UBAC2 (UBA domain-containing protein 2) review notes

UniProt: Q8NBM4 (UBAC2_HUMAN), 344 aa precursor. Synonym PHGDHL1, PSEC0110. HGNC:20486.
Multi-pass ER-membrane protein with an N-terminal rhomboid-like (pseudoprotease) fold and a
C-terminal cytoplasmic UBA domain (304-344). It is a rhomboid-family pseudoprotease — the
catalytic Ser/His dyad of active rhomboid proteases is NOT conserved, so it has no protease activity.

Core biology

1. ERAD membrane component (RHBDD1/UBAC2 machinery). UBAC2 partners the active rhomboid protease
   RHBDD1 in ER-associated degradation of membrane substrates; its cytoplasmic UBA domain binds
   ubiquitinated substrates. (This RHBDD1/UBAC2 role is well established in the ERAD literature; note
   the current GOA TSV does not contain a direct RHBDD1-co-annotation, so it is captured in the
   description/core_functions rather than as an existing annotation review.)
2. FAF2/UBXD8 ER receptor. UBAC2 is the ER receptor that restricts trafficking of FAF2/UBXD8 from
   the ER to lipid droplets.
   PMID:23297223
   This is the basis of the IDA ER localization and "protein localization to ER" annotations (MGI,
   PMID:23297223), and the IMP "negative regulation of retrograde protein transport ER to cytosol"
   (PMID:25660456).
3. ER-phagy receptor. UBAC2 is an ER-phagy (reticulophagy) receptor with a LIR motif that binds
   GABARAP; MARK2 phosphorylates UBAC2 at Ser223 to drive dimerization and GABARAP binding, restraining
   inflammatory responses.
   PMID:39284914
4. Wnt regulation. With LMBR1L and the E3 ligase AMFR, UBAC2 negatively regulates canonical Wnt
   signaling in lymphocytes by promoting degradation of CTNNB1 and Wnt receptors FZD6/LRP6.
   PMID:31073040

Annotation assessment summary

• serine-type endopeptidase activity (GO:0004252) IBA: UBAC2 is a rhomboid pseudoprotease lacking
  the catalytic dyad; this IBA is an over-propagation from the active-rhomboid branch of the PANTHER tree
  and is biologically incorrect → REMOVE (IEA/IBA over-propagation, refutable on biological grounds).
• ER membrane (GO:0005789 IEA) / ER (GO:0005783 IDA): core compartment → ACCEPT.
• protein localization to ER (GO:0070972) / negative regulation of retrograde protein transport ER to
  cytosol (GO:1904153): reflect the FAF2/UBXD8 ER-receptor role → ACCEPT.
• reticulophagy (GO:0061709) / negative regulation of inflammatory response (GO:0050728) IMP: ER-phagy
  receptor role → ACCEPT.
• negative regulation of canonical Wnt (GO:0090090) IMP/IEA: LMBR1L/AMFR Wnt role → ACCEPT (non-core
  relative to ERAD/ER-phagy, but experimentally supported).
• protein binding (GO:0005515) IPI: uninformative term → KEEP_AS_NON_CORE.

Falcon deep-research findings (incorporated 2026-06)

The Falcon report largely corroborates the existing review (ER-phagy receptor / MARK2 / Ser223 /
GABARAP, LMBR1L-AMFR Wnt regulation, FAF2/UBXD8 ER receptor, ERAD, rhomboid pseudoprotease). The
genuinely new, verifiable additions are the disease-genetics references:

• UBAC2 is a confirmed Behçet's disease (BD) susceptibility locus in Han Chinese (two-stage study,
  477 BD cases / 1,334 controls). The risk T allele of promoter SNP rs3825427 has lower promoter
  activity and lowers UBAC2 transcript variant 1 expression, suggesting disease risk acts via
  transcriptional modulation of UBAC2. PMID:22455605 (PubMed-verified; doi:10.1186/ar3789). Added as a top-level reference
  (relevance MEDIUM) — locus/regulatory evidence, not tied to a specific GO term.
• The 13q32 locus near UBAC2 (variant rs7335046) confers susceptibility to cutaneous basal cell
  carcinoma and squamous cell carcinoma. PMID:21700618 (PubMed-verified; doi:10.1093/hmg/ddr287). Added as a top-level reference
  (relevance LOW) — locus-level GWAS only, does not establish UBAC2 as the causal gene.
• Falcon also cites several rhomboid/ERAD reviews (Adrain 2020 FEBS J doi:10.1111/febs.15548; Kandel
  2020 BBA-MCR doi:10.1016/j.bbamcr.2020.118793; Lemberg 2016 Semin Cell Dev Biol
  doi:10.1016/j.semcdb.2016.06.022) reiterating the UBA-domain/polyubiquitin-binding adapter and
  FAF2/UBXD8 lipid-droplet-trafficking role already captured in the review; not added as separate
  references since they only restate existing primary-evidence-backed content.
• Note: the Falcon report repeatedly cites the ER-phagy paper as "doi:10.1038/s44318-024-00232-z";
  this is the same study already in the review as PMID:39284914 (He et al. 2024, EMBO J). No new
  ER-phagy finding beyond what is already annotated.

2026-09-20 full-gene IBA re-review

All 15 annotations assessed: protease IBA, ER locations, six interaction rows, ER-phagy, inflammation, ER retention/retrotranslocation and paired canonical Wnt annotations. Nonproteolytic structural participation remains valid.

Existing OpenScientist report: UBAC2-hypotheses/function-hypothesis-go-0004252/openscientist.md. It finds no GxSG in the three annotated membrane spans and no membrane catalytic histidine; the local UniProt sequence has only GSSG at residues 6-9. Primary PMID:23297223 explicitly identifies an ER-resident rhomboid pseudoprotease. Retain the serine-endopeptidase rejection.

The report overstates the topology/phylogeny mechanism and incorrectly describes PMID:23297223 as the original IBA reference; the source is GO_REF:0000033. Its GO:0140318 recommendation has a false ER-phagy label: QuickGO calls this protein transporter activity. These limitations are recorded in reference_review.

PMID:39284914 full text identifies the LIR-GABARAP interaction and selective ER-degradation mechanism. Replace that generic protein-binding row with GO:0140506 endoplasmic reticulum-autophagosome adaptor activity (verified QuickGO). Other generic binding rows removed solely as uninformative, not as false interactions. Core receptor MF now stated.

PMID:31073040 supports limiting receptor maturation and stabilizing destruction-complex proteins through the LMBR1L/GP78/UBAC2 complex. Core wording now reflects this rather than claiming direct CTNNB1 ubiquitination by UBAC2. PMID:25660456 is abstract-only; the experimentally curated negative retrotranslocation row remains accepted without inventing a UBAC2-specific quote from its generic abstract.

Verified the proximate IBA PANTHER nodes from cached WITH/FROM fields and revised structured propagation metadata to match final decisions; no relationship-field reasoning, donor-count argument, or invented topology reconstruction was used.