AGFG1 is a two-part protein: an N-terminal ArfGAP domain (residues 11-135) built around a C4-type zinc finger, followed by roughly 430 residues of intrinsically disordered, low-complexity sequence carrying ten phenylalanine-glycine (FG) repeats of the kind found in nucleoporins, three FxxFxxF motifs that bind AP-2 appendage domains, and Asn-Pro-Phe (NPF) motifs that bind EH domains. The disordered tail is where its best-characterised activity lies: a 40-residue stretch immediately after the ArfGAP domain wraps around the folded longin domain of the R-SNARE VAMP7 and selects it as cargo for clathrin-mediated endocytosis, so AGFG1 is a cargo-selective clathrin adaptor (a CLASP) that retrieves post-fusion VAMP7 from the plasma membrane. Because VAMP7's own SNARE helix competes for the same groove, AGFG1 engages VAMP7 only once it is part of a cis-SNARE complex, which couples retrieval to prior fusion. The same NPF and FxxFxxF motifs place AGFG1 in the endocytic coat alongside EPS15, EPS15R, intersectin, FCHO1 and AP-2. In spermiogenesis the protein coats Golgi-derived proacrosomal vesicles and is required for their docking and fusion into a single acrosomic vesicle; loss of the mouse orthologue blocks acrosome formation, yields round-headed spermatozoa and male infertility, and leaves the acroplaxome deficient in keratin 5. AGFG1 is also the cellular cofactor, historically named hRIP/RIP/Rab, that HIV-1 Rev requires to move RRE-containing viral RNAs from the nuclear periphery into the cytoplasm; that role is specific to Rev-directed RNA, since bulk host mRNA distribution is unaffected when AGFG1 is ablated, and the Rev contact is indirect, bridged by CRM1 through the FG repeats. The protein is ubiquitously expressed and distributes between the nucleus and cytoplasmic vesicles including clathrin-coated pits and vesicles. Its ArfGAP domain is structurally genuine - the four zinc-coordinating cysteines hold a zinc ion in both solved structures and the arginine finger is present at Arg57 - but two of the three residues required for GTP hydrolysis are substituted, as they are throughout the AGFG proteins examined, so the domain is predicted to bind Arf GTPases without efficiently accelerating their hydrolysis. A family-level review states that some member of every ArfGAP subfamily except ADAP has shown in-vitro GAP activity, but names no AGFG protein or assay; the AGFG-specific section reports none. Together with direct human ARF proximity, the sequence evidence is therefore most consistent with AGFG1 acting as an Arf-binding effector with substantially reduced GAP activity, pending a direct catalytic assay.
Definition: The directed movement of intron-retaining viral RNA from the host cell nucleus to the cytoplasm, mediated by a viral RNA-binding adaptor protein such as HIV-1 Rev acting on a cis-acting RNA element, in cooperation with host nuclear export machinery.
Justification: GO's only viral nuclear-RNA-export term, GO:0046784 viral mRNA export from host cell nucleus, specifies intronless viral mRNA in its definition and in all four of its synonyms ("intronless viral mRNA transport from nucleus to cytoplasm", "intronless viral mRNA export out of nucleus", "intronless viral mRNA export from host nucleus", "intronless viral mRNA-nucleus export"). That differentia excludes the Rev/RRE pathway, which exists precisely to export unspliced and partly spliced, intron-retaining RNA past the host's splicing-dependent export machinery. The consequence is visible in this gene's record: AGFG1's role, established over three papers, is currently carried either by a host term that its own data contradict (GO:0006406 mRNA export from nucleus, which this review moves away from) or by a general viral-transport parent (GO:0075733). Either a new sibling of GO:0046784 or a relaxation of GO:0046784's definition to cover intron-retaining viral RNA would let this and the other Rev cofactors be annotated for what they do. The same gap applies to HTLV-1 Rex, whose activation domain AGFG1 also binds.
Parent term: viral gene expression
Supporting Evidence:
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005737 cytoplasm | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: Kept as non-core. AGFG1 is genuinely cytoplasmic and the propagation chain is sound, but cytoplasm is the least informative of this protein's locations - the ones that carry information are the clathrin-coated pit and vesicle, proposed separately. Reason: Every donor token was resolved. FB:FBgn0020304 is Drosophila drongo, which has no Swiss-Prot entry, so seven TrEMBL accessions were used and reported as such; the longest, E1JHR0, holds GO:0005737 by its own IDA. MGI:MGI:1333754 is mouse Agfg1 (Q8K2K6, Swiss-Prot, 561 aa) whose experimental evidence in this branch is a cytoplasmic-vesicle IDA rather than one at cytoplasm itself. UniProtKB:P52594 is AGFG1 itself, so this row is partly self-referential: that records a PAINT curator judging the location core for this gene and is valid, not circular. The same WITH/FROM field appears verbatim on the paralogue AGFG2, where P52594 makes it paralogue-derived instead - the same bytes with a different evidential status. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: PANTHER:PTN002353603 · PANTHER tree node, not a protein NOT RELEVANT An internal tree node; it is the inference structure, not an evidence source. FB:FBgn0020304 · Drosophila melanogaster drongo (no Swiss-Prot entry; E1JHR0 and six other TrEMBL accessions) SUPPORTS TRANSFER Holds GO:0005737 by IDA from PMID:27654348. An unreviewed entry is weaker support and that is stated rather than hidden. MGI:MGI:1333754 · Mus musculus Agfg1 (Q8K2K6, Swiss-Prot, 561 aa) SUPPORTS TRANSFER Experimental evidence in this branch is GO:0031410 IDA (PMID:11711676), i.e. at a descendant rather than at cytoplasm itself. UniProtKB:P52594 · AGFG1 itself - self-referential IBA SUPPORTS TRANSFER A self-referential WITH/FROM records the curator's own judgement for this gene; it is valid and is not circular propagation. Supporting Evidence: file:human/AGFG1/AGFG1-uniprot.txt CC -!- SUBCELLULAR LOCATION: Nucleus PMID:10613896 Hrb was present both in the nucleus and in extranuclear compartments (Fig. 4 A). |
| GO:0031410 cytoplasmic vesicle | IBA GO_REF:0000033 | ACCEPT | Summary: Accepted. This is where AGFG1 works, in both of its characterised roles: on clathrin-coated endocytic vesicles in somatic cells and on Golgi-derived proacrosomal vesicles in spermatids. Reason: The IBA lands on exactly the term its donor holds, so no downward correction is warranted: mouse Agfg1 carries GO:0031410 by IDA from PMID:11711676, which places the protein on the cytosolic surface of proacrosomal transport vesicles. AGFG1's own human EXP row from PMID:10613896 agrees independently, and PMID:18775314 refines the compartment to clathrin-coated pits and vesicles - proposed as separate, more specific rows rather than by modifying this one, since the general term is true of both the somatic and the spermatid pools. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002919572 · PANTHER tree node covering both the AGFG1 and AGFG2 subfamilies NOT RELEVANT A tree node rather than an evidence source. Its reach is discussed on the GO:0001675 row, where it matters. MGI:MGI:1333754 · Mus musculus Agfg1 (Q8K2K6, Swiss-Prot) SUPPORTS TRANSFER Holds GO:0031410 by IDA from PMID:11711676 - the same term, not a more specific one, so the propagation does not land above its donor. UniProtKB:P52594 · AGFG1 itself - self-referential IBA SUPPORTS TRANSFER Valid; and note the identical field on AGFG2 is paralogue-derived. Supporting Evidence: PMID:11711676 In wild-type spermatids, Hrb is associated with the cytosolic surface of proacrosomic transport vesicles that fuse to create a single large acrosomic vesicle at step 3 of spermiogenesis. PMID:18819912 In the cytoplasm, HRB colocalized with clathrin-, AP-2-, EPS15-, and transferrin receptor-containing vesicles. |
