Oncofusion

start / graph / gene-sesn2

Genestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q106-isr-stress-programme

SESN2 / sestrin-2 (P58004) - the ISR's exit into mTORC1: a leucine sensor that falls 1.4 log2 under either inhibitor, and the one panel member with a real structural record

also called SESN2 · Hi95 · SEST2 · Sestrin-2 · sestrin 2 · SESN2_HUMAN

Sestrin-2 is the cell's leucine detector: when the amino acid is scarce it grabs a protein complex and switches off the cell's main growth signal. It falls about three-fold in KMT2A-rearranged leukaemia cells under either of two unrelated drugs, which links the stress programme to the growth-signalling machinery.

The same thing, for a scientist

SESN2 (P58004, 480 aa) is an **intracellular leucine sensor** that, in the absence of leucine, binds the GATOR2 subcomplex and prevents mTORC1 signalling - the largest single body of evidence on this protein, carrying nine or more experimental citations in UniProt and **six PDB entries, the most of any gene in this beat's panel** [E3]. It was originally characterised as Hi95, a p53-regulated cysteine sulfinyl reductase that regenerates overoxidised peroxiredoxins [E4]. In RS4;11 it falls **-1.32 and -1.21 log2 on two independent GPL570 probes under MI-2-2** (p 0.003 and 0.020) and **-1.42 under VTP-50469 at day 7, with EPZ5676 at -1.53** [E1]. The KMT2A wild-type control is clean (REH +0.15 and -0.04) and SEM agrees in direction (-0.89, -0.76) though not significantly. **Why it matters more than its effect size: SESN2 is the point where the ISR hands off to mTORC1**, so an ISR-output collapse that includes SESN2 predicts mTORC1 **de-repression** - a growth signal going UP under a drug intended to stop growth. Nothing in either dataset or in this graph tests that. DepMap 26.06: -0.0199 mean across 1,258 screens, 0.2% below -0.5 - not a dependency [E2]. PubMed: **0** fusion-scoped and **0** menin-drug-scoped records, both with valid querytranslations and 'No items found.' [E5].

What it looks like
loading AF-P58004-F1…
AlphaFold AF-P58004-F1 ↗
colourshape

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.

very high (90+)confident (70–90)low (50–70)very low (<50)
I have never read one of these pictures before

A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.

Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.

Anything in yellow sticks is a bound small molecule, which is usually the drug.

Go deeper
The measurement
datasetcellarmDMSOlog2FC
array 223196_s_atRS4;11MI-2-29.43-1.32 (p 0.003)
array 223195_s_atRS4;11MI-2-26.77-1.21 (p 0.020)
array 223196_s_atSEMMI-2-29.05-0.89 (p 0.28)
array 223196_s_atREHMI-2-29.99-0.04
array 223195_s_atREHMI-2-26.85+0.15
RNA-seqRS4;11 d7VTP-504697.33-1.42
RNA-seqRS4;11 d7EPZ56767.33-1.53
RNA-seqRS4;11 d2VTP-504696.09-0.34
RNA-seqMOLM13 d7VTP-504697.30-0.87

Two probes agreeing to 0.11 log2 across a 2.7 log2 baseline range, with a clean wild-type control on both, is the second-best internal consistency in this beat's panel after ATF5's four probes. SESN2 is also the only panel member well expressed in every line and dataset - no arm sits near a detection floor - which makes its fold changes the most reliable numerically even though they are not the largest.

The mTORC1 consequence nobody has tested

SESN2's dominant function is as a leucine sensor upstream of mTORC1: with leucine absent, sestrin-2 binds GATOR2 and mTORC1 signalling is suppressed; with leucine bound, sestrin-2 releases GATOR2 and mTORC1 proceeds 3. Sestrin-2 is therefore an mTORC1 brake whose level sets how strongly amino-acid scarcity is transduced.

