Oncofusion

start / graph / cell-state-mtap-deleted-mta-high

Cell statestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q73-vocabulary

The MTAP-deleted, MTA-high state - a genotype-imposed metabolic state that this project's KMT2A-rearranged arm turns out NOT to have, and that its own screen could never have detected directly

also called MTAP deletion · MTAP-null state · MTA-high state · 9p21 co-deletion state · hypomorphic PRMT5 state

When a tumour deletes the tumour-suppressor gene CDKN2A it usually deletes a neighbouring housekeeping gene, MTAP, by accident. Losing MTAP lets a waste metabolite build up that partly jams an enzyme called PRMT5, which leaves those cells unusually easy to finish off by blocking that enzyme - and it also makes drug-sensitivity numbers in mixed cell-line panels hard to interpret.

The same thing, for a scientist

MTAP sits at 9p21 beside CDKN2A and is co-deleted as a passenger in a large share of cancers. Deletion causes intracellular accumulation of methylthioadenosine, MTA is a potent and selective inhibitor of PRMT5, and the resulting hypomorphic PRMT5 state sensitises the cell to further inhibition of PRMT5, of its partner WDR77, of the SAM-producing enzyme MAT2A, and of the PRMT5 co-complex protein RIOK1 [E1][E2][E3]. This graph needs the term for a specific and immediate reason: `gene-mat2a` recorded it as a declined vocabulary gap, and cycle 40 raised the possibility that MAT2A's selectivity rank in this project's KMT2A-selective scan was a property of the eleven-line panel rather than of the fusion. **That was tested this cycle and rejected** - PRMT5 (rank 17,006 of 17,728) and WDR77 (15,455) are both *less* depleted in the rearranged arm, the opposite of what MTAP deletion predicts [E4]. The node also records the methodological point that made the question hard: **a homozygously deleted gene has nothing for CRISPR to cut and scores near zero on Chronos regardless of biology**, so MTAP's own dependency score is uninformative by construction and the state has to be detected through its synthetic-lethal partners or by copy number [E4]. MTAP lesions are not exotic in this disease's neighbourhood: MTAP and CDKN2A are homozygously co-deleted at 35-70% in several tumour types including T-cell ALL [E5], and MTAP rearrangements were found in 9.47% of one 338-patient paediatric ALL cohort, enriched in T-ALL and associated with inferior outcome [E6].

Go deeper
Why a genotype gets a cell-state node at all

This graph's cell-state nodes are conditions a cell is in. MTAP deletion is a genotype, but what it does is impose a metabolic condition - MTA accumulation and a partially inhibited PRMT5 - that persists for the life of the clone and changes which drugs work. gene-mat2a made exactly this point when it declined the term: the MTAP-deletion vulnerability 'is a genotype rather than a cell state and also has no node'. The resolution taken here is to write it as a state and say in the entry that it is genotype-imposed, because the alternative - no term at all - is what stopped the question being posed for two cycles.

The practical value is that the term lets a selectivity claim be challenged. Any panel-based dependency on MAT2A, PRMT5, WDR77 or RIOK1 in mixed cell lines has to exclude this state before it is a finding about the biology under study.

The question this node was written to let someone ask, and the answer

Cycle 40 flagged that MAT2A's rank of 911 of 17,728 in this project's KMT2A-selective DepMap scan could be an artefact of the eleven-line panel if those lines were MTAP-enriched. CDKN2A deletion is among the commonest lesions in ALL and MTAP travels with it, so the worry was well founded.

Tested against the cached full scan (11 KMT2A-rearranged lines against 43 other leukaemia lines): PRMT5 rank 17,006 and WDR77 rank 15,455, both LESS depleted in the rearranged arm - the opposite direction to what MTAP deletion predicts. MAT2A (911) and RIOK1 (969) are consistent with a requirement, but the two genes that would report the deletion point away from it. No evidence of enrichment 4.

Two honest limits on that answer. It infers genotype from phenotype rather than measuring copy number, and a direct lookup on the named lines has not been done. And all four axis genes are pan-essential on a compressed scale with no computable interval from the cached arm means, so the magnitudes are not to be quoted as measured effects.

The detection trap, which generalises well past MTAP

A homozygously deleted gene scores near zero in a CRISPR dependency screen regardless of biology, because there is nothing to cut. MTAP's own Chronos rank in this project's scan is 7,038 with positive gene effects in both arms, and that number carries no information about MTAP status in either arm 4.

