ARFGEF3 (BIG3) is a 2177-residue cytoplasmic protein of the Sec7/BIG family that, unlike its paralogues BIG1 (ARFGEF1) and BIG2 (ARFGEF2), does not act as a guanine nucleotide exchange factor. Its Sec7 domain has lost the invariant "glutamic finger" glutamate that Sec7 GEFs use to destabilise ARF-bound GDP, carrying asparagine at the equivalent position, and the loss is shared with the mouse orthologue. The protein instead works through an armadillo/HEAT-type alpha-helical repeat region in its N-terminal third, which binds prohibitin 2 (PHB2). By holding PHB2 in the cytoplasm and occluding the surface through which PHB2 engages karyopherin-alpha import receptors, ARFGEF3 prevents oestradiol-driven nuclear entry of PHB2 and so relieves PHB2-mediated repression of oestrogen receptor alpha. A second scaffolding role operates on the same complex: ARFGEF3 is an A-kinase anchoring protein that binds protein kinase A through predicted RII-binding regions and docks the protein phosphatase 1 catalytic subunit PP1C-alpha through a canonical RVxF motif, inhibiting PP1C-alpha until PKA phosphorylation of ARFGEF3 relieves that inhibition. Expression is largely restricted to pancreatic islets, brain and salivary gland, with strong ectopic expression in breast carcinoma. In rodent islets the protein sits on immature insulin- and glucagon-containing secretory granules at the trans-Golgi network and restrains granule biogenesis, so that its loss raises hormone content and stimulated secretion and disturbs systemic glucose homeostasis. Reported localisation is cell-type dependent: diffusely cytoplasmic in breast carcinoma, granule- and TGN-associated in islet cells, lysosomal in hippocampal neurons, and mitochondrial in osteosarcoma lines.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005085 guanyl-nucleotide exchange factor activity | IBA GO_REF:0000033 | REMOVE | Summary: ARFGEF3 has never been shown to catalyse nucleotide exchange on any ARF, and its Sec7 domain lacks the invariant glutamate that performs the reaction. The ancestral node is correctly placed and the donors are genuine GEFs; ARFGEF3 is the descendant that lost the catalytic residue. Reason: Sec7-domain exchange factors act through a single invariant glutamate, the glutamic finger, which inserts into the ARF nucleotide pocket to destabilise bound GDP. Re-deriving this from live sequence rather than taking it on the literature's word, with the comparison panel built from this row's own WITH/FROM field: all 17 donors that have an annotated SEC7 domain carry glutamate at the derived glutamic-finger column, whereas ARFGEF3 carries asparagine at Q5TH69 position 692 and matches the Sec7 catalytic motif zero times anywhere in its SEC7 domain. Mouse Arfgef3 carries serine at the equivalent position, so the loss is clade-level and not a human annotation artefact. The check is controlled in both directions - a verified-active paralogue scores positive, and the same sequence with its glutamic finger mutated to alanine scores negative - so it is capable of reporting the opposite result and does not here. This reproduces the published alignment of PMID:24997568 and matches what the group that characterised the protein states in two further papers. Note the asymmetry being relied on: losing the catalytic residue is strong evidence against the activity, whereas retaining it would not have been evidence for it, so only the negative direction is claimed. REMOVE rather than MARK_AS_OVER_ANNOTATED because the term asserts a specific chemical activity that the protein is structurally unable to perform, which is a different situation from a term that merely exceeds its evidence. The node itself should not be changed: the other members really are GEFs, and this row is the single member that should not inherit. Propagation Review Root cause: PROPAGATION BAD Failure modes: PSEUDO OR SUBACTIVITY LOSS Sources checked: PANTHER:PTN008950430 SUPPORTS SOURCE BUT NOT TARGET PAINT ancestral node, not a protein. The node placement is sound - the clade it covers really did have exchange activity - but ARFGEF3 is a descendant in which the catalytic residue has been lost. UniProtKB:Q9Y6D6 · ARFGEF1 SUPPORTS SOURCE BUT NOT TARGET Human BIG1, the closest paralogue. Holds GO:0005085 by its own IDA and retains the glutamic finger at E793, so it is exactly the contrast that makes the target's loss visible. UniProtKB:Q9Y6D5 · ARFGEF2 SUPPORTS SOURCE BUT NOT TARGET Human BIG2. Own IDA for the term; retains the glutamic finger at E738. SGD:S000005241 · MON2 SOURCE WEAK OR INFERRED The one donor UniProt annotates with no SEC7 domain at all and no IPR000904, and one of only three without an IDA (IGI/IPI/ISS). Mon2 shares IPR015403 with ARFGEF3, so