AIG1 is a polytopic integral membrane threonine hydrolase that degrades fatty acid esters of hydroxy fatty acids (FAHFAs), a class of endogenous bioactive lipids in which a fatty acid is esterified to a hydroxyl group on a second fatty-acid backbone. It hydrolyses the internal ester bond to release a free fatty acid and a hydroxy fatty acid, preferring isomers whose branch point lies distal from the carboxylate head group, and it does not act on the common phospholipid or neutral lipid classes. Catalysis uses an unusual Thr/His dyad rather than a serine hydrolase fold: Thr43 is the nucleophile and His134 the general base, and both residues are predicted to sit within transmembrane helices, so the chemistry is thought to occur inside the bilayer where the FAHFA ester bond is buried. AIG1 accounts for the majority of FAHFA hydrolysis in human prostate carcinoma cells and primary T cells, and together with its only paralogue ADTRP it constitutes the FAHFA-hydrolase arm of bioactive lipid turnover in mammals. Because FAHFAs are insulin-sensitising, degrading them is metabolically consequential: in mouse adipocytes AIG1 acts downstream of IRF3 to blunt insulin signalling in a manner requiring its catalytic residues. Separately from its lipid chemistry, AIG1 binds the RING E3 ubiquitin ligase RCHY1/Pirh2 and promotes Pirh2-dependent ubiquitination of p53; whether that second activity needs the hydrolase site is untested. The gene symbol records its discovery as a dihydrotestosterone-inducible transcript in hair-follicle dermal papilla cells and says nothing about this activity.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005515 protein binding | IPI PMID:21988832 Toward an understanding of the protein interaction network o... | MODIFY | Summary: A real, multiply-assayed partner rather than screen noise, so the uninformative parent term can be replaced. Q96PM5 is RCHY1/Pirh2, a RING-type E3 ubiquitin transferase (EC 2.3.2.27), which makes GO:0031625 ubiquitin protein ligase binding the directly licensed child. Reason: Partner resolution: UniProtKB:Q96PM5 is reviewed Swiss-Prot ZN363_HUMAN, 261 aa, 'RING finger and CHY zinc finger domain-containing protein 1', EC 2.3.2.27 - the canonical entry at canonical length, not a TrEMBL ORFeome fragment. Because the partner is itself a ubiquitin-protein ligase, GO:0031625 'Binding to a ubiquitin protein ligase enzyme, any of the E3 proteins' is entailed by the interaction rather than inferred from a downstream model. Evidence strength was checked by expanding IntAct rather than trusting UniProt's NbExp=4, because a single screen can be logged as several sub-methods: the four IntAct records for this pair (experiment EBI-3895963, PMID:21988832) are four different assay types - '2 hybrid', 'pull down', 'anti tag coip' and 'confocal microscopy' - so the count survives expansion. For contrast, 141 of AIG1's 161 IntAct interactions come from one publication logged as 'two hybrid array' + 'two hybrid prey pooling approach' + 'validated two hybrid' (47 each), and none of those reached GOA. Independent corroboration comes from a 2025 study that co-immunoprecipitated endogenous AIG1 and Pirh2 from cardiomyocytes and confirmed it by proximity ligation. Two provenance caveats are recorded rather than glossed. First, the two yeast two-hybrid papers (PMID:21988832 and PMID:21622095) share an author and are both Fudan University work, so they are not two independent screens; PMID:40303337 is a different department 14 years later. Second, PMID:40303337 maps the interaction to AIG1 residues 35-93 by deletion, a span that removes TOPO_DOM 31..44, the whole of TRANSMEM 45..67 and TOPO_DOM 68..87, so it cannot distinguish a contact surface from loss of folding or membrane insertion; no binding-site claim is made here. Proposed replacements: ubiquitin protein ligase binding Supporting Evidence: PMID:40303337 Co-IP results similarly indicated that endogenous AIG1 strongly interacted with Pirh2 in cardiomyocytes under steady-state conditions PMID:27018888 later found to interact with the E3 ligase Pirh2 by yeast two-hybrid screening |
| GO:0005886 plasma membrane | EXP PMID:27018888 AIG1 and ADTRP are atypical integral membrane hydrolases tha... | MARK AS OVER ANNOTATED | Summary: The cited reference contains no localisation experiment of any kind. What it shows is recovery of activity in a crude 100,000 g membrane pellet, which supports GO:0016020 membrane - separately annotated from the same paper - but not a plasma-membrane compartment call. The only study that did examine AIG1's localisation places it in the ER membrane. Reason: A phrase scan of the cached full text (full_text_available true) returns zero occurrences of 'plasma membrane', 'cell surface', 'immunofluoresc', 'confocal', 'subcellular', 'localization' and 'localisation', against positive controls in the same file of 'membrane fraction' 3, 'membrane lysates' 9, 'transmembrane' 29, 'HEK293T' 36 and 