| GO:0001675 acrosome assembly | IBA GO_REF:0000033 | ACCEPT | Summary: Accepted for AGFG1, which is the true orthologue of the mouse gene whose null phenotype defines this term. The defect on this row is not the term and not the evidence - it is that the node also carries it to the paralogue AGFG2, which has no supporting evidence in any species. Reason: The donor is mouse Agfg1 (Q8K2K6), which holds this term by two independent IMPs: the null is male-sterile with round-headed, acrosome-less spermatozoa and proacrosomal vesicles that form but cannot fuse, and a second study shows the mutant builds only a flattened pseudoacrosome. UniProt records the same conclusion for human AGFG1 by similarity. So for AGFG1 this is an orthologue transfer from a decisive phenotype and it stands. The node, however, is PTN002919572, whose 36 recipients were enumerated: 21 AGFG1-type, 12 AGFG2-type and 3 unnamed gene products across 17 species, and whose human reach is exactly AGFG1 plus AGFG2. Mouse and rat Agfg2 are themselves IBA recipients, so no AGFG2 orthologue anywhere carries experimental evidence for this term, while human AGFG2's own characterised function is von Willebrand factor secretion from endothelial Weibel-Palade bodies. The term belongs at the AGFG1-orthologue node (PTHR46134:SF1), not at the shared ancestral node; that recommendation is filed once in suggested_questions with both genes named rather than repeated on each affected row. Propagation Review Root cause: NO FAILURE CORE Sources checked: MGI:MGI:1333754 · Mus musculus Agfg1 (Q8K2K6, Swiss-Prot) SUPPORTS TRANSFER Two independent IMPs at the exact term, PMID:11711676 and PMID:14724135. PANTHER:PTN002919572 · PANTHER tree node whose human reach is exactly AGFG1 + AGFG2 SUPPORTS SOURCE BUT NOT TARGET Sound for AGFG1; the same node carries the term to 12 AGFG2-named recipients with no AGFG2 evidence in any species. The comment records that asymmetry; the row-level verdict for AGFG1 is unaffected. Supporting Evidence: PMID:11711676 Here we report that male mice with a null mutation in Hrb are infertile and display round-headed spermatozoa that lack an acrosome. PMID:11711676 We conclude that Hrb is required for docking and/or fusion of proacrosomic vesicles during acrosome biogenesis. file:human/AGFG1/AGFG1-uniprot.txt CC -!- FUNCTION: Required for vesicle docking or fusion during acrosome file:human/AGFG1/AGFG1-bioinformatics/RESULTS.md | `GO:0001675` acrosome assembly | 121 | 5 | `PTN002919572` | 36 | **AGFG1 + AGFG2** | |
| GO:0007289 spermatid nucleus differentiation | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: Kept as non-core. The mouse mutant does have a spermatid nuclear phenotype, but it is a consequence of failed acrosome biogenesis and of a keratin-deficient acroplaxome rather than an independent function - and the single reference attached to the donor annotation reports a negative result for this gene. Reason: The sole donor evidence is one mouse IMP citing PMID:16765935, and that paper's result for hrb is that H1T2 polar localisation is unaltered in hrb-null spermatids, which the authors use to argue that acrosome development is not required for nuclear polarity. The phenotype that does support the term - round spermatid nuclei with protrusions the authors name nucleopodes - is reported in PMID:14724135, and the same paper attributes it to an acroplaxome that lacks keratin 5 and therefore cannot withstand the tension of nuclear shaping. So the term is defensible for the orthologue but is downstream of acrosome assembly and is attached to the wrong reference. Per the campaign's rule the row is not punished for a citation defect that lives upstream; the reference problem is filed as a question for MGI/GO Central instead. As with the other two rows from this node, the human paralogue AGFG2 receives it with no evidence of its own. Propagation Review Root cause: NO FAILURE NON CORE Failure modes: SOURCE MISCITATION Sources checked: MGI:MGI:1333754 · Mus musculus Agfg1 (Q8K2K6, Swiss-Prot) SUPPORTS TRANSFER Holds GO:0007289 by IMP, but the cited paper (PMID:16765935) reports no change in hrb-null spermatids for the marker it assayed; PMID:14724135 carries the supporting phenotype. PANTHER:PTN002919572 · PANTHER tree node whose human reach is exactly AGFG1 + AGFG2 SUPPORTS SOURCE BUT NOT TARGET Same node-reach asymmetry as the acrosome-assembly row. Supporting Evidence: PMID:16765935 In acrosome-deficient round spermatids of hrb -/- and gopc -/- mice, H1T2 localization is not altered, indicating that proper acrosome development is not required for specifying nuclear polarity. PMID:14724135 As spermiogenesis advances, round-shaped spermatid nuclei of the mutant display several nuclear protrusions, designated nucleopodes. |
| GO:0045109 intermediate filament organization | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: Kept as non-core. Better supported than it looks: the mouse mutant's acroplaxome, a cytoskeletal plate, retains F-actin but is deficient in keratin 5, an intermediate-filament protein. It remains a specific consequence of failed acrosome biogenesis rather than a general cytoskeletal function. Reason: The donor's IMP cites PMID:14724135, and that paper does support the term at the level GO states it: the keratin-5 deficiency of the mutant acroplaxome is an intermediate-filament organisation defect, and the authors use it to explain the nuclear protrusions. Nothing about a broader role in intermediate-filament organisation follows, and there is no human evidence, so the row is kept out of the core set. AGFG1's node contributes 36 of the 1124 IBA annotations to this term, and once again half of that reach is the AGFG2 subfamily. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: MGI:MGI:1333754 · Mus musculus Agfg1 (Q8K2K6, Swiss-Prot) SUPPORTS TRANSFER IMP at the exact term from PMID:14724135, whose keratin-5 result supports it directly. PANTHER:PTN002919572 · PANTHER tree node whose human reach is exactly AGFG1 + AGFG2 SUPPORTS SOURCE BUT NOT TARGET Same node-reach asymmetry as the acrosome-assembly row. Supporting Evidence: PMID:14724135 Yet the acroplaxome in the mutant contains F-actin but is deficient in keratin 5. PMID:14724135 We suggest that nucleopodes develop at a site where a keratin 5-deficient acroplaxome may not withstand tension forces operating during spermatid nuclear shaping. |
| GO:0005096 GTPase activator activity | IEA GO_REF:0000002 | MARK AS OVER ANNOTATED | Summary: Over-annotated. The zinc finger and the arginine finger are intact, but the field identifies three residues as catalytically required and AGFG1 retains only one: it has Tyr39 where a tryptophan is needed and Thr71 where the aspartate that contacts the Arf catalytic glutamine is needed. Every GAP-competent ArfGAP tested keeps all three. Direct human proximity data place AGFG1 with ARF1, ARF3 and ARF6, supporting retained Arf binding but not catalysis. Although a family review makes an uncited subfamily-level statement that encompasses AGFG, it names no AGFG protein or assay and its AGFG-specific section reports none; the combined evidence therefore predicts substantially reduced GAP activity rather than proving its complete absence. Reason: This row reverses an earlier verdict in this same review, and the reversal is the finding. The first pass tested the one residue the consensus-nomenclature paper names - the arginine finger of CX2CX16CX2CX4R - found it intact at Arg57, and reported the pseudoenzyme hypothesis as NOT confirmed. That was an artefact of testing one residue. PMID:23433073 names three positions as catalytically required (W451, R469 and D484 in ASAP3, the subfamily with a solved Arf6 complex) and states that mutating any one of them severely impairs activity: R469 is the arginine finger, D484 contacts the Arf6 catalytic glutamine Q67 and stabilises switch 2, and W451 sits in the Arf-ArfGAP interface. Measured by alignment to the ASAP3 domain, with every GAP-competent member of the panel required to recover all three before any absence is reported: ARFGAP1, ARFGAP3, ASAP1, ASAP3 and SMAP1 all score 3/3, while human AGFG1 scores 1/3 - Tyr39 for the tryptophan, Arg57 for the arginine, Thr71 for the aspartate. Human AGFG2, mouse Agfg1 and Drosophila drongo are identical in pattern, and the same alignment independently reproduces the arginine position that the motif scan gives, so two methods agree. This is the subfamily-wide loss the source paper predicts from 40 AGFG sequences, of which only two