AKAP12 bioinformatics — results

The PKC-binding motifs coincide with two of the three CaM-binding motifs

akap12_motifs.py searched human Q02952 (1782 aa) with the PKC-binding consensus
published for rodent SSeCKS (PMID:21903576), without assuming the rodent
coordinates transfer. Two matches, and both fall inside a UniProt-annotated AKAP
CaM-binding (WSK) motif:

So the human protein has three WSK motifs, of which two also match the
PKC-binding consensus, and the third (801-821) does not.

This is a structural rationale for an observation the review records as
rodent-only: PKC phosphorylation antagonises calmodulin binding
(PMID:11820772). If the PKC-binding determinant physically overlaps the
CaM-binding motif, the two ligands compete for the same segment, and
phosphorylation within it would be expected to disfavour CaM. The overlap is a
sequence fact about the human protein; the antagonism itself has still only been
measured in rodent, and the review hedges it accordingly.

Caveats. These are regex matches to a published consensus, not binding
measurements — they establish that the motif is present and where, not that it is
used. The UniProt motif annotations are PROSITE-ProRule (PRU01241) inferences,
not experimental. Nothing here demonstrates human PKC binding or human CaM
binding.

Open discrepancy: three human motifs vs four rodent sites

PMID:11820772 reports four 1-5-10 CaM-binding sites in rodent SSeCKS;
human UniProt annotates three WSK motifs, and this search finds three WSK
cores. Whether the fourth rodent site is genuinely absent from human or merely
unannotated is unresolved, and is recorded as a knowledge_gaps entry on the
calmodulin core_functions entry rather than silently reconciled.

Document integrity

akap12_audit.py passes: 42 entries (38 matching GOA rows 1:1, plus 4 NEW),
actions ACCEPT 11 / KEEP_AS_NON_CORE 17 / MODIFY 3 / MARK_AS_OVER_ANNOTATED 7 /
NEW 4; 53 reference_id occurrences parse to 53; 13 propagation_review blocks
checked; the retracted PMID:27683220 retains is_invalid: true and is cited by
no supported_by.

akap12_audit_selftest.py passes: eight mutations each trip their named guard with the
expected message, and two controls leave the audit silent, so a guard that fires on
everything would be caught too. Mutants are audited in a tempfile — the curated review file
is never written, which the script asserts at the end.