'FAHFA' 67, so the zeros are real absences rather than a broken scan. The only sentence in the paper that resembles a compartment claim is explicitly speculative about where the chemistry might occur, and the paper's topology statement is six prediction programs rather than an experiment. This is the same reasoning the merged ADTRP review applied to its byte-identical row, and the scan reproduces that review's counts exactly; the two genes then diverge on the conclusion, and the divergence is the point. ADTRP's plasma-membrane call stands independently on PMID:21868574, which imaged ADTRP with TFPI and caveolin-1 in endothelial lipid rafts, so that review could ACCEPT the term with only an attribution caveat. AIG1 has no equivalent paper: its UniProt SUBCELLULAR LOCATION line reads 'Cell membrane {ECO:0000269|PubMed:27018888}' and cites nothing else, so the entire plasma-membrane claim on this gene rests on a reference that does not make it. Meanwhile PMID:27040980 used a protease-protection assay to place AIG1 in the ER membrane, with an ER Ca2+ functional readout. Not REMOVE: AIG1 is unambiguously a membrane protein and the term is unsupported rather than refuted, the cached text excludes supplementary figures, and the ER paper is a single study whose full text is not available. Marked over-annotated, with the compartment conflict recorded as a knowledge gap and a suggested experiment. Supporting Evidence: PMID:27018888 Both KC01 and JJH260, but not THL or ABC34 also inhibited the FAHFA hydrolase activity of LNCaP cell lysates, which was mostly found in the membrane fraction PMID:27018888 That these residues are generally predicted to be embedded within transmembrane domains of AIG1 and ADTRP indicates these enzymes could have evolved to perform hydrolytic chemistry within the cell membrane environment PMID:27040980 Analyzing the topology of AIG1 in the ER membrane using a protease-protection assay suggested |
| GO:0005886 plasma membrane | IEA GO_REF:0000044 | MARK AS OVER ANNOTATED | Summary: Mechanically correct mapping of SL-0039 'Cell membrane', but the upstream assertion it maps is the same unsupported one as the EXP row: UniProt's SUBCELLULAR LOCATION line for this protein cites PubMed:27018888 alone, and that paper performed no localisation experiment. The route is sound, the input is not. Reason: The mapping names its source explicitly and is not an unattributed bulk import, so this is not a defect in GO_REF:0000044. The defect is one level up. AIG1's UniProt entry reads 'SUBCELLULAR LOCATION: Cell membrane {ECO:0000269|PubMed:27018888}; Multi-pass membrane protein {ECO:0000255}', i.e. a single experimental citation for the compartment, and the cached full text of that citation contains no localisation experiment. This is where AIG1 and ADTRP genuinely differ: ADTRP's equivalent line cites two papers, one of which (PubMed:21868574) did image the protein at the cell surface, so the SL-0039 mapping has real input there and none here. Because the annotation is a faithful projection of a UniProt statement, the fix belongs in UniProt rather than in GO, and is raised in suggested_questions as an evidence-attribution correction. GO:0016020 membrane, which this protein certainly satisfies, is annotated separately and is accepted. Propagation Review Root cause: SOURCE WEAK OR INFERRED Failure modes: SOURCE EVIDENCE WEAK Sources checked: UniProtKB-SubCell:SL-0039 · Swiss-Prot subcellular location 'Cell membrane' SOURCE WEAK OR INFERRED The keyword-to-GO mapping itself is correct. The UniProt SUBCELLULAR LOCATION statement it reads from is supported on this protein by one reference, PubMed:27018888, whose cached full text contains no localisation experiment. Supporting Evidence: PMID:27018888 That these residues are generally predicted to be embedded within transmembrane domains of AIG1 and ADTRP indicates these enzymes could have evolved to perform hydrolytic chemistry within the cell membrane environment |
| GO:0016020 membrane | IDA PMID:27018888 AIG1 and ADTRP are atypical integral membrane hydrolases tha... | ACCEPT | Summary: This is the compartment the cited paper actually establishes. Endogenous AIG1 activity partitions to the membrane fraction of human LNCaP cells, and the recombinant protein is recovered in the HEK293T membrane pellet. Reason: The paper separates membrane and soluble fractions by ultracentrifugation at 100,000 g and finds the FAHFA hydrolase activity of LNCaP lysates mostly in the membrane fraction, with the endogenous enzyme detected there by activity-based probe labelling. That is a direct assay on the untransfected human cell line, so IDA is the right code even though the mutagenesis work was done by overexpression. Six predicted transmembrane helices and the family-specific InterPro signature corroborate it. This annotation carries all the localisation information the reference supports, which is precisely why the two plasma-membrane rows from the same reference do not. Supporting Evidence: PMID:27018888 Both KC01 and JJH260, but not THL or ABC34 also inhibited the FAHFA hydrolase activity of LNCaP cell lysates, which was mostly found in the membrane fraction PMID:27018888 The membrane lysates of hAIG1- and hADTRP-transfected HEK293T cells showed negligible hydrolytic activity above a mock-transfected control proteome with the majority of tested lipid substrates |