This is the same category error this graph has now refused three times for CDKN2A, SIRT1 and ATM in different guises: reading a screen's output on a gene as though the screen could see the gene's genomic state. A dependency screen measures what happens when you cut a gene; it cannot report a gene that is already gone. The state has to be detected either by copy number, or through the synthetic-lethal partners that remain cuttable.

How common the lesion is where this disease lives

MTAP lies at 9p21 next to CDKN2A, and the two are homozygously co-deleted with a frequency of 35 to 70% in lung and pancreatic cancer, glioblastoma, osteosarcoma, soft-tissue sarcoma, mesothelioma and T-cell acute lymphoblastic leukaemia 5. In a retrospective 338-patient paediatric ALL cohort, 32 patients (9.47%) carried MTAP rearrangements - most commonly MTAP::CDKN2B-AS1, a fusion architecture consistent with a 9p21 deletion breakpoint - with enrichment in T-ALL, higher presenting WBC and LDH, more intermediate/high-risk assignment and inferior outcome 6.

Neither source reports the KMT2A-rearranged subset. The frequency in infant KMT2A-rearranged B-ALL specifically is, as far as this beat could establish, unpublished - and infant KMT2A-rearranged disease is characterised by a famously silent mutational landscape, which argues against but does not exclude a common 9p21 deletion.

Facts and figures
ontology noteNo Cell Ontology term; this is a genotype-imposed metabolic condition rather than a cell identity. GO:0019509 (L-methionine salvage from methylthioadenosine) is the process this state removes.
state definitionA condition, created by biallelic loss of MTAP (usually as a 9p21 passenger of CDKN2A deletion), in which methylthioadenosine accumulates intracellularly and partially inhibits PRMT5, leaving the cell in a hypomorphic PRMT5 state that is selectively sensitised to further inhibition of PRMT5, WDR77, MAT2A or RIOK1.
detection trapA homozygously deleted gene has nothing for CRISPR to cut and scores near zero on Chronos whatever the biology. MTAP's rank in this project's scan is 7,038 with positive gene effects in both arms, and that number carries no information about MTAP status in either arm. A dependency screen measures the consequence of cutting a gene; it cannot report a gene that is already absent. The same error has been refused three times in this graph in other guises (CDKN2A, SIRT1, ATM).
shown for fusionNO, and the negative is the point: tested against this project's 11-line KMT2A-rearranged DepMap arm and rejected. Nobody has measured MTAP copy number in a KMT2A-rearranged patient cohort.
why this node existsgene-mat2a named it as a declined vocabulary gap - 'the best-replicated MAT2A-dependent state in cancer is MTAP-DELETION-driven PRMT5 vulnerability, which is a genotype rather than a cell state and also has no node'. Without the term the confounder could be worried about but not recorded, linked or answered.
how to detect it
[
  "copy number at 9p21 covering MTAP, with biallelic status - the primary and only direct route",
  "MTAP protein loss by immunohistochemistry or western blot",
  "intracellular MTA by LC-MS, which is the metabolite that does the work",
  "symmetric dimethylarginine (SDMA) marks as a functional readout of the hypomorphic PRMT5 state",
  "NOT by MTAP's own score in a CRISPR dependency screen - see detection_trap"
]
synthetic lethal axis
[
  "PRMT5",
  "WDR77",
  "MAT2A",
  "RIOK1"
]
status in this projects index arm
{
  "answer": "NOT ENRICHED - no evidence of MTAP deletion in the 11 KMT2A-rearranged DepMap lines",
  "basis": "PRMT5 rank 17,006 and WDR77 rank 15,455 of 17,728, both LESS depleted in the rearranged arm, the opposite of what MTAP deletion predicts",
  "held_in": "ana-the-kmt2a-arm-is-not-mtap-deleted-and-chronos-cannot-see-a-deletion",
  "what_would_settle_it_outright": "a copy-number lookup on the 11 named lines; the line identifiers were not recorded in the source scan's outputs, which is a gap in that analysis rather than in this entry",
  "consequence": "MAT2A's rank in this project's scan is not an MTAP artefact, so the SAM-supply argument in cell-state-sam-methionine-dependence stands on its own evidence"
}
frequency where known
[
  {
    "population": "lung and pancreatic cancer, glioblastoma, osteosarcoma, soft-tissue sarcoma, mesothelioma, T-cell ALL",
    "measure": "homozygous MTAP/CDKN2A co-deletion",
    "value": "35-70%",
    "pmid": "19478948",
    "caveat": "quoted range across tumour types in a 2009 introduction, not a measurement in that paper"
  },
  {
    "population": "paediatric ALL, single-centre cohort of 338",
    "measure": "MTAP REARRANGEMENT by whole-transcriptome sequencing (not deletion by copy number)",
    "value": "9.47% (32/338), enriched in T-ALL, 75% of them MTAP::CDKN2B-AS1",
    "pmid": "41261957"
  },
  {
    "population": "infant KMT2A-rearranged B-ALL",
    "measure": "MTAP deletion",
    "value": "NOT FOUND in the literature searched by this beat",
    "note": "bounded absence: neither cited source stratifies by KMT2A rearrangement, and this beat found no cohort reporting MTAP status in infant KMT2A-rearranged disease. The silent mutational landscape of this subtype argues against a common 9p21 deletion but does not exclude one."
  }
]
How this could bite us back
  • This state is a reason to distrust a selectivity number, not only a reason to give a drug. Any panel-based dependency on MAT2A, PRMT5, WDR77 or RIOK1 in mixed cell lines is confounded by MTAP status, and the confound produces a clean, plausible rank rather than an error.
  • MTAP deletion is a passenger of CDKN2A deletion, so a therapeutic strategy built on it is a strategy aimed at a lesion the tumour did not select for and could in principle retain heterogeneously; subclones with an intact MTAP allele are pre-existing resistance.
  • PRMT5 is required for normal haematopoiesis and for splicing fidelity across tissues, so the therapeutic window in an MTAP-deleted tumour rests entirely on the modest additional sensitisation MTA provides - and the founding paper calls that sensitisation modest.
  • MTA is a diffusible metabolite. An MTAP-null clone in a marrow full of MTAP-competent normal cells may not accumulate MTA to the concentration an isogenic cell-line experiment reports, which is a straightforward way for this state to be present genetically and absent functionally in a patient.
  • If a future cohort does find MTAP deletion in KMT2A-rearranged disease, the immediate consequence is that this project's MAT2A result would need re-testing, not that a new drug arrives - and the re-test would have to be done in lines whose MTAP status is measured rather than inferred.
Still open
  1. What is MTAP copy number in the eleven KMT2A-rearranged DepMap lines? The phenotype-based answer says not deleted, but the direct measurement has not been made, and the line identifiers are not recorded in the source scan's outputs. Recovering the list and making one copy-number call closes this outright.
  2. What is the frequency of MTAP deletion in infant KMT2A-rearranged B-ALL? No cohort reports it. Infant KMT2A-rearranged disease has a famously silent mutational landscape, so the prior is low, but CDKN2A deletion is common enough in paediatric ALL generally that the number is worth having before any PRMT5- or MAT2A-directed agent is considered in this population.
  3. Why do MAT2A and RIOK1 move together while PRMT5 and WDR77 move the other way in this project's arm? SAM supply and SAM consumption diverging within one pathway is not MTAP biology and has no explanation. Noise at n = 11 is the null; a PRMT5-independent SAM-supply requirement in KMT2A-rearranged cells is the interesting alternative.
  4. Does the SAM-supply dependency reported in KMT2A-rearranged ALL interact with this state? An MTAP-deleted, KMT2A-rearranged cell would have both a raised SAM demand and a partially inhibited PRMT5, and nothing predicts whether those are additive, redundant or opposed.
Every source behind this entry (6)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    ACROSS CANCER CELL LINES of multiple lineages, integrating genomic profiling with functional dependency data showed that loss of methylthioadenosine phosphorylase confers a selective dependence on PRMT5 and its binding partner WDR77; MTAP-deleted cells have increased intracellular methylthioadenosine, MTA specifically inhibits PRMT5 enzymatic activity, and either MTA or a small-molecule PRMT5 inhibitor modestly and preferentially impairs viability of MTAP-null lines relative to isogenic MTAP-expressing counterparts.

    MTAP deletion confers enhanced dependency on the PRMT5 arginine methyltransferase in cancer cells. · Kryukov GV, Wilson FH, Ruth JR, Paulk J, Tsherniak A, Marlow SE, et al., Garraway LA · Science 2016
    PMID 26912360DOI
    What this evidence cannot show

    Pan-cancer cell lines with no leukaemia stratification reported in the abstract and no KMT2A-rearranged arm. The authors' own word for the viability effect is 'modest', which is a weaker claim than the synthetic-lethal framing the field adopted afterwards; a modest isogenic difference is exactly the size of effect a mixed panel can turn into a confident selectivity rank. Read: abstract only.

