IRE1 review notes

2026-09-30 -- apoptosis IBA and broad-MF cleanup for APOPTOSIS

The only death-related yeast IRE1 annotation is the GO_Central IBA to
GO:0070059 intrinsic apoptotic signaling pathway in response to endoplasmic reticulum stress. I left this as MARK_AS_OVER_ANNOTATED and tightened the
propagation review: the current local PANTHER cache places conserved ER
localization, RNA endonuclease activity, Ser/Thr kinase activity, and
IRE1-mediated UPR on the pan-eukaryotic IRE1-family node PTN000359335, but the
ER-stress apoptosis term now appears on PTN000359344 with mouse ERN seeds. Yeast
IRE1 is an adaptive IRE1-HAC1 UPR sensor/kinase/RNase, not a mediator of the
metazoan IRE1/ERN apoptosis branch.

Also cleaned the whole review before marking it complete: broad direct molecular
function ancestors such as nucleotide binding, catalytic activity, kinase
activity, transferase activity, RNA nuclease activity, hydrolase activity, and
metal ion binding now MODIFY to ATP binding, protein Ser/Thr kinase activity,
RNA endonuclease activity, or magnesium ion binding as appropriate. The generic
IntAct protein binding row for DCR2 was changed to REMOVE; the DCR2-Ire1
interaction is real PMID:16990850, but the specific phosphatase activity belongs to Dcr2 and
GO:0005515 adds no useful IRE1-side activity.

2026-09-02 Update: nuclear-localization annotation (GO:0005634, IDA, PMID:17035634)

Audited the existing review for oversights. The IDA annotation of GO:0005634 (nucleus)
from PMID:17035634 had been marked UNDECIDED with the stated reason "Unable to access
PMID:17035634 to verify the nuclear localization claim." This was incorrect: the
publication is cached in this repository (publications/PMID_17035634.md) and its
abstract directly and unambiguously supports the annotation.

Goffin et al. 2006 (Mol Biol Cell) show that Ire1p's cytoplasmic linker region contains
an 18-residue nuclear localization sequence (NLS) recognized by both importin alpha
(Kap60p) and multiple importin beta homologues, that this NLS drives Ran-GTPase-dependent
nuclear import of Ire1p (or an NLS-GFP reporter) in vivo, and that NLS-disrupting point
mutations impair ER-stress-induced HAC1 mRNA splicing:

PMID:17035634

PMID:17035634

PMID:17035634

Changed the annotation's action from UNDECIDED to KEEP_AS_NON_CORE: the nuclear
pool and NLS-dependent import are real and evidence-backed (so UNDECIDED, reserved for
cases where the evidence genuinely cannot be assessed, was not appropriate), but IRE1's
best-established, defining catalytic activities (kinase trans-autophosphorylation and
HAC1 pre-mRNA endoribonuclease splicing) are ER-membrane events, so nuclear import is
kept as a non-core regulatory/trafficking aspect rather than promoted into
core_functions.

All other existing annotations, core_functions, and the top-level description were
reviewed and found sound and well-supported; no other changes made.

2026-09-04 — SFT review consistency after the nuclear-localization re-review

The SFT prediction review still classified GO:0005634 (nucleus) as NPI and asserted
that IRE1 is never found in the nucleus after the main gene review retained the IDA
annotation from PMID:17035634 as KEEP_AS_NON_CORE. The benchmark policy treats an
exact term retained with a positive AIGR action as CNN (correct but not novel), even
when the function or location is non-core. The standard SFT audit identified this
single deterministic category conflict across its 95-gene cohort.

The categorical exclusion is also too strong for the cached abstract, which reports
NLS-dependent localization and mutational effects on signaling PMID:17035634. This consistency
repair follows the existing annotation review; it does not infer that nuclear
localization is IRE1's principal location or resolve the mechanism of trafficking of
the intact membrane protein. The reference cache remains abstract-only.

2026-09-30 IBA rereview

Rechecked the six IRE1 IBA rows against the current cached PTHR13954 PAINT export: