouse genetics; co-IP (ENL-BMI1, non-leukaemic); proteomics of PRC1 composition
shown for fusionMLL-AF9 and non-Hox-activating fusions in mouse; nothing in KMT2A::AFF1 or B-lineage
epigenetic readoutH2AK119ub1 (deposited directly)
cell context evidence classmaintains_a_state (explicit, with an epistasis arm) + required_in (for leukaemogenic potential). Both at the RING1A/RING1B pair level, not at the RNF2 level.
cell context shown for fusionPARTIAL and indirect. Hox-activating fusions in mouse, which in this literature means MLL-AF9 and equivalents, plus non-Hox-activating fusions; and MOZ-TIF2, which is a KAT6A fusion and not a KMT2A fusion. Measured 2026-09-21: (RNF2[tiab] OR RING1B[tiab]) AND (KMT2A[tiab] OR MLL[tiab]) AND ("B-ALL"[tiab] OR "lymphoblastic"[tiab] OR AF4[tiab] OR AFF1[tiab]) returns 0. Zero in titles and abstracts; supporting figures not searched. There is NOTHING in KMT2A::AFF1 and nothing in B-lineage.
cell context tension and its resolutionRNF2 is the MOST anti-selective gene in this tranche - 94.87th percentile of 17,728 on this project's KMT2A-selectivity scan, -0.028 in the 11 KMT2A-rearranged lines against -0.125 in 43 other leukaemia lines, difference +0.096, pan-line mean -0.047. A single-gene CRISPR screen finds RNF2 to be essentially nothing anywhere. The resolution is straightforward and worth stating because it applies to several genes in this graph: BOTH published requirement results delete Ring1A and Ring1B together, and RING1 is a functional paralogue of RNF2 in the same complex. A single-gene screen cannot detect a requirement carried redundantly by a paralogue pair, so the DepMap null and the genetic requirement are compatible - but it also means the DepMap arm provides NO independent support, and the whole case rests on two mouse double-knockout studies in fusions that are not the index fusion.
cell context symbol cautionThe registry records three traps on this symbol: RNF2 lists BAP1 as a gene synonym, the protein is universally called RING1B while the UniProt entry name is RING2_HUMAN, and RING1 is a strict PREFIX of RING1B so any substring count conflates two different genes. All three matter here, because the load-bearing experiments are Ring1A/Ring1B double deletions and a careless reading of 'Ring1' in those papers assigns the result to the wrong gene.
aliases retired or alternate
[
  "RING1B",
  "RING2",
  "DING",
  "HIPI3"
]
complex partners
[
  "gene-pcgf1",
  "gene-bcor",
  "gene-bcorl1",
  "gene-kdm2b",
  "gene-skp1",
  "gene-rybp"
]
domains
[
  {
    "name": "RING (RING-type zinc finger)",
    "start": 51,
    "end": 91,
    "source": "UniProt Q99496, fetched 2026-09-19"
  },
  {
    "name": "Nucleosome acidic-patch binding region",
    "start": 93,
    "end": 98,
    "source": "UniProt Q99496, fetched 2026-09-19"
  }
]
localization
{
  "subcellular": "nucleus, chromatin-bound",
  "subnuclear": "not established in any KMT2A-rearranged system",
  "chromatin_targeting": "indirect - via whichever PCGF/targeting module it is bound to; PRC1.1 targets CpG islands via KDM2B"
}
essentiality
{
  "depmap_scan": "analyses/depmap-kmt2a-selective-scan-full",
  "pan_essentiality_all_lines": -0.0474,
  "n_all_lines": 1208,
  "mean_kmt2a_rearranged": -0.0283,
  "mean_other_leukemia": -0.1247,
  "kmt2a_selectivity_difference": 0.0964,
  "rank": 16819,
  "n_scanned": 17728,
  "percentile": 94.873,
  "interpretation": "no KMT2A selectivity as a single gene; the RING1A paralogue is untested in this assay and almost certainly masks the requirement"
}
cell type
{
  "required_in": "Leukaemic cells of several fusion classes, in mouse. Genetic depletion of the PRC1 H2A ubiquitin ligase subunits Ring1a and Ring1b showed PRC1 activity is required for the leukaemogenic potential of both Hox-activating and non-Hox-activating fusions. Separately, deleting Ring1A/B from MOZ-TIF2 AML cells diminished self-renewal capacity.",
  "critical_caveat": "BOTH studies delete RING1A AND RING1B TOGETHER. Single RNF2/RING1B loss is not the tested perturbation anywhere in this literature, so every requirement statement here is a statement about the PRC1 catalytic pair, not about RNF2. That is also the most economical explanation of the DepMap result below.",
