Oncofusion

start / graph / gene-shmt2

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q109-serine-one-carbon

SHMT2 (P34897) - the mitochondrial one-carbon step, and the paralogue that rises while it falls is why its loss may not matter

also called serine hydroxymethyltransferase, mitochondrial · GLYM_HUMAN · GLYA

SHMT2 converts serine into glycine inside mitochondria and in doing so supplies most of the one-carbon units a dividing cell needs for DNA. A menin inhibitor halves it in KMT2A-rearranged leukaemia cells - but its cytosolic twin SHMT1 goes up at the same time, and in leukaemia both have to be removed before growth suffers.

The same thing, for a scientist

SHMT2 (P34897, 504 aa, exactly one reviewed human record, no synonyms) cleaves serine to glycine with 5,10-methylene-THF, providing the major source of folate one-carbon units, and feeds de novo mitochondrial thymidylate synthesis [E3]. **Measured here:** RS4;11 day 7, **VTP-50469 -1.09** from a baseline of 8.10 log2 CPM, EPZ5676 -0.67, margin **-0.42** (61% reproduced); on the array all three GPL570 probes fall in RS4;11 (-0.62, -0.29, -0.71) and SEM (-0.60, -0.35, -0.57) and all three RISE in REH (+0.17, +0.04, +0.15) from well-detected baselines of 10.89-13.09 log2 RMA, which makes SHMT2 **the cleanest wild-type discriminator in the panel**; MV4;11 -0.94 against SGC0946 -0.33; MOLM13 -0.41, flat [E1]. **The paralogue moves the other way in the same samples: SHMT1 +0.78** [E1]. That matters because in T-ALL **loss of BOTH SHMT1 and SHMT2 was necessary for impaired growth and cell-cycle arrest**, the phenotype was rescued by formate, and the dual inhibitor RZ-2994 remained effective in methotrexate resistance [E4]. DepMap: mean **-0.1095, 1.7% of 1,258 below -0.5, leukaemia -0.1842** [E2]. Germline biallelic loss causes a neurodevelopmental disorder with **congenital microcephaly and progressive hypertrophic cardiomyopathy** [E3].

Go deeper
SHMT2 is where the subcellular dimension is real

P34897 carries five SUBCELLULAR LOCATION statements, all ECO:0000269: Mitochondrion (PubMed 21876188, 24075985), Mitochondrion matrix / mitochondrion nucleoid (18063578), Mitochondrion inner membrane (21876188), Cytoplasm (24075985) and Nucleus (24075985) 3.

So this is not a formality in this beat: the menin-inhibitor serine response spans compartments. PHGDH, PSAT1 and GPT2 have no curated compartment at all; SHMT2, MTHFD2 and ALDH1L2 are mitochondrial; and SHMT2 additionally has a curated nuclear pool. Compartment does not predict compound-selectivity - the mitochondrial members (ALDH1L2 -0.49, SHMT2 -0.42, MTHFD2 -0.03) carry smaller menin-selective margins than cytosolic PHGDH (-1.72), which is the opposite of what a single mitochondrial-stress explanation would give.

The mitochondrion-nucleoid annotation rests on PubMed 18063578 and the nuclear one on 24075985; neither paper was opened, so per rule 91 these are recorded as UniProt's pointers and not as experiments this entry has read.

The paralogue is the reason not to nominate this

In the same RS4;11 day-7 samples, SHMT2 -1.09 and SHMT1 +0.78 1. SHMT1's baseline is low (3.02 log2 CPM) so its rise is small in absolute terms, but the direction is opposite and it is reproduced by EPZ5676 (+1.05).

A genome-scale CRISPR screen in T-ALL put the one-carbon folate, purine and pyrimidine pathways at the top of the required metabolic pathways, and then found that both SHMT1 and SHMT2 had to be suppressed before growth was impaired; the effect of the dual inhibitor RZ-2994 was rescued by formate, and suppression of both in vivo inhibited leukaemia progression 4.