retain the aspartate and none the tryptophan. Two things the reversal does NOT do. It does not contradict the ARF-proximity result: the same paper is explicit that predicted catalytic loss "should not be confused with consequent changes in the ability to bind Arf family GTPases", and it proposes that such proteins act as Arf effectors - so binding retained plus catalysis lost is one coherent picture rather than two conflicting ones. And it does not make the fly genetics irrelevant, but it does reinterpret it: drongo's "GTPase-activating function" was assayed genetically, drongo itself scores 1/3, and the same paper's alternative - an Arf effector that antagonises an Arf-GEF without hydrolysing GTP - fits that genetics as well. MARK_AS_OVER_ANNOTATED rather than REMOVE, because the domain is genuine, Arf binding is real, and no direct assay has been run on the human protein either way. The consensus review's statement that in-vitro GAP activity has been shown for at least one member of every subfamily except ADAP raises a countervailing prior for catalysis, but it is uncited, names no species, protein or assay, and is not substantiated in that paper's AGFG-specific section; it cannot establish activity for human AGFG1. The annotation route is also specific and filable, since of AGFG1's four InterPro signatures only IPR001164, the pan-ArfGAP catalytic-domain entry, maps to GO:0005096, while the AGFG-specific family entry IPR052248 maps to nothing - so InterPro2GO is capable of restraint here and the claim comes from the one entry that cannot discriminate subfamilies. Note also that GO has merged the nine substrate-specific GAP terms into GO:0005096 (they are its secondaryIds), so there is no more specific term to fall back to. The concurrent review of the paralogue AGFG2 reached the same verdict on this identical row from the same primary source, which is recorded here because an inconsistency between two independently-produced reviews of one row would itself be a defect. Propagation Review Root cause: PROPAGATION BAD Failure modes: PSEUDO OR SUBACTIVITY LOSS FUNCTIONAL DIVERGENCE Sources checked: InterPro:IPR001164 · Arf GTPase activating protein domain - the pan-ArfGAP catalytic-domain signature, spanning 60678 proteins SUPPORTS SOURCE BUT NOT TARGET The signature is sound and AGFG1 genuinely matches it over residues 11-135, but it cannot discriminate subfamilies. It maps to GO:0005096 while the AGFG-specific family entry IPR052248 maps to nothing, so the claim arrives from the one entry that cannot see the subfamily-wide loss of two of the three catalytically required residues. Supporting Evidence: PMID:18809720 They contain a characteristic C4-type zinc finger motif and a conserved arginine that is required for activity, within a particular spacing (CX2CX16CX2CX4R). PMID:18809720 Arf GAP activity has been demonstrated in vitro for at least one member of each subfamily, with the exception of the ADAPs, which appear to lack in vitro GAP activity. PMID:18809720 The 31 predicted human ArfGAPs have been classified into 10 subfamilies, based on sequence similarities of their ArfGAP domains PMID:18809720 some ArfGAPs use their GAP domain to bind Arf without promoting GTP hydrolysis PMID:31533044 Notably, we found that the ArfGAP Drongo and its GTPase-activating function are essential for the initial detachment of the border cell cluster from the basal lamina. PMID:38606629 AGFG1 was identified with class I classical ARFs (i.e. ARF1 and ARF3), the class III ARF (ARF6), as well as ARL8B and ARL13A. file:human/AGFG1/AGFG1-bioinformatics/RESULTS.md | **AGFG1 human (P52594)** | 562 | **29, 32, 49, 52** | **Arg57** | intact | file:human/AGFG1/AGFG1-bioinformatics/RESULTS.md **12/12 panel members retain CX2CX16CX2CX4R.** PMID:23433073 Mutation of any one of these three residues leads to severe loss in Arf GAP activity PMID:23433073 Only two of the 40 AGFG sequences contain an aspartate at the position homologous to D47 in the other subfamilies PMID:23433073 The AGFG consensus also uniquely lacks W14, which we predict to play a role in hydrophobic interactions with Arfs. PMID:23433073 the ArfGAP is a very highly conserved structural domain that is predicted to have lost substantial levels of GAP activity in at least one subfamily (AGFG) PMID:23433073 These predicted changes (including complete loss, potentially) in GAP activity or its regulation should not be confused with consequent changes in the ability to bind Arf family GTPases. file:human/AGFG1/AGFG1-bioinformatics/RESULTS.md | **AGFG1 human (SUBJECT)** | Y39 | R57 | T71 | **1/3** | file:human/AGFG1/AGFG1-deep-research-affinage.md Beyond these endocytic and Rev-export roles, no further mechanistic detail has been characterized in the available corpus. |
| GO:0005634 nucleus | IEA GO_REF:0000044 | ACCEPT | Summary: Accepted. A keyword-route restatement of the same UniProt SUBCELLULAR LOCATION line that the two EXP rows record from the primary papers. Reason: SL-0191 Nucleus derives from UniProt's curated SUBCELLULAR LOCATION statement, which is itself evidenced by PubMed:10613896 and PubMed:7634337. Nuclear localisation is well established and is required by the Rev-cofactor role, in which AGFG1 sits at the nuclear periphery where Rev-directed RNAs accumulate when it is ablated. Redundant with the EXP rows but not wrong. Supporting Evidence: file:human/AGFG1/AGFG1-uniprot.txt CC -!- SUBCELLULAR LOCATION: Nucleus PMID:14701878 We find, unexpectedly, that in the absence of functional hRIP, Rev-directed RNAs mislocalize and aberrantly accumulate at the nuclear periphery, where hRIP is localized. |
| GO:0005737 cytoplasm | IEA GO_REF:0000117 | KEEP AS NON CORE | Summary: Kept as non-core, for the same reason as the IBA cytoplasm row. The ARBA rule was fetched and evaluated: it fires through a single condition set built from two FunFams named for AGFG1 itself, so this is effectively a gene-specific rule and not a fold generalisation. Reason: All 2414 condition sets of ARBA00026971 were evaluated against AGFG1's own signature and taxon profile. Exactly one fires, and it requires both FunFam 1.10.220.150:FF:000005 and FunFam 3.30.450.50:FF:000005 - both named "Arf-GAP domain and FG repeat-containing protein 1". One further set mentions IPR001164 but does not fire, because it also requires IPR051718, which AGFG1 does not have. The rule grants exactly one term, GO:0005737. So there is no rule defect here; the row is simply the least informative statement of a location the gene has by experiment. Worth recording that the second FunFam belongs to the CATH Longin-domain superfamily, which AGFG1 acquires only through a fusion-construct crystal structure (see suggested_questions) - the artefact reaches this rule but does not change what it asserts. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: ARBA:ARBA00026971 · ARBA rule; 2414 condition sets, one of which fires for AGFG1 SUPPORTS TRANSFER The firing set is a pair of FunFams named for AGFG1 itself, so the rule is gene-specific in practice. It grants only GO:0005737. Supporting Evidence: PMID:18819912 In HeLa cells, HRB localized both in the nucleus and in the cytoplasm. |
| GO:0031410 cytoplasmic vesicle | IEA GO_REF:0000044 | ACCEPT | Summary: Accepted. The keyword-route restatement of the same curated SUBCELLULAR LOCATION line that the EXP row records from PMID:10613896. Reason: SL-0088 Cytoplasmic vesicle derives from UniProt's SUBCELLULAR LOCATION statement, evidenced by PubMed:10613896. The compartment is confirmed independently by colocalisation with clathrin, AP-2, EPS15 and transferrin-receptor vesicles, and is refined by the coated-pit and coated-vesicle rows proposed below. Supporting Evidence: PMID:18819912 In the cytoplasm, HRB colocalized with clathrin-, AP-2-, EPS15-, and transferrin receptor-containing vesicles. |
