ADIRF (adipogenesis regulatory factor, formerly C10orf116, also known as APM2 and AFRO) is a 76-residue nuclear and cytosolic protein of vertebrates, originally identified as one of the most abundant transcripts in human adipose tissue. It is a small, alanine- and glutamine-rich polypeptide with no recognisable domain, no resolved structure and no known molecular activity: its only sequence feature is the mature chain itself. Introduced into preadipocytes it raises the levels of the master adipogenic transcription factors PPARG and CEBPA during the early phase of differentiation and accelerates adipogenic conversion, placing it upstream of the core adipogenic transcriptional program, though the mechanism by which it does so - and whether it acts at transcription directly - is unresolved. Immunofluorescence places the protein in both the nucleoplasm and the cytosol, a distribution consistent with its being far smaller than the nuclear-pore diffusion limit, and it is also recovered from extracellular fluids: it is present in human serum at microgram-per-millilitre concentrations and in urinary and prostatic exosome preparations, despite lacking any signal peptide or transmembrane segment. Expression is highest in arterial tissue and in adipose tissue, with additional expression in heart, cornea, liver, kidney and spleen, and the protein is detected in vascular and non-vascular smooth muscle. Elevated ADIRF is associated with obesity and, in tumour cells, with resistance to cisplatin. The gene is present across vertebrates: orthologues in teleost fish, birds and mammals all align across essentially the whole 76-residue protein. It was nonetheless lost in the ancestor of mice, rats and hamsters through a 43-kilobase deletion that removed its promoter and first exon, and it has no paralogue, so rodent models of the standard kind cannot report on its function.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005634 nucleus | IBA GO_REF:0000033 | ACCEPT | Summary: Accepted. This is the PAN-GO curator's core-function statement for ADIRF, and its WITH/FROM names ADIRF itself (UniProtKB:Q15847) as the IBD seed alongside the node, so it is a self-referential IBA recording a human curator's judgement rather than a transfer from another species. The nucleus is independently supported by the gene's own IDA and by HPA immunofluorescence, and it is where a factor acting upstream of PPARG and CEBPA would have to act. Reason: PANTHER:PTN008674116 reaches seven mammalian ADIRF orthologues by IBA (human, chimpanzee, gorilla, dog, pig, cow, opossum) and every one of them lists UniProtKB:Q15847 as the seed, because human ADIRF is the only member of the family with any experimental annotation. Mouse and rat are absent from that list for a concrete reason: Muroidea have no ADIRF gene at all. I note but do not act on the fact that `is_active_in` asserts the molecular function is executed in the nucleus, which is a stronger claim than `located_in` for a protein whose molecular function is entirely unknown, and that HPA calls cytosol a main location alongside nucleoplasm. Both points are raised as questions for GO_Central rather than used to downgrade a curator's deliberate core-function call. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN008674116 Β· PANTHER ancestral node for the ADIRF family SUPPORTS TRANSFER A PANTHER internal tree node, not a protein. It reaches 7 mammalian ADIRF orthologues by IBA, all seeded from human Q15847. UniProtKB:Q15847 Β· human ADIRF (the gene under review; self-referential IBD seed) SUPPORTS TRANSFER Self-referential: the seed is the target. Per campaign convention this is valid and records a PAN-GO curator judging nuclear residence core, not a circular chain. Q15847 carries its own IDA to GO:0005634. Supporting Evidence: PMID:23239344 Our data demonstrated that C10orf116 is highly expressed in adipose tissue and is localized primarily within the nucleus. file:human/ADIRF/ADIRF-bioinformatics/RESULTS.md IBD seeds named in the WITH/FROM, with the `db` field read rather than flattened away: UniProtKB:Q15847 x14. The seed is the gene under review, so both IBA rows are **self-referential**: **True**. |
| GO:0045600 positive regulation of fat cell differentiation | IBA GO_REF:0000033 | ACCEPT | Summary: Accepted. The PAN-GO curators selected positive regulation of fat cell differentiation as ADIRF's core biological process, seeded from the gene's own IDA. This is the best-supported process claim for the gene and is the one UniProt's FUNCTION line leads with. Reason: Same self-referential node as the nucleus row: PANTHER:PTN008674116 plus UniProtKB:Q15847 as the IBD seed, propagating to seven mammalian orthologues. The term is correct for human ADIRF on its own experimental evidence, so the IBA adds curatorial weight rather than new evidence. Note the boundary the underlying experiment sets: it was ectopic expression in mouse 3T3-L1 preadipocytes, a lineage that has no ADIRF gene, so it shows ADIRF can promote adipogenic differentiation and not that adipogenesis requires it. That distinction is recorded as a knowledge gap rather than used to reject the term. Notably PAN-GO chose this term and not GO:0045944 from the same paper, which is why GO:0045944 is treated as non-core here. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN008674116 Β· PANTHER ancestral node for the ADIRF family SUPPORTS TRANSFER Internal tree node reaching 7 mammalian gene products by IBA. Six are genuine ADIRF orthologues of 75-76 aa aligning over 73-76 of the 76 human residues at 81.6-100% identity, so the transfer to human is sound and no granularity problem arises for this gene. The seventh, UniProtKB:A0A5F8H3S4 (Monodelphis domestica), is NOT an orthologue - a 447-aa tandem-repeat protein aligning over 62 of 76 residues at 15.8% identity - and it receives both of this node's terms. That is a defect in the node's membership, not in the transfer to human, and its root cause is upstream of PAINT: opossum has a real ADIRF gene (NCBI Gene 100020286) but UniProt's proteome for the species contains no ADIRF-sized member of the family, so the tree was given the wrong sequence. Raised in suggested_questions. UniProtKB:Q15847 Β· human ADIRF (the gene under review; self-referential IBD seed) SUPPORTS TRANSFER Self-referential seed carrying its own IDA to GO:0045600. Valid; records a core-function judgement, not a circular transfer. Supporting Evidence: PMID:23239344 Over-expression studies in 3T3-L1 cells indicated that it up-regulates the levels of CCAAT/enhancer binding protein Ξ± (C/EBPΞ±) and PPARΞ³ and promotes adipogenic differentiation starting from the early stage of adipogenesis. file:human/ADIRF/ADIRF-uniprot.txt CC -!- FUNCTION: Plays a role in fat cell development; promotes adipogenic |
| GO:0005634 nucleus | IEA GO_REF:0000120 | ACCEPT | Summary: Accepted as correct but redundant. The term is right - ADIRF has its own IDA to nucleus and a more specific HPA IDA to nucleoplasm - but this row carries no independent evidence, because both tokens in its WITH/FROM trace back to the same experiment the IDA row already records. Reason: InterPro's IPR034450 is a single-family signature whose entire curated basis is this gene: its InterPro description is written from human ADIRF's own papers, and it has exactly one reviewed Swiss-Prot member, human ADIRF itself. interpro2go maps IPR034450 to exactly two terms, GO:0045600 and GO:0005634. The second WITH/FROM token, UniProtKB-SubCell:SL-0191, is UniProt's own Nucleus subcellular-location term, and UniProt's SUBCELLULAR LOCATION line attributes it to ECO:0000269|PubMed:23239344 - the same paper as the IDA row. So both routes into this row originate in PMID:23239344. Correct term, zero added information; ACCEPT rather than REMOVE because nothing false is asserted. The WITH/FROM names no ARBA rule id, so there was no rule to fetch at rest.uniprot.org/arba/ despite the GO_REF:0000120 reference. Propagation Review Root cause: EVIDENCE CIRCULAR OR REDUNDANT Failure modes: CIRCULAR PROPAGATION Sources checked: InterPro:IPR034450 Β· InterPro family 'Adipogenesis regulatory factor' CIRCULAR OR REDUNDANT Family entry whose description and GO mapping derive from human ADIRF's own literature, and whose only reviewed member is human ADIRF. Mapping it back onto human ADIRF adds no evidence. UniProtKB-SubCell:SL-0191 Β· UniProt subcellular location 'Nucleus' CIRCULAR OR REDUNDANT UniProt's own Nucleus annotation for Q15847 is attributed to ECO:0000269|PubMed:23239344, the same paper behind the IDA row. Supporting Evidence: file:human/ADIRF/ADIRF-uniprot.txt CC -!- SUBCELLULAR LOCATION: Nucleus {ECO:0000269|PubMed:23239344}. |
| GO:0045600 positive regulation of fat cell differentiation | IEA GO_REF:0000002 | ACCEPT | Summary: Accepted as correct but redundant. Pure InterPro2GO from IPR034450, a signature built from this gene's own papers, so for human ADIRF the row restates the IDA. The term is right; the route is circular. The same mapping over-reaches badly on other species, which is filed as an upstream question rather than acted on here. Reason: InterPro's IPR034450 is a single-family signature whose entire curated basis is this gene: its InterPro description is written from human ADIRF's own papers, and it has exactly one reviewed Swiss-Prot member, human ADIRF itself. interpro2go maps IPR034450 to exactly two terms, GO:0045600 and GO:0005634. Measured over the whole propagation (withFrom=InterPro:IPR034450, fully paginated, 1512 annotations asserted complete): GO:0045600 reaches 723 gene products, of which only 130 are ADIRF-sized (60-90 aa) while 504 are larger than 200 aa and therefore cannot be orthologues of a 76-aa protein, and 237 are non-vertebrate metazoans. Human ADIRF is not one of the mis-hit recipients, so this row's action turns on the human evidence alone and is ACCEPT. The upstream defect - an Ala/Gln-rich 76-aa signature matching long tandem-repeat proteins in molluscs, insects and nematodes and carrying a mammal-derived adipocyte term to them - is raised in suggested_questions for InterPro and GO. Propagation Review Root cause: EVIDENCE CIRCULAR OR REDUNDANT Failure modes: CIRCULAR PROPAGATION Sources checked: InterPro:IPR034450 Β· InterPro family 'Adipogenesis regulatory factor' CIRCULAR OR REDUNDANT For human ADIRF the mapping is circular: the entry's description cites this gene's own papers and its only reviewed member is human ADIRF. Separately, and not a reason to change this row, the same entry carries GO:0045600 to 723 gene products of which 504 exceed 200 aa - a composition-driven match on a low-complexity 76-aa sequence, quantified in the bioinformatics results. Supporting Evidence: PMID:23239344 Over-expression studies in 3T3-L1 cells indicated that it up-regulates the levels of CCAAT/enhancer binding protein Ξ± (C/EBPΞ±) and PPARΞ³ and promotes adipogenic differentiation starting from the early stage of adipogenesis. file:human/ADIRF/ADIRF-bioinformatics/RESULTS.md | GO:0045600 positive regulation of fat cell differentiation | 723 | 130 | 504 | 89 | file:human/ADIRF/ADIRF-bioinformatics/RESULTS.md Recipients that are both non-vertebrate and larger than 