STK3 (MST2) review notes

Automated deep research was unavailable for this review (no deep-research provider
keys), so no *-deep-research-*.md file exists. These notes are compiled by hand from
the cached publications in publications/, the UniProt record (Q13188) and the
cached GO-CAM index. Quotes are verbatim from the cached files.

Identity

Catalytic activity (core)

Regulation / complexes

Localization

Processes

GO-CAM

gocams/index.tsv contains STK3 (UniProtKB:Q13188) in 8 human models:
- 7 Hippo models (e.g. 65a1f4f800003391 "Hippo signaling core components (Human).",
65c57c3400001478 "Negative regulation of Hippo signaling by SLMAP-STRIPAK complex")
all as GO:0004674 protein serine/threonine kinase activity, part of GO:0035329
hippo signaling, occurs in GO:0005737 cytoplasm.
- 65039e8700001110 "TRIM69 activation of STK3 by ubiquitination leading to centrosome
separation" as GO:0004674, part of GO:0071539 protein localization to centrosome,
in GO:0005813 centrosome (NEK2 carries centrosome separation in that model).
This matches the core functions chosen here.

Premetazoan context (ancestral vs animal-specific)

Tetramerization row (InterPro2GO IPR036674)

Human STK3 carries the same IEA GO:0051262 protein tetramerization from IPR036674
(p53 tetramerisation superfamily) as coHpo. The structural evidence is that the MST2
SARAH domain forms homodimers and 1:1 heterodimers (PMID:23972470, PMID:25004971,
PMID:17517604). Unlike coHpo, human MST2 has one documented tetrameric assembly, the
2:2 SAV1-MST2 complex (PMID:29063833), but the dimer-of-dimers interface there is made by
SAV1 WW domains, not by STK3. So the InterPro2GO basis (a p53-like tetramerisation fold)
is still wrong; I marked it MARK_AS_OVER_ANNOTATED rather than REMOVE because a
STK3-containing tetramer does exist in human cells.

Protein binding rows

93 IPI protein binding rows. Policy: MODIFY where the cited paper supports a more
informative MF (SARAH heterodimerization with SAV1 / RASSF5 shown in focused studies;
self-interaction -> protein homodimerization activity), otherwise REMOVE as uninformative
(the interactions themselves are not disputed). No GO term for "Atg8-family protein
binding" was found in QuickGO, so the ATG8-family rows are removed rather than modified.