Oncofusion

start / graph / cell-state-sam-methionine-dependence

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

SAM / methionine dependence - the one metabolic state in this graph demonstrated in KMT2A-rearranged lymphoblastic leukaemia itself, with a drug and a PDX behind it

also called S-adenosylmethionine addiction · SAM addiction · methionine dependence · methionine-cycle vulnerability · MAT2A dependence

Every chemical tag a cell puts on DNA or on histones is paid for with a molecule called SAM, which is made from the amino acid methionine in food. KMT2A-rearranged leukaemia carries an unusually heavy load of those tags, so it has an unusually large SAM bill - and taking methionine away, or blocking the enzyme that makes SAM, kills it.

The same thing, for a scientist

This node fills a vocabulary gap that `gene-mat2a` named explicitly: the graph had `cell-state-oxphos-dependence` and `cell-state-lipid-pufa-dependence` but no methionine-cycle state, so MAT2A's entry recorded 'NO CELL STATE IN THIS GRAPH'S VOCABULARY IS MAINTAINED BY MAT2A' rather than approximate. The state is well evidenced **in the index disease**, which is rare for a metabolic claim here: KMT2A-rearranged ALL is particularly vulnerable to perturbation of the methionine cycle, the dependency is attributed to an increased SAM requirement for maintaining the hypermethylated state, important pro-survival KMT2A-r target genes are repressed under methionine restriction, and the orally active MAT2A inhibitor FIDAS-5 impaired leukaemia progression in patient-derived xenografts with synergy against HDAC inhibitors [E1]. The state is therefore a **consequence** of the chromatin programme rather than an independent metabolic lesion, which is the argument `mechanism-sam-supply-sustains-kmt2a-r-hypermethylation` already holds and which this node does not restate. Two things bound the claim. First, elevated methionine requirement is a general property of transformed cells, described since the 1970s, so 'KMT2A-r cells are methionine-dependent' needs a comparator to be a finding rather than a base rate [counter-evidence on E1]. Second, the best-replicated MAT2A dependency in cancer is genotype-conditional on MTAP deletion [E2] - and this project has now tested and rejected the possibility that its own KMT2A-rearranged DepMap arm is MTAP-deleted [E3].

Go deeper
What makes this state different from the other two metabolic states here

cell-state-oxphos-dependence splits by lineage and reverses direction between AML and B-ALL. cell-state-lipid-pufa-dependence rests on KMT2A-rearranged AML. This state is the only one of the three whose founding evidence is in paediatric KMT2A-rearranged acute lymphoblastic leukaemia, with a patient-derived xenograft endpoint and an orally active tool compound 1.

It is also the only one with a stated causal direction from the chromatin programme to the metabolism: the paper's own claim is that the methionine dependency is driven by an increased need for SAM to maintain the hypermethylated state, and that methionine restriction represses pro-survival KMT2A-r target genes 1. That is a consequence-of-the-programme dependency, which is the useful kind, because it predicts that anything raising the methylation load raises the SAM bill - and the reported synergy with HDAC inhibitors is consistent with combining two pressures on the same chromatin state.

The base rate this claim has to clear

Methionine dependence is not a KMT2A discovery. Transformed cells in general have an elevated methionine requirement relative to normal cells and have been called methionine-dependent since the 1970s; revertants selected for methionine independence concurrently lose other properties of malignancy, which is how tightly the phenotype tracks transformation [counter-evidence on E1].

So the load-bearing part of 1 is not 'these cells need methionine' - almost every cancer line does - but the comparative part: that KMT2A-rearranged leukaemia is particularly vulnerable among ALL subtypes, and that the mechanism runs through SAM supply to a hypermethylated chromatin state rather than through a generic biosynthetic requirement. A reader quoting this node should quote that, not the bare dependence.

The genotype confounder, raised and settled

The most replicated route to a MAT2A dependency in cancer has nothing to do with chromatin: MTAP deletion causes accumulation of methylthioadenosine, MTA partially inhibits PRMT5, and MAT2A depletion then reduces growth and PRMT5 methylation selectively in MTAP-deleted cells 2. MTAP is co-deleted with CDKN2A, which is one of the commonest lesions in ALL, so the confounder was a live one: if this project's eleven KMT2A-rearranged DepMap lines were MTAP-enriched, MAT2A's selectivity rank would be a property of the panel.

That was tested this cycle and the answer is no. PRMT5 (rank 17,006 of 17,728) and its obligate partner WDR77 (15,455) are both less depleted in the rearranged arm - the opposite of what MTAP deletion predicts - and MTAP's own Chronos score is uninformative by construction because a deleted gene has nothing to cut 3. The SAM-supply argument in 1 therefore stands on its own evidence, and the two routes to MAT2A are separate: see cell-state-mtap-deleted-mta-high for the genotype state and its detection.

How to score the state

Methionine withdrawal or restriction in defined medium with homocysteine replacement (the classical test that separates true methionine dependence from a transmethylation-cycle defect); intracellular SAM and SAH by LC-MS with the SAM:SAH ratio as the methylation-potential index; sensitivity to a MAT2A inhibitor such as FIDAS-5 with SAM depletion confirmed pharmacodynamically rather than assumed; and bulk H3K4me3, H3K79me2 and DNA methylation as the downstream readouts the state is supposed to be paying for.

The trap specific to this state: a growth effect of methionine withdrawal is not evidence of a SAM-for-chromatin requirement. Methionine is also a proteinogenic amino acid, a polyamine precursor and the initiator of every translated protein, so any of those can produce the same growth curve. The discriminating measurement is whether the methylation marks and the target-gene repression move, which is what 1 reports and what a reader should check before transferring this state to a new cell.

