MAT2A - the enzyme that makes the methyl donor, and the one metabolic dependency in this disease with two independent papers and a lymphoid PDX result

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.

• Dietary methionine restriction is not deliverable in an infant with leukaemia. Methionine is an essential amino acid and these patients are under one year old, often already failing to thrive; the nutritional arm of this strategy cannot be used in the population the finding is most exciting for.
• MAT2A supplies the methyl donor for the entire methylome, including in normal haematopoietic stem cells, gut epithelium and the developing brain. The therapeutic window rests entirely on a quantitative difference in SAM demand, with no qualitative selectivity step.
• The proposed partners - DOT1L, PRMT5, HDAC inhibitors - are all agents with their own dose-limiting toxicity and, in DOT1L's case, a clinical failure in this exact disease. A synergy that requires two marginal drugs is not obviously better than one.
• Resistance has an obvious route: the methionine salvage pathway and MTAP status. A cell that can regenerate methionine from methylthioadenosine is partly insulated from MAT2A blockade, and MTAP deletion status is not reported in either study.
• MAT2A is depleted in 95.0% of DepMap lines at a Chronos mean of -1.349. Whatever the comparative cell-line pharmacology shows, the genetic evidence says this enzyme is required by almost every dividing human cell tested, which is the definition of a narrow therapeutic index rather than a targeted therapy.
• PRMT5 and DOT1L, the two combination partners named in the earlier work, are themselves broadly required in DepMap, so each arm of the proposed combination carries its own broad toxicity on top of the other's.
• MAT2A's pan-essentiality is -1.3485 across 1,208 lines, with the KMT2A-rearranged arm at -1.5424 and other leukaemias at -1.4310. The therapeutic index is set by the -1.43 that every other leukaemia also has, not by the -0.11 difference, and SAM is required by every dividing cell in the body.
• The selectivity may be reporting the wrong lesion. MAT2A is a validated vulnerability in MTAP-DELETED cells through MTA-mediated PRMT5 inhibition (PMID 27068473), and MTAP is co-deleted with CDKN2A. Until MTAP status is genotyped across the KMT2A-rearranged DepMap lines, the rank-911 placement cannot be attributed to the fusion.
• If MAT2A acts on chromatin as a corepressor rather than only as a cytosolic SAM source (PMID 21362551), then methionine restriction and MAT2A inhibition are NOT interchangeable interventions, and a trial designed on one cannot be justified by preclinical data from the other.

1. Which KMT2A fusion partners were in the FIDAS-5 PDX panel? If infant KMT2A::AFF1 was not among them, the single most translatable metabolic result in this beat has not actually been tested in the index disease.
2. Is the hypermethylated state that consumes the SAM a DNA-methylation state or a histone-methylation state? The two imply completely different partner drugs - hypomethylating agents versus DOT1L/PRMT5/menin inhibitors - and the abstracts do not say.
3. Does MAT2A inhibition retain activity in menin-inhibitor-resistant cells carrying MEN1 mutations? If yes, this is a genuinely orthogonal route and should be prioritised over yet another interface binder.
4. MAT2A and DOT1L inhibition synergise. DOT1L inhibition alone was clinically modest in this disease. Is the MAT2A combination a way to rescue a failed drug, or a way to reproduce its failure with added toxicity?
5. Does MAT2A dependency in KMT2A-rearranged lines survive a comparison against lineage-matched non-KMT2A leukaemia rather than against healthy controls? (carried forward from the superseded cycle-48 entry)
6. Is the DOT1L combination effect additive on a shared cytotoxic axis, or synergistic through SAM limitation specifically? Methionine-depletion rescue would separate them. (carried forward from the superseded cycle-48 entry)
7. Can the functional MLLr-enhanced response be reconciled with a DepMap Chronos of -1.349 at 95% of lines depleted, or is the comparative result an artefact of the comparator lines chosen?
8. Are the eleven KMT2A-rearranged DepMap lines MTAP-deleted? One genotype lookup would say whether this entry's selectivity signal is about the fusion or about a co-deletion at 9p21.
9. Does MAT2A occupy chromatin in a KMT2A-rearranged cell, and if so at which loci? The only chromatin-occupancy result for this protein is at one gene in murine hepatoma; a CUT&RUN in one leukaemia line would establish whether the local-channelling model applies here at all.
10. Why does UniProt annotate no compartment for a protein with a published nuclear pool and a published chromatin recruitment? Either the primary result has not been replicated or the curation is behind, and knowing which changes how much weight the localisation argument can carry.

