STC2 / stanniocalcin-2 (O76061) - the second largest effect in the signature (-4.40 log2), a SECRETED glycoprotein, and its localisation rests on curator inference alone
also called STC2 · Stanniocalcin-2 · stanniocalcin 2 · STC2_HUMAN
Stanniocalcin-2 is a hormone-like protein the cell makes and sends out, and it is switched on by the stress programme. It shows the second biggest fall of any gene when KMT2A-rearranged leukaemia cells meet a menin-blocking drug, dropping about twenty-fold.
The same thing, for a scientist
STC2 (O76061, 302 aa) is a **SECRETED** glycoprotein with anti-hypocalcaemic action on calcium and phosphate homeostasis, and an established ATF4/ISR-responsive gene. It carries the **second largest effect in the 24-gene two-compound intersection: -4.40 log2 under VTP-50469 in RS4;11 at day 7 and -1.43 under MI-2-2** [E1]. Two features distinguish it from the rest of the panel. **First, it is the only signature member that leaves the cell** - every other panel gene acts intracellularly, so STC2 is the one whose loss could change the marrow microenvironment rather than only the leukaemic cell, and no experiment has looked at that in any fusion context. **Second, its expression pattern across the four lines is the most informative in the panel**: array DMSO baselines are SEM 11.94, RS4;11 8.62 and **REH 4.89 - at background** [E1]. So STC2 is highly expressed in both t(4;11) lines and essentially absent from the KMT2A-wild-type line, which is a cleaner fusion association than any other panel member shows - but it also means REH's flat response (-0.33) is a **non-measurement**, not a negative control, and this entry says so. Effect is **shared with DOT1L inhibition** (EPZ5676 -4.16) [E1]. **The localisation record is the weakest in the beat: UniProt's only location statement is Secreted with evidence code ECO:0000305 - curator inference, not experiment** [E3]. DepMap 26.06: +0.0593 across 1,258 screens, 0.0% below -0.5 [E2]. PubMed: **0** fusion-scoped, **0** menin-drug-scoped [E4].
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.
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.
The measurement, and why REH is not a control here
| dataset | cell | arm | DMSO | log2FC |
|---|---|---|---|---|
| array 203438_at | SEM | MI-2-2 | 11.94 | -1.07 (p 0.09) |
| array 203438_at | RS4;11 | MI-2-2 | 8.62 | -1.43 (p 0.005) |
| array 203438_at | REH | MI-2-2 | 4.89 (background) | -0.33 (non-measurement) |
| array 203439_s_at | RS4;11 | MI-2-2 | 4.95 (background) | -0.23 |
| RNA-seq | RS4;11 d7 | VTP-50469 | 5.02 | -4.40 |
| RNA-seq | RS4;11 d7 | EPZ5676 | 5.02 | -4.16 |
| RNA-seq | RS4;11 d2 | VTP-50469 | 2.77 | -2.77 (floor collision) |
| RNA-seq | MOLM13 d7 | VTP-50469 | 6.15 | -0.80 |
| RNA-seq | MOLM13 d2 | VTP-50469 | 4.55 | +0.76 |
Three things this table says that a single number would not.
- The expression pattern is the fusion association. SEM 11.94 and RS4;11 8.62 against REH 4.89 - a 7 log2 spread, with the KMT2A-wild-type line at background. That is a stronger statement about fusion dependence than any fold change in this panel.
- REH therefore cannot serve as the negative control for STC2. A gene at 4.89 log2 RMA is not measurably expressed, so its -0.33 is an absence of signal, not an absence of response. The prior entry's argument that the 24 genes are "expressed in the wild-type line, unresponsive in it" holds for TSC22D3, TRIB3, SESN2 and CHAC1, which it names - it does not hold for STC2, and this entry records that exception.
- The day-2 RS4;11 value is a floor collision. -2.77 with a DMSO baseline of exactly 2.77 means the treated arm is zero counts; the analysis flags it and it is not a measured fold change.
MOLM13 disagrees in sign between days (+0.76 at day 2, -0.80 at day 7), so the KMT2A::MLLT3 evidence for STC2 is weak.