| GO:0016020 membrane | IEA GO_REF:0000002 | ACCEPT | Summary: InterPro2GO from IPR006838, the family-specific ADTRP/AIG1 signature rather than a bare fold. It maps to a cellular component only, so the classic signature-implies-catalysis error is absent, and the location it asserts is independently confirmed by IDA on this gene. Reason: IPR006838 is the entry integrated with PANTHER family PTHR10989 (ANDROGEN-INDUCED PROTEIN 1-RELATED) and Pfam PF04750 (Far-17a_AIG1), so it is orthogroup-specific and not a promiscuous fold signature. interpro2go maps it to GO:0016020 and to no molecular function, which is the correct restraint for an entry spanning thousands of unstudied proteins even though two of its members are now characterised hydrolases. Kept as a separate row from the IDA because the two have different provenance: one is a signature-based prediction that happens to be right, the other is a measurement on this protein. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: InterPro:IPR006838 · InterPro ADTRP/AIG1 family entry, integrated with PANTHER PTHR10989 and Pfam PF04750 SUPPORTS TRANSFER Family-specific signature, not a bare fold. Maps only to a cellular component, and the location is independently confirmed by IDA on this gene. |
| GO:0016787 hydrolase activity | IBA GO_REF:0000033 | MODIFY | Summary: A stale projection, not a curator judgement. PANTHER replaced this node's molecular function with the specific GO:0120573 FAHFA hydrolase activity on 2026-06-03 and already projects the new term onto this protein; GOA has not ingested that release and still ships the superseded general term. Reason: Checked against PANTHER's authoritative files rather than against GOA's own WITH/FROM column, because the WITH/FROM column can only report what GOA has already projected. In the current IBD.gaf at data.pantherdb.org, node PTN001659973 carries exactly two annotations - 'PTN001659973 GO:0120573 IBD UniProtKB:Q9NVV5|UniProtKB:Q96IZ2 F 20260603' and 'PTN001659973 GO:0042758 IBD UniProtKB:Q96IZ2|UniProtKB:Q9NVV5 P 20251127'. GO:0016787 is not among them. In the current leaf projection gene_association.paint_uniprot.gaf.gz both Q9NVV5 and Q96IZ2 already carry GO:0120573 by IBA from that node, dated 20260603, and neither carries a GO:0016787 IBA at all. QuickGO meanwhile returns zero IBA annotations to GO:0120573 anywhere in GOA. So PAINT placed the general term on 2026-05-28, refined it to the specific term six days later, and the GOA row under review is a snapshot of the earlier state. The repo's own cached slice records the transition: interpro/panther/PTHR10989/PTHR10989-paint.tsv carried GO:0016787 dated 20260528 before commit 5d1348b100 and carries GO:0120573 dated 20260603 after it. The replacement is well-founded on its own terms as well as being what upstream has already done: the node has exactly two seeds, human AIG1 and human ADTRP, both of which hold GO:0120573 by their own IMP from the same characterisation paper, so the donors agree completely and no heterogeneous clade is forcing a broad parent. This supersedes the KEEP_AS_NON_CORE verdict reached on the byte-identical row in the merged ADTRP review, which argued from the GOA projection date that the general term was deliberate; that was correct on the data available in July 2026 and is no longer correct. ADTRP's row should be revisited, and the point is raised once in suggested_questions with both genes named rather than repeated per gene. Propagation Review Root cause: SOURCE STALE OR MISSING Failure modes: GRANULARITY MISMATCH Sources checked: PANTHER:PTN001659973 · PANTHER IBD node, taxon:2759 Eukaryota, seeded by human AIG1 and human ADTRP SOURCE STALE OR MISSING The node no longer carries GO:0016787. The live IBD.gaf gives it GO:0120573 (IBD, 20260603) and GO:0042758 (IBD, 20251127) and nothing else, and PANTHER's leaf GAF already projects GO:0120573 onto this protein. UniProtKB:Q96IZ2 · human ADTRP, reviewed (Swiss-Prot), the only paralogue and a node seed SUPPORTS TRANSFER A genuine paralogue co-seed carrying its own IMP to GO:0120573 and GO:0042758 from PMID:27018888, so this is not a family-level guess. It supports the specific term, which is why the row is modified rather than removed. UniProtKB:Q9NVV5 · human AIG1 (the target itself) SUPPORTS TRANSFER Self-referential IBD seed recording a PAINT curator's judgement that the hydrolase function is a family-level character. Valid, not circular. Proposed replacements: FAHFA hydrolase activity Supporting Evidence: PMID:27018888 We show that AIG1, and the sequence-related homologous protein ADTRP, possess conserved Thr and histidine (His) residues required for FP reactivity and find that both enzymes hydrolyze the fatty-acid esters of hydroxy-fatty acid (FAHFA) class of lipids in vitro and in human cells. |