| GO:0005515 protein binding | IPI PMID:18775314 Molecular basis for the sorting of the SNARE VAMP7 into endo... | MODIFY | Summary: Modified to SNARE binding. The partner is VAMP7 and the interaction is the best-characterised molecular event on this gene - mapped to residues 136-176, measured by ITC, resolved in a crystal structure, and abolished by mutations on both sides - so bare protein binding discards everything that was established. Reason: P51809 resolves to reviewed VAMP7_HUMAN at its canonical length of 220 aa, so this is not a TrEMBL or ORFeome substitution. The evidence is unusually strong for a single publication: IntAct holds seven records from this one paper, but across three orthogonal methods - yeast two-hybrid, GST pull-down and isothermal titration calorimetry - rather than three sub-methods of one screen. The binding region was narrowed to Hrb residues 136-176 by truncation, the affinity measured, the complex crystallised, and paired substitutions in either partner abolished binding. VAMP7 is an R-SNARE and GO:0000149 SNARE binding states exactly what was measured. The cargo-selection function that this binding serves is proposed separately as GO:0035615, because it is a different claim: binding a SNARE is the measurement, acting as its clathrin adaptor is the interpretation, and the latter rests additionally on the depletion phenotype. Proposed replacements: SNARE binding Supporting Evidence: PMID:18775314 A series of truncated, C-terminally His6-tagged Hrb constructs (Figure 1E) indicated that the sequence of Hrb that bound VAMP7 longin domain was between residues 136 and 176. PMID:18775314 the KD for the interaction was measured as 10.5 μM PMID:18775314 ITC (Figures 4C and 4D) and GST pull-down (Figures S2D and S2E) experiments showed that, as predicted, all the soluble mutations strongly inhibited the interaction between the two proteins |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | MARK AS OVER ANNOTATED | Summary: Over-annotated. The partner is POLE2, a DNA polymerase epsilon subunit, and the support is a single high-throughput yeast two-hybrid screen logged in IntAct as three sub-methods of the same experiment. Reason: P56282 resolves to reviewed DPOE2_HUMAN at its canonical 527 aa, so the partner identity is sound. The evidence is not. Expanding the IntAct records rather than trusting a replicate count shows three records from one publication, whose methods are two hybrid array, two hybrid prey pooling approach and validated two hybrid - the same screen recorded three ways - with an IntAct MI-score of 0.56 and no orthogonal assay. The promiscuity argument does not apply and is reported as a negative: POLE2 has 37 distinct IntAct partners against AGFG1's 37, so neither is a hub, and both proteins are nucleocytoplasmic so there is no topological objection either. The case rests on single-screen evidence alone, which is the established treatment for unreplicated HuRI rows in this corpus. Nothing in AGFG1's literature connects it to replicative DNA polymerase. Supporting Evidence: file:human/AGFG1/AGFG1-bioinformatics/RESULTS.md | POLE2 (P56282) | 3 | **1** (`PMID:25416956`) | two hybrid array + two hybrid prey pooling + validated two hybrid | 0.56 | 37 | |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | MARK AS OVER ANNOTATED | Summary: Over-annotated, for the same reason as the POLE2 row: NYAP2 is a singleton hit from one yeast two-hybrid screen counted three times. Judged separately from POLE2 because per-partner verdicts are the point, and it happens to reach the same verdict. Reason: Q9P242 resolves to reviewed NYAP2_HUMAN at its canonical 653 aa. As with POLE2, the three IntAct records come from one publication under two hybrid array, two hybrid prey pooling approach and validated two hybrid, MI-score 0.56, with no orthogonal assay. NYAP2 has only 8 distinct IntAct partners, so it is not a promiscuous hub - that arm of the argument is negative here too - and it is a neuronal phosphoinositide-3-kinase adaptor with no reported connection to endocytic cargo selection, acrosome biogenesis or nuclear export. Supporting Evidence: file:human/AGFG1/AGFG1-bioinformatics/RESULTS.md | NYAP2 (Q9P242) | 3 | **1** (`PMID:25416956`) | two hybrid array + two hybrid prey pooling + validated two hybrid | 0.56 | 8 | |
| GO:0005634 nucleus | EXP PMID:10613896 The eps15 homology (EH) domain-based interaction between eps... | ACCEPT | Summary: Accepted. Direct immunofluorescence with an affinity-purified antibody, in the paper that also establishes the EPS15 interaction. Reason: The paper images endogenous AGFG1 with an affinity-purified anti-peptide serum that recognises a single band on Western blot, and reports it in the nucleus and in punctate extranuclear structures that co-stain for EPS15. Nuclear localisation is also required by the Rev-cofactor role and is independently reported for HeLa cells. Supporting Evidence: PMID:10613896 Hrb was present both in the nucleus and in extranuclear compartments (Fig. 4 A). PMID:10613896 The extranuclear fraction of Hrb displayed a punctate morphology. |
| GO:0005634 nucleus | EXP PMID:7634337 Identification of a novel cellular cofactor for the Rev/Rex ... | ACCEPT | Summary: Accepted. The discovery paper's localisation, which UniProt curates as the source of the Nucleus SUBCELLULAR LOCATION line and which later work confirms. Reason: This is the one thing this reference is cited for that it does support - unlike its RNA-binding and mRNA-export rows, which are reviewed separately. UniProt's SUBCELLULAR LOCATION statement cites this paper together with PubMed:10613896, and the localisation has been reproduced in HeLa cells and is required by the Rev-cofactor mechanism. Supporting Evidence: file:human/AGFG1/AGFG1-uniprot.txt CC -!- SUBCELLULAR LOCATION: Nucleus PMID:18819912 In HeLa cells, HRB localized both in the nucleus and in the cytoplasm. |
| GO:0031410 cytoplasmic vesicle | EXP PMID:10613896 The eps15 homology (EH) domain-based interaction between eps... | ACCEPT | Summary: Accepted. The punctate extranuclear structures imaged in this paper are the cytoplasmic-vesicle pool, and later work identifies them as clathrin-coated pits and vesicles. Reason: Endogenous AGFG1 forms punctate extranuclear structures the majority of which co-stain for EPS15, an endocytic coat component. Two independent later studies place the same pool on clathrin- and AP-2-positive coated pits and vesicles, which is why the more specific compartments are proposed as additional rows rather than as a replacement for this one. Supporting Evidence: PMID:10613896 The extranuclear fraction of Hrb displayed a punctate morphology. PMID:18775314 However, Hrb did show extensive colocalization with clathrin and AP2 in clathrin-coated pits (CCPs) and CCVs (Figure 2B). |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8863718 | KEEP AS NON CORE | Summary: Kept as non-core. This entry and the fifteen that follow are one claim: sixteen GOA rows that differ only in the Reactome reaction id, all asserting cytosol. They are analysed here as a set and each retains its own entry so the review reconciles one-to-one with the GOA file. Reason: Reactome assigns a compartment to each participant of each reaction, so pathway membership generates one row per reaction rather than one per finding. This reaction, "AGFG1 binds VAMP7", is the only one of the sixteen that is about AGFG1; the other fifteen are generic downstream clathrin-mediated-endocytosis steps in which AGFG1 has no described role. The term itself is not wrong - AGFG1 is recruited to forming coated pits from a soluble pool - but cytosol is the least informative of its locations and is not independently curated by UniProt. Supporting Evidence: PMID:18775314 Taken together, these experiments demonstrate that Hrb functions as a cargo/clathrin adaptor in clathrin-mediated endocytosis of VAMP7 through the interaction of the VAMP7 longin domain with Hrb. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8867754 | KEEP AS NON CORE | Summary: Kept as non-core. One of the fifteen generic clathrin-pathway reactions; see the R-HSA-8863718 entry for the shared analysis. Reason: F- and N-BAR domain proteins binding the coated pit is not a step in which AGFG1 has any described role; the row exists because AGFG1 is a participant of the parent pathway. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8867756 | KEEP AS NON CORE | Summary: Kept as non-core. This is the one generic reaction of the fifteen whose description does cover AGFG1's role, since AGFG1 is a CLASP; see the R-HSA-8863718 entry for the shared analysis. Reason: "CLASP proteins and cargo are recruited to the nascent clathrin-coated pit" is a correct description of what AGFG1 does, and two independent papers classify it as a CLASP. The row is still only asserting cytosol, which is why the verdict matches the rest of the set; the informative content is captured by the proposed coated-pit and cargo-adaptor rows. Supporting Evidence: PMID:22484487 In addition, Dab2, Hrb and CALM, three tandem Asn-Pro-Phe-containing CLASPs bind the μHDs in a concentration-dependent manner (Fig. 2c). |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8868071 | KEEP AS NON CORE | Summary: Kept as non-core. Generic clathrin-pathway reaction; see the R-HSA-8863718 entry. Reason: Clathrin recruiting PIK3C2A is not a step involving AGFG1. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8868072 | KEEP AS NON CORE | Summary: Kept as non-core. Generic clathrin-pathway reaction; see the R-HSA-8863718 entry. Reason: A lipid-kinase step with no described AGFG1 involvement. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8868230 | KEEP AS NON CORE | Summary: Kept as non-core. Generic clathrin-pathway reaction; see the R-HSA-8863718 entry. Reason: SNX9-dependent actin recruitment is not a step involving AGFG1. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8868236 | KEEP AS NON CORE | Summary: Kept as non-core. Generic clathrin-pathway reaction; see the R-HSA-8863718 entry. Reason: BAR-domain recruitment of dynamin is not a step involving AGFG1. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8868648 | KEEP AS NON CORE | Summary: Kept as non-core. Generic clathrin-pathway reaction; see the R-HSA-8863718 entry. Reason: Synaptojanin phosphoinositide hydrolysis is not a step involving AGFG1. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8868651 | KEEP AS NON CORE | Summary: Kept as non-core. Generic clathrin-pathway reaction; see the R-HSA-8863718 entry. Reason: Endophilin recruitment of synaptojanins is not a step involving AGFG1. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8868658 | KEEP AS NON CORE | Summary: Kept as non-core. Generic clathrin-pathway reaction; see the R-HSA-8863718 entry. Reason: HSPA8-mediated uncoating is not a step involving AGFG1. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8868659 | KEEP AS NON CORE | Summary: Kept as non-core. Generic clathrin-pathway reaction; see the R-HSA-8863718 entry. Reason: Auxilin recruitment to the coated vesicle is not a step involving AGFG1. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8868660 | KEEP AS NON CORE | Summary: Kept as non-core. Generic clathrin-pathway reaction; see the R-HSA-8863718 entry. Reason: Auxilin recruitment of HSPA8:ATP is not a step involving AGFG1. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8868661 | KEEP AS NON CORE | Summary: Kept as non-core. Generic clathrin-pathway reaction; see the R-HSA-8863718 entry. Reason: Dynamin-mediated scission is not a step involving AGFG1. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8869438 | KEEP AS NON CORE | Summary: Kept as non-core. Generic clathrin-pathway reaction; see the R-HSA-8863718 entry. Reason: Dissociation of clathrin-associated proteins is a step AGFG1 would be released in rather than one it drives; nothing beyond cytosol is asserted. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8871193 | KEEP AS NON CORE | Summary: Kept as non-core. Generic clathrin-pathway reaction; see the R-HSA-8863718 entry. Reason: AAK1 dissociation and AP-2 mu2 dephosphorylation is not a step involving AGFG1. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8871194 | KEEP AS NON CORE | Summary: Kept as non-core. Generic clathrin-pathway reaction; see the R-HSA-8863718 entry. Reason: RAB5 and GAPVD1 binding AP-2 is not a step involving AGFG1. |
| GO:0005515 protein binding | IPI PMID:11545741 Two closely related human nuclear export factors utilize ent... | UNDECIDED | Summary: Undecided. The row's validity is not in question, but the paper cannot be read, so the informative molecular function that should replace bare protein binding cannot be identified. Reason: Q9H4D5 resolves to reviewed NXF3_HUMAN at its canonical 531 aa, so the partner identity is sound. The reference-projection test argues this is considered curation rather than pipeline output: the paper produced ten annotations over four entities - AGFG1, NXF3, NXF1 and XPO1 - with five different terms, including nuclear RNA export factor complex. IntAct holds no record of the pair, consistent with a UniProt curator reading the paper directly. But the full text is not retrievable (Europe PMC reports isOpenAccess N, inEPMC N and no PMC id), so what was assayed cannot be established. A plausible reading is that AGFG1's FG repeats were tested against NXF3, since the paper's central point is that NXF3 lacks the nuclear-pore-targeting domain that NXF1/TAP uses; that is a hypothesis, not a finding, and it is not used to pick a replacement term. Per CLAUDE.md, UNDECIDED is the correct action when the relevant publication cannot be accessed. Supporting Evidence: file:human/AGFG1/AGFG1-bioinformatics/RESULTS.md | `PMID:11545741` | 10 | 4 (AGFG1, NXF3, NXF1, XPO1) | 5 terms incl. `GO:0042272` | **coherent multi-protein curation, not a bulk import** | |
| GO:0003723 RNA binding | TAS PMID:7634337 Identification of a novel cellular cofactor for the Rev/Rex ... | REMOVE | Summary: Removed. The direction of the interaction is inverted: in the cited paper Rev binds the RNA and AGFG1 binds Rev's protein activation domain. AGFG1 has no RNA-binding module, UniProt declines the claim, and the mouse orthologue carries no such annotation. Reason: Five independent grounds, all checkable. First, the cited reference does not support the term: its own abstract states that Rab (AGFG1) binds the Rev activation domain when Rev is assembled onto its RNA target, so the RNA is bound by Rev, and the paper reports only protein-protein interactions with Rev and Rex activation domains. Second, the AGFG1-Rev contact is not even direct - the family reference states it is indirect and probably bridged by CRM1 - so AGFG1 is two steps from the RNA. Third, AGFG1 has no RNA-binding domain: its only modules are the ArfGAP domain, whose C4 zinc finger is buried, fully occupied by structural zinc and described by the family reference as architectural rather than catalytic, and the nucleoporin-like FG repeats, which are protein-interaction motifs that bind CRM1 and EH domains. Fourth, UniProt records only a hedge, "May play a role in RNA trafficking or localization", and assigns no RNA-binding keyword. Fifth, the provenance is a legacy bulk import: this reference annotates exactly one entity in all of GOA, with three rows, and the mouse orthologue Agfg1 carries zero GO:0003723 rows. A recorded Europe PMC query for direct RNA-binding evidence returned no such study - that is a statement about the query, not a proof of absence, and it is the weakest of the five grounds. This is the same shape as annotating a protein that carries a sugar as a lectin: a partner-of-an-RNA-binding-protein study read as a binding activity. Supporting Evidence: PMID:7634337 Rab binds the Rev activation domain when Rev is assembled onto its RNA target and can significantly enhance Rev activity when overexpressed. PMID:7634337 We report the identification of a novel human gene product that binds to not only the HIV-1 Rev activation domain in vitro and in vivo but also to functionally equivalent domains in other Rev and Rex proteins. PMID:18809720 The Rev–AGFG1 interaction is indirect and possibly bridged by the nuclear export receptor CRM1. PMID:18809720 The zinc finger has an architectural rather than catalytic role file:human/AGFG1/AGFG1-bioinformatics/RESULTS.md **Mouse Agfg1 carries zero `GO:0003723` rows, no `GO:0005643`, and no `GO:0006406`.** file:human/AGFG1/AGFG1-bioinformatics/RESULTS.md | `PMID:7634337` | 3 | **1** (AGFG1) | `GO:0003723`, `GO:0005634`, `GO:0006406` | legacy single-gene ProtInc import | |