The only secreted member of the signature
Every other gene in this beat's panel acts inside the cell. STC2 is secreted, which changes what its loss could mean 3:
- A 20-fold fall in a secreted factor is a change to the extracellular milieu, not only to the leukaemic cell's internal state. In a marrow niche that means stromal cells, osteoblasts and other haematopoietic cells stop receiving a signal they were receiving.
- STC2's annotated physiology is calcium and phosphate homeostasis, and menin inhibitors already have an endocrine signature in this graph -
endocrine/is an existing node folder and differentiation syndrome plus QTc are recorded adverse events for this drug class. A drug-induced 20-fold fall in a calcium-regulating secreted factor in a bone-marrow-resident tumour is a mechanistically plausible route to a calcium or phosphate disturbance, and nothing in the clinical record has been checked against it here. That is a specific, cheap, unperformed audit - a FAERS or trial-safety query on hypocalcaemia and hypophosphataemia under revumenib and ziftomenib - and it is recorded as an open question rather than a claim. - The caveat that matters: STC2's tumour-cell contribution to circulating stanniocalcin-2 in a patient is unknown and likely small relative to normal tissue, so the systemic argument is weak. The LOCAL marrow argument is the defensible one.
And the localisation record cannot support even that. UniProt's sole location statement for STC2 is Secreted with evidence code ECO:0000305 - curator inference from sequence and homology, not an experimental observation 3. Per rule 91's clause 4, ECO:0000305 is a different standing from ECO:0000269, and for the one panel member whose compartment carries the argument, the compartment is the least well evidenced.
| uniprot | O76061 |
|---|---|
| length AA | 302 |
| entry name | STC2_HUMAN |
| protein name | Stanniocalcin-2 |
| ensembl gene ID | ENSG00000113739 |
| PDB note | UniProt O76061 cross-references four PDB entries and NONE was fetched. No chain, resolution or residue range is claimed from them. |
| alphafold ID | AF-O76061-F1 |
| domains note | UniProt O76061's feature table carries NO Domain features. No stanniocalcin fold boundary is annotated and none is invented. Lane deliberately EMPTY per the integer-residue rule. |
| accession resolution | gene_exact:STC2 AND organism_id:9606 AND reviewed:true returns EXACTLY ONE reviewed row, O76061, STC2_HUMAN, 302 aa, no synonyms; re-fetched alone with name and length compared back: PASS. |
| localization | SECRETED - AND THE EVIDENCE CODE IS ECO:0000305, CURATOR INFERENCE, NOT AN EXPERIMENT. UniProt O76061's location block contains exactly one statement, 'Secreted', with no ECO:0000269 anywhere. Per rule 91 clause 4 this is a materially weaker standing than an experimental code, and STC2 is the one panel member whose compartment carries an argument (a secreted factor changes the microenvironment, an intracellular one does not). No STC2 localisation experiment exists in any haematopoietic cell. Treat 'secreted' as a signal-peptide prediction. |
| subnuclear localization | NOT APPLICABLE AND NOT ASKABLE. STC2 has no annotated nuclear pool, coded or uncoded, and is a predicted secreted protein. There is no subnuclear question and none is invented. Recorded as a bounded absence resting on a prior absence. |
| cell type | SEM (11.94) and RS4;11 (8.62) - both t(4;11) KMT2A::AFF1 B-ALL and both expressing it highly - against REH (4.89, KMT2A wild type, AT BACKGROUND), plus MOLM13 (KMT2A::MLLT3 AML, 6.15 at d7 but sign-discordant between days) [E1]. UniProt's tissue list includes peripheral blood leukocytes. No primary leukaemia sample and no normal haematopoietic cell has been assayed. |