200 aa: **217**. |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Marked as over-annotated. The interaction with IL18 is a single unreplicated yeast-two-hybrid hit from the HuRI systematic screen, its apparent triplicate support is one screen logged under three assay versions, and `protein binding` conveys nothing about what ADIRF does. No informative replacement term is proposed because no functional consequence of the interaction has ever been measured. Reason: UniProt records NbExp=3 for this pair, but IntAct returns a single interaction record (EBI-11784632, MI-score 0.56) logged under three sub-methods - two hybrid array, two hybrid prey pooling approach, and validated two hybrid - and the source paper states it screened with a panel of three Y2H assay versions. So NbExp counts assay versions, not independent experiments. Neither ADIRF nor IL18 is named in the paper's text; the pair comes from the supplementary interaction list. There is no orthogonal assay and no follow-up literature linking the two proteins. Three checks that would have strengthened the case came back NEGATIVE and are reported as such rather than omitted: (1) the partner accession resolves to reviewed canonical Swiss-Prot IL18 at the full 193 aa, so there is no TrEMBL or partial-ORFeome substitution; (2) neither protein is a promiscuous hub - IL18 has 15 distinct IntAct partners and ADIRF has 4; (3) the interaction is not topologically impossible, because IL18 is a leaderless cytokine whose precursor UniProt describes as cytosolic and ADIRF is nucleoplasmic and cytosolic, so both could meet in the cytosol. The verdict therefore rests only on the absence of replication and the uninformativeness of the term, not on any argument that the observation is impossible. Supporting Evidence: PMID:32296183 We screened this search space a total of nine times with a panel of three Y2H assay versions file:human/ADIRF/ADIRF-uniprot.txt CC Q15847; Q14116: IL18; NbExp=3; IntAct=EBI-7162516, EBI-3910835; |
| GO:0005654 nucleoplasm | IDA GO_REF:0000052 | ACCEPT | Summary: Accepted, and it is the most specific supported nuclear term for this gene. HPA's immunofluorescence for antibody HPA026810 has reliability 'Supported', and the same antibody has an unusually clean specificity control: it gives no signal in any mouse tissue, and mouse has no ADIRF gene. Reason: GO:0005654 is a verified descendant of GO:0005634, so this row is a refinement of the nucleus rows rather than an independent claim, and it is the term core_functions should carry for the nuclear compartment. The antibody control is worth stating: PMID:31945134 used Sigma HPA026810 - the same antibody underlying this HPA call - for immunohistochemistry across mouse tissues and saw no ADIRF staining anywhere, which is the expected result for a species carrying a 43-kb deletion of the locus. An antibody that stains human tissue and is blank in the species lacking the gene is about as well controlled as a localisation reagent gets. Separately, HPA's own record gives TWO main locations, Nucleoplasm and Cytosol, and GOA imported only the nucleoplasm half; the missing GO:0005829 row is proposed as a NEW annotation below rather than treated as a defect in this row. Supporting Evidence: PMID:31945134 Finally, mouse did not stain for LPHN2/ADGRL2 in lung non-vascular SMC or for ADIRF in any tissue. file:human/ADIRF/ADIRF-bioinformatics/RESULTS.md HPA record for ADIRF (ENSG00000148671), IF reliability **Supported**, main subcellular locations **Nucleoplasm, Cytosol**. |
| GO:0070062 extracellular exosome | HDA PMID:23533145 In-depth proteomic analyses of exosomes isolated from expres... | KEEP AS NON CORE | Summary: Kept as non-core. A real high-throughput detection in exosomes purified from expressed prostatic secretions in urine, but one of 1046 proteins given this single term from this one paper, with no ADIRF-specific validation - and the authors themselves caution that abundant contaminating proteins in such preparations must be verified before their presence is generalised. Reason: The reference-projection test shows this PMID annotates 1046 distinct gene products with GO:0070062 and nothing else, all HDA, all assigned by UniProt. That is a bulk import, not 1046 independent findings; ADIRF is not mentioned anywhere in the paper's narrative and the row derives from a supplementary identification list. One count I could not reconcile and am recording rather than explaining away: the paper's text says close to 900 proteins were identified in the two pools, while GOA imported 1046 entities from it, and the identifications live in a supplemental table absent from the cached text. ADIRF's presence outside the cell is independently real and better supported than the exosome-specific claim: it is measured in human serum by ELISA at microgram-per-millilitre concentrations across 139 patients (PMID:33737617) and detected as an intact protein by top-down mass spectrometry in lipoaspirate fluid (PMID:26719138). Neither is in GOA. That corroboration is why this row is kept rather than marked over-annotated. It supports the parent - GO:0070062 is a verified descendant of GO:0005576 extracellular region - and not the exosome mechanism, which requires release by multivesicular-body fusion and has never been tested for ADIRF. I did not propose GO:0005576 separately because this row already entails it. The protein has no signal peptide and no transmembrane segment, so the export route is unknown; that is filed as a knowledge gap. Supporting Evidence: PMID:23533145 Certainly the presence of high abundant contaminating proteins, in exosome preparations from cancer-related biofluids such as EPS-urine, must be taken into account and further verified before generalizing their presence to a clinical association with the cancerous condition. PMID:23533145 in total, close to 900 proteins were identified in the two EPS-urine exosome pools PMID:33737617 cut-off value was determined as 18.7 Β΅g/mL, with a sensitivity and specificity of 84.0% and 71.7%, respectively |
