Oncofusion

start / graph / gene-chac1

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

CHAC1 / BOTCH (Q9BUX1) - the largest single effect in the whole menin-inhibitor response (-5.11 log2) and a glutathione-degrading pro-apoptotic enzyme, named nowhere in this graph until now

also called CHAC1 · BOTCH · botch · MGC4504 · ChaC glutathione-specific gamma-glutamylcyclotransferase 1 · CHAC1_HUMAN

CHAC1 is an enzyme that chews up glutathione, the cell's main antioxidant, and it is switched on when the cell is under stress to help push it towards death. It shows the biggest change of any gene when KMT2A-rearranged leukaemia cells are treated with a menin-blocking drug, falling about thirty-fold.

The same thing, for a scientist

CHAC1 (Q9BUX1, 222 aa) is a glutathione-specific gamma-glutamylcyclotransferase that cleaves glutathione into 5-oxo-L-proline and Cys-Gly, and acts as the pro-apoptotic arm of the unfolded protein response downstream of the ATF4-ATF3-DDIT3/CHOP cascade - activated by that cascade and specifically NOT by the parallel XBP1 or ATF6 pathways [E2]. **It carries the single largest effect size in the entire two-compound menin-inhibitor intersection: -5.11 log2 under VTP-50469 in RS4;11 at day 7, and -1.38 under MI-2-2** [E1]. Measured here across every arm both datasets contain, three things sharpen that: (a) **the fall is fusion-linked** - CHAC1 sits at 8.95 log2 RMA in DMSO in the KMT2A-wild-type line REH and moves **+0.49** there under the same compound at the same dose [E1]; (b) **it is not menin-specific** - EPZ5676, a DOT1L inhibitor, gives **-5.03** in the same samples [E1]; (c) **CHAC1 is not a dependency** - DepMap 26.06 mean gene effect **+0.0133** across 1,257 screens, 0.4% of lines below -0.5, leukaemia mean +0.0178 [E3], so its loss is a readout and not a mechanism of killing. **The graph named CHAC1 zero times across 1,993 entries before this cycle**, and PubMed holds **zero** records for CHAC1 jointly with any of eight KMT2A-fusion operands and zero with a thirteen-name menin-inhibitor list [E4].

What it looks like
loading AF-Q9BUX1-F1…
AlphaFold AF-Q9BUX1-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 number, and what makes it the largest

Every CHAC1 row in the two datasets 1:

datasetcellarmDMSOlog2FC
GSE63664 array, 219270_atRS4;11MI-2-2 10 uM8.32-1.38 (p 0.006)
GSE63664 arrayRS4;11EPZ0047778.32-0.40
GSE63664 arraySEMMI-2-28.82-0.38
GSE63664 arrayREH (KMT2A wild type)MI-2-28.95+0.49
GSE127506RS4;11 d2VTP-504693.28-1.45
GSE127506RS4;11 d7VTP-504696.35-5.11
GSE127506RS4;11 d7EPZ56766.35-5.03
GSE127506MOLM13 d7VTP-504695.59-1.89

REH is the discriminator and it is a strong one: CHAC1 is more highly expressed in the wild-type line than in either t(4;11) line, and the compound moves it the wrong way there. So this is not an expression artefact and not a general toxicity response.

But the SEM value is a problem for the story and is reported as such. SEM is the other KMT2A::AFF1 line in the same batch, and MI-2-2 moves CHAC1 only -0.38 there against -1.38 in RS4;11. Under the potent compound only RS4;11 was assayed. So the strongest version of this result rests on ONE cell line under ONE compound, with the weak compound agreeing in one t(4;11) line and not the other.

What CHAC1 loss would do, and why it is a bite-back rather than a win

CHAC1 degrades glutathione, and glutathione depletion is a proximal event in apoptosis initiation 2, 5. So a 30-fold loss of CHAC1 is predicted to RAISE intracellular reduced glutathione. Two consequences, both of which argue against reading this as part of the drug's benefit:

  1. A cell with more glutathione is harder to kill with anything that works through oxidative or ferroptotic stress. Several agents a menin inhibitor might be combined with in KMT2A-r leukaemia - anthracyclines, arsenicals, ferroptosis inducers, and cysteine/GSH-axis drugs generally - depend on that stress.
  2. CHAC1 is pro-apoptotic by direct test: overexpression enhances apoptosis, siRNA suppresses it 2. Removing it removes a death signal.