1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In a human CRISPR/Cas9-MLL-rearranged leukaemia model, MAT2A was significantly overexpressed relative to healthy controls, consistent with publicly available patient data; using non-MLLr and MLLr cell lines and the model, an MLLr-specific enhanced response was detected to the MAT2A inhibitor PF-9366 and to siRNA-mediated MAT2A knockdown, altering proliferation, viability, differentiation, apoptosis, cell cycling and histone methylation; combining PF-9366 with chemotherapy or with inhibitors of the SAM-dependent methyltransferases DOT1L and PRMT5 produced more pronounced effects.

   MAT2A as Key Regulator and Therapeutic Target in MLLr Leukemogenesis. · Secker KA, Bloechl B, Keppeler H, ..., Schneidawind C · Cancers (Basel) 2020
   PMID 32456310DOI
   ▸What this evidence cannot show

   Abstract read this session; full text not retrieved, so the size of the MLLr-versus-non-MLLr response difference and the number of lines in each arm were not checked. The 'publicly available patient data' comparison is a database association whose cohort and multiple-testing handling were not verified here. PF-9366 is an allosteric MAT2A inhibitor with limited independent characterisation in haematopoietic cells; an MLLr-enhanced response to one tool compound is not a demonstrated genotype selectivity. The CRISPR/Cas9-MLLr model is an engineered system, which is a strength for causality and a weakness for representing patient disease.

   ▸Evidence that argues against this (1)

   • SAM is the universal methyl donor and methionine restriction has broad antiproliferative effects across cancer types, so an MLLr-enhanced response to MAT2A inhibition must be separated from a general proliferation effect; the later KMT2A-r ALL study addresses this by tying the dependency to the hypermethylated state and to repression of specific KMT2A-r target genes rather than to growth rate. Tee T et al. 2025
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   KMT2A-rearranged leukaemia is particularly vulnerable to perturbations of the methionine cycle; the methionine dependency is driven by an increased need for S-adenosylmethionine to maintain the hypermethylated state of KMT2A-rearranged leukaemias; important pro-survival KMT2A-r target genes are repressed under methionine restriction, which combined with downstream metabolic changes results in rapid cell death; the orally active MAT2A inhibitor FIDAS-5 impaired leukaemia progression in patient-derived xenograft models, and a drug screen revealed 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, ..., Van Leeuwen FN · Haematologica 2025
   PMID 40270205DOI
   ▸What this evidence cannot show

   Abstract read this session; full text not retrieved. The abstract does not state which KMT2A fusion partners the PDX models carry, so 'KMT2A-rearranged ALL' must not be read as infant t(4;11)/KMT2A::AFF1 without checking - recorded here as an unresolved gap. 'Hypermethylated state' is not defined in the abstract as DNA methylation versus histone methylation, and the two have different SAM budgets and different therapeutic implications; that ambiguity is not resolvable from the abstract. FIDAS-5 is an investigational compound with no clinical data. The methionine-restriction arm is a dietary intervention whose deliverability in growing infants is not addressed.

   ▸Evidence that argues against this (1)

   • An accompanying editorial in the same journal frames this as a promising metabolic-epigenome strategy rather than an established one; and the underlying manoeuvre, methionine restriction, has decades of broad antiproliferative literature across tumour types, so the specificity claim rests on the comparative experiments rather than on the manoeuvre itself. Heckl D et al. 2025
3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   UniProt reviewed human record P31153 (METK2_HUMAN) has primary gene name MAT2A, gene synonyms AMS2 and MATA2, length 395 aa, EC 2.5.1.6, recommended protein name 'S-adenosylmethionine synthase isoform type-2'.