  "systems": "mouse; MOZ-TIF2 is not a KMT2A fusion at all",
  "nodes": []
}
cell state
{
  "maintains_state": "Undifferentiated leukaemic identity, stated in those terms by the source. PRC1 activity and H2AK119Ub deposition prevent the differentiation of leukaemic cells INDEPENDENTLY of the Cdkn2a locus, and are described as critical factors maintaining the undifferentiated identity of cancer cells across different types of leukaemia. In the MOZ-TIF2 model Ring1A/B maintain AML stem cells in part by repressing Glis2: Glis2 overexpression makes the cells differentiate into mature cells, and Glis2 knockdown in Ring1A/B-deficient cells blocks that differentiation.",
  "reading": "This is one of only two entries in the tranche where a source states identity maintenance in its own words rather than leaving it to be inferred from a growth curve. The Glis2 epistasis is what makes it a mechanism rather than a phenotype.",
  "nodes": [
    "cell-state-differentiation-block",
    "cell-state-stem-like-self-renewal"
  ]
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.0283,
  "chronos_mean_other_leukaemia": -0.1247,
  "difference": 0.0964,
  "pan_essentiality_all_lines": -0.0474,
  "selectivity_rank": 16819,
  "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."
}
How this could bite us back
  • H2AK119ub is a global, developmentally essential mark. A RING1B catalytic inhibitor is a whole-organism Polycomb inhibitor and would carry the toxicity profile of one, not of a leukaemia-selective agent.
  • RING1A redundancy means a selective RING1B inhibitor is likely to be buffered in exactly the tissues where it works, giving apparent tolerability and apparent inefficacy together.
  • Because PRC1.1 loss confers menin-inhibitor resistance, an agent that lowers PRC1 output could antagonise the drug this project is built around if the two were combined.
  • Rossi's result is that PRC1 maintains the undifferentiated leukaemic identity - so removing it releases differentiation, which in a KMT2A-rearranged patient is the same physiology that produces differentiation syndrome.
Still open
  1. Does dual RING1A/RING1B loss, or a catalytic-dead RING1B knock-in, impair KMT2A::AFF1 B-ALL as it impairs MLL-AF9 AML? The experiment has been done in myeloid mouse models and not in B-lineage.
  2. If PRC1 catalysis is required for leukaemic identity (Rossi) but PRC1.1 loss confers menin-inhibitor resistance (Zhou), which PRC1 flavour supplies which function? A PCGF-paralogue-resolved H2AK119ub map in a KMT2A-rearranged line would answer it.
  3. RING1B protein is destabilised by BCOR loss. Is RING1B protein level, rather than PRC1.1 genotype, the thing that predicts menin-inhibitor response?
  4. Are there RING1B catalytic inhibitors with enough selectivity over RING1A to test the redundancy question pharmacologically?
  5. Every requirement result for RNF2 in leukaemia comes from a Ring1A/Ring1B double deletion, and the single-gene CRISPR screen is flat. Is RING1 sufficient to carry H2AK119Ub in a KMT2A-rearranged B-ALL blast? If it is, then RNF2 alone is not a target in this disease and the entry's degree overstates its standing; if it is not, the screen null needs another explanation.
  6. PRC1 maintains undifferentiated identity in Hox-activating fusions in mouse and nothing has been done in KMT2A::AFF1 or in any B-lineage system. Does H2AK119Ub occupancy at the HOXA cluster in KMT2A::AFF1 B-ALL look like the myeloid case at all? This graph's PRC1 layer currently generalises across a lineage boundary it has not tested.
Every source behind this entry (8)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Each of the six major PRC1 groups contains a distinct PCGF subunit together with a RING1A/B ubiquitin ligase; association with RYBP or YAF2 prevents incorporation of canonical CBX, PHC and SCM subunits, and although both RYBP- and CBX/PHC/SCM-containing complexes compact chromatin, only RYBP stimulates RING1B activity toward H2AK119ub1.