Neither claim has been tested under a menin inhibitor. No glutathione measurement, no ROS measurement and no CHAC1 rescue experiment exists in any KMT2A-rearranged cell; a GSH/GSSG assay on drug-treated RS4;11 is a one-week experiment.

A specificity note that matters for interpreting any such assay: CHAC2, the paralogue, degrades glutathione in the cytosol with 10-20-fold lower catalytic efficiency than CHAC1 at comparable Km, so CHAC2 does not functionally substitute at physiological turnover and cannot be assumed to compensate 5.

Identifier trap: BOTCH, and the alias that is also an English verb

Rule 88 first: no CHAC1 or BOTCH row exists in data/symbol-collisions.json (157 rows read 2026-09-22). New measurement.

Measured as ID sets 4: CHAC1[tiab] returns 286 records, BOTCH[tiab] returns 24, and they share 2 - Jaccard 0.0065. The BOTCH-only records are overwhelmingly the English word used in clinical and surgical prose, not the gene. Fusion-scoped, BOTCH AND (KMT2A OR MLL OR ...) returns 0 with a non-empty querytranslation and "No items found.", so widening to the alias recovers nothing here - but a beat that searched only BOTCH would be reading a 24-record corpus of which 22 are not about this gene.

On the array annotation, GPL570 carries one CHAC1 probe and no probe whose own symbol field is BOTCH, so the array side is unambiguous. The accession side is clean: gene_exact:CHAC1 AND organism_id:9606 AND reviewed:true returns exactly one reviewed record, Q9BUX1, CHAC1_HUMAN, 222 aa, "Glutathione-specific gamma-glutamylcyclotransferase 1", with the name and length read back 5.