    Evidence that argues against this (1)
    • The effect size is the weak point, and it matters for how this state should be used. The same paper describes the preferential impairment of MTAP-null viability as modest, and the mechanism is partial inhibition of PRMT5 rather than its loss - a 'hypomorphic' state, in the sibling paper's word. A state that shifts a dependency modestly can still confound a rank-based selectivity scan while being too small to carry a therapeutic hypothesis, and those two uses of the same finding should not be conflated. Read: abstract only. Mavrakis KJ et al. 2016
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    IN A LARGE-SCALE shRNA SCREEN ACROSS 390 CANCER CELL LINE MODELS, viability of MTAP-deficient cancer cells was impaired by depletion of PRMT5; MTAP-deleted cells accumulate methylthioadenosine which inhibits PRMT5 methyltransferase activity, deleting MTAP in MTAP-proficient cells rendered them sensitive to PRMT5 depletion, and reconstituting MTAP in an MTAP-deficient line rescued PRMT5 dependence - establishing that MTA accumulation creates a hypomorphic PRMT5 state selectively sensitised to further PRMT5 inhibition.

    Disordered methionine metabolism in MTAP/CDKN2A-deleted cancers leads to dependence on PRMT5. · Mavrakis KJ, McDonald ER, Schlabach MR, Billy E, Hoffman GR, deWeck A, et al., Sellers WR · Science 2016
    PMID 26912361DOI
    What this evidence cannot show

    shRNA rather than CRISPR, so partial knockdown with known off-target seed effects; 390 lines across all lineages with no reported leukaemia or KMT2A breakdown. The isogenic deletion-and-rescue pair is the strongest part and is what makes the causal direction safe. Read: abstract only.

    Evidence that argues against this (1)
    • The dependency does not stop at PRMT5, which complicates using PRMT5 alone as the reporter for this state. The same lesion creates vulnerability to MAT2A, which supplies PRMT5's substrate SAM, and to the PRMT5 co-complex protein RIOK1 - so a panel can show an MTAP-driven MAT2A signal with the PRMT5 signal muted or absent depending on which nodes the screen covers and how pan-essential each is. Detection of the state by any single axis member is therefore unreliable. Read: abstract only. Marjon K et al. 2016
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    IN MTAP-DELETED CANCER CELLS, the vulnerability extends beyond PRMT5 to the SAM-producing enzyme MAT2A and to the PRMT5 co-complex protein RIOK1: MAT2A depletion reduces growth and PRMT5 methylation activity selectively in MTAP-deleted cells, defining a MAT2A/PRMT5/RIOK1 axis vulnerable in CDKN2A/MTAP-deleted cancers.

    MTAP Deletions in Cancer Create Vulnerability to Targeting of the MAT2A/PRMT5/RIOK1 Axis. · Marjon K, Cameron MJ, Quang P, Clasquin MF, Mandley E, Kunii K, et al., Marks KM · Cell Rep 2016
    PMID 27068473DOI
    What this evidence cannot show

    shRNA screening in cancer cell lines, no leukaemia stratification in the abstract, no KMT2A arm. This is the citation that makes MAT2A a member of this state's axis and is therefore the reason the MAT2A rank in this project's scan needed checking at all. Already cited in gene-mat2a. Read: abstract only.

    Evidence that argues against this (1)
    • MAT2A has a second, genotype-independent route to being required, and in this disease that is the route with the evidence. In KMT2A-rearranged paediatric ALL the methionine-cycle vulnerability is attributed to an increased SAM requirement for maintaining a hypermethylated chromatin state, with target-gene repression under restriction and efficacy of a MAT2A inhibitor in patient-derived xenografts - no MTAP lesion invoked. A MAT2A dependency is therefore not diagnostic of this state. Read: abstract only. Tee T et al. 2025
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    IN THIS PROJECT'S OWN CACHED DepMap SCAN of 11 Cellosaurus-annotated KMT2A-rearranged leukaemia lines against 43 other leukaemia lines, there is no evidence that the rearranged arm is enriched for MTAP deletion: PRMT5 (rank 17,006 of 17,728) and WDR77 (rank 15,455) are both LESS depleted in the rearranged arm, the opposite direction to what MTAP deletion predicts, while MTAP's own Chronos score is uninformative by construction because a homozygously deleted gene has nothing for CRISPR to cut.