| cell state | Measured in the day-6/day-7 drug-arrested state. STC2's RNA-seq DMSO baseline rises 2.25 log2 from day 2 to day 7 in RS4;11, the largest control-arm drift of any panel member, so the day-7 effect size is substantially a failure-to-induce rather than a repression. As a calcium/phosphate-homeostasis factor its physiological input is the extracellular ion state, which neither experiment measured or controlled, and as an ER-stress-responsive secreted protein its level also depends on secretory load. |
| epigenetic readout | No writer, eraser or reader activity. Its fall is reproduced by the DOT1L inhibitor EPZ5676 to within 0.24 log2, so its expression tracks H3K79-methylation-dependent fusion output rather than the menin interface. NOTE the expression pattern is the epigenetically interesting part: 7 log2 higher in both t(4;11) lines than in the KMT2A-wild-type line, which is what a locus under direct fusion-complex control would look like - but no H3K79me2, H3K4me3 or KMT2A/menin ChIP at the STC2 locus exists in any cell. |
| shown for fusion | MEASURED in KMT2A::AFF1 (RS4;11 -1.43 array / -4.40 RNA-seq; SEM -1.07, not significant) and weakly and inconsistently in KMT2A::MLLT3 (MOLM13 +0.76 at d2, -0.80 at d7). THE WILD-TYPE COMPARISON IS AN EXPRESSION DIFFERENCE, NOT A RESPONSE DIFFERENCE - REH does not express STC2 measurably. Transcript only; no STC2 protein, no secretion assay, no ChIP and no functional test in any fusion context. |
| absence class | literature-absence-bounded |
| absence class note | RECLASSIFIED 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, with the wild-type comparison reclassified as an EXPRESSION difference rather than a response difference; BOUNDED-UNPUBLISHED in prose (0 fusion-scoped, 0 menin-drug-scoped, valid querytranslations and 'No items found.'), and this is the lowest-alias-exposure zero in the beat.» |
PDB IDS
[ "7Y5Q", "8A7D", "8A7E", "8HGH" ]
domains
[]
the measurement
{
"sem_array_mi22_lfc": -1.07,
"sem_array_dmso_log2_rma": 11.94,
"rs411_array_mi22_lfc": -1.428,
"rs411_array_dmso_log2_rma": 8.62,
"reh_array_mi22_lfc": -0.33,
"reh_array_dmso_log2_rma": 4.89,
"rs411_rnaseq_d7_vtp_lfc": -4.4,
"rs411_rnaseq_d7_epz5676_lfc": -4.163,
"rs411_rnaseq_d2_vtp_lfc": "-2.77, FLOOR COLLISION (DMSO 2.77, treated arm zero counts) - not a measured fold change",
"molm13_rnaseq_d2_vtp_lfc": 0.757,
"molm13_rnaseq_d7_vtp_lfc": -0.8,
"rank": "second largest effect in the 24-gene two-compound concordant intersection",
"menin_specificity": "SHARED WITH DOT1L INHIBITION (VTP -4.40, EPZ5676 -4.16)",
"the_expression_pattern_is_the_finding": "SEM 11.94, RS4;11 8.62, REH 4.89 log2 RMA - a 7 log2 spread with the KMT2A-wild-type line at background. Stronger evidence of fusion association than any fold change here.",
"correction_to_a_graph_argument": "REH is NOT a usable negative control for STC2: at 4.89 log2 RMA the gene is not measurably expressed, so its flat response is an absence of signal. The prior entry's 'expressed in the wild-type line, unresponsive in it' holds for TSC22D3, TRIB3, SESN2 and CHAC1 and NOT for STC2."
}essentiality
{
"depmap_release": "26.06 (read back from meta{dataVersion})",
"depmap_screens": 1258,
"chronos_mean": 0.0593,
"pct_lines_below_minus_0_5": 0,
"leukaemia_screens": 57,
"leukaemia_mean_gene_effect": 0.0399,
"reading": "not a dependency - BUT a pooled CRISPR screen in shared medium is the worst possible assay for a SECRETED factor, because knockout cells are rescued by their neighbours' secretion. This null is close to uninformative for this gene."