Facts and figures
uniprotQ9BUX1
length AA222
entry nameCHAC1_HUMAN
protein nameGlutathione-specific gamma-glutamylcyclotransferase 1
ensembl gene IDENSG00000128965
PDB noteUniProt Q9BUX1 cross-references NO PDB entries. No structure exists for human CHAC1.
alphafold IDAF-Q9BUX1-F1
domains noteUniProt Q9BUX1's feature table carries NO Domain features, so no ChaC/gamma-glutamylcyclotransferase fold boundary is annotated and none is invented. Lane deliberately EMPTY per the integer-residue rule.
accession resolutiongene_exact:CHAC1 AND organism_id:9606 AND reviewed:true returns EXACTLY ONE reviewed row, Q9BUX1, CHAC1_HUMAN, 222 aa, synonym BOTCH, re-fetched alone with name and length compared back: PASS.
localizationCYTOSOL, on one experimental citation: UniProt Q9BUX1, Cytoplasm/cytosol, ECO:0000269 from PubMed 19109178 - and that paper IS the CHAC1 paper, which is the good rule-91 case. A trans-Golgi-network pool is annotated ECO:0000250 BY SIMILARITY to mouse Q8R3J5 only, and is recorded as an uncoded inference. No CHAC1 localisation experiment has ever been done in a haematopoietic cell.
subnuclear localizationNOT APPLICABLE AND NOT ASKABLE. CHAC1 has no annotated nuclear pool at all - coded or uncoded - so there is no subnuclear question to answer. Recorded as a bounded absence resting on a prior absence, not as an unmeasured compartment: a cytosolic glutathione-degrading enzyme is not expected in the nucleus and nobody has reported it there.
cell typeRS4;11 and SEM (t(4;11) KMT2A::AFF1 infant/paediatric B-ALL), REH (KMT2A wild-type ETV6::RUNX1 B-ALL, negative control) and MOLM13 (KMT2A::MLLT3 AML) - all measured here [E1]. The founding functional work is in human aortic endothelial cells [E2] and the paralogue kinetics in purified human and mouse protein [E5]. NO primary leukaemia sample, NO normal haematopoietic cell and NO patient material anywhere in CHAC1's record in this disease.
cell stateMeasured in the drug-arrested state at day 6-7 of continuous menin- or DOT1L-inhibitor exposure. CHAC1 is itself a STATE-DEPENDENT gene - it is induced by chemical UPR inducers and not by other cell stressors [E2] - and its DMSO baseline in the RNA-seq rises from 3.28 to 6.35 log2 CPM between day 2 and day 7, i.e. the untreated cells are progressively inducing it as the culture proceeds. So part of the day-7 effect is failure to induce rather than repression, and the amino-acid and ER-load state of the medium was uncontrolled in both experiments.
epigenetic readoutNo writer, eraser or reader activity. Its epigenetic content here is that its 30-fold fall is reproduced to within 0.09 log2 by EPZ5676, a DOT1L inhibitor that removes H3K79me2/me3 - so CHAC1 expression tracks the fusion's H3K79-methylation-dependent elongation output rather than the menin interface. NO H3K4me3, H3K79me2 or H3K27me3 measurement at the CHAC1 locus exists in any cell type, KMT2A-rearranged or otherwise; that is a bounded absence, not a claim.
shown for fusionMEASURED in KMT2A::AFF1 (RS4;11 strongly, SEM weakly) and KMT2A::MLLT3 (MOLM13, -1.89), with KMT2A wild-type REH as an internal negative control that moves the OTHER WAY (+0.49). Transcript only. No CHAC1 protein, no glutathione measurement, no ChIP and no functional rescue in any KMT2A-rearranged cell.
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-LARGE in transcript; BOUNDED-UNPUBLISHED in prose (0 fusion-scoped and 0 menin-drug-scoped PubMed records, both with valid querytranslation and 'No items found.'). Note the instructive self-contradiction: the -5.11 value comes from a published paper's own supplementary matrix, so the datum is public and only the sentence is missing.»
PDB IDS
[]
domains
[]
the measurement
{
  "rs411_array_mi22_lfc": -1.383,
  "rs411_array_dmso_log2_rma": 8.32,
  "sem_array_mi22_lfc": -0.376,
  "reh_array_mi22_lfc": 0.487,
  "reh_array_dmso_log2_rma": 8.95,
  "rs411_rnaseq_d7_vtp_lfc": -5.108,
  "rs411_rnaseq_d7_epz5676_lfc": -5.034,
  "rs411_rnaseq_d2_vtp_lfc": -1.446,
  "molm13_rnaseq_d7_vtp_lfc": -1.891,
  "rank": "largest single effect in the 24-gene two-compound concordant intersection",
  "menin_specificity": "SHARED WITH DOT1L INHIBITION (VTP -5.11, EPZ5676 -5.03, difference 0.09 log2)"
}
essentiality
{
  "depmap_release": "26.06 (dataVersion year 26 month 06, read back in session)",
  "depmap_screens": 1257,
  "chronos_mean": 0.0133,
  "pct_lines_below_minus_0_5": 0.4,
  "leukaemia_screens": 57,
  "leukaemia_mean_gene_effect": 0.0178,
  "reading": "NOT a dependency in any lineage. The largest effect in the signature is a readout, not a mechanism of killing."
}
identifier traps
{
  "registry_status": "NOT in data/symbol-collisions.json as of 2026-09-22 (157 rows read). New measurement.",
  "BOTCH": "24 PubMed records against CHAC1's 286, sharing 2 (Jaccard 0.0065). The BOTCH-only records are largely the English verb in clinical prose. Fusion-scoped BOTCH returns a valid 0.",
  "array": "GPL570 carries one CHAC1 probe and no probe whose own symbol field is BOTCH.",
  "uniprot": "clean - exactly one reviewed human record"
}
How this could bite us back
  • Losing the cell's glutathione-degrading enzyme 30-fold is predicted to RAISE reduced glutathione, which would make the cell harder to kill with anything working through oxidative or ferroptotic stress - so a menin inhibitor may be antagonising a partner agent on this axis while appearing to be doing something useful.
  • CHAC1 is pro-apoptotic by direct overexpression and knockdown test. Removing it removes a death signal, so the single largest gene-expression change in the whole response points the wrong way for the drug's purpose.
  • The paralogue CHAC2 is 10-20-fold less catalytically efficient, so it will not compensate - any therapeutic reasoning that treats the ChaC family as redundant is wrong at physiological turnover.
Still open
  1. Does intracellular reduced glutathione RISE in RS4;11 under VTP-50469, as a 30-fold CHAC1 loss predicts? A GSH/GSSG ratio on drug-treated cells is a one-week experiment and would decide whether a menin inhibitor is protecting the cell against oxidative and ferroptotic partner agents.
  2. Why does MI-2-2 move CHAC1 -1.38 in RS4;11 and only -0.38 in SEM, when both are t(4;11) KMT2A::AFF1? Either the two lines differ in a way that matters for this axis, or the RS4;11 value is partly an artefact of its n=2 DMSO arm.
  3. Is the day-7 effect repression or failure to induce? The DMSO baseline itself rises 3.07 log2 between day 2 and day 7, and separating those requires a CHAC1 time course with a matched-density control.
  4. Does re-expressing CHAC1 in drug-treated RS4;11 restore apoptosis, and does CHAC1 knockout blunt the menin inhibitor? Either result would move CHAC1 from readout to participant.
Every source behind this entry (5)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    CHAC1 falls 5.11 log2 under VTP-50469 (330 nM, day 7) and 1.38 log2 under MI-2-2 (10 uM, 6 d) in RS4;11, the largest effect in the 24-gene two-compound intersection; it rises 0.49 log2 in the KMT2A-wild-type line REH where its DMSO baseline is higher (8.95 vs 8.32); and EPZ5676 gives -5.03 in the same RS4;11 day-7 samples.