the same node may be donating exchange activity to a second member that does not perform it. Worth checking on the PAINT side. UniProtKB:Q15438 · CYTH1 SUPPORTS SOURCE BUT NOT TARGET Cytohesin-1; own IDA, glutamic finger retained at E157. UniProtKB:O43739 · CYTH3 SUPPORTS SOURCE BUT NOT TARGET Cytohesin-3; own IDA, glutamic finger retained. UniProtKB:Q9UIA0 · CYTH4 SUPPORTS SOURCE BUT NOT TARGET Cytohesin-4; own IDA, glutamic finger retained. UniProtKB:A5PKW4 · PSD SUPPORTS SOURCE BUT NOT TARGET EFA6A; own IDA, glutamic finger retained. UniProtKB:Q5JU85 · IQSEC2 SUPPORTS SOURCE BUT NOT TARGET BRAG1; own IDA and IMP, glutamic finger retained. MGI:MGI:1334257 · Cyth1 SUPPORTS SOURCE BUT NOT TARGET Mouse cytohesin-1; own IDA. Ortholog of the human CYTH1 donor, so not independent of it. MGI:MGI:1918215 · Psd3 SUPPORTS SOURCE BUT NOT TARGET Mouse EFA6C; own IDA, glutamic finger retained. AGI_LocusCode:AT1G13980 · GNOM SUPPORTS SOURCE BUT NOT TARGET Arabidopsis GNOM; own IDA, glutamic finger retained. Shows the activity predates the plant/animal split. FB:FBgn0028538 · Sec71 SUPPORTS SOURCE BUT NOT TARGET Drosophila Sec71; own IDA. TrEMBL entry, so its protein name is not independent evidence, but its GO evidence is curated. FB:FBgn0264560 · garz SUPPORTS SOURCE BUT NOT TARGET Drosophila Gartenzwerg; own IGI rather than IDA. TrEMBL entry, and this MOD id maps to four UniProt entries, so the resolution is reported rather than treated as unique. PomBase:SPAC11E3.11c SUPPORTS SOURCE BUT NOT TARGET Fission yeast PH/SEC7 protein; own IMP rather than IDA. Gene symbol not asserted because the entry carries none. SGD:S000002577 · SEC7 SUPPORTS SOURCE BUT NOT TARGET The founding Sec7 protein; own IDA, glutamic finger retained at E923. SGD:S000000748 · GEA2 SUPPORTS SOURCE BUT NOT TARGET Budding yeast Gea2; own IDA, glutamic finger retained. SGD:S000003792 · GEA1 SUPPORTS SOURCE BUT NOT TARGET Budding yeast Gea1; own IDA, glutamic finger retained at E636. SGD:S000006299 · SYT1 SUPPORTS SOURCE BUT NOT TARGET Budding yeast Syt1; own IDA, glutamic finger retained. Supporting Evidence: PMID:24997568 they showed that BIG3 lacked the conserved motif and critically, the essential glutamate residue PMID:24997568 This region is conserved among the BIG3 orthologues, all of which lack the functional motif. PMID:24711543 It is worth noting that the single known functional domain in BIG3 by sequence analysis, the Sec7 domain, has a non-functional catalytic motif PMID:27696409 we recently identified BIG3 as a novel ARF GEF protein with a non-functional catalytic motif in the SEC7 domain file:human/ARFGEF3/ARFGEF3-bioinformatics/RESULTS.md | known-active donors (own experimental GEF evidence) | 17 | **E in 17/17** | |
| GO:0005085 guanyl-nucleotide exchange factor activity | IEA GO_REF:0000002 | REMOVE | Summary: The InterPro Sec7 signature matches ARFGEF3, but the signature detects the fold, not the catalytic residue, and ARFGEF3's copy of the fold has lost it. A textbook domain-implied catalytic over-annotation. Reason: IPR000904 is genuinely present, so the match is not in dispute; what is wrong is the inference that presence of the domain implies exchange activity. The residue that performs the reaction is absent from ARFGEF3 and from its mouse orthologue (asparagine at Q5TH69 692, serine at Q3UGY8 688, against glutamate in 17 of 17 catalytically verified relatives), and no experiment has measured exchange activity for this protein. The published alignment reached the same conclusion, noting that the Sec7 region of BIG3 does not even align significantly to those of BIG1 and BIG2. This is the same defect as the IBA row and is removed for the same reason, but the mechanism differs and is worth separating: the IBA inherits from a correctly placed ancestral node, whereas this row comes from a domain-to-activity rule that cannot see catalytic competence. The rule is the thing that needs amending, which is why the propagation review points at IPR000904 itself. Propagation Review Root cause: PROPAGATION BAD Failure modes: PSEUDO OR SUBACTIVITY LOSS Sources checked: InterPro:IPR000904 · Sec7 domain SUPPORTS SOURCE BUT NOT TARGET The signature match is real: ARFGEF3 residues 583-796 are an annotated SEC7 domain. The interpro2go line mapping IPR000904 to GO:0005085 is what over-reaches, because a profile match to the domain does not establish that the catalytic glutamate is present. Amending that mapping, or gating it on the motif, would fix every Sec7-fold pseudo-GEF at once. Supporting Evidence: PMID:24997568 they showed that BIG3 lacked the conserved motif and critically, the essential glutamate residue PMID:24997568 This region is conserved among the BIG3 orthologues, all of which lack the functional motif. PMID:24997568 BIG3, unlike BIG1 and BIG2, potentially lacks GEF activity, despite being annotated to contain the Sec7 domain (based on sequence similarity). PMID:24997568 BIG3 showed only 21% identities to BIG1 and BIG2, with ~30% identity in DUF1981 and no significant similarity (i.e., a BLAST e-value of > 10) found in the Sec7 domain. |