A 1.4 log2 loss of the brake predicts mTORC1 de-repression. Combined with the rest of this beat's panel the prediction sharpens, because several other mTORC1-relevant genes move in the same direction in the same samples 1:

genefunction relative to mTORC1RS4;11 d7, VTP-50469
SESN2leucine sensor, mTORC1 brake-1.42
RHEBL1RHEB-like GTPase, mTORC1 activator family-1.34
SLC7A5leucine importer (with SLC3A2)-1.34
SLC3A2heavy chain of the SLC7A5 transporter-1.26
ASNSasparagine synthetase, ATF4 target-1.72

Brake down, leucine importer down, and an mTORC1-activator-family GTPase down, all together. Those do not compose into a single prediction - losing the importer lowers intracellular leucine, which would activate sestrin-2's brake function, while losing sestrin-2 removes it. The net direction of mTORC1 signalling in a menin-inhibitor-treated RS4;11 cell is genuinely unpredictable from these data, and it has never been measured. A phospho-S6 or phospho-4E-BP1 western is the readout, and it is the same blot as the phospho-eIF2alpha experiment this beat proposes.

Why this is the most therapeutically actionable gap in the beat: mTOR inhibitors are clinically available, and if a menin inhibitor de-represses mTORC1 then a combination is rational and nobody has looked; if it suppresses mTORC1 the combination is redundant. One blot decides which.