| GO:0005643 nuclear pore | TAS PMID:7637788 A human nucleoporin-like protein that specifically interacts... | REMOVE | Summary: Removed. The term's definition asserts membership of the nuclear pore complex, while the cited paper claims only sequence homology with nucleoporins - and a primary paper states outright that AGFG1's status as an authentic NPC constituent is not established. Reason: GO:0005643 is defined as a protein complex providing a discrete opening in the nuclear envelope, so part_of asserts that AGFG1 is a component of the NPC. The support offered is the title claim of PMID:7637788, and its abstract makes the nature of that claim explicit: the protein "has homology with nucleoporins". The only nucleoporin-like feature is the FG repeats, which are protein-interaction motifs that also occur outside nucleoporins, and AGFG1 carries no nucleoporin signature - PANTHER places it in PTHR46134 and InterPro in the ArfGAP domain family. A primary paper states the negative directly. UniProt, curating this same paper together with PubMed:10613896, records Nucleus and Cytoplasmic vesicle and neither nuclear envelope nor nuclear pore. Provenance is the same legacy import as the RNA-binding row: this reference annotates exactly one entity in all of GOA, with this one row, and the mouse orthologue has no such annotation. Nothing is proposed in its place, because nuclear localisation is already recorded by four other rows and no finer compartment has been demonstrated. Supporting Evidence: PMID:7637788 This hRIP protein has homology with nucleoporins, a class of proteins that mediate nucleocytoplasmic transport. PMID:10613896 Second, it is not yet certain whether Hrb is an authentic constituent of the nuclear pore complex (Bogerd et al. 1995; Fritz et al. 1995). file:human/AGFG1/AGFG1-uniprot.txt CC -!- SUBCELLULAR LOCATION: Nucleus file:human/AGFG1/AGFG1-bioinformatics/RESULTS.md | `PMID:7637788` | 1 | **1** (AGFG1) | `GO:0005643` | legacy single-gene ProtInc import | |
| GO:0006406 mRNA export from nucleus | TAS PMID:7634337 Identification of a novel cellular cofactor for the Rev/Rex ... | MODIFY | Summary: Modified. The evidence is entirely Rev-dependent export of RRE-containing viral RNA, and the same laboratory later showed that bulk host mRNA distribution is unaffected - so a host-process term reads much more broadly than the data. Moved to the viral-process branch, with the term that is actually needed filed as a request. Reason: GO:0006406's definition does not exclude viral mRNA, so the row is imprecise rather than false; what makes it the wrong branch is that AGFG1's requirement is specific to Rev-directed RNA. Ablating AGFG1 leaves the distribution of cellular poly(A)+ mRNA, nuclear proteins and even NES-containing proteins unaffected while Rev-directed RNAs accumulate at the nuclear periphery. The obvious viral term cannot be used: GO:0046784 viral mRNA export from host cell nucleus specifies intronless viral mRNA in its definition and in all four of its synonyms, whereas Rev exports unspliced and partly spliced, intron-retaining RNA - the term's differentia excludes this pathway. So the replacement is GO:0075733, whose definition, "the directed movement of a virus, or part of a virus, within the host cell", is satisfied by what RNA in situ hybridisation actually measured. This is a sideways move between the host-process and viral-process branches rather than a generalisation, and it was verified as such: GO:0075733 is NOT a descendant of GO:0006406 over is_a/part_of. The missing term is filed under proposed_new_terms. Proposed replacements: intracellular transport of virus Supporting Evidence: PMID:14701878 We further show that the RNA mislocalization pattern resulting from loss of hRIP activity is highly specific to Rev function: the intracellular distribution of cellular poly(A)(+) mRNA, nuclear proteins, and, most important, NES-containing proteins, are unaffected. PMID:14701878 Thus, hRIP is an essential cellular Rev cofactor, which acts at a previously unanticipated step in HIV-1 RNA export: movement of RNAs from the nuclear periphery to the cytoplasm. PMID:15749819 We find that ablation of hRIP activity by a dominant-negative mutant or RNA interference inhibits virus production by mislocalizing Rev-directed RNAs to the nuclear periphery. |
| GO:0035615 clathrin-cargo adaptor activity | IMP PMID:18775314 Molecular basis for the sorting of the SNARE VAMP7 into endo... | NEW | Summary: Proposed. This is AGFG1's characterised molecular function and GOA records nothing like it: the protein selects the R-SNARE VAMP7 for incorporation into endocytic clathrin-coated vesicles, and depleting it leaves VAMP7 on the cell surface exactly as depleting clathrin does. Reason: The term's definition is "bringing together a cargo protein with clathrin, responsible for the formation of endocytic vesicles", and it was verified to sit under GO:0140312 cargo adaptor activity and hence under GO:0060090. The functional support is a depletion phenotype, which is why the evidence code is IMP rather than IDA: roughly 90% knockdown of AGFG1 doubles surface VAMP7, and clathrin knockdown does the same, while a VAMP7 longin-domain mutant that folds correctly but cannot bind AGFG1 also accumulates at the surface - and does so even when fused to an unrelated reporter, which shows the longin domain alone is the sorting determinant. The authors state the conclusion in exactly these terms. A second, independent laboratory classifies AGFG1 as a CLASP and shows concentration-dependent binding to the FCHO1 mu-homology domain with controls excluding an indirect route. One restraint: the AGFG1-clathrin and AGFG1-AP-2 contacts themselves are motif-based inference in the 2008 paper - its own supporting figure is a sequence figure and the citation is to a review - so GO:0030276 clathrin binding is deliberately NOT proposed. The functional term rests on cargo binding, coated-pit colocalisation and the depletion phenotype, not on a measured clathrin contact. The entailment between the two runs one way and is worth stating rather than eliding: because GO:0035615 sits under GO:0140312, asserting it does commit to bridging cargo to the coat. Retreating to that parent would not avoid the commitment - GO:0140312's own definition reads "Binding directly to the structural scaffolding elements of a vesicle coat (such as clathrin or COPII), and bridging the membrane, cargo receptor, and membrane deformation machinery", so it names the coat too and adds a cargo-receptor clause that is wrong here, since VAMP7 is the cargo and not a receptor. What discharges the commitment is measurement rather than the motif argument: concentration-dependent binding to the FCHO1 mu-homology domain with controls excluding an indirect route, direct colocalisation with clathrin and AP-2 in coated pits and vesicles, and a depletion phenotype that phenocopies clathrin depletion for this cargo. GO:0030276 remains withheld because it would assert something narrower and stronger - a directly assayed AGFG1-clathrin contact - which no experiment provides. Supporting Evidence: PMID:18775314 Taken together, these experiments demonstrate that Hrb functions as a cargo/clathrin adaptor in clathrin-mediated endocytosis of VAMP7 through the interaction of the VAMP7 longin domain with Hrb. PMID:18775314 Quantification using an 125I-labeled monoclonal anti-HA antibody indicated that there was a doubling of VAMP7-HA on the cell surface following PMID:22484487 In addition, Dab2, Hrb and CALM, three tandem Asn-Pro-Phe-containing CLASPs bind the μHDs in a concentration-dependent manner (Fig. 2c). |
| GO:0072583 clathrin-dependent endocytosis | IMP PMID:18819912 Role of HRB in clathrin-dependent endocytosis. | NEW | Summary: Proposed. Two independent knockdown studies place AGFG1 in clathrin-mediated endocytosis. They disagree about how general the requirement is, and this row records only what both support. Reason: In HRB knockdown cells the endocytosis of both fluorescent transferrin and pHLuorin-tagged TI-VAMP is strongly reduced by FACS uptake assay, and AGFG1 colocalises with clathrin-, AP-2-, EPS15- and transferrin-receptor-positive vesicles. Independently, AGFG1 depletion causes surface accumulation of VAMP7 in the same way clathrin depletion does. The two studies are not fully consistent - the second reports no effect of AGFG1 depletion on EGF internalisation or degradation and reads AGFG1 as a cargo-selective rather than general adaptor - so this row asserts involvement in clathrin-dependent endocytosis, which both support, and does not assert that AGFG1 is required for bulk clathrin-dependent uptake. The discrepancy is recorded as a knowledge gap rather than adjudicated. Evidence code IMP: both studies are siRNA knockdowns. Supporting Evidence: PMID:18819912 uptake experiments followed by fluorescence-activated cell sorting showed that the endocytosis of fluorescent transferrin and pHLuorin-TI-VAMP is strongly reduced in HRB knockdown cells PMID:18775314 Consistent with Hrb functioning as a SNARE-specific adaptor, depleting HeLaM cells of Hrb by siRNA had no effect on the internalization and intracellular accumulation of EGF nor on its degradation when compared to mock transfected cells (Figure S5). |