So SHMT2's DepMap null (1.7% of 1,258) is most likely a redundancy null rather than a dispensability null, and a drug-induced fall in SHMT2 alone, with SHMT1 rising, is unlikely to have a metabolic consequence. This is the clearest paralogue case in the beat and it is measured rather than assumed - which is the standard this project adopted after an earlier beat's paralogue-redundancy claim was decided by a proxy.

Facts and figures
gene symbolSHMT2
uniprotP34897
length AA504
entry nameGLYM_HUMAN
protein nameSerine hydroxymethyltransferase, mitochondrial
ensembl gene IDENSG00000182199
ensembl readbackdisplay_name read back as SHMT2 before use
accession resolutiongene_exact:SHMT2 AND organism_id:9606 AND reviewed:true returns EXACTLY ONE reviewed row (rule 62 asserted): P34897, GLYM_HUMAN, gene_primary SHMT2, no synonyms, 504 aa.
localizationMITOCHONDRION, experimental, with a curated nuclear and cytoplasmic pool as well. UniProt P34897 carries five ECO:0000269 statements: Mitochondrion (PubMed 21876188, 24075985), Mitochondrion matrix / mitochondrion nucleoid (18063578), Mitochondrion inner membrane (21876188), Cytoplasm (24075985), Nucleus (24075985). Rule 91: none of those papers was opened, so each is UniProt's pointer and not an experiment read here. No localisation experiment on SHMT2 has been performed in any KMT2A-rearranged cell and no co-localisation with a fusion oncoprotein exists.
subcellular localizationMITOCHONDRIAL MATRIX AND INNER MEMBRANE, plus cytoplasm and nucleus - the best-annotated compartment record in this beat's panel, and the reason the subcellular dimension is substantive here rather than a formality. PHGDH, PSAT1 and GPT2 have no curated compartment at all; ALDH1L2 and MTHFD2 are mitochondrial.
subnuclear localizationNOT RESOLVED BELOW 'NUCLEUS'. UniProt records a Nucleus statement (ECO:0000269, PubMed 24075985) but no subnuclear compartment - no speckle, nucleolar or body assignment. Recorded as an unresolved POSITIVE rather than a negative: SHMT2 is annotated nuclear, and where in the nucleus is unmeasured. This graph holds localization-nuclear-speckles-kmt2a-r and an open AFF1/AFF4 speckle question, so a subnuclear assignment for a nuclear-annotated metabolic enzyme is a real and unperformed experiment.
cell typeRS4;11 and SEM (KMT2A::AFF1 B-ALL), REH (KMT2A wild type, ETV6::RUNX1 - and for SHMT2 this is a clean control, all three probes well above floor in all three lines), MOLM13 (KMT2A::MLLT3 AML) and MV4;11 (KMT2A::AFF1 myeloid) measured here. T-ALL cell lines and in vivo models in PMID 34341479. AML leukaemia stem cells in PMID 40752747 (title only). NO primary infant KMT2A-rearranged sample and NO normal human pro-B cell.
cell stateMeasured in the drug-treated arresting state at day 7 of continuous menin-inhibitor exposure under uncontrolled medium serine and glycine supply. At day 2 SHMT2 carries only 20% of its day-7 fall (-0.22 of -1.09), the LATEST profile of any moving serine gene and closer to the late ISR-output genes (TRIB3 13%, ATF5 22%) than to the serine arm's median of 38% - a second reason SHMT2 does not behave like PHGDH. The state in which SHMT2 has published function is S/G2 progression: SHMT1/2 inhibition induces S/G2 arrest rescued by formate.
epigenetic readoutNONE MEASURED at the SHMT2 locus in any system in the menin or KMT2A literature. PMID 29672864's H3K4me3 ChIP is at the PHGDH promoter only and its SSP gene set was PHGDH, PSAT1 and PSPH - SHMT2 is not in it. No H3K4me3, H3K79me2, KMT2A or menin ChIP at SHMT2 exists anywhere. A bounded absence, and the locus would come free with any ChIP-seq this beat's other entries request.