    PCGF homologs, CBX proteins, and RYBP define functionally distinct PRC1 family complexes · Gao Z, Zhang J, Bonasio R, Strino F, Sawai A, Parisi F, Kluger Y, Reinberg D · Molecular Cell 45(3):344-56 2012
    PMID 22325352DOI
    What this evidence cannot show

    Abstract fetched and read on 2026-09-19; full text not read. HEK293 and mouse embryonic stem cells; no leukaemia system. 'Six major groups' is a proteomic classification and later work has added variant compositions.

    Evidence that argues against this (1)
    • The clean in-vitro division of labour does not survive intact in cells: BCOR PUFD alone fails to assemble stably with KDM2B and needs additional N-terminal residues and a prior PCGF1 contact, so complex membership measured on purified fragments overstates how discrete these assemblies are in vivo. Wong SJ et al. 2020
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Genetic depletion of the H2A ubiquitin ligase subunits Ring1a and Ring1b in mice showed that HOXA9 activation cannot repress the Cdkn2a locus in the absence of PRC1 and its H2AK119 monoubiquitination, and that PRC1 activity is required for the leukemogenic potential of both Hox-activating fusions (such as MLL-AF9) and non-Hox-activating fusions, preventing differentiation of leukaemic cells independently of Cdkn2a expression.

    Maintenance of leukemic cell identity by the activity of the Polycomb complex PRC1 in mice · Rossi A, Ferrari KJ, Piunti A, Jammula S, Chiacchiera F, Mazzarella L, Scelfo A, Pelicci PG, Pasini D · Science Advances 2(10):e1600972 2016
    PMID 27730210DOI
    What this evidence cannot show

    Abstract fetched and read on 2026-09-19; full text not read. Mouse models only, and both Ring1a and Ring1b are deleted together, so the result establishes a requirement for PRC1 catalysis and not for RNF2 as a single gene. The fusion is MLL-AF9 (myeloid), not KMT2A::AFF1. The PubMed abstract text is gene-name-stripped in places, so the identity of the specific loci was read from context.

    Evidence that argues against this (1)
    • The unbiased single-gene dependency data point the other way: RNF2 ranks 16,819 of 17,728 for KMT2A-rearranged selectivity with a difference of +0.096, and is barely essential anywhere (pan-essentiality -0.047 across 1,208 lines). Deleting one paralogue of a redundant pair in a proliferation assay is not the same experiment as deleting both in vivo, but it does mean 'inhibit RING1B' is not supported as a KMT2A-selective strategy. Broad Institute DepMap / Open Targets Platform; analyses/depmap-kmt2a-selective-scan-full 2026
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Bcor loss-of-function in murine myeloid cells produced significantly higher proliferation and differentiation with upregulated Hox gene expression; mutation of Bcor reduced protein levels of RING1B, the H2A ubiquitin ligase subunit of PRC1 family complexes, and reduced H2AK119ub upstream of the upregulated HoxA genes.