   UniProtKB entry P31153 (METK2_HUMAN), S-adenosylmethionine synthase isoform type-2 · UniProt Consortium · UniProt 2026
   ▸What this evidence cannot show

   Resolved 2026-09-20 with gene_exact:MAT2A AND organism_id:9606 AND reviewed:true; exactly one reviewed row and gene_primary read back as MAT2A. The human MAT family contains MAT1A (liver-specific catalytic subunit) and MAT2B (regulatory subunit) - MAT2B is NOT a catalytic paralogue and a paper naming 'MAT2' without a letter is ambiguous between MAT2A and the MAT2A/MAT2B holoenzyme. Neither cited paper uses that ambiguous form; both write MAT2A.

   ▸Evidence that argues against this (1)

   • A reviewed record fixes sequence and symbol on the date read. It cannot establish that MAT2A rather than MAT1A or the regulatory subunit MAT2B carries the dependency in leukaemic cells, and neither cited study reports a MAT1A or MAT2B comparison in its abstract. Secker KA et al. 2020
4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   In a phase 1 study of the DOT1L inhibitor pinometostat in 51 adults with advanced acute leukaemia, particularly MLL-rearranged disease, a maximum tolerated dose was not established, the most common adverse events of any cause were fatigue (39%), nausea (39%), constipation (35%) and febrile neutropenia (35%), and 2 patients - both with t(11;19) - experienced complete remission at 54 mg/m2 per day by continuous intravenous infusion; the authors describe efficacy as a single agent as modest and frame the study as groundwork for future combination approaches.

   The DOT1L inhibitor pinometostat reduces H3K79 methylation and has modest clinical activity in adult acute leukemia. · Stein EM, Garcia-Manero G, Rizzieri DA, ..., Tallman MS · Blood 2018
   PMID 29724899DOI
   ▸What this evidence cannot show

   Abstract read this session; identifiers re-fetched from PubMed rather than copied from the superseded sibling entry, per this project's rule against re-using identifiers from other graph entries. Adults with advanced disease, not infants; the population is the one in which almost every agent underperforms, so a modest single-agent result there does not settle the value of the target in newly diagnosed infant disease. Cited here only to bound the enthusiasm for MAT2A-plus-DOT1L combinations, not as evidence about MAT2A.

   ▸Evidence that argues against this (1)

   • The same study reports proof of concept for clinically meaningful responses through DOT1L targeting and explicitly lays groundwork for combination approaches, so it is not a refutation of the pathway - it is a refutation of single-agent sufficiency, which is precisely what a SAM-supply combination is proposing to fix. Secker KA et al. 2020
5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

   In this project's own DepMap-derived dependency analysis, MAT2A has a Chronos mean of -1.349 with 95.0% of lines depleted, and a KMT2A-selectivity difference of -0.158 in the myeloid split and -0.014 in the lymphoid split - values that sit inside the floor-compression range this project treats as uninterpretable rather than small.

   Age-restricted vulnerability panel: PLK1, AURKA, MAT2A and FFAR2 dependency in KMT2A-rearranged leukaemia lines · Oncofusion knowledge graph internal analysis · Oncofusion knowledge graph (internal analysis, ana-age-restricted-vulnerability-panel) 2026
   ▸What this evidence cannot show

   Carried forward from the superseded cycle-48 gene-mat2a entry; the underlying numbers were computed by that earlier analysis and were NOT recomputed by this beat, so they are reported on that entry's authority and should be re-derived before being used in a decision. A DepMap number is not a result until it is lineage-split, and this one is - both halves are given. A gene depleted in 95% of lines has almost no dynamic range for selectivity to appear in, which is why the correct verdict is 'uninterpretable', not 'not selective'.