    What this evidence cannot show

    A count inside two named public cohorts. THE WEAK LINK IS SEM: MI-2-2 gives only -0.38 there against -1.38 in RS4;11, and the potent compound was never run in SEM, so the -5.11 rests on one cell line under one compound. The RS4;11 array DMSO arm is n=2 (submitter QC exclusion). Two platforms, two doses, two durations. The RNA-seq DMSO baseline rises from 3.28 at d2 to 6.35 at d7, i.e. CHAC1 is being induced in the DMSO arm over the culture, so part of the day-7 effect size is the control arm moving up rather than the treated arm moving down - the log2FC is correct but its interpretation is 'failure to induce' as much as 'repression'.

    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 is not a readout of the menin-KMT2A interface. It marks disruption of the fusion's transcriptional output by either route, and a menin-specific interpretation of it is unsupported. NCBI GEO, re-analysed here 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    CHAC1 is a pro-apoptotic component of the unfolded protein response activated by the ATF4-ATF3-CHOP cascade and specifically not by the parallel XBP1 or ATF6 pathways; CHAC1 overexpression enhances apoptosis and CHAC1 siRNA suppresses it, by TUNEL, PARP cleavage and AIF nuclear translocation.

    CHAC1/MGC4504 is a novel proapoptotic component of the unfolded protein response, downstream of the ATF4-ATF3-CHOP cascade. · Mungrue IN, Pagnon J, Kohannim O, Gargalovic PS, Lusis AJ · Journal of Immunology 182(1):466-76 2009
    PMID 19109178DOI
    What this evidence cannot show

    HUMAN AORTIC ENDOTHELIAL CELLS with chemically induced UPR - not leukaemia, not a fusion, not a menin inhibitor. The apoptosis results are from overexpression and knockdown, which bracket the physiological range rather than sitting inside it. This paper is already cited once in the graph (rule 82); what is new here is the gene-level entry and the KMT2A-context measurement.