| GO:0005905 clathrin-coated pit | IDA PMID:18775314 Molecular basis for the sorting of the SNARE VAMP7 into endo... | NEW | Summary: Proposed. GOA records cytoplasmic vesicle and cytosol for this protein but not the compartment two independent studies actually image it in. Reason: Immunofluorescence confocal microscopy shows extensive colocalisation of AGFG1 with clathrin and AP-2 in coated pits, and no colocalisation with the lysosomal marker lgp120 or the late-endosomal marker M6PR - so the assignment is discriminating rather than a general vesicular signal. A second study independently reports colocalisation with clathrin-, AP-2-, EPS15- and transferrin-receptor-containing structures in HeLa cells. The location is also what the FxxFxxF AP-2-appendage and clathrin-box motifs predict, and what the cargo-selection function requires. Supporting Evidence: PMID:18775314 However, Hrb did show extensive colocalization with clathrin and AP2 in clathrin-coated pits (CCPs) and CCVs (Figure 2B). PMID:18819912 In the cytoplasm, HRB colocalized with clathrin-, AP-2-, EPS15-, and transferrin receptor-containing vesicles. |
| GO:0030136 clathrin-coated vesicle | IDA PMID:18775314 Molecular basis for the sorting of the SNARE VAMP7 into endo... | NEW | Summary: Proposed alongside the coated-pit row, and from the same imaging: AGFG1 is on coated vesicles as well as coated pits, which is where cargo selection has to be completed. Reason: The same confocal experiment that places AGFG1 in coated pits places it in clathrin-coated vesicles, again with negative controls against lysosomal and late-endosomal markers. Proposed as a separate row because GO keeps the two compartments distinct and the functional argument uses both: recruitment happens at the pit, and retrieval of VAMP7 requires AGFG1 to remain associated as the vesicle forms. Supporting Evidence: PMID:18775314 However, Hrb did show extensive colocalization with clathrin and AP2 in clathrin-coated pits (CCPs) and CCVs (Figure 2B). |
| GO:0008270 zinc ion binding | IDA file:human/AGFG1/AGFG1-bioinformatics/RESULTS.md | NEW | Summary: Proposed. Both structures of AGFG1's ArfGAP domain resolve a zinc ion held by Cys29/32/49/52. UniProt asserts the term through the retired keyword route, which GOA no longer imports, so a structurally verified annotation is currently missing from GOA. Reason: Computed from the deposited coordinates of both AGFG1 ArfGAP-domain entries: 2OLM (X-ray, 1.48 Angstrom) and 2D9L (NMR, 20 models) each contain one zinc ion, and in both it is coordinated by the thiolates of Cys29, Cys32, Cys49 and Cys52 at 2.26-2.40 Angstrom. Author numbering is not UniProt numbering - 2D9L is offset by +6 - so the offset was taken from SIFTS and each converted position checked against the UniProt sequence. The four cysteines are exactly the ones predicted independently by the published CX2CX16CX2CX4R spacing, so the structural result and the sequence scan corroborate each other. UniProt already carries the term as "GO:0008270; F:zinc ion binding; IEA:UniProtKB-KW", but that route was withdrawn from GOA, so the term is absent from all 35 GOA rows. This is the mirror of the DNA-binding keyword on the same entry, which is not supported and is filed as a UniProt correction request instead. On the choice of reference: neither structure has a primary publication. PDBe reports 2OLM as "ArfGap domain of HIV-1 Rev binding protein" and 2D9L as "Solution structure of the ArfGap domain of human RIP", both "To be published" with a null PubMed id and a null DOI, and UniProt records 2D9L only as submitted to the PDB. So no PMID exists to carry original_reference_id. The evidence code stays IDA because the assay is the two deposited structures themselves; the file: reference points at the committed computation over their coordinates, which is the only citable record of the coordination shell. Supporting Evidence: file:human/AGFG1/AGFG1-bioinformatics/RESULTS.md | 2OLM | X-ray 1.48 Å | 1 | +0 | Cys29, Cys32, Cys49, Cys52 | 2.30–2.40 Å | file:human/AGFG1/AGFG1-uniprot.txt DR GO; GO:0008270; F:zinc ion binding; IEA:UniProtKB-KW. PMID:18809720 The side chains of only the conserved arginine (Arg50, on the right) and the four zinc finger cysteines (center; Cys22, 25, 42, and 45) are displayed along with the coordinated Zn2+ (gray sphere). |
| GO:0120211 proacrosomal vesicle fusion | ISS PMID:11711676 Lack of acrosome formation in Hrb-deficient mice. | NEW | Summary: Proposed as the specific process the mouse experiments actually measured. The accepted GO:0001675 acrosome assembly IBA is correct but general; what the null blocks is precisely the fusion step, and GO has a term for it. Reason: GO:0120211's definition is "Fusion of the membrane of proacrosomal vesicle with the membrane of another proacrosomal vesicle to form the acrosome", which is the mouse phenotype verbatim: the vesicles form, coat with AGFG1, and cannot fuse. It was verified to be a descendant of GO:0001675 over is_a/part_of, so this row is additive and asserts strictly more than - not something different from - the IBA it sits under. Evidence code ISS with the mouse orthologue in supporting_entities, because the measurement is on mouse Agfg1 (Q8K2K6) and human AGFG1's support is orthology; ISS takes the sequence-similar entity, not the interactor. This row also records the decision on the acrosome-associated cellular component, which was raised in review: GO:0001669 acrosomal vesicle is NOT proposed, on two grounds. Its definition denotes the mature organelle - "A structure in the head of a spermatozoon that contains acid hydrolases ... derived from the lysosome" - whereas AGFG1 is documented on the cytosolic surface of the precursor vesicles, and in the null the mature acrosome never forms at all; and the evidence is mouse-only, so a human CC row would rest on orthology for a compartment whose human counterpart has not been imaged. Searching for that CC is what surfaced GO:0120211, which is the better annotation. Supporting Evidence: PMID:11711676 Although proacrosomic vesicles form in spermatids that lack Hrb, the vesicles are unable to fuse, blocking acrosome development at step 2. PMID:11711676 We conclude that Hrb is required for docking and/or fusion of proacrosomic vesicles during acrosome biogenesis. |
Loading supporting content…
Download this section (compressed HTML)Q: PAINT/GO Central: should the three spermiogenesis terms on PANTHER node PTN002919572 - GO:0001675 acrosome assembly, GO:0007289 spermatid nucleus differentiation and GO:0045109 intermediate filament organization - be moved down to the AGFG1-orthologue subfamily node (PTHR46134:SF1)? The node's 36 recipients comprise 21 AGFG1-type, 12 AGFG2-type and 3 unnamed gene products across 17 species, so its human reach is exactly AGFG1 plus AGFG2. The entire experimental basis for all three terms is mouse Agfg1 (Q8K2K6), by IMP; mouse and rat Agfg2 are themselves IBA recipients, so no AGFG2 orthologue in any species holds experimental evidence for any of them. Human AGFG2's own characterised function is Weibel-Palade body sizing and von Willebrand factor secretion in endothelium, which is unrelated. One node edit would retract three incorrect terms from 12 AGFG2-named recipients while leaving all 21 AGFG1-named recipients correct. Stated once here for both genes rather than repeated per row; a concurrent review of AGFG2 is in progress and should treat this as a claim to verify rather than a finding.