| GO:0070062 extracellular exosome | HDA PMID:19056867 Large-scale proteomics and phosphoproteomics of urinary exos... | KEEP AS NON CORE | Summary: Kept as non-core. An independent high-throughput detection, this time in exosomes from normal human urine rather than a cancer-related biofluid, which is a useful corroboration of the other GO:0070062 row - but again one of over a thousand proteins from a single study, with no ADIRF-specific validation. Reason: This row's value relative to the PMID:23533145 row is that the source material is healthy-donor urine, so the contaminating-tumour-protein caveat that the other paper raises about itself does not apply here. Against that, the paper is cached abstract-only (full_text_available: false), so I can characterise only what the abstract states, and the reference-projection test shows it annotates 1016 distinct gene products with GO:0070062 and no other term - a bulk import. ADIRF is a small, very abundant protein and both exosome preparations are from the urogenital tract where ADIRF is well expressed, so detection is expected on abundance grounds alone. ADIRF's presence outside the cell is independently real and better supported than the exosome-specific claim: it is measured in human serum by ELISA at microgram-per-millilitre concentrations across 139 patients (PMID:33737617) and detected as an intact protein by top-down mass spectrometry in lipoaspirate fluid (PMID:26719138). Neither is in GOA. That corroboration is why this row is kept rather than marked over-annotated. It supports the parent - GO:0070062 is a verified descendant of GO:0005576 extracellular region - and not the exosome mechanism, which requires release by multivesicular-body fusion and has never been tested for ADIRF. I did not propose GO:0005576 separately because this row already entails it. The protein has no signal peptide and no transmembrane segment, so the export route is unknown; that is filed as a knowledge gap. Supporting Evidence: PMID:19056867 Overall, the analysis identified 1132 proteins unambiguously, including 177 that are represented on the Online Mendelian Inheritance in Man database of disease-related genes, suggesting that exosome analysis is a potential approach to discover urinary biomarkers. PMID:26719138 adipogenesis regulatory factor, perilipin-1 fragments, and S100A6, along with their PTMs |
| GO:0005634 nucleus | IDA PMID:23239344 A Novel pro-adipogenesis factor abundant in adipose tissues ... | ACCEPT | Summary: Accepted. Direct observation of nuclear localisation in the paper that first characterised the gene, independently corroborated by HPA immunofluorescence with a well-controlled antibody. Unlike the two process rows from the same paper, this one is a direct assay and IDA is the correct evidence code for it. Reason: The abstract states the localisation directly. The full text is not available (full_text_available: false) so the method is not visible to me, but the result is corroborated by an independent antibody-based dataset and by UniProt's curated SUBCELLULAR LOCATION line. One caveat is recorded as a knowledge gap rather than used against the row: at 76 aa and 7855 Da ADIRF is far below the nuclear-pore passive-diffusion limit and has no annotated NLS, so nuclear presence is the default expectation for a protein this small and carries little information. The paper says 'primarily within the nucleus', which implies enrichment rather than mere presence, and enrichment of a freely diffusible protein requires a retention mechanism that nobody has identified. Supporting Evidence: PMID:23239344 Our data demonstrated that C10orf116 is highly expressed in adipose tissue and is localized primarily within the nucleus. file:human/ADIRF/ADIRF-uniprot.txt CC -!- SUBCELLULAR LOCATION: Nucleus {ECO:0000269|PubMed:23239344}. |
| GO:0045600 positive regulation of fat cell differentiation | IDA PMID:23239344 A Novel pro-adipogenesis factor abundant in adipose tissues ... | ACCEPT | Summary: Accepted as the term, with a correction to the evidence code: the supporting experiment is over-expression, so this should be IMP rather than IDA. This is ADIRF's only characterised biological role and the one UniProt's FUNCTION line leads with, but the experiment shows that ADIRF can promote adipogenic differentiation, not that adipogenesis requires it. Reason: The abstract names the assay outright as over-expression in 3T3-L1 cells, and GO's IMP definition explicitly covers over-expression and ectopic expression of wild-type genes while IDA is reserved for a direct assay of the gene product's own activity or location. The mis-coding is therefore decidable from the abstract alone and needs no full text. I did not change the term, which is well supported. A second boundary on the interpretation, measured rather than assumed: 3T3-L1 is a mouse line, and ADIRF is absent from the entire Muroidea clade - 0 genes in mouse, rat and Muroidea in NCBI Gene against positive controls of 1, 1 and 30 for ADIPOQ, with the sister rodent clade Sciuridae retaining 9 ADIRF genes so the loss is localised rather than a query artefact. PMID:31945134 