| GO:0042758 long-chain fatty acid catabolic process | IBA GO_REF:0000033 | ACCEPT | Summary: Correct for this gene, which holds the identical term by its own IMP from the same paper, and the substrate chemistry was verified rather than assumed. Unlike the molecular-function row from the same node, this term is still current in PANTHER's IBD.gaf. Reason: The suspicion worth testing was a direction error: AIG1 releases free long-chain fatty acids rather than degrading one, so 'long-chain fatty acid catabolic process' could have been an inversion of the kind GO:2000738 produces. It is not. ChEBI classifies the substrate CHEBI:83670 9-PAHSA(1-) as a long-chain fatty acid anion - its hierarchical ancestors include 'fatty acid anion', 'long-chain fatty acid anion' and 'lipid' - so the FAHFA is itself a long-chain fatty acid and hydrolysing it is literally its breakdown. Every one of UniProt's twelve catalytic activity lines for this protein gives PhysiologicalDirection=left-to-right, i.e. the curated direction is hydrolysis. Node status was checked independently of the molecular-function row and differs from it: GO:0042758 is still present at PTN001659973 in the current IBD.gaf, dated 20251127, so this row is not stale. The node reaches well beyond the two characterised seeds, which is a legitimate upstream question about a substrate-level claim propagated on dyad conservation, and is recorded in suggested_questions rather than as a defect in this gene's row. Verdict matches the merged ADTRP review on the byte-identical row. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN001659973 · PANTHER IBD node, taxon:2759 Eukaryota, seeded by human AIG1 and human ADTRP SUPPORTS TRANSFER Still current at this node in the live IBD.gaf (IBD, 20251127), unlike the molecular-function annotation from the same node. UniProtKB:Q96IZ2 · human ADTRP, reviewed (Swiss-Prot), paralogue co-seed SUPPORTS TRANSFER Carries its own IMP to GO:0042758 from PMID:27018888. UniProtKB:Q9NVV5 · human AIG1 (the target itself) SUPPORTS TRANSFER Self-referential IBD seed; the target independently holds the same term by IMP from the same paper. Supporting Evidence: PMID:27018888 These results, taken together, indicate that AIG1 functions as a major FAHFA hydrolase in human cells. PMID:32152231 The levels of other lipid classes were unaltered indicating that AIG1 and ADTRP specifically hydrolyze FAHFAs. |
| GO:0042758 long-chain fatty acid catabolic process | IMP PMID:27018888 AIG1 and ADTRP are atypical integral membrane hydrolases tha... | ACCEPT | Summary: Supported by shRNA knockdown of AIG1 in human LNCaP cells, which cut hydrolysis of an isotopically labelled PAHSA substrate by about 70 percent. The experiment is human, so IMP is the species-correct code. Reason: Two independent human loss-of-function arms support it: two stable shRNA lines against AIG1 in LNCaP reduced conversion of doubly labelled 13C,2H-PAHSA to its free fatty acid and hydroxy fatty acid products by roughly 70 percent relative to a scrambled control, and two structurally distinct covalent inhibitors reduced the same conversion by 70-80 percent while their matched inactive control compounds did not. The same inhibitor pattern reproduced in primary human T cells. All systems are human, so IMP rather than ISS is correct. The process term is chemically sound for this reaction because ChEBI classifies the FAHFA substrate as a long-chain fatty acid anion. It remains the only process term GO can offer; there is no FAHFA-level biological process, which is recorded as an ontology knowledge gap. Supporting Evidence: PMID:27018888 We next treated shAIG1 and shControl cells with 13C,2H-PAHSA and observed a ~70% reduction in hydrolysis products in the shAIG1 cells compared to shControl or uninfected cell lines PMID:27018888 These results, taken together, indicate that AIG1 functions as a major FAHFA hydrolase in human cells. |
| GO:0120573 FAHFA hydrolase activity | IEA GO_REF:0000116 | ACCEPT | Summary: The gene's core molecular function, reached by an automatic route that is nonetheless anchored on twelve explicitly curated reactions, each carrying experimental evidence from the characterisation paper. Fully congruent with the IMP row. Reason: The arithmetic closes exactly on both sides. GO:0120573 carries twelve RHEA cross-references (52048, 52052, 52056, 52060, 52064, 52068, 52072, 52076, 52080, 52084, 52092, 52096); this annotation's WITH/FROM lists all twelve; and UniProt curates twelve CATALYTIC ACTIVITY lines for Q9NVV5, one per reaction, each tagged ECO:0000269|PubMed:27018888 with PhysiologicalDirection=left-to-right. So the IEA is a re-expression of experimental UniProt content through the Rhea mapping rather than a similarity inference, which is why it agrees with the IMP row instead of competing with it. The twelve reactions span 9-, 12-, 13- and 5-branched FAHFAs with palmitoyl, palmitoleoyl, oleoyl and stearoyl acyl donors, matching the substrate panel in the paper. Note that ADTRP's equivalent row lists eleven, being without RHEA:52092; recorded as an observed difference between the two records, not interpreted. Supporting Evidence: PMID:27018888 AIG1 and ADTRP displayed a preference for FAHFAs with branching distal from the carboxylate head group of the lipids |