}identifier traps
{
"registry_status": "NOT in data/symbol-collisions.json as of 2026-09-22 (157 rows read) - and it does not need a row.",
"lowest_exposure_in_the_panel": "'stanniocalcin 2' and 'stanniocalcin-2' return IDENTICAL 223-record id sets sharing 188 with STC2[tiab]'s 456 - Jaccard 0.3829, the HIGHEST overlap measured in this beat. The symbol and the spelled-out name are used together, so STC2's bounded absence has almost no name-list exposure and is the most trustworthy zero in the set.",
"array": "GPL570 carries two STC2 probes; the second (203439_s_at) is at background in every line (4.14-8.40) and is reported as background, not as a null."
}- STC2 is the only secreted member of this signature, so a menin inhibitor is changing what the leukaemic cell tells its niche - and the niche's response is entirely unmeasured. A stromal compartment that stops receiving a signal may become more supportive rather than less.
- STC2 regulates calcium and phosphate. A 20-fold fall in a marrow-resident tumour is a mechanistically plausible route to an electrolyte disturbance on a drug class that already carries differentiation syndrome and QTc as recorded adverse events, and no one has checked the pharmacovigilance record against it.
- The DepMap null on this gene is an assay artefact waiting to be quoted: pooled CRISPR screens cannot detect secreted-factor dependencies because neighbours rescue each other through the medium. Anyone citing STC2's flat gene effect as evidence it does not matter would be citing the screen's design.
- Has hypocalcaemia or hypophosphataemia been reported under revumenib or ziftomenib? STC2 is an anti-hypocalcaemic secreted factor that falls 20-fold under a menin inhibitor in a marrow-resident tumour, and a FAERS or trial-safety query on those two terms is a one-day audit that nobody has run.
- Does STC2 protein actually leave a KMT2A-rearranged cell? The secretion call is curator inference (ECO:0000305) with no experimental code, and every microenvironmental argument about this gene depends on it.
- Why is STC2 expressed 7 log2 higher in both t(4;11) lines than in the KMT2A-wild-type line? A KMT2A, menin or H3K79me2 ChIP at the STC2 locus would say whether this is a direct fusion target, and it is the panel member where that question is most worth asking.
- Is the day-7 effect repression or failure to induce? The DMSO baseline drifts up 2.25 log2 between day 2 and day 7, the largest control-arm drift in the panel.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset
STC2 falls 4.40 log2 under VTP-50469 in RS4;11 at day 7 (EPZ5676 -4.16) and 1.43 under MI-2-2 (p 0.005), the second largest effect in the 24-gene intersection; its array DMSO baselines are SEM 11.94, RS4;11 8.62 and REH 4.89, i.e. the KMT2A-wild-type line expresses it at background; the RS4;11 day-2 value of -2.77 is a floor collision with a treated arm of zero counts; and MOLM13 disagrees in sign between days (+0.76 at d2, -0.80 at d7).
What this evidence cannot show
THE KEY CAVEAT, AND IT CORRECTS AN ARGUMENT ELSEWHERE IN THE GRAPH: REH is NOT a usable negative control for STC2, because at 4.89 log2 RMA the gene is not measurably expressed there - its -0.33 is an absence of signal, not an absence of response. The prior entry's claim that all 24 concordant genes are 'expressed in the wild-type line, unresponsive in it' is supported by the four genes it names and does NOT hold for STC2. The day-2 RS4;11 figure is a floor collision and is flagged as such in the analysis output. SEM's -1.07 is not significant (p 0.09, n=2 arms). RS4;11 array DMSO arm n=2.
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
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
In DepMap 26.06, STC2 has a mean CRISPR gene effect of +0.0593 across 1,258 screens with 0.0% of lines below -0.5 and a leukaemia mean of +0.0399 across 57 leukaemia screens.
DepMap CRISPR (Chronos) gene effect via the Open Targets Platform GraphQL depMapEssentiality field; dataVersion year 26 month 06, read back from meta{dataVersion} in the same session · DepMap / Open Targets Platform 2026What 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 CRISPR screen in monoculture is the WORST possible assay for a secreted factor: autocrine effects are diluted into the medium and paracrine effects are absent by construction in a pooled screen where neighbouring cells carry different guides. This null is close to uninformative for this particular gene.