    BCOR regulates myeloid cell proliferation and differentiation · Cao Q, Gearhart MD, Gery S, Shojaee S, Yang H, Sun H, Lin DC, Bai JW, Mead M, Zhao Z, Chen Q, Chien WW, Alkan S, Alpermann T, Haferlach T, Muschen M, Bardwell VJ, Koeffler HP · Leukemia 30(5):1155-65 2016
    PMID 26847029DOI
    What this evidence cannot show

    Abstract fetched and read on 2026-09-19; full text not read. Conditional Bcor alleles in murine myeloid cells plus expression profiling of BCOR-mutant AML patient samples; KMT2A rearrangement is not a selection criterion, so the HoxA link here is about normal and BCOR-mutant myelopoiesis, not about a KMT2A fusion driving HOXA.

    Evidence that argues against this (1)
    • The direction is uncomfortable for a simple 'PRC1.1 restrains the leukaemia' story: BCOR loss here increases BOTH proliferation and differentiation, whereas the KMT2A-rearranged menin-resistance work reports PRC1.1 loss reducing monocytic differentiation signatures. Either the phenotype depends on the driver or on the lineage stage, and neither paper tests the other's system. Zhou X et al. 2024
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    ATM phosphorylates the MLL fusion partner ENL (MLLT1) at conserved SQ sites after DNA double-strand breaks; this increases ENL's interaction with the PRC1 E3-ubiquitin-ligase complex via BMI1, enriching PRC1 at transcription elongation sites near breaks and ubiquitinating H2A to repress transcription, with ENL SQ sites and BMI1 both required for KU70 accumulation and cellular resistance to double-strand breaks.

    PMID 25921070DOI
    What this evidence cannot show

    Abstract fetched and read on 2026-09-19; full text not read. This is a DNA-damage-response phenomenon in non-leukaemic human cells, with wild-type ENL, not an MLL-ENL fusion, and no KMT2A fusion protein was tested. The PRC1 involved is the BMI1 (PCGF4) canonical complex, so it is a RING1B contact by way of the canonical arm and not evidence about PRC1.1.

    Evidence that argues against this (1)
    • The authors' own follow-up review frames this as a collaboration that 'may be link to the MLL-ENL/AF9 leukemia' - explicitly a conjecture, not a result. No experiment in either paper puts RING1B on an MLL fusion protein, and this entry does not treat it as one. Ui A et al. 2016
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    In the project's genome-wide KMT2A-rearranged selectivity scan of 17,728 genes, RNF2 ranks 16,819 (94.873rd percentile) with a difference of +0.0964 (KMT2A-rearranged mean -0.0283 against other-leukaemia mean -0.1247) and a pan-essentiality of -0.0474 across 1,208 lines.

    Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap Chronos via Open Targets) · Broad Institute DepMap / Open Targets Platform; analyses/depmap-kmt2a-selective-scan-full 2026
    What this evidence cannot show

    Numbers read out of analyses/depmap-kmt2a-selective-scan-full/results.json on 2026-09-19. Eleven KMT2A-rearranged against 43 other leukaemia lines, none infant-derived, AML-heavy. Critically, RNF2 has a functional paralogue (RING1/RING1A) that a single-gene CRISPR knockout does not touch, so a near-zero score is the expected reading for a redundant pair and is weak evidence of dispensability.

    Evidence that argues against this (1)
    • Rossi et al. had to delete Ring1a AND Ring1b to reveal the requirement, and when they did, PRC1 catalysis was necessary for MLL-AF9 leukaemogenesis. The single-gene scan therefore cannot settle whether PRC1 catalysis matters here; it only says RNF2 alone is not a line-level vulnerability. Rossi A et al. 2016
  6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB Q99496 (RING2_HUMAN, E3 ubiquitin-protein ligase RING2) is 336 aa and annotates a RING-type zinc finger at aa 51-91 and a region at aa 93-98 that interacts with nucleosomes via the acidic patch on histones H2A and H2B.