   ▸Evidence that argues against this (1)

   • Pan-essentiality in a screen measures what happens when a gene is deleted outright in a cell line over weeks. It does not measure what happens under partial, reversible pharmacological occupancy, and the functional literature reports a comparative MLLr-enhanced response to a MAT2A inhibitor against non-MLLr lines, which is a different and more drug-like measurement. Tee T et al. 2025
6. E_q71_loc1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In MURINE HEPATOMA CELLS, MATII-alpha (the MAT2A product) WAS LOCALIZED IN NUCLEI and found to form a dense network with chromatin-related proteins including the Swi/Snf and NuRD complexes; it was RECRUITED TO THE MAF RECOGNITION ELEMENT (MARE) at the HO-1 gene as a transcriptional COREPRESSOR of MafK, and knocking MATII-alpha down derepressed HO-1 at both basal and induced levels - with the catalytic activity of MATII-alpha and its interacting factors MATII-beta, BAF53a, CHD4 and PARP1 all required for that repression.

   Methionine adenosyltransferase II serves as a transcriptional corepressor of Maf oncoprotein. · Katoh Y, Ikura T, Hoshikawa Y, Tashiro S, Ito T, Ohta M, Kera Y, Noda T, Igarashi K · Mol Cell 2011
   PMID 21362551DOI
   ▸What this evidence cannot show

   Abstract read in full; full text not read. MURINE hepatoma cells - not leukaemia, not haematopoietic, not human, not KMT2A-rearranged (brief rule 73). A SEPARATE AND STRIKING POINT: UniProt P31153 carries NO SUBCELLULAR LOCATION comment and no cellular-component keyword at all (fetched 2026-09-21) - it is the ONLY gene in this tranche of sixteen with no annotated compartment whatsoever. The nuclear pool described here is therefore a primary-literature finding that the reference database has not taken up, and the default assumption in the field is that MAT2A is a cytosolic metabolic enzyme.

   ▸Evidence that argues against this (1)

   • The best-replicated MAT2A vulnerability in cancer has nothing to do with chromatin localisation, with leukaemia or with KMT2A: shRNA screening identified MAT2A and PRMT5 as vulnerable in cells with MTAP DELETION, because the MTAP substrate MTA accumulates and potently and selectively inhibits PRMT5, so MAT2A depletion reduces growth and PRMT5 methylation activity SELECTIVELY IN MTAP-DELETED CELLS. MTAP is co-deleted with CDKN2A, a lesion this graph already carries as a cooperating event. That is a genotype-conditional explanation for MAT2A dependency that is available without invoking the fusion, and any MAT2A-in-KMT2A argument has to say what it adds to it - and whether the KMT2A-rearranged lines in the scan are MTAP-deleted. Marjon K et al. 2016
7. E_q71_ct1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   In a HUMAN CRISPR/Cas9 MLL-REARRANGED LEUKAEMIA MODEL GENERATED IN CD34+ CELLS, MYC was identified by RNA sequencing as one of the most important key drivers, and combining the BRD4 inhibitor JQ-1 with drugs targeting MAT2A or PRMT5 - which the authors and others had previously reported promote cell death in MLLr cells - produced augmented anti-leukaemic effects, alongside activation of T, NK and iNKT cells, release of immunomodulatory cytokines and downregulation of the PD-1/PD-L1 axis.

   Targeting MYC in combination with epigenetic regulators induces synergistic anti-leukemic effects in MLLr leukemia and simultaneously improves immunity. · Fitzel R, Secker-Grob KA, Keppeler H, Korkmaz F, Schairer R, Erkner E, Schneidawind D, Lengerke C, Hentrich T, Schulze-Hentrich JM, Schneidawind C · Neoplasia 2023
   PMID 37148657DOI
   ▸What this evidence cannot show

   Abstract read in full; full text not read. A HUMAN CD34+-DERIVED CRISPR MLLr model plus cell lines - human and MLL-rearranged, which is closer to this graph's disease than most evidence here, but the model's partner gene is not stated in the abstract and it is an engineered rather than a patient-derived system (brief rule 73). The MAT2A contribution is a COMBINATION result: the readout is synergy with JQ-1, not a MAT2A monotherapy effect, and 'previously reported by we and others' is a back-reference rather than a measurement made here. The immune readouts are in a model system and are not a clinical claim.