Evidence that argues against this (1)
- Pooled CRISPR screens systematically miss secreted-factor dependencies because the medium is shared - a knockout cell is rescued by its neighbours' secretion. STC2's flat gene effect is therefore expected regardless of its biology, and quoting it as evidence that STC2 does not matter would be an assay-artefact argument. UniProt Knowledgebase 2026
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
UniProt O76061 is the single reviewed human STC2 record, 302 aa with no listed synonyms, annotated as having an anti-hypocalcaemic action on calcium and phosphate homeostasis, expressed in muscle, heart, pancreas, kidney, spleen, prostate, small intestine, colon and peripheral blood leukocytes, with four PDB entries - and its ONLY subcellular-location statement is 'Secreted' carrying evidence code ECO:0000305, curator inference, with no experimental evidence code anywhere in the location block.
UniProtKB - O76061, STC2_HUMAN, stanniocalcin-2 · UniProt Knowledgebase 2026What 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. RULE 91 CLAUSE 4 APPLIES IN FULL: the location is ECO:0000305 - a curator's inference, not an experiment - which is the weakest localisation standing in this beat's whole panel, and it is the one panel member whose compartment carries an argument. The four PDB entries were NOT fetched and nothing is claimed from them. UniProt's tissue list includes peripheral blood leukocytes, which is the only haematopoietic statement available and is an expression statement, not a localisation one.
Evidence that argues against this (1)
- A secretion call by curator inference is a sequence-signal-peptide argument, and signal peptides are not always used - retained, mislocalised or intracellularly acting forms of nominally secreted proteins are common. For STC2 in a leukaemia cell, where no experiment has been done, 'secreted' should be treated as a prediction rather than an observation, which weakens every microenvironmental argument built on it including this entry's. UniProt Knowledgebase 2026
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
PubMed returns zero records for STC2 jointly with an eight-operand KMT2A-fusion list and zero for STC2 with a thirteen-name menin-inhibitor list, both with non-empty querytranslations and warninglist 'No items found.'; the spelled-out forms 'stanniocalcin 2' and 'stanniocalcin-2' return identical 223-record id sets of which 188 also mention STC2 (Jaccard 0.3829, the highest alias overlap in this beat's panel), and the fusion-scoped spelled-out query also returns a valid zero.
NCBI PubMed esearch, run 2026-09-22, with querytranslation and warninglist read back from the response for every query quoted · NCBI PubMed 2026What this evidence cannot show
Rule 3 satisfied on all zeros. The 0.38 Jaccard is the HIGHEST alias overlap measured in this beat - unlike CHOP, PERK or SGK, the spelled-out stanniocalcin name and the symbol are largely used together, so this symbol has almost no name-list exposure and its zero is correspondingly trustworthy. Rule 94 form; rule 95 observed. Rule 80: titles and abstracts only, and for a gene whose entire presence in this story is a row in a supplementary count matrix, that limit is doing real work.
Evidence that argues against this (1)
- STC2 has a substantial cancer literature as a hypoxia- and ER-stress-induced factor in solid tumours, so the zero is specifically about the KMT2A field and not about STC2's relevance to cancer biology. A hypoxia-responsive secreted factor in a marrow-resident tumour is a priori interesting, and the absence reflects that nobody in this field has looked rather than that looking failed. NCBI PubMed 2026
STC2 is an ATF4/ISR-responsive gene and the second largest effect in the signature (-4.40 log2) while ATF4's own transcript is flat
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
expressed 7 log2 higher in both t(4;11) lines (SEM 11.94, RS4;11 8.62) than in KMT2A-wild-type REH (4.89, at background) and falls 4.40 log2 under VTP-50469
- ← regulatesPinometostat (EPZ-5676) — the DOT1L inhibitor, and the most instructive failure in this field
EPZ5676 -4.16 against VTP-50469 -4.40 in the same RS4;11 day-7 samples
UNCHECKED SAFETY LEAD: a 20-fold fall in an anti-hypocalcaemic secreted factor in a marrow-resident tumour; no pharmacovigilance query on hypocalcaemia or hypophosphataemia under a menin inhibitor has been run
REH is not a usable negative control for STC2 - its REH baseline is 4.89 log2 RMA, at background, so its flat response is an absence of signal