Oncofusion

start / graph / ana-unstated-system-rnf20-h2bub-resolved

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q18b-unstated-systems

"Diverse models of MLL-rearranged leukaemia" is one mouse cell line: the RNF20 H2Bub requirement resolves to murine RN2 (MLL-AF9/NrasG12D)

also called GSE43725 provenance · RNF20 MLL-rearranged system · RN2 provenance

A paper this graph cites for the claim that the H2B-tagging enzyme RNF20 is needed by MLL-rearranged leukaemia never says which leukaemia it used. Its deposited sequencing is three samples from a single mouse cell line, RN2, which carries MLL-AF9 - not the infant fusion this project studies.

The same thing, for a scientist

Three entries (`gene-rnf20`, `hyp-h2bub-direction-is-unknown-in-kmt2a`, `mark-h2bk120ub`) carry a caveat that PMID 23412334's fusions are *not specified in the abstract* [E1]. The deposited sequencing settles it. GEO series **GSE43725** holds exactly **three** samples, all *Mus musculus*, strain C57BL/6, `cell type: Acute myeloid leukemia`, `cell line: RN2` - a doxycycline-inducible shRenilla control and two independent Rnf20 hairpins (shRen.713, shRnf20.3277, shRnf20.3595) [E2]. The same laboratory's Mediator depositions annotate the identical RN2 line with `genotype: MLL-AF9/NrasG12D` [E3]. Of the deposited sequencing, the RNF20 requirement is therefore **one murine MLL-AF9 myeloid line carrying a cooperating NRAS G12D allele**, one sample per hairpin. Three imports stack - fusion (MLLT3, not AFF1), species, lineage - plus a fourth axis the caveat never raised: a cooperating RAS lesion that infant KMT2A::AFF1 typically lacks at diagnosis.

Go deeper
What GSE43725 actually contains
GSM titleOrganismStrainCell typeCell line
RN2/shRen.713Mus musculusC57BL/6Acute myeloid leukemiaRN2
RN2/shRnf20.3277Mus musculusC57BL/6Acute myeloid leukemiaRN2
RN2/shRnf20.3595Mus musculusC57BL/6Acute myeloid leukemiaRN2

The shared growth protocol reads "The Tet-on-competent leukemia cells (RN2) were cultured in RPMI1640 media supplemented with 10%FBS and 1%P/S" and the treatment protocol "RN2 cells harboring a Doxyclycline-inducible shRNA (shRen.713/shRNF20.3277/shRNF20.3595) were treated with doxycycline for 4 days and harvested." That is the entire deposited roster 2.

What this does and does not change

Does not change: the paper may well have run further models whose data were never deposited - western blots, colony assays, a second fusion. The bound here is of the deposited sequencing, not of the study. gene-rnf20's own caveat already notes that the H3K79me rescue experiments are at two loci rather than genome-wide, and that limit is untouched.

Does change: the phrase "diverse models" cannot be supported from the deposition, which is singular in fusion, species, lineage and cell line. Any entry that reads the RNF20 requirement as a general property of KMT2A rearrangement is generalising from n = 1 line. And RN2's NRAS G12D matters independently: H2B ubiquitination sits downstream of transcriptional output that RAS signalling also drives, so a cooperating oncogene is not a neutral background for this particular mark.