    UniProtKB Q99496 (RING2_HUMAN), E3 ubiquitin-protein ligase RING2 · UniProt Knowledgebase, reviewed (Swiss-Prot) 2026
    What this evidence cannot show

    Record fetched 2026-09-19; primary gene name read back as RNF2 and length 336 confirmed from the same fetched record, not from recall. Note that the UniProt protein name is RING2 while the HGNC symbol is RNF2 and the literature overwhelmingly writes RING1B - three names for one protein, which is exactly the retired-symbol trap this project has been caught by before.

    Evidence that argues against this (1)
    • The naming divergence is not cosmetic: a PubMed query on the current HGNC symbol alone under-counts this literature badly. '(MLL OR KMT2A) AND (RNF2 OR RING1B OR RING1A OR RING1)' returns 8 records, whereas the informative papers are indexed under RING1B and BMI1 rather than RNF2. Any counted-absence claim about this gene run on the symbol RNF2 would be wrong. NCBI PubMed, queried 2026-09-19 2026
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    PRC1 activity maintains the undifferentiated identity of leukaemic cells: genetic depletion of the H2A ubiquitin ligase subunits Ring1a and Ring1b in mice shows that PRC1 and its deposition of H2AK119 monoubiquitination are required for the leukaemogenic potential of both Hox-activating and non-Hox-activating fusion proteins, preventing the differentiation of leukaemic cells independently of Cdkn2a expression, and that Hoxa9 activation cannot repress Cdkn2a in the absence of PRC1.

    Maintenance of leukemic cell identity by the activity of the Polycomb complex PRC1 in mice. · Rossi A, Ferrari KJ, Piunti A, Jammula S, Chiacchiera F, Mazzarella L, Scelfo A, Pelicci PG, Pasini D · Sci Adv 2016
    PMID 27730210DOI
    What this evidence cannot show

    Mouse, and the perturbation is a Ring1a/Ring1b DOUBLE depletion - nothing here isolates RNF2/RING1B from its paralogue. The fusions are described as Hox-activating and non-Hox-activating classes rather than named individually in the abstract, and KMT2A::AFF1 is not among the systems discussed. Read: abstract only.

    Evidence that argues against this (1)
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In a mouse model of MOZ-TIF2 AML, Ring1A and Ring1B are required to maintain leukaemia stem cells: their deletion diminishes self-renewal capacity and induces expression of numerous genes including Glis2; Glis2 overexpression causes the AML cells to differentiate into mature cells, and Glis2 knockdown in Ring1A/B-deficient cells blocks that differentiation.

    Ring1A and Ring1B inhibit expression of Glis2 to maintain murine MOZ-TIF2 AML stem cells. · Shima H, Takamatsu-Ichihara E, Shino M, Yamagata K, Katsumoto T, Aikawa Y, Fujita S, Koseki H, Kitabayashi I · Blood 2018
    PMID 29371181DOI
    What this evidence cannot show

    Mouse, and MOZ-TIF2 is a KAT6A::NCOA2 fusion - NOT a KMT2A fusion - so this is an analogy to the index disease rather than evidence within it. Again a Ring1A/Ring1B double deletion, so it does not isolate RNF2. The Glis2 epistasis is the strongest part of the result and is the reason this is graded as identity maintenance rather than as a growth phenotype. Read: abstract only.

    Evidence that argues against this (1)
    • Nothing in the RNF2 cell-context literature reaches KMT2A::AFF1 or B-lineage leukaemia. Measured 2026-09-21: (RNF2[tiab] OR RING1B[tiab]) AND (KMT2A[tiab] OR MLL[tiab]) AND ("B-ALL"[tiab] OR "lymphoblastic"[tiab] OR AF4[tiab] OR AFF1[tiab]) returns 0 records, with large operands on both sides. Zero in titles and abstracts; supporting figures not searched. Two mouse models of two non-index fusions is the entire cell-state case. Rossi A et al. 2016
Connected to