Oncofusion

start / graph / gene-msi2

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q6-interactome-rna-layer

MSI2 (Musashi-2) — sustains the MLL stem-cell programme in mouse, and the myeloid DepMap arm disagrees in sign

also called Musashi-2 · Musashi2 · MSI2H · musashi RNA binding protein 2

Musashi-2 keeps the MLL leukaemia stem-cell programme running by making sure key messages such as HOXA9 and MYC are translated. In human myeloid KMT2A-rearranged cell lines, though, the dependency data point the wrong way.

The same thing, for a scientist

Conditional Msi2 deletion in the haematopoietic compartment delays leukaemogenesis, reduces disease burden and causes loss of leukaemia-stem-cell function in a murine MLL leukaemia model; MSI2 directly maintains the MLL self-renewal programme by retaining efficient translation of Hoxa9, Myc and Ikzf2 mRNAs, and an Msi2-deficient LSC gene signature correlates with improved survival in AML patients [E1]. It is also the hub whose interactome screen produced SYNCRIP [E2]. **The DepMap data disagree in the arm where the literature is**: MSI2 is -0.284 in four B-lymphoid KMT2A-rearranged lines and **+0.154** in the seven myeloid ones, lineage-discordant, with a Chronos mean of -0.159 and only 5.7% of 1,208 lines depleted [E3].

Go deeper
The disagreement, and which way to read it

The published requirement is myeloid and murine. The myeloid half of this project's index arm gives MSI2 a positive difference - the seven KMT2A-rearranged myeloid lines are, if anything, less dependent on MSI2 than 43 other leukaemia lines 3.

Seven cell lines do not overturn a conditional-knockout mouse experiment, and the entry does not claim they do. But the disagreement is recorded rather than omitted, for the same reason this graph records METTL3's lineage discordance: quoting the pooled eleven-line number would have produced a small negative figure that read as agreement with the literature and is nothing of the kind.

The most likely explanation is the one that recurs through this beat: MSI2's reported phenotype is leukaemia-stem-cell function and self-renewal, measured in vivo over months. A two-to-three-week proliferation dropout in established lines cannot see it, and a gene depleted in 5.7% of 1,208 lines is essentially invisible to that assay.

MSI2 as the hub rather than the target

MSI2's most durable contribution to this field may be methodological. Proteomic analysis of the MSI2-interacting RNA-binding protein network, followed by functional shRNA screening, identified 24 genes required for in vivo leukaemia and produced SYNCRIP 2.

That is an interactome-first enumeration of the RNA layer, and it is the one route this beat did not run: it started from a protein hub and asked which of its partners were functionally required in vivo. The equivalent experiment has never been done from IGF2BP3 in a B-lymphoid KMT2A::AFF1 context, and it would be the natural way to enumerate that operon's protein partners rather than its transcript targets.