shown for fusionREADOUT ONLY, and it is the panel's best-controlled readout: all three array probes fall in both KMT2A::AFF1 B-lymphoid lines and all three rise in the KMT2A-wild-type line from well-detected baselines, which no other serine gene in this beat achieves on every probe. But the wild-type arm uses the WEAK compound MI-2-2 and the effects are tenths of a log2. FLAT in KMT2A::MLLT3 (MOLM13 -0.41). No mechanism, no chromatin measurement and no dependency has been shown in any fusion; the functional work is in T-ALL, which carries no KMT2A rearrangement.
paralogue moves the other waySHMT1 (P34896, 483 aa, cytosolic, and itself carrying an ECO:0000269 Nucleus statement) is +0.78 under VTP-50469 and +1.05 under EPZ5676 in the same RS4;11 day-7 samples from a baseline of 3.02 log2 CPM, and is flat on all four array probes in all three lines. In T-ALL both SHMT1 and SHMT2 had to be lost before growth was impaired (PMID 34341479), so a SHMT2-only fall with SHMT1 rising is unlikely to have a metabolic consequence, and SHMT2's DepMap null is most likely a redundancy null.
inhibitorRZ-2994, a dual SHMT1/SHMT2 inhibitor, tool compound only; no clinical programme and no trial. ClinicalTrials.gov returns 0 studies for 'SHMT inhibitor' (queried 2026-09-22).
absence classmeasured-not-analysed
absence class noteThe transcript response was in deposited data and unanalysed. Never-measured: a chromatin mark at the SHMT2 locus in a KMT2A-rearranged cell, and whether SHMT1 induction compensates for SHMT2 loss in this disease.
mv411 independent lab
menin_vtp50469_500nM-0.940menin_vtp50469_500nM: -0.94margin-0.610margin: -0.61dot1l_sgc0946_5uM-0.330dot1l_sgc0946_5uM: -0.33dmso_baseline_log2cpm5.990dmso_baseline_log2cpm: 5.99
the measurement rs411 d7 log2fc
{
  "vtp50469_330nM": -1.09,
  "epz5676_1uM": -0.67,
  "vtp_minus_epz": -0.423,
  "epz_fraction_of_vtp": 0.61,
  "dmso_baseline_log2cpm": 8.1,
  "detected": true,
  "floor_collision": false,
  "d2_vtp": -0.222,
  "d2_share_of_d7": 0.2
}
molm13 kmt2a mllt3
{
  "vtp": -0.408,
  "epz": -0.246,
  "dmso_baseline_log2cpm": 7.88,
  "detected": true
}
mv411 independent lab — the numbers behind the chart
{
  "menin_vtp50469_500nM": -0.94,
  "dot1l_sgc0946_5uM": -0.33,
  "margin": -0.61,
  "dmso_baseline_log2cpm": 5.99
}
the measurement array gse63664 every probe
{
  "214095_at": {
    "rs411_dmso_log2_rma": 11.74,
    "rs411_mi22": -0.62,
    "sem_dmso": 11.39,
    "sem_mi22": -0.6,
    "reh_dmso": 10.95,
    "reh_mi22": 0.17
  },
  "214096_s_at": {
    "rs411_dmso_log2_rma": 13.41,
    "rs411_mi22": -0.29,
    "sem_dmso": 13.22,
    "sem_mi22": -0.35,
    "reh_dmso": 13.09,
    "reh_mi22": 0.04
  },
  "214437_s_at": {
    "rs411_dmso_log2_rma": 11.72,
    "rs411_mi22": -0.71,
    "sem_dmso": 11.48,
    "sem_mi22": -0.57,
    "reh_dmso": 10.89,
    "reh_mi22": 0.15
  },
  "detection_floor_used": "6.0 log2 RMA; all three probes far above it in all three lines, so every value is a measurement and the REH nulls are real nulls rather than absences of signal"
}
essentiality
{
  "depmap_release": null,
  "depmap_release_note": "Rule 83: no release identifier in the Open Targets payload. Fetched 2026-09-22.",
  "depmap_screens": 1258,
  "depmap_tissue_blocks_flattened": 29,
  "chronos_mean": -0.1095,
  "chronos_median": -0.0963,
  "pct_lines_below_minus_0_5": 1.7,
  "leukaemia_screens": 57,
  "leukaemia_mean_gene_effect": -0.1842,
  "depmap_common_essential": "not asserted",
  "paralogue_caveat": "SHMT1 mean -0.1535, 2.1% below -0.5, leukaemia -0.1201 in the same fetch. Both are single-gene nulls and the T-ALL screen shows the double is what matters, so neither number is informative alone."