| GO:0120573 FAHFA hydrolase activity | IMP PMID:27018888 AIG1 and ADTRP are atypical integral membrane hydrolases tha... | ACCEPT | Summary: The gene's core molecular function, established by mutating both catalytic residues in the human protein and by knocking the enzyme down in human cells. This is a measured activity on this protein, not a domain- or family-name inference. Reason: The predicted defect class for a membrane hydrolase - a catalytic term inherited from a domain or family name while the residues are gone - is absent here, and the check was run rather than assumed. UniProt's feature table carries SITE 43 and SITE 134 'Important for catalytic activity', both ECO:0000269|PubMed:27018888, and both have MUTAGEN entries reading 'Loss of hydrolase activity'. The underlying experiments: of fifteen individual Ser/Thr-to-Ala mutants, only T43A abolished fluorophosphonate probe labelling; H134A did the same while two non-conserved His mutants had no effect; and both mutants lost FAHFA hydrolase activity outright. Substrate specificity was tested against a panel, with negligible activity on lyso- and phospholipids and on mono-, di- and tri-glycerides. Endogenous enzyme was then knocked down in human LNCaP cells with two shRNAs. Every system is human. Note also how thin the term is elsewhere: GO:0120573 has 495 annotations across all species and exactly two of them are experimental, this one and ADTRP's, both from this paper. The FlyBase assignment on a human annotation is unusual but not a defect - GO permits cross-database annotation and FlyBase curates the Drosophila family member. Supporting Evidence: PMID:27018888 The FAHFA hydrolase activities of AIG1 and ADTRP were abolished by mutating their putative catalytic nucleophilic residues Thr-43 and Thr-47, respectively PMID:27018888 Mutation of His134 to Ala in AIG1 (H134A) eliminated FP-labeling PMID:27018888 Among the 15 individual Ser/Thr-to-Ala mutant proteins evaluated, only the T43A mutant showed a complete loss of FP-labeling |
| GO:0046627 negative regulation of insulin receptor signaling pathway | ISO PMID:38816388 Inflammation causes insulin resistance in mice via interfero... | NEW | Summary: A coverage gap: a catalysis-dependent effect of AIG1 on insulin signalling was demonstrated in adipocytes and produced no GO annotation on any species. Proposed as ISO from mouse Aig1 because every AIG1-specific experiment in the paper is murine. Reason: In mouse adipocytes, AIG1 overexpression lowered insulin-stimulated AKT phosphorylation and glucose uptake, AIG1 knockdown reversed the same readouts in the opposite genetic background, and a selective covalent AIG1 inhibitor restored glucose homeostasis in vivo on a high-fat diet. Critically for a catalytic gene, the T43A and H134A alleles - the same two residues whose mutation abolishes FAHFA hydrolysis - failed to reproduce the wild-type effect, so the phenotype runs through the enzyme activity rather than through an unrelated scaffolding role, which is what makes it a function rather than a disease association. The evidence code is not IMP because the species check was run explicitly: FI3KO and FI3OE adipocytes are murine, the activity-based profiling was in mouse brain and kidney membranes, and the high-fat-diet work is in mice. The paper does use human SGBS adipocytes, but only for IRF3; the sole hAIG1 strings in the full text are inside the shRNA construct name shAIG1. ISO rather than the broader ISS, with the source sequence named in supporting_entities as UniProtKB:Q9D8B1 (AIG1_MOUSE, reviewed), because the transfer is from the 1:1 orthologue - Q9D8B1 and Q9NVV5 sit in the same PANTHER subfamily and both are reviewed under the same protein name - rather than from generic sequence similarity. Note that on GO's own conventions the primary annotation belongs on mouse Aig1 with IMP and the human row is the orthology projection of it; neither exists in GOA today, so both are proposed here and the mouse row is the one a curator should make first. Supporting Evidence: PMID:38816388 Adipocyte-specific knockout of IRF3 protects male mice against high-fat diet-induced insulin resistance, whereas overexpression of IRF3 or AIG1 in adipocytes promotes insulin resistance on a high-fat diet. PMID:38816388 As before, FI3KO adipocytes exhibited elevated p-AKT level and insulin-stimulated glucose uptake, which was fully suppressible by WT AIG1 overexpression, but not two different alleles of Aig1 bearing mutations in the catalytic domain (T43A and H134A) |