Facts and figures
uniprotP58004
length AA480
entry nameSESN2_HUMAN
protein nameSestrin-2
ensembl gene IDENSG00000130766
PDB noteUniProt P58004 cross-references six PDB entries - the most of any gene in this beat's panel - and NONE was fetched. No chain, resolution, residue range, leucine-binding pocket or GATOR2 interface is claimed from them.
alphafold IDAF-P58004-F1
domains noteUniProt P58004's feature table carries NO Domain features, so no sestrin fold or leucine-binding-pocket boundary is annotated and none is invented, despite six available structures. Lane deliberately EMPTY per the integer-residue rule; the structures would supply boundaries but were not fetched.
accession resolutiongene_exact:SESN2 AND organism_id:9606 AND reviewed:true returns EXACTLY ONE reviewed row, P58004, SESN2_HUMAN, 480 aa, synonyms Hi95 and SEST2; re-fetched alone with name and length compared back: PASS.
localizationCYTOPLASM, on a single coded citation. UniProt P58004 carries exactly one location statement, Cytoplasm, ECO:0000269 from PubMed 15105503. RULE 91 FLAG: that paper is the p53/sestrin peroxireduxin-regeneration Science paper, not a localisation study, so the code establishes that a curator read an experiment and not that this is the founding or best observation of sestrin-2's compartment. Functionally the protein must reach the lysosomal GATOR2 pool, which the annotation does not capture. No SESN2 localisation experiment exists in any haematopoietic cell.
subnuclear localizationNOT APPLICABLE AND NOT ASKABLE. SESN2 has no annotated nuclear pool, coded or uncoded. There is no subnuclear question to answer, and none is invented. Recorded as a bounded absence resting on a prior absence.
cell typeRS4;11 and SEM (t(4;11) KMT2A::AFF1 B-ALL), REH (KMT2A wild type, clean negative control on both probes) and MOLM13 (KMT2A::MLLT3 AML), measured here [E1]. UniProt records SESN2 as widely expressed. The leucine-sensor work is in non-haematopoietic culture and in vitro [E3]; the founding redox work in purified protein and cell culture [E4]. No primary leukaemia sample, no normal haematopoietic cell.
cell stateMeasured in the day-6/day-7 drug-arrested state. SESN2 is definitionally an AMINO-ACID-STATE sensor - its function is read out only when leucine is limiting - and NEITHER public experiment controlled or reported the amino-acid composition of the medium or the degree of medium exhaustion at day 6-7. So an amino-acid-state-dependent gene was measured under uncontrolled amino-acid supply, in cultures whose density differs between arms because the drug arrests growth. This is the single largest uncontrolled variable across this beat's whole panel.
epigenetic readoutNo writer, eraser or reader activity. Its fall is reproduced by the DOT1L inhibitor EPZ5676 to within 0.11 log2, so its expression tracks H3K79-methylation-dependent fusion output rather than the menin interface. No histone-mark measurement at the SESN2 locus exists in any KMT2A-rearranged cell. SESN2 is p53-regulated, and p53 status differs across the four lines measured here - which was not controlled for and is a real confounder for this gene specifically.
shown for fusionMEASURED in KMT2A::AFF1 (RS4;11 on two agreeing probes plus RNA-seq, SEM in the same direction but not significantly) with a clean KMT2A-wild-type REH control, and in KMT2A::MLLT3 (MOLM13 -0.87). Transcript only. No sestrin-2 protein, no mTORC1 readout, no ChIP and no functional test in any fusion context.
absence classliterature-absence-bounded
absence class noteRECLASSIFIED 2026-09-23 (cycle 76). This field held free text, which `scripts/check-absence-class.py` rejects; the prose is preserved here VERBATIM and unsummarised because it is more informative than the token that replaced it. Token chosen: literature-absence-bounded - the two-part prose says MEASURED-AND-<something> for the data arm and BOUNDED-UNPUBLISHED for the prose arm. The data arm is a RESULT, not an absence; the absence is the bounded literature, and the bound's own limits (titles and abstracts only, per rule 80) are stated in the preserved prose. ORIGINAL absence_class, verbatim: «MEASURED-AND-RELIABLE in transcript; BOUNDED-UNPUBLISHED in prose (0 fusion-scoped, 0 menin-drug-scoped, all with valid querytranslations and 'No items found.') - WITH THE IMPORTANT LIMIT that the operand list contained no mTOR, mTORC1, rapamycin or GATOR term, so this zero bounds the sestrin literature and not the mTORC1-in-KMT2A literature.»
PDB IDS
[
  "5CUF",
  "5DJ4",
  "5T0N",
  "6N0M",
  "9LVJ",
  "9LWF"
]
domains
[]
the measurement
{
  "rs411_array_mi22_lfc_223196_s_at": -1.32,
  "rs411_array_mi22_lfc_223195_s_at": -1.212,
  "rs411_array_dmso_log2_rma": 9.43,
  "sem_array_mi22_lfc": -0.892,
  "reh_array_mi22_lfc_223196_s_at": -0.043,
  "reh_array_mi22_lfc_223195_s_at": 0.152,
  "rs411_rnaseq_d7_vtp_lfc": -1.418,
  "rs411_rnaseq_d7_epz5676_lfc": -1.528,
  "rs411_rnaseq_d2_vtp_lfc": -0.34,
  "molm13_rnaseq_d7_vtp_lfc": -0.87,
  "note": "the only panel member well expressed in EVERY arm of both datasets with no value near a detection floor, so numerically the most reliable fold changes in the set",
  "menin_specificity": "SHARED WITH DOT1L INHIBITION (VTP -1.42, EPZ5676 -1.53)"
}
the mtorc1 handoff
{
  "why_it_matters": "SESN2 is the point where the ISR hands off to mTORC1. Losing the brake predicts mTORC1 DE-REPRESSION - a growth signal rising under a drug meant to stop growth.",
  "co_moving_genes_rs411_d7_vtp": {
    "SESN2 (mTORC1 brake)": -1.42,
    "RHEBL1 (RHEB-like GTPase)": -1.34,
    "SLC7A5 (leucine importer)": -1.34,
    "SLC3A2 (importer heavy chain)": -1.26,
    "ASNS (ATF4 target)": -1.72
  },
  "why_the_prediction_does_not_resolve": "Brake down predicts mTORC1 up; leucine importer down lowers intracellular leucine, which would ACTIVATE the brake. The net direction is unpredictable from transcript data.",
  "the_readout": "phospho-S6 or phospho-4E-BP1 western on drug-treated RS4;11 - the same blot as the phospho-eIF2alpha experiment this beat proposes",
  "therapeutic_stake": "mTOR inhibitors are clinically available. If a menin inhibitor de-represses mTORC1 the combination is rational and untested; if it suppresses mTORC1 the combination is redundant.",
  "framing_status": "POST-HOC. The five genes were pre-registered as the panel's amino-acid arm, but grouping them as an mTORC1 story was not pre-registered and is labelled as post-hoc."
}
essentiality
{
  "depmap_release": "26.06 (read back from meta{dataVersion})",
  "depmap_screens": 1258,
  "chronos_mean": -0.0199,
  "pct_lines_below_minus_0_5": 0.2,
  "leukaemia_screens": 57,
  "leukaemia_mean_gene_effect": -0.0317,
  "reading": "not a dependency; paralogue redundancy with SESN1 and SESN3 was NOT tested and is a standard explanation for a single-gene null in a three-member family"
}
identifier traps
{
  "registry_status": "NOT in data/symbol-collisions.json as of 2026-09-22 (157 rows read).",
  "sestrin_2_spelling": "'sestrin 2' and 'sestrin-2' return IDENTICAL 199-record id sets, because PubMed normalises the hyphen into a phrase. So the hyphenation rule that matters for compound codes (VTP-50469 versus VTP50469) does NOT apply to spelled-out protein names, and trying both is free but adds nothing.",
  "Hi95": "the founding alias, from the 2004 p53 paper; not queried as a token here because 'Hi95' is not a plausible clean search string.",
  "array": "GPL570 carries two SESN2 probes, both well above the detection floor at 9.43 and 6.77 log2 RMA in RS4;11 DMSO."
}
How this could bite us back
  • SESN2 is an mTORC1 BRAKE. A drug that removes it may be de-repressing the cell's principal growth-signalling pathway while suppressing the fusion's transcriptional programme - an escape route that no MEN1 sequencing panel would detect and that an mTOR inhibitor would close.
  • SESN2 also regenerates overoxidised peroxiredoxins in its founding description, so losing it should raise oxidative burden - the opposite prediction to losing CHAC1, which should raise glutathione. The redox net effect of this drug on a KMT2A-rearranged cell is unknown and the two largest panel members point in opposite directions.
  • Sestrin-2's function is only read out when leucine is scarce. A leukaemia cell in a bone-marrow niche is not in leucine-replete tissue-culture medium, so the in-vitro measurement may understate or invert what happens in a patient - and asparaginase, a backbone drug in this disease, deliberately strips an amino acid.
Still open
  1. Is mTORC1 signalling UP or DOWN in menin-inhibitor-treated RS4;11? The brake (SESN2), the leucine importer (SLC7A5/SLC3A2) and an mTORC1-activator-family GTPase (RHEBL1) all fall together and the net direction is unpredictable. One phospho-S6 blot decides whether an mTOR inhibitor combination is rational or redundant.
  2. Was the medium's amino-acid state controlled in either public experiment? A leucine sensor measured at day 6-7 in cultures of differing density, with no reported medium composition, is the largest uncontrolled variable in this beat's panel - and it applies to every ATF4 amino-acid-response gene in it.
  3. Do SESN1 and SESN3 change in the same samples, and would family redundancy explain SESN2's flat DepMap gene effect? Neither paralogue was queried.
  4. Does p53 status confound this gene? SESN2 is p53-regulated and the four lines measured here differ in p53 status, which was not controlled.
Every source behind this entry (5)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    SESN2 falls 1.32 and 1.21 log2 on two independent GPL570 probes under MI-2-2 in RS4;11 (p 0.003 and 0.020) and 1.42 log2 under VTP-50469 at day 7 (EPZ5676 -1.53), while the KMT2A-wild-type line REH moves -0.04 and +0.15; in the same RS4;11 day-7 samples RHEBL1 falls 1.34, SLC7A5 1.34, SLC3A2 1.26 and ASNS 1.72.