}
How this could bite us back
  • **Germline biallelic SHMT2 loss causes a neurodevelopmental disorder with congenital microcephaly, spastic paraparesis and PROGRESSIVE HYPERTROPHIC CARDIOMYOPATHY in childhood.** The cardiac phenotype is the distinctive one and it lands on a population already receiving anthracyclines, whose dose-limiting toxicity is cardiomyopathy. Any SHMT-directed agent in this disease inherits an anticipated cardiac interaction and nothing has been measured.
  • **One-carbon units from SHMT2 are required for thymidylate and purine synthesis in every dividing cell**, including the marrow that must recover between blocks of chemotherapy and the gut epithelium already injured by it. The in vitro rescue is formate, which is not a clinical manoeuvre.
  • **The paralogue makes a single-gene intervention futile and a double-gene intervention more toxic.** The published leukaemia result requires loss of both SHMT1 and SHMT2, removing cytosolic and mitochondrial one-carbon supply at once - closer to an antifolate than to a selective agent, and antifolates are already in this protocol.
  • **Nothing here is a fusion-selective vulnerability.** The functional work is in T-ALL, the fusion-scoped literature is a correctly formed zero, the transcript fall is 1.09 log2 with 61% of it reproduced by a DOT1L inhibitor, and 1.7% of 1,258 DepMap lines require the gene.
Still open
  1. Does SHMT1 induction compensate for the SHMT2 fall in KMT2A-rearranged cells? The T-ALL screen says both must go before growth suffers, and in these samples one falls while the other rises. Measuring glycine and formate output under a menin inhibitor, with and without SHMT1 knockdown, decides whether SHMT2's fall means anything at all.
  2. Is a SHMT inhibitor ANTAGONISTIC with methotrexate in this disease? Methotrexate is backbone therapy in infant KMT2A-rearranged ALL and acts on the same folate pool; RZ-2994 remained effective in methotrexate-RESISTANT T-ALL cells, which is a resistance claim and not a combination claim. Nobody has asked about antagonism in a methotrexate-containing protocol.
  3. Where in the nucleus is SHMT2? UniProt carries an ECO:0000269 Nucleus statement with no subnuclear resolution, and this graph already holds a nuclear-speckle question for AFF1 and AFF4. A nuclear pool of a one-carbon enzyme in a fusion-driven cell is a cheap and unperformed imaging experiment.
  4. Does the mitochondrion-nucleoid annotation matter here? PubMed 18063578 places SHMT2 in the mitochondrial nucleoid and the paper was not opened. If SHMT2 is part of a mitochondrial-DNA-associated complex, a 1.09 log2 fall has implications for mitochondrial genome maintenance that a one-carbon framing misses.
Every source behind this entry (5)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    On a pre-registered panel, SHMT2 falls 1.09 log2 in RS4;11 at day 7 under VTP-50469 330 nM from a detected baseline of 8.10 log2 CPM against 0.67 under EPZ5676 (margin -0.423, 61% reproduced), carries only 20% of that effect at day 2 (-0.222), is -0.408 in MOLM13 from a detected baseline of 7.88 and -0.94 in MV4;11 against -0.33 under SGC0946; on the GPL570 array all three probes fall under MI-2-2 in RS4;11 (-0.62 at baseline 11.74, -0.29 at 13.41, -0.71 at 11.72) and SEM (-0.60, -0.35, -0.57) and all three rise in REH (+0.17 at 10.95, +0.04 at 13.09, +0.15 at 10.89); and in the same RS4;11 day-7 samples the cytosolic paralogue SHMT1 moves in the opposite direction, +0.78 under VTP-50469 and +1.05 under EPZ5676 from a baseline of 3.02 log2 CPM.