Q: MGI/GO Central: mouse Agfg1's GO:0007289 spermatid nucleus differentiation IMP cites PMID:16765935, but that paper's result for hrb-null spermatids is a negative - H1T2 polar localisation is unaltered, which the authors use to argue that acrosome development is not required for nuclear polarity. The phenotype that does support the term, round spermatid nuclei bearing protrusions the authors call nucleopodes, is reported in PMID:14724135. Should the reference on that annotation be corrected? The human IBA descends from it.
Q: UniProt: the AGFG1 entry (P52594) carries "KW DNA-binding" and "DR GO; GO:0003677; F:DNA binding; IEA:UniProtKB-KW", but AGFG1 has no DNA-binding module. Its only zinc site is the ArfGAP C4 finger, which two structures show is buried and fully occupied by structural zinc on Cys29/32/49/52, and which the family's consensus reference describes as architectural rather than catalytic. The keyword appears to follow InterPro's remark that the ArfGAP zinc finger resembles the C4-type GATA finger, which is a fold resemblance rather than a measured activity. GOA no longer imports the keyword route, so there is no GO row to act on and no GO action is proposed; this is a request to withdraw the keyword upstream. The same entry's "GO:0008270; F:zinc ion binding" keyword is the mirror case - correct, dropped from GOA with the route, and proposed here as a NEW annotation.
Q: CATH/Gene3D and UniProt: AGFG1 is assigned G3DSA:3.30.450.50 "Longin domain" over residues 154-234, a region UniProt itself annotates as disordered (MobiDB-lite 145-193) and polar-biased (COMPBIAS 176-191). The cause is traceable: PDB 2VX8's chains A-D are each a single 169-residue polypeptide that PDBe maps to two UniProt entries, AGFG1 136-175 and mouse VAMP7 1-120, and the authors describe building exactly that fusion. The longin fold belongs to VAMP7 and the domain call has crossed the fusion junction. The negative control is the paralogue: AGFG2, with the identical InterPro signature set and 71.2% identity over the ArfGAP domain, has no 3.30.450.50 assignment - so the error is keyed on having a chimeric PDB entry, not on family membership. It currently produces no wrong GO term, since the derived FunFam is one condition of an ARBA rule that grants only cytoplasm, but it is worth correcting before it propagates.
Q: GOA/UniProt: UniProt's SUBUNIT line records "Interacts with EPS15R and EPS15. Interacts with FCHO1." and none of the three has a GOA counterpart. All are from low-throughput studies with mapped determinants - the EPS15 interaction is NPF/EH-dependent and its disruption abolishes the synergy with Rev, and the FCHO1 interaction is concentration-dependent with controls excluding an indirect route. Is there a reason these were not annotated?
Q: Has AGFG1 or AGFG2 ever been assayed for Arf GTPase-activating activity? The consensus nomenclature paper states that GAP activity has been demonstrated in vitro for at least one member of each subfamily except the ADAPs, but its own AGFG section reports none for either paralogue and says only that much less information is available on AGFG2. If no such assay exists, the family-level statement should not be read as covering the AGFG subfamily.
Experiment: Measure Arf GAP activity of recombinant human AGFG1 directly, using the standard single-turnover GTP-hydrolysis assay on myristoylated ARF1 and ARF6 loaded with GTP on liposomes. Compare full-length AGFG1, the isolated ArfGAP domain (residues 1-141, the construct already crystallised as 2OLM), loss-of-function Arg57Ala, and restoration Tyr39Trp and Thr71Asp single and double mutants. ARFGAP1 is the positive control and ADAP1 the negative control, since the ADAP subfamily is reported to lack in-vitro activity despite an intact motif. Rescue by the restoration mutants would directly test the residue-level pseudoenzyme model; appreciable wild-type activity would instead show that AGFG1 uses a different catalytic solution and support retaining GO:0005096.
Hypothesis: Human AGFG1 retains Arf binding but has substantially reduced GAP activity because two catalytically required residues are substituted.
Type: in vitro GTP hydrolysis assay with structure-guided mutagenesis
Experiment: Ask whether VAMP7 is mislocalised in Agfg1-null mouse spermatids, and whether the AGFG1 substitutions that abolish VAMP7 longin-domain binding in vitro (L160S/L163S, L163P/L164P, L171S/L173S) fail to rescue acrosome formation when expressed in Agfg1-null spermatids. A positive result would unify the two literatures on this gene under one molecular function; a negative one would show the acrosomal role uses a different cargo.
Hypothesis: The acrosomal and somatic activities of AGFG1 are one mechanism - selection of SNARE-bearing vesicles through the VAMP7 longin domain.
Type: knock-in rescue with binding-deficient mutants plus immunolocalisation
Experiment: Resolve the transferrin/EGF discrepancy with a single cargo panel in one cell line. Knock AGFG1 down and measure uptake of transferrin, EGF, VAMP7 and a clathrin-independent control in parallel, with clathrin knockdown as the reference. The two 2008 studies used different cargoes in different cells and reached different conclusions about whether AGFG1 is a general clathrin-mediated-endocytosis component or a cargo-selective adaptor, and the answer determines whether GO:0072583 should carry a general or a cargo-restricted framing.
Hypothesis: AGFG1 is a cargo-selective adaptor rather than a general component of clathrin-mediated endocytosis.
Type: parallel cargo-uptake panel after siRNA depletion
Experiment: Screen globozoospermia cohorts for AGFG1 loss-of-function variants. The mouse null phenotype is unambiguous and human AGFG1 is expressed in sperm, but the human acrosome annotations rest entirely on orthology; a human loss-of-function allele would convert them from similarity-based to direct evidence, and AGFG1 is absent from the current globozoospermia gene panels.
Hypothesis: Loss of AGFG1 function causes globozoospermia in humans, as it does in mice.
Type: targeted human genetics in an infertility cohort
Experiment: Because the VAMP7-binding stretch lies immediately C-terminal to the ArfGAP domain, and because Arf1/Arf6 recruit and release coat components, test whether Arf GTP loading modulates AGFG1-VAMP7 binding in vitro, and whether an Arg57Ala AGFG1 still supports VAMP7 retrieval in cells.
Hypothesis: AGFG1's ArfGAP domain and its adjacent cargo-adaptor tail act on the same membrane event rather than being two independent activities that share a polypeptide.
Type: in vitro binding under defined Arf nucleotide states plus cellular rescue
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)