    What this evidence cannot show

    SESN2 is well expressed in every arm of both datasets with no value near a detection floor, which makes these the most numerically reliable fold changes in the panel. SEM agrees in direction (-0.89) but not significantly (p 0.28) and its arms are n=2. RS4;11 array DMSO arm n=2. The mTORC1 gene set listed here was PRE-REGISTERED as the amino-acid arm of the panel, not selected after seeing the SESN2 result - but it was not pre-registered as an mTORC1 hypothesis, and grouping these five genes under mTORC1 is a post-hoc framing that the entry labels as such.

    Evidence that argues against this (1)
    • The same collapse is produced almost exactly by EPZ5676, a potent DOT1L inhibitor, in the same cells at the same timepoint, so the effect marks disruption of the fusion's transcriptional output by either route and not the menin-KMT2A interface specifically. NCBI GEO, re-analysed here 2026
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    In DepMap 26.06, SESN2 has a mean CRISPR gene effect of -0.0199 across 1,258 screens with 0.2% of lines below -0.5 and a leukaemia mean of -0.0317 across 57 leukaemia screens.

    What this evidence cannot show

    Release recorded per rule 83; gene-specific screen count; leukaemia screens by substring match with ids persisted per rule 90. A mean of -0.02 is indistinguishable from zero. Sestrins are a three-member family (SESN1, SESN2, SESN3) and paralogue redundancy is a standard explanation for a single-gene CRISPR null in such a family - neither SESN1 nor SESN3 was queried here, so redundancy was not tested.