    The KMT2A-rearranged DepMap arm is not MTAP-deleted, and the gene that looked like the test cannot answer it · Oncofusion Knowledge Graph coordinator · internal analysis node ana-the-kmt2a-arm-is-not-mtap-deleted-and-chronos-cannot-see-a-deletion; analyses/mtap-deletion-in-the-kmt2a-arm/ 2026
    What this evidence cannot show

    Internal, unpublished, and it infers genotype from phenotype: no copy number was measured and the 11 line identifiers were not recorded in the source scan's outputs, so a direct lookup remains undone. All four axis genes are pan-essential on a compressed scale (roughly -1.00 to -1.35) and the cached file carries arm means only, so no confidence interval is computable and the magnitudes must not be repeated as measured effect sizes (rule 87). n = 11 in the index arm.

    Evidence that argues against this (1)
    • The same analysis contains a pattern its own conclusion does not explain: MAT2A and RIOK1 move one way (-0.111, -0.108) while PRMT5 and WDR77 move the other (+0.106, +0.060). Supply and consumption of SAM diverging within one pathway is not MTAP biology. If part of that split is noise at n = 11, then the PRMT5/WDR77 direction the negative conclusion rests on comes from the same noisy instrument, and only a copy-number lookup on the named lines settles the question outright. Marjon K et al. 2016
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    MTAP lies at 9p21 close to CDKN2A, and MTAP and CDKN2A are homozygously co-deleted with a frequency of 35 to 70% in lung and pancreatic cancer, glioblastoma, osteosarcoma, soft-tissue sarcoma, mesothelioma and T-cell acute lymphoblastic leukaemia; in normal MTAP-positive cells MTAP cleaves methylthioadenosine to adenine and 5-methylthioribose-1-phosphate, a reaction MTAP-null tumour cells cannot perform.

    PMID 19478948DOI
    What this evidence cannot show

    The 35-70% figure is a range quoted across several tumour types in this paper's introduction rather than a measurement made in it, and it predates modern copy-number calling; the paper's own subject is a purine-analogue killing strategy, not the frequency. T-ALL is named, B-ALL is not, and KMT2A-rearranged disease is not stratified anywhere. Use it as an order-of-magnitude statement about how common the lesion is, not as a rate for any specific cohort. Read: abstract only.

    Evidence that argues against this (1)
    • A cohort-level measurement in the right disease gives a much lower and lineage-skewed number. In 338 paediatric ALL patients, 32 (9.47%) carried MTAP rearrangements, significantly enriched in T-ALL - so in paediatric lymphoblastic leukaemia as a whole the lesion is roughly one patient in ten rather than one in three, and it is concentrated in the lineage this graph is not about. Read: abstract only. Zhang S et al. 2026
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    IN A RETROSPECTIVE COHORT OF 338 PAEDIATRIC ACUTE LYMPHOBLASTIC LEUKAEMIA PATIENTS analysed by whole-transcriptome sequencing, 32 (9.47%) harboured MTAP rearrangements, most commonly MTAP::CDKN2B-AS1 (75.0% of the rearranged cases); the MTAP-rearranged group had a significantly higher proportion of T-ALL, higher presenting white cell counts and LDH, more frequent intermediate- or high-risk assignment, enrichment of NOTCH1, PTEN, FBXW7, SRCAP and USP7 mutations, and significantly inferior outcomes.

    PMID 41261957DOI
    What this evidence cannot show

    This measures MTAP REARRANGEMENT detected in RNA, not MTAP deletion measured by copy number, and the two are not the same variable - MTAP::CDKN2B-AS1 is consistent with a 9p21 deletion breakpoint but the abstract does not report MTAP protein loss, MTA levels or biallelic status. Single-centre retrospective cohort; the authors themselves state that independent prognostic impact requires multivariable analysis they did not perform. No KMT2A-rearranged breakdown is reported, and infants are not separated. Read: abstract only.

    Evidence that argues against this (1)
    • The functional state this node describes requires biallelic MTAP loss with MTA accumulation and a measurable fall in PRMT5 activity; a rearrangement called from RNA establishes none of those. The founding work defines the state by intracellular MTA concentration and by isogenic deletion and rescue, and an MTAP-rearranged transcript with retained function on the other allele would not produce it. A cohort frequency for the rearrangement is therefore an upper bound on the frequency of the state, not a measurement of it. Read: abstract only. Kryukov GV et al. 2016
Connected to