Facts and figures
ontology noteNo Cell Ontology term; this is a metabolic condition, not a cell identity. The nearest structured vocabulary is GO:0006555 (methionine metabolic process) and GO:0006556 (S-adenosylmethionine biosynthetic process), both process terms.
state definitionA condition in which proliferation and survival require exogenous methionine at a level normal cells do not, because the cell's transmethylation load exceeds what the methionine salvage and de novo remethylation routes can supply; operationally scored by growth arrest under methionine withdrawal with homocysteine replacement, by a fall in the SAM:SAH ratio, and by sensitivity to MAT2A inhibition.
why this node existsgene-mat2a recorded `declined_vocabulary`: there was no 'SAM / methionine dependence' metabolic state in the graph even though cell-state-oxphos-dependence and cell-state-lipid-pufa-dependence set the pattern, and MAT2A's entry therefore wrote 'NO CELL STATE IN THIS GRAPH'S VOCABULARY IS MAINTAINED BY MAT2A' rather than approximate to one of them.
shown for fusionYES for the disease class - demonstrated in KMT2A-rearranged paediatric ALL including patient-derived xenografts (PMID 40270205). The abstract does not name the partner genes, so whether KMT2A::AFF1 specifically is represented is unresolved; treat it as KMT2A-rearranged ALL rather than as the index fusion.
the trapA growth effect of methionine withdrawal is not evidence of a SAM-for-chromatin requirement. Methionine is also proteinogenic, the initiator of translation and a polyamine precursor, so any of those can produce the same growth curve. Require the marks and the target programme to move.
operational readouts
[
  "methionine withdrawal in defined medium with homocysteine replacement (separates methionine dependence from a remethylation defect)",
  "intracellular SAM and SAH by LC-MS; SAM:SAH ratio as the methylation-potential index",
  "MAT2A inhibitor sensitivity (FIDAS-5 in the cited work) with pharmacodynamic confirmation of SAM depletion",
  "bulk H3K4me3, H3K79me2 and DNA methylation as the downstream marks the SAM is being spent on",
  "expression of the KMT2A-r target programme under restriction, which is what distinguishes a chromatin-load dependency from a generic amino-acid requirement"
]
two independent routes to mat2a
{
  "chromatin_load_route": "SAM demand set by a hypermethylated chromatin state; KMT2A-r ALL (PMID 40270205)",
  "genotype_route": "MTAP deletion raises MTA, MTA partially inhibits PRMT5, and the MAT2A/PRMT5/RIOK1 axis becomes vulnerable (PMID 27068473)",
  "which_applies_here": "the chromatin-load route; the genotype route was tested against this project's index arm and rejected - see cell-state-mtap-deleted-mta-high and ana-the-kmt2a-arm-is-not-mtap-deleted-and-chronos-cannot-see-a-deletion"
}
sibling metabolic states
[
  "cell-state-oxphos-dependence",
  "cell-state-lipid-pufa-dependence",
  "cell-state-hypoxic-glycolytic"
]
How this could bite us back
  • Methionine restriction in an infant is not a benign intervention. Methionine is an essential amino acid and the initiator of every translated protein; a dietary arm in a child under one year old collides directly with growth, and the cited efficacy was obtained in xenografted mice whose growth requirements are not a baby's.
  • If the dependency is a consequence of the fusion's methylation load, then any drug that lowers that load - a menin inhibitor being the obvious one - is predicted to REDUCE the SAM requirement. The intuitive combination of a menin inhibitor with a MAT2A inhibitor could therefore be antagonistic, and nothing has tested it.
  • SAM is the universal methyl donor: a MAT2A inhibitor does not choose between H3K4, H3K79, DNA, RNA cap, phospholipid and small-molecule methylation. On-target toxicity is distributed across every methylation-dependent process in the body, and the normal tissue with the highest transmethylation flux is the liver.
  • The reported synergy is with HDAC inhibitors, a class with its own cardiac and marrow toxicity in children, so the most supported combination is also the one with the least headroom in a heavily pre-treated infant.
  • Methionine dependence is close to a general property of transformed cells, so a positive result in a new KMT2A-rearranged model has a high prior of being non-specific. Any claim of selectivity needs a fusion-negative leukaemia comparator run in the same medium on the same day.
Still open
  1. Which KMT2A fusion partners were in the models behind the SAM-addiction result? The abstract reports 'KMT2A-rearranged ALL' without naming partners, and this graph's recurring finding is that evidence is partner-imported. If none of the models is KMT2A::AFF1, the index fusion inherits this state by class membership rather than by measurement.
  2. Is the 'hypermethylated state' DNA methylation or histone methylation? The two predict different combination partners - a DNA methyltransferase inhibitor versus a histone methyltransferase inhibitor - and the abstract does not disambiguate. The measurement is a SAM-depletion time course with parallel WGBS and histone-mark ChIP in the same cells.
  3. Does menin inhibition change the SAM bill? If the dependency is set by the chromatin load the fusion imposes, then revumenib, which collapses that load, should REDUCE methionine dependence - making MAT2A inhibition and menin inhibition antagonistic rather than additive. Nobody has measured SAM, SAH or methionine sensitivity before and after a menin inhibitor, and it is a one-plate experiment on a drug already in patients.
  4. MAT2A and RIOK1 move together in this project's scan while PRMT5 and WDR77 move the other way. SAM supply and SAM consumption diverging within one pathway is not MTAP biology and has no explanation; is it noise at n = 11, or a SAM-supply requirement in KMT2A-rearranged lines that is independent of PRMT5?
Every source behind this entry (3)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In PAEDIATRIC KMT2A-REARRANGED ACUTE LYMPHOBLASTIC LEUKAEMIA, the disease is particularly vulnerable to perturbation of the methionine cycle; the methionine dependency is driven by an increased need for S-adenosylmethionine to maintain the hypermethylated state of KMT2A-rearranged leukaemia, important pro-survival KMT2A-r target genes are repressed under methionine restriction with rapid cell death, the orally active MAT2A inhibitor FIDAS-5 impaired leukaemia progression in patient-derived xenograft models, and a drug screen showed strong synergy between MAT2A inhibition and histone deacetylase inhibitors.

    S-adenosylmethionine addiction confers sensitivity to methionine restriction in KMT2A-rearranged acute lymphoblastic leukemia. · Tee T, Ruiter TJJ, Wu S, Zhang W, Schenau DVI, Rodionova M, et al., Van Leeuwen FN · Haematologica 2025
    PMID 40270205DOI
    What this evidence cannot show

    The abstract does not state which KMT2A partner genes the models carry, so whether the index fusion KMT2A::AFF1 is represented and at what weight is unresolved from the abstract alone; nor does it give the comparator ALL subtypes or the effect sizes. 'Hypermethylated state' is ambiguous between DNA methylation and histone methylation and the abstract does not disambiguate - a caveat also carried by mechanism-sam-supply-sustains-kmt2a-r-hypermethylation. Methionine restriction as a clinical intervention in an infant is a different proposition from methionine restriction in a xenograft. Read: abstract only. Already cited in three entries here (gene-mat2a, mechanism-sam-supply-sustains-kmt2a-r-hypermethylation, and a metabolic-dependency concept entry); cited here to define the state, not to restate their conclusions.

    Evidence that argues against this (1)
    • The null for this claim is high. Cancer cells in general have an elevated methionine requirement relative to normal cells and have been termed methionine-dependent since the 1970s; methionine-independent revertants selected in methionine-restricted medium concurrently revert several other properties of malignancy, and the dependence is not explained by reduced endogenous methionine synthesis. So 'KMT2A-rearranged cells die under methionine restriction' is close to a base rate for transformed cells, and only the comparative and mechanistic parts of E1 - particular vulnerability among ALL subtypes, SAM supply to a hypermethylated chromatin state, repression of KMT2A-r target genes - carry information. Read: abstract only; this source is a book-chapter review used as a map of the older literature, not as a primary result. Hoffman RM et al. 2019
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In MTAP-DELETED CANCER CELL LINES (a genotype, not a lineage), shRNA screening identified MAT2A and PRMT5 as vulnerable enzymes: the MTAP substrate methylthioadenosine accumulates on MTAP loss and is a potent and selective inhibitor of PRMT5, MTAP-deleted cells have reduced PRMT5 methylation activity and increased sensitivity to PRMT5 depletion, and MAT2A depletion reduces growth and PRMT5 methylation activity selectively in MTAP-deleted cells because MAT2A produces the PRMT5 substrate SAM.

    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

    Cancer cell lines across lineages with no leukaemia stratification reported in the abstract, and no KMT2A-rearranged arm. This is cited here as the alternative, genotype-conditional route to the same enzyme, so that a MAT2A result in a new panel is not automatically read as a chromatin-load result. Read: abstract only.

    Evidence that argues against this (1)
    • This route does not explain the KMT2A result, because the genotype it requires is absent from this project's index arm. In the cached full DepMap scan over 11 KMT2A-rearranged and 43 other leukaemia lines, PRMT5 (rank 17,006) and WDR77 (rank 15,455) are both LESS depleted in the rearranged arm, the opposite direction to what MTAP deletion predicts, so there is no evidence of MTAP enrichment in the index panel. Note also that all four axis genes are pan-essential on a compressed scale and the arm means carry no computable interval, so the magnitudes are not to be quoted as measured effects. Oncofusion Knowledge Graph coordinator 2026
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    In this project's own cached DepMap scan comparing 11 Cellosaurus-annotated KMT2A-rearranged leukaemia lines against 43 other leukaemia lines, the synthetic-lethal partners that would report MTAP deletion point away from it - PRMT5 rank 17,006 and WDR77 rank 15,455 of 17,728, both less depleted in the rearranged arm - 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 2026
    What this evidence cannot show

    An internal analysis of this project's own scan, not a published result, and it infers genotype from phenotype rather than measuring copy number - a direct copy-number lookup on the 11 lines has not been done. All four axis genes are pan-essential between roughly -1.00 and -1.35, so every arm difference sits on a compressed scale, and the cached file holds arm means only, so no confidence interval is computable and the magnitudes must not be repeated as measured effect sizes. Cited here as the reason the genotype confounder is set aside, not as positive evidence for the SAM-supply mechanism.

    Evidence that argues against this (1)
    • Phenotype is a weak instrument for genotype here, and the same analysis says so: the four axis genes split in a way MTAP biology does not predict, with MAT2A and RIOK1 moving one way and PRMT5 and WDR77 the other. If part of that pattern is noise at n = 11, the conclusion drawn from the PRMT5/WDR77 direction is drawn from the same noisy instrument. Only a copy-number lookup on the named lines settles it, and that call has not been made. Kryukov GV et al. 2016
Connected to