| GO:0031398 positive regulation of protein ubiquitination | ISO PMID:40303337 AIG1 protects against doxorubicin-induced cardiomyocyte ferr... | NEW | Summary: The second, non-lipid half of AIG1's biology, currently represented in GO only by a bare protein binding row. AIG1 binding to Pirh2/RCHY1 promotes Pirh2-dependent ubiquitination of p53. Proposed as ISO from mouse Aig1 because the functional work is murine. Whether it requires the hydrolase activity is untested, so it is not claimed to be catalysis-independent. Reason: AIG1 knockdown reduced Pirh2 protein and Pirh2-p53 association, AIG1 overexpression increased both, and deleting the AIG1 region required for Pirh2 binding reduced p53 ubiquitination - a consistent gain- and loss-of-function pair with a readout on the ubiquitination step itself rather than only on p53 levels. This is the natural biological-process partner of the GO:0031625 binding term proposed on the GO:0005515 row, and neither has any representation in GOA today. The code is not IMP because the experiments are in C57BL/6 mice, AAV9-transduced hearts and the murine HL-1 cardiomyocyte line; the HEK293T truncation co-IPs do not state the species of the constructs, so they cannot upgrade it. ISO rather than ISS, with the source sequence named as UniProtKB:Q9D8B1 (AIG1_MOUSE, reviewed), because the transfer is from the 1:1 orthologue. GO has no p53-specific ubiquitination child, so GO:0031398 is the most specific available term. A deliberately weaker claim than the paper's: no residue- or surface-level mechanism is asserted, because the 35-93 deletion that maps the interaction removes an entire transmembrane helix and cannot separate a contact surface from loss of folding. Supporting Evidence: PMID:40303337 Importantly, the deletion of the 35-93 aa domain of AIG1 resulted in significant reductions in Pirh2-p53 interaction and p53 ubiquitination PMID:40303337 Mechanistically, AIG1 directly interacted with the Pirh2 E3 |
| GO:1901800 positive regulation of proteasomal protein catabolic process | ISO PMID:40303337 AIG1 protects against doxorubicin-induced cardiomyocyte ferr... | NEW | Summary: The degradation outcome is demonstrated separately from the ubiquitination step and with a discriminating control, so it earns its own term rather than being folded into GO:0031398. Added after a reviewer pointed at experiments in this paper that the first pass had not used. Reason: Added on reviewer challenge, and the challenge was right: the first pass was about to decline this term on the grounds that the paper's proteasome statement is hedged ('likely regulated through a ubiquitin-proteasome mechanism'), without having read the experiments behind it. Those experiments are properly controlled. A cycloheximide chase shows AIG1 silencing decelerates p53 turnover and AIG1 overexpression accelerates it, which measures degradation rate directly rather than steady-state level. A three-inhibitor panel then discriminates the route: the drop in p53 caused by AIG1 overexpression is reversed by the proteasome inhibitor MG132 but unaffected by the autophagy inhibitor 3-methyladenine or the autolysosome inhibitor bafilomycin A1. Having both a positive and a negative arm is what makes this a route assignment rather than an inference from the ubiquitination result, and it is why the term is not redundant with GO:0031398: that term covers the modification step, this one the catabolic outcome, and each was measured on its own. Same species reasoning and same ISO source as the two rows above. Supporting Evidence: PMID:40303337 In fact, a cycloheximide (CHX) chase assay conducted in HL-1 cardiomyocytes revealed that AIG1 silencing significantly decelerated p53 protein degradation PMID:40303337 Our results showed that protein level of p53 was reduced in response to AIG1 overexpression, an effect reversed by MG132 treatment but unaffected by 3-MA or Baf A1 under DOX stress |
Loading supporting content…
Download this section (compressed HTML)Q: GOA is shipping a superseded PAINT annotation for both members of PTHR10989's characterised clade. Node PTN001659973 carried GO:0016787 hydrolase activity dated 20260528, but PANTHER replaced it with GO:0120573 FAHFA hydrolase activity on 20260603; the current IBD.gaf gives the node only GO:0120573 and GO:0042758, and gene_association.paint_uniprot.gaf.gz already projects GO:0120573 by IBA onto both UniProtKB:Q9NVV5 (AIG1) and UniProtKB:Q96IZ2 (ADTRP). QuickGO nevertheless returns zero IBA annotations to GO:0120573 anywhere in GOA, and both genes still carry the GO:0016787 IBA. Is this simply an ingest lag, and if so when will the refresh land? Stated once for both genes: the merged ADTRP review reached KEEP_AS_NON_CORE on the byte-identical row from the pre-refresh data and should be revisited.
