EFR3A: paired evidence notes

Human reference: https://www.uniprot.org/uniprotkb/Q14156/entry

Primary source: PMID:23229899 PtdIns4P synthesis by PI4KIIIα at the plasma membrane and its impact on plasma membrane identity..

Accordingly, a bioorthogonal metabolic labeling approach (Hang et al., 2011) revealed that FLAG-tagged EFR3A and EFR3B were palmitoylated at the N-terminal Cys-rich motif (Fig. 4 B). Importantly, palmitoylation-deficient mutants (C6S/C7S/C8S/C9S for EFR3A and C5S/C7S/C8S for EFR3B) were localized in the cytosol (Fig. 4 C), and the EFR3B palmitoylation-deficient mutant was unable to recruit tagged TTC7B and M1-PI4KIIIα to the plasma membrane (Fig. S2).

The horse sequence comparison is in EFR3A-bioinformatics/RESULTS.md. Transfer is assessed for this exact accession; human biochemical evidence is not equine experimental validation.

2026-10-20 — Full re-review, independent of the ProtNLM2 predictions

• Deep research: just deep-research-falcon HORSE EFR3A --fallback perplexity-lite failed. Edison returned "Failed to authenticate" and the perplexity-lite fallback also failed, so there is no deep-research file. I worked from primary literature cached in publications/: PMID:23229899 (full text), PMID:24360784 (full text, newly cached), PMID:26571211 (full text, newly cached) and PMID:25380825 (abstract only).
• Identity: the benchmark accession is an alternative-first-exon product, not the canonical product. New analysis EFR3A-bioinformatics/compare_horse_models_rereview.py produced RESULTS-rereview.md and horse_models_rereview.json. All four horse EFR3A UniProt entries come from one Ensembl gene, ENSECAG00000014883 [file:HORSE/EFR3A/EFR3A-bioinformatics/RESULTS-rereview.md "All horse EFR3A models derive from Ensembl gene(s): ENSECAG00000014883.4."].
• A0A9L0S4L8 (796 aa, transcript ENSECAT00000116354) starts from a first exon at chr9:74894800-74895054, which lies between canonical exons 2 and 3. Its N-terminus is MIDKDGLVKA. Human residues 2-26 are unpaired, and nothing aligns to the Cys6-Cys9 palmitoylation motif.
• A0A3Q2GU27 (821 aa, ENSECAT00000044066) is the full-length horse product: MPTRVCCCCS..., 98.54% identical to human Q14156 over all 821 residues. It also carries the IBA rows (GO:0005886, GO:0072659) that the benchmark accession lacks.
• F6UA29 (848 aa) has an upstream extension. A0A3Q2KQA4 (794 aa) has an internal deletion (human residues 259-285).
• This is not human isoform 2 (Q14156-2, which lacks residues 1-36). The horse short form keeps about 10 more N-terminal residues and has its own start exon.
• The mouse EFR3A basic-patch residues (K50/R53/R61/R68/R70, PMID:24360784) are all retained, at horse K25/R28/R36/R43/R45 in this model.
• Evolutionary grounding: PANTHER PTHR12444:SF1 (EFR3 homolog A) is the same subfamily as human Q14156 and mouse Q8BG67. EFR3B is SF4; fly/yeast-type Efr3 are in other subfamilies. The family review interpro/panther/PTHR12444/PTHR12444-review.yaml also places the horse accession in SF1, so there is no disagreement. That review lists A0A9L0S4L8 as an SF1 representative without noting that it is a truncated-N-terminus form; the full-length A0A3Q2GU27 would be a better representative. No PAINT tsv is cached for PTHR12444. Human IBA rows come from PTN000278197.
• Annotation: the only GOA row is GO:0005886 plasma membrane (IEA, ARBA + SubCell). I kept it UNDECIDED, now with a full rationale. Mammalian palmitoylation-deficient EFR3A (C6S-C9S) is cytosolic PMID:23229899. Palmitoylation is required for mammalian membrane association PMID:24360784. The basic patch is also required, but nobody has shown it is sufficient. So the term is right for the gene but unverified for this product form. No NEW terms. I did not propose protein localization to plasma membrane (GO:0072659) for this form, because the EFR3B palmitoylation mutant cannot recruit TTC7B/PI4KA. The AT1R/GPCR desensitization phenotype (PMID:25380825) is a depletion phenotype, not participation.
• Description rewritten as a project-independent summary. I added a core function with no GO term (complex scaffold, membrane recruitment not established for this form) and one BIOLOGY knowledge gap.
• Relative to review next: the canonical product of the same gene, horse A0A3Q2GU27 (same_gene_canonical). The repo has no review of it; its folder would be genes/HORSE/EFR3A__A0A3Q2GU27. The human ortholog review genes/human/EFR3A exists but is status: IN_PROGRESS.
• Modules: modules/phosphoinositide_biosynthesis.yaml models the PM PI4KA (Stt4) step with an "Efr3-Ypp1 scaffold". No GO-CAM in gocams/index.tsv includes EFR3A; the PI4KA model 68fac5ed00003196 is relevant context.