identifies the mechanism as a 43-kb deletion removing the promoter and first exon. So the only functional experiments on this gene are gain-of-function in a background with no endogenous orthologue, murine adipogenesis demonstrably proceeds without any ADIRF, and no loss-of-function experiment exists in any system that has the gene. That does not make the term wrong - it bounds what it means - so the requirement claim is filed as a knowledge gap rather than asserted or used to reject the row. Supporting Evidence: PMID:23239344 Over-expression studies in 3T3-L1 cells indicated that it up-regulates the levels of CCAAT/enhancer binding protein Ξ± (C/EBPΞ±) and PPARΞ³ and promotes adipogenic differentiation starting from the early stage of adipogenesis. PMID:31945134 The mouse locus corresponding to human ADIRF harbors a deletion of close to 43 kb (Fig 5A). This deletion is predicted to remove the promoter and first exon of ADIRF, a sequence that encodes the first 42 amino acids of the 76 amino acid gene product. file:human/ADIRF/ADIRF-bioinformatics/RESULTS.md ADIRF absent from Muroidea: **True**; retained in the sister rodent clade Sciuridae: **True** (9 genes). Knowledge gap: It is unknown whether ADIRF is required for adipocyte differentiation in any cell that expresses it endogenously. Every functional experiment on the gene is ectopic over-expression in mouse 3T3-L1 preadipocytes, and the mouse lineage carries a 43-kb deletion of the ADIRF locus, so the assay system has no endogenous orthologue and murine adipogenesis proceeds without any ADIRF. OPEN BIOLOGYCURATION RESIDUAL_SUBGAP "Over-expression studies in 3T3-L1 cells indicated that it up-regulates the levels of CCAAT/enhancer binding protein Ξ± (C/EBPΞ±) and PPARΞ³ and promotes adipogenic differentiation starting from the early stage of adipogenesis." β PMID:23239344 "deletion of this segment (and loss of ADIRF expression) occurred in the evolutionary predecessor to mouse, rat, and hamster." β PMID:31945134 |
| GO:0045944 positive regulation of transcription by RNA polymerase II | IDA PMID:23239344 A Novel pro-adipogenesis factor abundant in adipose tissues ... | KEEP AS NON CORE | Summary: Kept as non-core. The term is the UniProt curator's reading of the full text, which I do not overrule, but it rests on the same single over-expression experiment as the GO:0045600 row - so it is a mechanistic interpretation of one result rather than an independent finding, and the PAN-GO curators reviewing the same paper selected GO:0045600 and not this term as ADIRF's core process. Reason: What the abstract reports is a change in the levels of C/EBP-alpha and PPAR-gamma, which does not by itself localise the effect to RNA polymerase II transcription. UniProt's curator, who read the full text I cannot see, wrote that ADIRF 'stimulates transcription initiation' of these factors, and per project policy an experimental call made from the full text is not overruled from an abstract - so the term stands and this is not a MODIFY. What keeps it non-core is that no mechanism has been established: ADIRF has no annotated DNA-binding activity, no measured promoter occupancy, and no identified cofactor, so whether it acts at Pol II transcription directly or several steps upstream is unknown. The evidence code has the same problem as the GO:0045600 row - the supporting experiment is over-expression, which GO codes as IMP, not IDA. Supporting Evidence: PMID:23239344 Over-expression studies in 3T3-L1 cells indicated that it up-regulates the levels of CCAAT/enhancer binding protein Ξ± (C/EBPΞ±) and PPARΞ³ and promotes adipogenic differentiation starting from the early stage of adipogenesis. PMID:23239344 C10orf16 manifested the characteristics of an adipocyte lineage-specific nuclear factor that can modulate the master adipogenesis transcription factors early during differentiation. Knowledge gap: The mechanism by which ADIRF increases C/EBP-alpha and PPAR-gamma expression is unknown. No DNA-binding activity, promoter occupancy, chromatin association or transcriptional cofactor has been reported for the protein, so it is not known whether it acts at transcription at all or at some earlier step whose output is a change in the levels of these factors. OPEN BIOLOGY MF_DARK "C10orf16 manifested the characteristics of an adipocyte lineage-specific nuclear factor that can modulate the master adipogenesis transcription factors early during differentiation." β PMID:23239344 |
| GO:0005829 cytosol | IDA GO_REF:0000052 | NEW | Summary: Proposed as missing. The Human Protein Atlas gives ADIRF TWO main subcellular locations, Nucleoplasm and Cytosol, from the same immunofluorescence experiment at reliability 'Supported', but GOA imported only the nucleoplasm half. GO:0005829 is absent from ADIRF's entire GOA record. Reason: This is a gap in the HPA import rather than a disagreement with it, and it was verified against a positive control so that a zero cannot be a broken query: GAPDH is also called Cytosol by HPA and does carry GO:0005829 in GOA, with the term id resolved through the same location-to-term mapping used for ADIRF. GO:0005829 is not a descendant of GO:0005634, so this adds genuinely new information rather than restating the nuclear rows. This row rests on the HPA immunofluorescence call alone, which is sufficient for it. It deliberately does NOT lean on the sub-diffusion-limit argument: at 7855 Da ADIRF is below the nuclear-pore passive-diffusion limit, but that predicts BOTH compartments and so discounts both equally - it cannot be evidence for one of them. That argument therefore appears only in the knowledge gap asking whether nuclear enrichment is active. Proposed as located_in only - no claim is made that any ADIRF function is executed in the cytosol, and cytosol is deliberately absent from core_functions locations for that reason. Supporting Evidence: file:human/ADIRF/ADIRF-bioinformatics/RESULTS.md HPA record for ADIRF (ENSG00000148671), IF reliability **Supported**, main subcellular locations **Nucleoplasm, Cytosol**. file:human/ADIRF/ADIRF-bioinformatics/RESULTS.md Terms expected from HPA's main locations: GO:0005654, GO:0005829. Missing from ADIRF's GOA record entirely: **GO:0005829**. |
Loading supporting contentβ¦
Download this section (compressed HTML)Q: Should the InterPro2GO mapping from IPR034450, or the signature itself, be restricted? Measured with a fully paginated QuickGO query (withFrom=InterPro:IPR034450, 1512 annotations, completeness asserted): GO:0045600 positive regulation of fat cell differentiation reaches 723 gene products, of which only 130 are ADIRF-sized (60-90 aa) while 504 exceed 200 aa and 237 are non-vertebrate metazoans - among them a 2304-aa Toxocara canis protein, a 1578-aa Melipona quadrifasciata protein and nine Mizuhopecten yessoensis proteins. The cause is that human ADIRF is a 76-aa low-complexity sequence whose three commonest residues are 43.4% of it, and PTHR39227 accordingly contains 556 members over 200 aa including 39 fungal, 16 bacterial and 15 plant entries. Aligning human ADIRF against five genuine orthologues and twelve oversized recipients separates them cleanly on coverage: orthologues align over 75-76 of 76 residues, oversized members over 23-65, and the oversized members are tandem-repeat proteins with 11/22/33-residue periods. Note this is a signature-specific problem, not a general indictment of InterPro2GO: the mapping consists of exactly two terms and both are correct for the gene the entry was built from. There is a reciprocal half worth fixing at the same time: the family has 50 teleost members in UniProtKB and not one of them is ADIRF-sized, so its entire fish content is oversized spurious matches, while the genuine teleost ADIRF proteins - for example NP_001373520.1 in Danio rerio, 81 aa, aligning over 71 of the 76 human residues - are annotated in RefSeq and absent from the family. The signature is thus simultaneously over-inclusive of unrelated repeat proteins and under-inclusive of real orthologues.
Suggested experts: InterPro curators, GO Central
Q: Should UniProtKB:A0A5F8H3S4 be removed from PANTHER subfamily PTHR39227:SF1, and is the Monodelphis domestica proteome missing its real ADIRF? PANTHER node PTN008674116 propagates both of human ADIRF's IBA terms - including is_active_in GO:0005634, which asserts the molecular function is executed there - to 7 gene products. Six are genuine orthologues of 75-76 aa aligning over 73-76 of the 76 human residues at 81.6-100% identity. The seventh, A0A5F8H3S4, is a 447-aa 'Uncharacterized protein' built from a 22-residue tandem repeat at 85.4% periodicity whose three commonest residues are 52.1% of the sequence; it aligns over only 62 of 76 residues at 15.8% identity, inside the same band as the unambiguously spurious IPR034450 matches. Importantly the root cause looks to be upstream of PAINT rather than in the tree curation: NCBI Gene has a real 3-exon ADIRF gene for this species (GeneID 100020286, ADIPOQ control non-zero for the same taxon), but UniProt's Monodelphis proteome contains no ADIRF-sized member of the family, so the only candidate the HMM had to match was the repeat protein. The actionable items are therefore (a) whether the family HMM should admit a 447-aa tandem-repeat protein at all, and (b) whether the missing opossum ADIRF protein should be added to the reference proteome. Stated as a question rather than a finding about PAINT, whose placement of the six real orthologues is correct.
Suggested experts: PANTHER curators, GO Central, UniProt proteomes
Q: Is GO:0045600's taxon constraint of only_in_taxon NCBITaxon:6072 (Eumetazoa) intended to admit invertebrates? The constraint machinery is demonstrably working - of the 723 recipients of GO:0045600 via IPR034450, zero are outside Metazoa, so the fungal, plant and bacterial family members are correctly excluded, and GO:0005634's Eukaryota constraint likewise excludes the bacterial and archaeal ones. But 237 recipients are invertebrate metazoans, and they pass the constraint. GO:0045444's definition is 'The process in which a relatively unspecialized cell acquires specialized features of an adipocyte, an animal connective tissue cell specialized for the synthesis and storage of fat' - so the question is whether an insect fat body cell or a mollusc storage cell satisfies that differentia. If not, the constraint may want tightening toward Vertebrata. Raised as a question because the answer determines whether the invertebrate annotations are a constraint problem or purely a signature problem.
Suggested experts: GO Central, GO taxon constraint working group
Q: Should the two PMID:23239344 process rows (GO:0045600 and GO:0045944) be recoded from IDA to IMP? The abstract states the assay as 'Over-expression studies in 3T3-L1 cells', and GO's IMP definition explicitly covers over-expression and ectopic expression of wild-type genes, whereas IDA is for a direct assay of the gene product's own activity or location. The GO:0005634 row from the same paper is a genuine direct observation and should stay IDA. Flagged rather than acted on because the review schema records term-level actions, not evidence-code corrections.