Suggested experts: GO Central / PAINT curators, GOA / UniProt-GOA pipeline
Q: UniProt evidence attribution for Q9NVV5. The SUBCELLULAR LOCATION line reads 'Cell membrane {ECO:0000269|PubMed:27018888}', but the cached full text of that paper contains no localisation experiment - zero occurrences of 'plasma membrane', 'cell surface', 'immunofluoresc', 'confocal', 'subcellular' or 'localization', against positive controls of 'membrane fraction' 3, 'membrane lysates' 9 and 'transmembrane' 29 in the same file. What that paper supports is 'Multi-pass membrane protein'. Note the contrast with the paralogue: ADTRP's equivalent line cites PubMed:21868574 as well, which did image the protein at the cell surface, so ADTRP's compartment call has independent support and AIG1's does not. Should the ECO:0000269 compartment assertion be withdrawn on AIG1, or re-sourced?
Suggested experts: UniProt curators
Q: Where is AIG1 actually located, and what is its topology? UniProt gives six transmembrane helices with a cytoplasmic N-terminus, all from prediction (ECO:0000255, with the topological domains ECO:0000305 from a paper that ran no topology experiment). PMID:27040980 measured five transmembrane domains with a luminal N-terminus and a cytosolic C-terminus by protease protection, in the ER membrane, and reported an ER calcium phenotype. The two agree only on the cytosolic C-terminus. The compartment matters for interpretation: a plasma-membrane enzyme and an ER enzyme sample different FAHFA pools, and a cytosolic E3 ligase partner is accessible from either but only via the cytosolic face.
Suggested experts: Membrane protein topology, Lipid signalling
Q: Should the substrate-level process term GO:0042758 sit at PTN001659973, which PAINT scopes to taxon:2759 Eukaryota? The character conserved across that node is the Thr/His catalytic dyad, which licenses a claim about mechanism. Substrate has been established only for the two human members, and the node's non-animal recipients include fungal and plant proteins whose only reviewed representatives are curated as uncharacterised. The molecular function has just been refined at this node; the question is whether the process term should instead sit on the subclade where the seeds are.
Suggested experts: GO Central / PAINT curators
Q: Do AIG1's short splice isoforms retain any activity? Isoform 6 substitutes His134 and truncates at 135, and isoform 5 truncates at 138, so both lose the catalytic general base; isoform 3 deletes residues 48-99. No GOA annotation on this gene carries an isoform qualifier and no experiment has tested isoform-specific activity, so GO currently asserts the hydrolase function of a gene product whose annotated isoforms are not all capable of it.
Suggested experts: UniProt curators, Lipid signalling
Experiment: Localise endogenous AIG1 in a cell line where it is abundant (LNCaP, or primary human T cells) by surface biotinylation with a membrane-impermeant reagent alongside immunofluorescence against ER, Golgi and plasma-membrane markers, and confirm by sucrose-gradient fractionation with compartment marker blots. Then repeat the in situ 13C,2H-PAHSA hydrolysis assay on surface-stripped versus permeabilised cells to ask where the activity is executed, not merely where the protein sits. A negative surface result would settle two GOA rows and a UniProt line at once.
Hypothesis: AIG1 resides in the ER membrane rather than the plasma membrane, and the current plasma-membrane annotation is an over-read of a crude membrane fraction.
Type: localisation and topology
Experiment: Co-immunoprecipitate Pirh2 with wild-type AIG1 and with the T43A and H134A catalytic mutants, and measure p53 ubiquitination in each case; in parallel, test whether a covalent AIG1 inhibitor that occupies the active site (KC01 or JJH260) alters Pirh2 binding. If catalytically dead AIG1 still binds Pirh2 and still supports p53 ubiquitination, the two functions are genuinely separable and should be curated as such. Use a point-substitution or small-epitope strategy rather than the 35-93 deletion, which removes a whole transmembrane helix.
Hypothesis: The AIG1-Pirh2 interaction is independent of the FAHFA hydrolase activity, so the two functions are separable.
Type: structure-function separation
Experiment: Measure the full FAHFA isomer panel by LC-MS in tissues of Aig1 single-knockout mice, with Adtrp single-knockout and double-knockout littermates as the comparison set. The published in vivo study reports elevated FAHFAs for Adtrp-KO and for the double knockout but not for the Aig1 single knockout, so what is missing is a design in which AIG1 rather than ADTRP is the dominant hydrolase in the tissue assayed. Pick those tissues by measuring rather than by citation: the candidate profiles disagree (PMID:27018888 cites biogps for brain and macrophages, PMID:11266118's Northern blot gives heart, ovary, testis, liver and kidney), so quantify AIG1 and ADTRP activity side by side across a mouse tissue panel by competitive ABPP first, then run the lipidomics where the ratio is most favourable.
Hypothesis: AIG1 alone controls tissue FAHFA levels in vivo, which the published double-knockout design leaves untested.