    Evidence that argues against this (1)
    • A growth-suppressive brake protein should show a POSITIVE gene effect on knockout if it were doing anything measurable in a rich-medium screen, and it shows -0.02. Either sestrin-2's mTORC1 brake is not load-bearing in these conditions - which is expected, since screen medium is leucine-replete - or the family compensates. Either way the null carries no information about the drug-treated state. UniProt Knowledgebase 2026
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt P58004 is the single reviewed human SESN2 record, 480 aa, synonyms Hi95 and SEST2, with six PDB entries, annotated as an intracellular leucine sensor that negatively regulates mTORC1 signalling through the GATOR complex and that in the absence of leucine binds the GATOR2 subcomplex and prevents mTORC1 signalling, on nine or more experimental citations; its only subcellular-location statement is Cytoplasm.

    UniProtKB - P58004, SESN2_HUMAN, sestrin-2 · UniProt Knowledgebase 2026
    What this evidence cannot show

    Resolved live 2026-09-22: exactly one reviewed record (rule 62 satisfied), re-fetched alone with name, length and primary symbol compared back: PASS. Graded established because the leucine-sensor function rests on a large, independently replicated body of work reflected in nine-plus UniProt experimental citations and six structures - unusually strong for this panel. RULE 91: the Cytoplasm location's ECO:0000269 points at PubMed 15105503, which is the p53/peroxiredoxin paper and NOT a localisation study; the function citations were not individually opened. The six PDB entries were NOT fetched and no residue, chain or interface is claimed from them.

    Evidence that argues against this (1)
    • The leucine-sensor model has been contested: an alternative reading places sestrin-2's mTORC1 effects downstream of AMPK and of redox signalling rather than of direct leucine binding, and the affinity of the leucine interaction has been argued to be too weak to be physiological in some cell types. UniProt's function line reflects the dominant model, not a settled one. Budanov AV et al. 2004
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Sestrins were identified as a p53-regulated family whose members are required for regeneration of peroxiredoxins containing overoxidised cysteine sulfinic acid, and purified Hi95 (sestrin 2) supports ATP-dependent reduction of overoxidised PrxI in vitro, making sestrins cysteine sulfinyl reductases rather than disulfide reductases.

    Regeneration of peroxiredoxins by p53-regulated sestrins, homologs of bacterial AhpD. · Budanov AV, Sablina AA, Feinstein E, et al. · Science 2004
    PMID 15105503DOI
    What this evidence cannot show

    In vitro biochemistry with purified protein plus cell-culture experiments, 2004, in non-haematopoietic systems - and this founding function has since been substantially displaced by the leucine-sensor/mTORC1 model as the dominant account of sestrin-2 biology. It is cited here because it is the origin of the Hi95 alias and because a redox function on a gene that falls alongside CHAC1 - a glutathione-degrading enzyme - is a coincidence worth recording. The catalytic claim has been disputed in later literature that this beat did not survey.

    Evidence that argues against this (1)
    • The direct sulfinyl-reductase activity attributed to sestrins here has not been robustly reproduced, and the field's centre of gravity has moved entirely to the leucine-sensing/GATOR2 model. Citing the 2004 redox function alongside the 2010s mTORC1 function risks presenting two accounts as complementary when one has largely superseded the other. UniProt Knowledgebase 2026
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    PubMed returns zero records for SESN2 jointly with an eight-operand KMT2A-fusion list and zero for SESN2 with a thirteen-name menin-inhibitor list, both with non-empty querytranslations and warninglist 'No items found.'; the hyphenated and unhyphenated spellings 'sestrin 2' and 'sestrin-2' both return 199 records of which 114 also mention SESN2 (Jaccard 0.2111), and the fusion-scoped spelled-out query also returns a valid zero.

    What this evidence cannot show

    Rule 3 satisfied on all zeros. Rule 72 clause observed: the spelled-out name was tried in hyphenated and unhyphenated forms and they return IDENTICAL id sets, because PubMed normalises the hyphen into a phrase - so for this protein the hyphen distinction that matters for compound codes like VTP-50469 does not apply. Rule 94 form; rule 95 observed. Rule 80: titles and abstracts only.

    Evidence that argues against this (1)
    • A SESN2/mTORC1 result in a KMT2A-rearranged cell would most likely be reported as an mTOR or rapamycin result, not as a sestrin result, and the operand list contains no mTOR, mTORC1, rapamycin, everolimus or GATOR term. The zero bounds the sestrin literature and not the mTORC1-in-KMT2A literature, which is substantial and was not searched. NCBI PubMed 2026
Connected to