Suggested experts: UniProt curators, GO Central
Q: Is is_active_in the right qualifier for GO_Central's GO:0005634 IBA on a gene with no known molecular function? is_active_in asserts that the molecular function is executed in the nucleus, but ADIRF has no measured molecular function of any kind, and the Human Protein Atlas gives it two main locations - Nucleoplasm and Cytosol - from one immunofluorescence experiment at reliability Supported. At 7855 Da the protein is well below the nuclear-pore passive-diffusion limit and has no annotated NLS, so dual distribution is the physical expectation. The row was accepted as a deliberate core-function judgement by curators who read the same literature; the question is whether located_in would be the safer qualifier until a molecular function is known.
Suggested experts: GO Central, PAN-GO curators
Q: Should GOA import the cytosol half of HPA's ADIRF immunofluorescence call? HPA gives Nucleoplasm and Cytosol as co-equal main locations for ENSG00000148671 (antibody HPA026810, IF reliability Supported), but the GO_REF:0000052 row carries only GO:0005654, and GO:0005829 is absent from ADIRF's whole GOA record. Verified against a positive control - GAPDH is also called Cytosol by HPA and does carry GO:0005829 - so this is a gap in the import rather than a broken query. Proposed here as a NEW row.
Suggested experts: HPA curators, GOA curators
Q: Should UniProt entry Q15847 drop or qualify its two keyword-derived GO cross-references? The entry carries 'DR GO; GO:0030154; P:cell differentiation; IEA:UniProtKB-KW.' and 'DR GO; GO:0006351; P:DNA-templated transcription; IEA:UniProtKB-KW.', neither of which is entailed by the gene's evidence. I fetched the relations rather than inferring them from labels: GO:0045944 is not a descendant of GO:0006351, and GO:0045600 is a descendant of neither GO:0030154 nor GO:0045444, because GO deliberately keeps regulation out of the is_a hierarchy. GO:0006351 is 'The synthesis of an RNA transcript from a DNA template', i.e. it says ADIRF performs transcription rather than regulating it. GOA no longer imports the Swiss-Prot keyword route - GO_REF:0000043 returns 0 human annotations against 139,714 for GO_REF:0000044 - so these claims are now invisible from GOA and there is no GO row to correct; it is a UniProt-side request only.
Suggested experts: UniProt curators
Q: Should UniProt entry Q15847 add the arterial expression data and the two missing primary references? Its TISSUE SPECIFICITY line leads with adipose tissue, but PMID:31945134 reports over 700 RKMP in tibial artery in GTEx and states the gene is more abundant in arterial than adipose tissue. That paper is also the only source for the 43-kb Muroid deletion that explains why the gene has no mouse orthologue - highly relevant to anyone planning experiments - and PMID:33737617 provides the only quantitative measurement of the protein in a body fluid. Neither is in the entry's reference list.
Suggested experts: UniProt curators
Q: Is ADIRF's real biology vascular rather than adipose, and if so what accounts for the adipose framing? The gene's approved name, its InterPro entry name, its PANTHER family name and all eleven of its GOA rows derive from its discovery as the second most abundant transcript in adipose tissue, but expression is higher in artery, and the one study that examined vascular smooth muscle found it there while noting its complete absence in mouse. No GO action is proposed because no vascular function has been assayed.
Suggested experts: vascular smooth muscle biologists, adipose tissue biologists
Experiment: CRISPR knockout and inducible degron-tagged depletion of ADIRF in a human preadipocyte model that expresses it endogenously (for example SGBS cells or primary human adipose-derived stromal cells), scoring differentiation by lipid accumulation and by PPARG and CEBPA induction. This is the experiment the entire existing literature lacks: every published functional result is ectopic expression in mouse 3T3-L1 cells, and the mouse lineage carries a 43-kb deletion of the ADIRF locus, so no loss-of-function has ever been tested in a background that has the gene. Reconstitution with the human protein should rescue.
Hypothesis: ADIRF is required for, and not merely sufficient to accelerate, adipocyte differentiation in cells that express it endogenously.
Type: loss-of-function genetics
Experiment: Proximity-dependent biotinylation (TurboID or BioID2) with ADIRF fused at each terminus, performed in parallel in differentiating human preadipocytes and in primary human vascular smooth muscle cells, with a size-matched inert bait as control. Because ADIRF is only 76 residues, run a free-tag control to subtract compartment background, and compare nuclear and cytosolic fractions separately. This directly targets the gene's largest gap - no molecular function of any kind is known - and the vascular arm addresses the tissue where the gene is most abundant but wholly unannotated.
Hypothesis: ADIRF acts through a protein partner rather than through an intrinsic biochemical activity.
Type: proximity labelling proteomics
Experiment: CUT&RUN or ChIP-seq for endogenous ADIRF (or a knock-in epitope tag to avoid over-expression artefacts) in differentiating human preadipocytes, with matched input and an IgG control, asking whether the protein occupies the PPARG and CEBPA regulatory regions. A negative result would be informative: it would argue that ADIRF acts upstream of transcription rather than at it, and would justify generalising GO:0045944 to a term that does not assert action at Pol II.
Hypothesis: ADIRF associates with chromatin at the PPARG and CEBPA loci, as its GO:0045944 annotation implies.