    What this evidence cannot show

    One line per platform, single doses, NO dose-response, n=2-3 per arm, day 7 of continuous exposure so arrest is a competing explanation for any coordinated biosynthetic fall. No series records the amino-acid composition of the medium, and this is an amino-acid-response panel. MI-2-2 in the array arm is the WEAK compound, so the REH wild-type contrast has little dynamic range. SHMT1's baseline of 3.02 log2 CPM is low, so its +0.78 is a small absolute change and the reciprocal-paralogue reading must not be over-stated from transcript data alone.

    Evidence that argues against this (1)
    • SHMT2's margin of -0.42 is smaller than the serine arm's median of -0.55 and far from PHGDH's -1.72, so SHMT2 is one of the panel members that would NOT support this beat's menin-selectivity conclusion on its own. A reader using SHMT2 as the exemplar of the serine arm would reach the opposite verdict. this repository 2026
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    SHMT2 DepMap gene effect is mean -0.1095, median -0.0963, with 1.7% of 1,258 screens below -0.5 and a leukaemia mean of -0.1842 over 57 screens, fetched with all 29 tissue blocks flattened and approvedSymbol read back as SHMT2; SHMT1 in the same fetch is -0.1535 with 2.1% below -0.5 and a leukaemia mean of -0.1201.

    What this evidence cannot show

    Rule 83: the payload carries NO release identifier, so depmap_release is recorded as null and absolute values drift quarterly; the cross-gene contrast is the durable part. Leukaemia screens are matched on the payload's own diseaseFromSource string against four substrings and are NOT split by KMT2A status. Rule 84: no median-split proportion is reported - the percentage is against an absolute -0.5 threshold. A CRISPR knockout in folate- and serine-replete medium understates a biosynthetic dependency.

    Evidence that argues against this (1)
    • This null is most likely a PARALOGUE null rather than a dispensability null: a genome-scale screen in T-ALL found that BOTH SHMT1 and SHMT2 had to be lost before growth was impaired, so a single-gene CRISPR score is the wrong instrument for this gene and DepMap's per-gene summary cannot show it. Pikman Y et al. 2022
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt reviewed human record P34897 (GLYM_HUMAN) has primary gene name SHMT2 with no synonyms, length 504 aa, catalyses cleavage of serine to glycine with production of 5,10-methylenetetrahydrofolate, provides the major source of folate one-carbon units and contributes to de novo mitochondrial thymidylate biosynthesis, carries FIVE ECO:0000269 subcellular-location statements (Mitochondrion, mitochondrion matrix / mitochondrion nucleoid, mitochondrion inner membrane, Cytoplasm and Nucleus), and records a neurodevelopmental disorder with cardiomyopathy, spasticity and brain abnormalities characterised by global developmental delay, moderate to severe intellectual disability, spastic paraparesis, congenital microcephaly, progressive hypertrophic cardiomyopathy in childhood and corpus callosum abnormalities in all patients.

    What this evidence cannot show

    Resolved 2026-09-22 with gene_exact:SHMT2 AND organism_id:9606 AND reviewed:true; exactly one reviewed row, gene_primary and length read back. Rule 91: NONE of the five cited localisation papers was opened, so each is recorded as UniProt's pointer rather than as an experiment this entry has read - including the mitochondrion-nucleoid and nuclear statements, which are the unusual ones and therefore the ones most worth reading. The disease description is a curated summary, not a cohort.