Type: genetic loss of function with lipidomics
Experiment: Express isoforms 1, 3, 5 and 6 individually in HEK293T cells and assay each for fluorophosphonate probe reactivity and 9-PAHSA hydrolysis, then co-express each short isoform with the full-length enzyme and ask whether hydrolysis is suppressed below the full-length-alone level. Human dermal papilla and liver, where the alternative transcripts were originally detected, would be the tissues in which the result matters.
Hypothesis: The short AIG1 isoforms that lack His134 are catalytically dead and may act as dominant negatives.
Type: isoform structure-function
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: Where AIG1 executes its hydrolase activity is unresolved, and the two available answers are incompatible: GOA and UniProt say plasma membrane, while the only study that examined the question places the protein in the ER membrane with a different topology.
OPEN BIOLOGYCURATION CC_DARK
What is known: Known: AIG1 is a polytopic membrane protein, and its activity partitions to the membrane fraction of human cells, so GO:0016020 membrane is secure. Unknown: which membrane. The plasma-membrane annotation is attributed to a paper containing no localisation experiment, and the ER assignment rests on one unreplicated protease-protection study whose full text is unavailable.
Significance: The compartment determines which FAHFA pool the enzyme can reach and whether a cytosolic E3 ligase partner is accessible, and it is the difference between a cell-surface drug target and an intracellular one. Two GOA rows and a UniProt ECO:0000269 statement currently assert an answer that their cited evidence does not contain.
Provenance (the field's own admissions):
Gap: GO can express FAHFA hydrolysis as a molecular function but not as a biological process, so FAHFA catabolism has to be annotated with the much broader GO:0042758 long-chain fatty acid catabolic process.
OPEN ONTOLOGY BP_DARK
What is known: Known and expressible: GO:0120573 FAHFA hydrolase activity exists, created 2026-03-14 and anchored on twelve RHEA reactions. Absent: any process term at the FAHFA level. A QuickGO text search for FAHFA returns GO:0120573 and nothing else, and walking the ontology rather than only searching it gives the same answer - GO:1901569 fatty acid derivative catabolic process has five children (fatty acid primary amide, ketone body, icosanoid, fatty-acyl-CoA and fatty alcohol catabolic processes) and none covers FAHFAs.
Significance: GO:0042758 is chemically correct here, because ChEBI classifies 9-PAHSA(1-) as a long-chain fatty acid anion, but it is far broader than the demonstrated specificity and it is the term PAINT propagates from a pan-eukaryotic node. GO already models the thioester analogue as GO:0036115 fatty-acyl-CoA catabolic process, so the precedent exists. The merged ADTRP review files the corresponding proposed_new_terms entry ('fatty acid ester of hydroxy fatty acid catabolic process', parent GO:0042758); this review concurs and records the gap rather than duplicating the request.
Provenance (the field's own admissions):
Gap: Whether AIG1 alone sets endogenous FAHFA tone in vivo is untested; the published genetic evidence comes from Adtrp single knockouts and AIG1/ADTRP double knockouts.
OPEN BIOLOGY RESIDUAL_SUBGAP
What is known: Known: in human cells AIG1 accounts for roughly 70 percent of PAHSA hydrolysis, and loss of both enzymes raises tissue FAHFAs in mice without altering other lipid classes. Unknown: the contribution of AIG1 on its own in any tissue, which matters because the two enzymes are reported to have distinct expression profiles. That reported difference is itself unsettled and should not be treated as established: PMID:27018888's discussion cites biogps for AIG1 being broadest in brain and macrophages with ADTRP restricted to metabolic organs, whereas the Northern blot in PMID:11266118 puts human AIG1 highest in heart, ovary, testis, liver and kidney, and Human Protein Atlas records AIG1 as tissue-enhanced in liver.
Significance: Every in vivo statement about AIG1's physiological role currently inherits its support from an experiment in which the paralogue was also deleted. A tissue-matched single-knockout series would separate them.
Provenance (the field's own admissions):
Gap: Whether AIG1's Pirh2/p53 function requires its hydrolase activity is unknown, so the gene has two functional descriptions with no established relation between them.
OPEN BIOLOGY RESIDUAL_SUBGAP
What is known: Known: AIG1 binds Pirh2 and promotes Pirh2-dependent p53 ubiquitination in murine cardiomyocytes; separately, it hydrolyses FAHFAs through Thr43 and His134. Unknown: whether the catalytically dead T43A or H134A protein still binds Pirh2 and still supports p53 ubiquitination. The mapped requirement, residues 35-93, includes Thr43 and a whole transmembrane helix, so the existing data cannot separate the two.
Significance: This is why core_functions carries only the hydrolase activity even though the review proposes three annotations for the Pirh2 arm. Splitting them into two co-equal core functions would assert a separation that nobody has tested; if the second turns out to require the first, they are one function and not two.
Provenance (the field's own admissions):
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)