Facts and figures
scriptnone - read from the GEO accession-display text endpoint
statusrun
result summary3 of 3 deposited samples are Mus musculus, cell line RN2, cell type AML. 0 human samples. 0 KMT2A::AFF1 samples. Fusion identified as MLL-AF9 by cross-deposition genotype annotation of the same line.
counts
samples_human0.000samples_human: 0samples_kmt2a_aff10.000samples_kmt2a_aff1: 0distinct_cell_lines1.000distinct_cell_lines: 1samples_total3.000samples_total: 3samples_mouse3.000samples_mouse: 3
inputs
[
  "PMID 23412334",
  "GSE43725",
  "GSE74651",
  "GSE74650",
  "GSE74536",
  "GSE78221",
  "GSE78224"
]
counts — the numbers behind the chart
{
  "samples_total": 3,
  "samples_human": 0,
  "samples_mouse": 3,
  "samples_kmt2a_aff1": 0,
  "distinct_cell_lines": 1
}
entries affected
[
  "gene-rnf20",
  "hyp-h2bub-direction-is-unknown-in-kmt2a",
  "mark-h2bk120ub"
]
import axes
[
  "fusion: KMT2A::MLLT3 not KMT2A::AFF1",
  "species: mouse not human",
  "lineage: myeloid not B-lymphoid",
  "cooperating lesion: NRAS G12D"
]
How this could bite us back
  • An RNF20 or H2Bub-directed intervention prioritised from this evidence would be prioritised from a single mouse myeloid line with a cooperating RAS mutation, and a null result in infant KMT2A::AFF1 disease would be uninterpretable between wrong fusion, wrong species and no RAS background.
  • RNF20 is required for homologous recombination in normal cells, so inhibiting the H2Bub axis risks a genome-instability toxicity that a leukaemia-line viability screen is not designed to see.
  • The deposition's n of one per hairpin means the effect size behind this dependency is unbounded by the public data; a combination designed around it is being designed around an unreplicated measurement.
Still open
  1. Does RNF20 loss deplete H2BK120ub and impair growth in a human KMT2A::AFF1 pro-B line (SEM, RS4;11), where there is no cooperating RAS allele?
  2. Is the RNF20 requirement in RN2 separable from the NRAS G12D background? An isogenic RN2 line without the RAS allele would answer it and does not appear in any deposition.
Every source behind this entry (3)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The published RNF20 paper reports that the H2B ubiquitin ligase RNF20 is required for MLL-rearranged leukaemia, and its abstract does not name the MLL fusions used.

    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

    Title and identifiers taken from the NCBI esummary record fetched on 2026-09-19. The full text could not be read: Europe PMC fullTextXML for PMC3593849 returns HTTP 500 on repeated attempts and NCBI efetch db=pmc returns front matter with no body element, so this entry asserts nothing about the paper's Methods text.

    Evidence that argues against this (1)
    • RNF20's H2B ubiquitination activity has substantial non-leukaemic and fusion-independent functions - homologous-recombination control among them - so a requirement demonstrated in one leukaemia line does not establish that the dependency is fusion-driven rather than a general proliferation requirement. Nakamura K et al. 2011
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    GEO series GSE43725, the only deposition linked to PMID 23412334, contains three Mus musculus samples, all annotated strain C57BL/6, cell type 'Acute myeloid leukemia' and cell line 'RN2': one shRenilla control and two independent Rnf20 hairpins.

    What this evidence cannot show

    Read on 2026-09-19 from the GEO accession-display text endpoint (acc.cgi?targ=gsm&form=text&view=brief) and enumerated sample by sample. GEO characteristics are submitter-entered and unverified. GSE43725 was found via elink dbfrom=pubmed&db=gds; the db=gds term query 23412334[PMID] returns zero records, as it does for every PMID this beat tested.

    Evidence that argues against this (1)
    • Sample-level GEO annotation is entered by submitters without curation, and a systematic assessment of cell-line and mouse-strain fields across genomics depositions found them inconsistent enough to warrant automated re-annotation - so a cell-line string is a submitter's assertion, not an authenticated identity. Rogic S et al. 2026
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    The RN2 murine AML line is annotated genotype 'MLL-AF9/NrasG12D' and source name 'MLL-AF9/NrasG12D murine cell line' across five GEO series from the same laboratory.

    What this evidence cannot show

    This is a cross-deposition inference: GSE43725 itself records only 'cell line: RN2' and does not restate the genotype. The identification rests on the same line name carrying an explicit MLL-AF9/NrasG12D genotype in GSE74651, GSE74650, GSE74536, GSE78221 and GSE78224, and on the shared protocol string naming 'the murine AML RN2 cell line'. A line name is not a genotype assay, and lines drift.

    Evidence that argues against this (1)
    • Identifying a cell line from a name string across depositions is exactly the practice the cell-line authentication literature warns against; a shared label can hide divergent sublines, and only STR or genotyping establishes identity. Souren NY et al. 2022
Connected to