SEC23A (human, Q15436) notes

2026-10-21 — Full re-review, independent of the ProtNLM2 predictions

Context: reviewed as the experimentally characterised reference for HORSE/SEC23A
(A0A3Q2H5Z6). ProtNLM2 output was not opened or used.

Deep research

just deep-research-falcon human SEC23A --fallback perplexity-lite failed: falcon
returned "Failed to authenticate" and the perplexity fallback returned HTTP 401
(insufficient quota). No deep-research file exists; the review is based on the cached
publications, the UniProt record and Reactome entries.

Evolutionary grounding

• UniProt Q15436 is in PANTHER PTHR11141:SF7 (PROTEIN TRANSPORT PROTEIN SEC23A) within
  PTHR11141 (SEC23). The paralog SEC23B is in a separate subfamily (PROTEIN TRANSPORT
  PROTEIN SEC23B).
• The four IBA rows (COPII vesicle coat, ER exit site, COPII-coated vesicle cargo
  loading, GTPase activator activity) come from PAINT node PTN000124617. Human Q15436 is
  itself a donor for the first three; the GTPase activator IBA rests on yeast Sec23
  donors only. The horse ortholog receives the same four IBA rows.
• PAINT did not propagate zinc ion binding, ER membrane, cytosol, or any binding term from
  the human experimental data; I took those omissions as deliberate and did not add them
  as inherited family functions.

What is experimentally established in human SEC23A

• Localisation to the transitional ER / ER exit sites and associated vesicles by
  immuno-EM; soluble cytosolic ~350 kDa complex; complementation of yeast sec23-1
  PMID:8898360
• CLSD F382L is a loss of function in cell-free budding PMID:16980979
  and the mutant poorly recruits SEC13-SEC31 PMID:17981132.
• Structures with SEC22B cargo at the Sec23/24 interface PMID:17499046 and PPP-motif
  binding by the gelsolin-like domain to TANGO1/cTAGE5 PRDs and SEC31 PMID:27551091.
• Preferential binding to SAR1B PMID:32358066.

Inferred, not shown directly in human (in the cached literature)

• SAR1 GTPase-activating activity: established for yeast Sec23p; human assignment is
  PAINT inference plus the conserved SAR1 interface. Accepted, flagged in a suggested
  question.
• Zinc: structure-derived UniProt binding sites (C61/C66/C85/C88), kept as non-core.

Decisions worth flagging

• 48 GO:0005515 rows: REMOVE per policy, except MODIFY to SNARE binding (SEC22B,
  PMID:17499046), MHC class I protein binding (PMID:20946353), protein kinase binding
  (CDK16, PMID:16091426) and proline-rich region binding (TANGO1/cTAGE5,
  PMID:27551091 and PMID:28442536).
• Endocytic recycling (ISO from mouse, GLUT4 trafficking): MARK_AS_OVER_ANNOTATED,
  because COPII acts at ER exit and the GLUT4 requirement reads as necessity, not
  participation.
• The zebrafish morpholino craniofacial phenotype in PMID:16980979 was not used as
  process evidence.
• No NEW terms. COPII vesicle coat assembly (GO:0048208) was considered, but budding and
  cargo loading already cover the process, so I did not add it.