   ▸Evidence that argues against this (1)

   • The cell of origin cuts against reading this as a fusion-specific dependency. The model is built by engineering a rearrangement into normal human CD34+ HSPCs - so the cells are, by construction, an HSPC-derived population, and SAM supply is something every proliferating cell needs. The general pattern across this tranche is that requirements documented in blood cells are lineage or housekeeping requirements: the Fanconi literature makes the same point directly, with an HSPC defect in people who carry no fusion at all. Ceccaldi R et al. 2012

• ← testsThe KMT2A-rearranged DepMap arm is not MTAP-deleted, and the gene that looked like the test cannot answer it

  THE ENTRY THE WORRY WAS ABOUT. MAT2A's selectivity rank of 911 was the number at risk: if the 11 KMT2A-rearranged lines were enriched for MTAP deletion, that rank would be a property of the panel rather than of the fusion. Tested and not supported - PRMT5 and WDR77, the two most direct read-outs of MTAP status, are both LESS depleted in the rearranged arm. The rank survives as not-an-artefact, and separately fails to be impressive: -0.11 on a -1.35 pan-essential requirement.

• ← related toWhere else in the body each target is expressed, and the two controls that say the method works
• supports →SAM supply sustains the KMT2A-rearranged hypermethylated state - the metabolic requirement that IS a consequence of the chromatin programme, and can be attacked as one

  MAT2A is the sole extrahepatic route to the methyl donor the state consumes

• depends on →KMT2A-rearranged acute lymphoblastic leukaemia

  KMT2A-rearranged ALL is particularly vulnerable to methionine-cycle perturbation; FIDAS-5 impaired PDX progression

• co-occurs with →DOT1L — the target where excellent chemistry met an indirect interaction

  MAT2A inhibition synergises with DOT1L inhibition; both depend on SAM

• related to →KMT2A / MLL1 — the gene that breaks

  supplies the methyl donor for the SET domain and for the fusion-driven hypermethylated state

• supports →Is the metabolic phenotype a consequence of the fusion programme or an independent dependency? The answer is that BOTH directions are published, and one of them points the other way entirely

  clearest published case of a metabolic requirement that is a consequence of the chromatin programme

• ← depends onSAM supply sustains the KMT2A-rearranged hypermethylated state - the metabolic requirement that IS a consequence of the chromatin programme, and can be attacked as one

  MAT2A is the enzyme node of the SAM supply requirement

• ← related toPRMT5 - the type II arginine methyltransferase this graph has reached for sixteen times, whose mark on H3R2 RECRUITS WDR5 to chromatin

  MAT2A supplies the SAM PRMT5 consumes; the MTAP/MAT2A/PRMT5 axis

• ← related toMTHFD1 (P11586) - the inverse case: the deepest leukaemia dependency in the serine/one-carbon panel, and the drug does not touch it

  MTHFD1 supplies one-carbon units to the methionine cycle that MAT2A consumes; MAT2A is pan-essential (95.2% of 1,258 lines) and also flat under the drug

• related to →The soluble-nucleoplasm versus chromatin-bound axis - the one localisation readout that keeps a size dimension, and nobody has run it on a KMT2A fusion

  MATII-alpha is nuclear and recruited to the MARE at HO-1 as a MafK corepressor, networking with Swi/Snf and NuRD - while UniProt P31153 annotates NO compartment at all; a chromatin-versus-soluble fractionation would settle whether the leukaemia mechanism is local channelling or bulk SAM supply (murine hepatoma; PMID 21362551)

• related to →Bone marrow haematopoietic stem cell — the postnatal comparator, and why the same fusion behaves differently here

  ORIGIN-OF-MODEL edge, not a normal-HSC requirement: the human MLLr model in which MAT2A inhibition kills and synergises with JQ-1 is built by CRISPR rearrangement of normal human CD34+ cells (PMID 37148657)

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

  MAT2A is the enzyme that makes SAM; the orally active MAT2A inhibitor FIDAS-5 impaired KMT2A-rearranged ALL progression in patient-derived xenografts

• ← depends onThe 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

  MAT2A supplies PRMT5's substrate SAM and is one of the four vulnerable nodes in MTAP-deleted cells - the reason MAT2A's rank in this project's scan needed a genotype check

• ← testsIf inhibiting one methyltransferase failed, constraining the methyl donor may not
• ← testsEvery age-restricted target nominated by the model is pan-essential or invisible