    Evidence that argues against this (1)
    • A protein with five curated compartments including a mitochondrial nucleoid and a nuclear pool is a protein whose 'mitochondrial' label is a simplification, and this beat's framing of a mitochondrial-versus-cytosolic split inside the serine pathway is correspondingly soft. SHMT1, the supposedly cytosolic paralogue, also carries an ECO:0000269 Nucleus statement. UniProt Consortium 2026
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    An unbiased genome-scale CRISPR-Cas9 screen in T-cell acute lymphoblastic leukaemia scored the one-carbon folate, purine and pyrimidine pathways as the top metabolic pathways required for proliferation; the SHMT1/SHMT2 inhibitor RZ-2994 differentially inhibited T-ALL lines and induced S/G2 arrest; the effects were rescued by formate supplementation; loss of BOTH SHMT1 and SHMT2 was necessary for impaired growth and cell-cycle arrest, and suppression of both inhibited leukaemia progression in vivo; and RZ-2994 decreased leukaemia burden in vivo and remained effective in the setting of methotrexate resistance in vitro.

    Targeting serine hydroxymethyltransferases 1 and 2 for T-cell acute lymphoblastic leukemia therapy. · Pikman Y, Ocasio-Martinez N, Alexe G, et al., Stegmaier K · Leukemia 2022
    PMID 34341479DOI
    What this evidence cannot show

    Abstract read this session; full text not retrieved, so the lines, the RZ-2994 concentrations and the in vivo model were not checked. **The system is T-ALL, not B-lineage KMT2A-rearranged ALL**, and the abstract states the lines were DIFFERENTIALLY sensitive without naming the determinant. RZ-2994 is a tool compound with no clinical programme; ClinicalTrials.gov returns 0 studies for 'SHMT inhibitor' (queried 2026-09-22). Formate rescue establishes the pathway but not the fusion relevance.

    Evidence that argues against this (1)
    • The methotrexate result cuts the other way for this disease: methotrexate is backbone therapy in KMT2A-rearranged infant ALL and acts on the same folate pool, so a SHMT-directed agent would be competing with or duplicating a drug these patients already receive, and this graph holds drug-methotrexate. The question that matters is antagonism, not addition, and nobody has asked it. Ducker GS et al. 2017
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    PubMed 2026-09-22, one tag per term with translations read back: SHMT2[tiab] returns 321 records; SHMT2[tiab] AND the eight-operand KMT2A-fusion set returns 0 with a non-empty querytranslation and warninglist 'No items found.'; the same against a nine-operand menin/menin-inhibitor set returns 0 with the same signature; SHMT2[tiab] AND leukemia[tiab] returns 6, all six titles read, of which the on-topic ones are the T-ALL SHMT1/2 targeting paper, an ALDH2-PKC-delta-SHMT2 axis in AML leukaemia stem cells and an NSUN2 serine-metabolism AML paper; and the alias GLYA[tiab] returns 282 with 0 fusion-scoped.

    What this evidence cannot show

    Bounded by titles and abstracts only; supporting figures not searched. The alias GLYA returns 282 records and was NOT id-set-intersected with SHMT2's 321, so whether it names the same object is unverified here - only its fusion-scoped zero is used, and a zero is safe either way. The six leukaemia-scoped titles were read but not their abstracts, except where cited separately.

    Evidence that argues against this (1)
    • A zero on the fusion operand set for a gene with 321 records means a well-studied enzyme has never been written about in this context, which makes a supplementary-table measurement more likely than a genuine absence of data. This beat's own RS4;11 and MOLM13 numbers are exactly such a measurement, extracted from a 2019 deposition, so the absence is one of reporting. this repository 2026
Connected to