Facts and figures
uniprotQ96DH6
length AA328
gene symbolMSI2
uniprot resolutiongene_exact:MSI2 AND organism_id:9606 AND reviewed:true; one hit; primary gene name read back as MSI2; length 328 aa checked
rna functionMusashi-family RNA-binding protein; retains efficient translation of HOXA9, MYC and IKZF2 mRNAs
essentiality
{
  "depmap_common_essential": false,
  "chronos_mean": -0.1589,
  "pct_lines_depleted": 5.7,
  "n_lines": 1208,
  "kmt2a_r_lymphoid_diff": -0.2843,
  "kmt2a_r_myeloid_diff": 0.1543,
  "null_percentile": 48.93,
  "lineage_discordant": true
}
pubmed counts
{
  "MSI2 AND leukemia": 71,
  "(MLL OR KMT2A) AND (MSI2 OR Musashi-2 OR Musashi2)": 3
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.291,
  "chronos_mean_other_leukaemia": -0.2858,
  "difference": -0.0051,
  "pan_essentiality_all_lines": -0.1589,
  "selectivity_rank": 8478,
  "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
  • The mechanism runs through HOXA9 translation and infant t(4;11) is HOXA-low, so the published route may be absent from the index disease.
  • MSI2 is required for normal HSC function as well as leukaemic self-renewal, which is the essential-function toxicity pattern this graph has repeatedly found in self-renewal targets.
  • An RNA-binding protein with no catalytic pocket and a reported requirement measured only in vivo is the hardest possible combination to develop chemistry against.
Still open
  1. Why does the myeloid half of the KMT2A-rearranged index arm disagree in sign with the murine MLL requirement?
  2. Has anyone run the MSI2 interactome-plus-in-vivo-screen design starting from IGF2BP3 in a B-lymphoid KMT2A::AFF1 context?
  3. Is MSI2 required in infant t(4;11) disease, which is HOXA-low, given that the mechanism runs through HOXA9 translation?
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Conditional deletion of Msi2 in the haematopoietic compartment delays leukaemogenesis, reduces disease burden and causes loss of leukaemia-stem-cell function in a murine MLL leukaemia model; MSI2 maintains the MLL self-renewal programme by interacting with and retaining efficient translation of Hoxa9, Myc and Ikzf2 mRNAs, and an Msi2-deficient LSC-like signature correlates with improved survival in AML patients.

    PMID 25664853DOI
    What this evidence cannot show

    Abstract read; full text not read. Murine, myeloid MLL leukaemia. A conditional deletion applied in the haematopoietic compartment rather than pharmacological inhibition of established disease.

    Evidence that argues against this (1)
    • In this project's DepMap scan the seven myeloid KMT2A-rearranged lines give MSI2 a positive difference of +0.154 against 43 other leukaemia lines - the opposite sign to the published myeloid requirement - and MSI2 is depleted in only 5.7% of 1,208 lines overall. Oncofusion knowledge graph (internal analysis) 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Proteomic analysis of the MSI2-interacting RNA-binding protein network combined with functional shRNA screening identified 24 genes required for in vivo leukaemia, of which SYNCRIP was the most differentially required between normal and myeloid leukaemia cells.

    PMID 28436985DOI
    What this evidence cannot show

    Abstract-level. Myeloid. Cited here for the enumeration method as much as for the biology.

    Evidence that argues against this (1)
    • An interactome-plus-screen enumeration inherits the biases of both: it finds partners of the chosen hub, and it scores them on the screen's readout. Starting from a different hub would produce a different 24 genes. Oncofusion knowledge graph (internal analysis) 2026
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.analysis

    DepMap CRISPR (Chronos) gives MSI2 a mean gene effect of -0.159 across 1,208 lines with 5.7% depleted below -0.5, and lineage-DISCORDANT index-arm differences of -0.284 (four B-lymphoid) and +0.154 (seven myeloid) against 43 other leukaemia lines, at the 48.9th percentile of a 466-gene null.

    What this evidence cannot show

    Four and seven index lines. Discordance across the split means the pooled number is meaningless and neither half is a result; both are reported because the myeloid half contradicts the literature it should agree with.

    Evidence that argues against this (1)
    • A gene depleted in 5.7% of lines has almost no signal for a lineage split to partition, so the discordance may be noise rather than biology. Reading either half as informative would be over-interpretation. Oncofusion knowledge graph (internal analysis) 2026
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB Q96DH6 (MSI2H_HUMAN) is human MSI2, 328 amino acids; resolved by gene_exact query with the primary gene name read back as MSI2 and the length checked.

    What this evidence cannot show

    Single reviewed human hit whose primary gene name equals the query symbol. This project has logged six cases of gene_exact returning a protein that merely lists the queried symbol as a synonym, so the read-back is the point, not the query.

    Evidence that argues against this (1)
    • An accession establishes identity and length. It says nothing about whether this protein matters in KMT2A-rearranged leukaemia, which is what the rest of this entry is for. UniProt Consortium 2026
Connected to