Type: chromatin occupancy
Experiment: Compare the nucleocytoplasmic distribution of ADIRF-GFP with ADIRF fused to a tandem multimeric tag that raises the fusion above the passive-diffusion limit (for example ADIRF-3xGFP, roughly 90 kDa), alongside free GFP and free 3xGFP controls, and measure exchange by fluorescence loss in photobleaching. If nuclear enrichment persists for the oversized fusion, retention is active and a retention factor exists to be found; if it collapses, the observed distribution is equilibration and the GO nuclear annotations describe an accessible compartment rather than a functional site.
Hypothesis: Nuclear enrichment of ADIRF is active rather than a consequence of free diffusion through the nuclear pore.
Type: live-cell imaging
Experiment: Fractionate conditioned medium from human adipocytes and vascular smooth muscle cells by size-exclusion chromatography and density gradient, and ask whether ADIRF co-migrates with CD9/CD63/TSG101-positive vesicles or with the free protein fraction; test protease protection with and without detergent to distinguish luminal cargo from surface-associated protein; and test brefeldin A insensitivity to exclude the classical pathway, which the absence of a signal peptide already predicts. Include a cell-death marker such as LDH release as the control that discriminates secretion from lysis. This decides whether GO:0070062 or a generic extracellular-region annotation is the supportable term.
Hypothesis: ADIRF reaches serum by unconventional secretion or in extracellular vesicles, not by release from damaged cells.
Type: secretion and vesicle biochemistry
Experiment: Test the interaction orthogonally: reciprocal co-immunoprecipitation of endogenous ADIRF and pro-IL18 from a cell type expressing both, in situ proximity ligation assay, and NanoBiT complementation with a non-interacting pair as negative control. Because pro-IL18 is cytosolic and ADIRF is cytosolic and nucleoplasmic the interaction is topologically possible, so this is a genuine open question rather than an implausible one - but the current annotation rests on a single screen logged under three assay versions and has never been replicated by any other method.
Hypothesis: The ADIRF-IL18 interaction reported by the HuRI two-hybrid screen occurs at native expression levels.
Type: interaction validation
What is not known β curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: ADIRF has no known molecular function of any kind. It is a 76-residue protein whose UniProt feature table contains a single CHAIN 1..76 and nothing else - no domain, no signal peptide, no transmembrane segment, no active or binding site - with no resolved experimental structure, and no catalytic, nucleic-acid-binding or cofactor-binding activity has ever been measured.
OPEN BIOLOGY MF_DARK
What is known: Established: subcellular distribution (nucleoplasm and cytosol), a pro-adipogenic effect on ectopic expression, and one two-hybrid interaction with IL18. Not established: anything about what the polypeptide itself does at the molecular level.
Significance: This is why core_functions asserts a biological process and locations but no molecular_function or contributes_to_molecular_function, and why the only MF row in GOA is the uninformative GO:0005515. It also means there is no fold from which an activity could have been mis-inferred - the campaign's usual fold-to-activity propagation lead has nothing to find on this gene.
Provenance (the field's own admissions):
Gap: The route by which ADIRF reaches the extracellular space is unknown. The protein is measured in human serum at microgram-per-millilitre concentrations, detected as an intact species in lipoaspirate fluid, and recovered from two independent urinary exosome proteomes - yet it has no signal peptide and no transmembrane segment, so classical secretion is unavailable to it.
OPEN BIOLOGY CC_DARK
What is known: Established: the protein is present in serum, lipoaspirate fluid and exosome preparations. Not established: whether it is packaged into extracellular vesicles, released by an unconventional secretion pathway, or liberated from damaged cells; and whether serum ADIRF is vesicle-associated or free.
Significance: Determines whether GO:0070062 is the right extracellular term or whether a generic extracellular-region annotation is all the data support. The two existing GO:0070062 rows already entail GO:0005576, so no additional term was proposed; resolving this gap is what would justify keeping or refining them.
Provenance (the field's own admissions):
Gap: ADIRF's function in vascular smooth muscle is entirely uncharacterised, despite artery being the tissue in which the gene is most abundant. Its GO record contains no vascular annotation of any kind.
OPEN BIOLOGYCURATION BP_DARK
What is known: Established: high arterial expression, and immunohistochemical detection in human vascular and non-vascular smooth muscle. Not established: any function, partner or process in that tissue.
Significance: The gene's name, its InterPro entry name, its PANTHER family name and all eleven of its GOA rows derive from the tissue where it was first found rather than the tissue where it is most abundant. No GO action is proposed because no functional vascular experiment exists to annotate; this is recorded so the asymmetry is visible to the next curator.
Provenance (the field's own admissions):
Gap: Whether ADIRF is actively retained in the nucleus is unknown. At 76 residues and 7855 Da it is far below the nuclear-pore passive-diffusion limit and has no annotated nuclear localisation signal, so its presence in the nucleus is the default expectation and carries little information; the reported enrichment there would require a retention mechanism that nobody has identified.
OPEN BIOLOGY CC_DARK
What is known: Established: the protein is detected in nucleoplasm and in cytosol by immunofluorescence, and described as localised primarily within the nucleus. Not established: whether nuclear accumulation is active, what would retain it, or whether the two compartments simply reflect free equilibration.
Significance: Bears directly on whether GO_Central's is_active_in GO:0005634 qualifier is the right one for this gene, which is raised as a question rather than acted on, and on why cytosol is proposed as located_in only and kept out of core_functions locations.
Provenance (the field's own admissions):
Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)