    Evidence that argues against this (1)
    • CHAC1 is not a dependency in any DepMap lineage - mean gene effect +0.0133 across 1,257 screens, 0.4% of lines below -0.5. A gene whose complete CRISPR knockout has no fitness consequence in 1,257 cancer cell lines is a poor candidate for a pro-apoptotic effector whose loss matters, and the pro-apoptotic role may be specific to acute chemical UPR rather than to steady state. DepMap / Open Targets Platform 2026
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    In DepMap 26.06, CHAC1 has a mean CRISPR gene effect of +0.0133 across 1,257 screens with 0.4% of lines below -0.5 and a leukaemia mean of +0.0178 across 57 leukaemia screens.

    What this evidence cannot show

    Release recorded (dataVersion year 26 month 06, read back in session) per rule 83; CHAC1's screen count is 1,257 and is gene-specific, so no corpus denominator is implied. A positive mean gene effect means knockout is marginally growth-PROMOTING on average, which for a pro-apoptotic gene is directionally sensible but is within noise. Leukaemia screens identified by substring match on diseaseFromSource and their ids persisted.

    Evidence that argues against this (1)
    • DepMap screens are run in nutrient-rich medium in the absence of the stress that induces CHAC1 at all. A gene only expressed under ER or amino-acid stress cannot show a dependency in a screen that never applies that stress, so this null is about the assay's conditions as much as about the gene. Mungrue IN et al. 2009
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    PubMed returns zero records for CHAC1 jointly with any of eight KMT2A-fusion operands and zero for CHAC1 jointly with a thirteen-name menin-inhibitor list, both with non-empty querytranslations and warninglist 'No items found.'; the alias BOTCH reaches 24 records of which 2 also mention CHAC1, and fusion-scoped BOTCH also returns a valid zero.

    What this evidence cannot show

    Rule 3 satisfied: both zeros carry a non-empty querytranslation AND the warninglist 'No items found.' Rule 94 form throughout; rule 95 observed. Positive control in the same session: CHAC1[tiab] AND ATF4[tiab] = 66, so the instrument can return a one on this symbol. RULE 80: bounded by titles and abstracts; a CHAC1 row inside a supplementary differential-expression table - which is exactly where it would be - is invisible to this query, and indeed the -5.11 measured here comes from precisely such a table.

    Evidence that argues against this (1)
    • This absence is self-undermining in an instructive way: the -5.11 log2 value that makes CHAC1 interesting was extracted from the supplementary count matrix of a published menin-inhibitor paper. So the datum HAS been published; only the sentence naming CHAC1 has not. A bibliographic zero on a differentially expressed gene measures what authors chose to discuss, not what was measured. NCBI GEO, re-analysed here 2026
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    UniProt Q9BUX1 is the single reviewed human CHAC1 record, 222 aa, synonym BOTCH, cytosolic (ECO:0000269, PubMed 19109178) with a trans-Golgi pool annotated only by similarity to mouse Q8R3J5; it cleaves glutathione specifically and not other gamma-glutamyl peptides, and the paralogue CHAC2 degrades cytosolic glutathione with 10-20-fold lower catalytic efficiency at comparable Km.

    What this evidence cannot show

    Resolved live 2026-09-22, exactly one reviewed record (rule 62), accession re-fetched alone with name and length compared back (rule 1). RULE 91 CHECK: the cytosol call's ECO:0000269 citation IS the CHAC1 paper itself (PubMed 19109178), which is the good case - the code points at a paper genuinely about this protein. The trans-Golgi call is ECO:0000250 by similarity to mouse and is recorded as an uncoded inference, not a measurement. The CHAC2 kinetics are from PubMed 27913623, whose abstract was read; the enzyme assays were not inspected.

    Evidence that argues against this (1)
    • No CHAC1 localisation experiment has ever been performed in a haematopoietic cell, let alone a KMT2A-rearranged one, so 'cytosolic' here rests on one endothelial-cell paper. A glutathione-degrading enzyme whose compartment is uncertain has an uncertain substrate pool, and the trans-Golgi annotation - if real in human - would place part of the activity somewhere quite different. Kaur A et al. 2017
Connected to