| GO:0032012 regulation of ARF protein signal transduction | IEA GO_REF:0000002 | MARK AS OVER ANNOTATED | Summary: Unlike the molecular function rows, this one is not refuted - a catalytically dead Sec7 domain could still engage ARF and modulate its signalling. But nothing has tested it, and the term reaches ARFGEF3 from the same domain rule that produced the false activity call. Reason: Kept distinct from the two GO:0005085 rows deliberately, because the evidence differs in kind. Those rows assert a chemical reaction that the residue analysis shows the protein cannot catalyse, so they are removed. This row asserts only that ARFGEF3 modulates ARF signalling somehow, which a dead Sec7 domain acting as a decoy or competitor plausibly could do. That is in fact the model the protein's own characterisers propose, and they are explicit that it is a hypothesis. No experiment has measured ARF-GTP loading, ARF binding, or any ARF-dependent readout in response to ARFGEF3 in any system, so there is no evidence either way. Marked as over-annotated rather than removed: it is unsupported rather than contradicted, and it comes from the same interpro2go inference whose premise about this protein has already failed once. If the competitive-decoy model is ever tested, this term is the right one to restore. Propagation Review Root cause: PROPAGATION BAD Failure modes: PSEUDO OR SUBACTIVITY LOSS Sources checked: InterPro:IPR000904 · Sec7 domain SOURCE WEAK OR INFERRED Same domain-to-process rule as the GO:0005085 IEA row. It fires on the fold alone, and for this protein the fold's catalytic machinery is absent, so the process inference inherits an assumption that has already been shown false at the activity level. Supporting Evidence: PMID:24711543 one of our hypotheses is that BIG3 acts as a competitive non-functional Arf-GEF, thereby negatively modulating granule production PMID:24997568 they showed that BIG3 lacked the conserved motif and critically, the essential glutamate residue |
| GO:0016192 vesicle-mediated transport | IBA GO_REF:0000033 | ACCEPT | Summary: Correct and core. ARFGEF3 sits on immature secretory granules at the trans-Golgi network and sets how many of them are made, which is participation in vesicular traffic even though it contributes no catalysis. The broad term is the right level here, not a failure to be specific. Reason: This is the one family function ARFGEF3 demonstrably retained. Immuno-EM places it on insulin granules and on clathrin-positive TGN compartments, and removing it raises granule number, hormone content and stimulated secretion in beta cells, with the same sign in alpha cells for glucagon. So the protein does part of the work of the pathway rather than merely being required for it. On whether the term is too general: resolving all 18 protein donors shows they hold thirteen different descendant processes between them, spanning anterograde ER-to-Golgi, retrograde, intra-Golgi, endosome-to-Golgi and exocytic steps, so GO:0016192 is the genuine least common ancestor of the donor set rather than a coarse stand-in for one of them. A granularity mismatch requires the donors to agree, and here they emphatically do not, so no specificity upgrade is proposed on this row. The specific, better-evidenced process is instead proposed separately as GO:1904410 with ISS from the mouse work, which keeps the phylogenetic inference and the species-specific evidence in their own rows. Supporting Evidence: PMID:24711543 BIG3 is predominantly localized to insulin- and clathrin-positive trans-Golgi network (TGN) compartments. PMID:24711543 Furthermore, BIG3 predominantly localized to insulin granules of islet β-cells, as revealed by immuno-EM PMID:24711543 BIG3-deficient insulin-secreting cells display increased insulin content and granule number and elevated insulin secretion upon stimulation. file:human/ARFGEF3/ARFGEF3-bioinformatics/RESULTS.md `GO:0016192` is therefore the genuine **LCA** of a heterogeneous donor set, not a |
| GO:0005737 cytoplasm | EXP PMID:19496786 Activation of an estrogen/estrogen receptor signaling by BIG... | ACCEPT | Summary: The only experimental annotation on this gene, and it is sound: endogenous human BIG3 detected in the cytoplasm of human breast carcinoma tissue and of human cell lines with a validated antibody. Reason: Immunohistochemistry on human tumour tissue and immunocytochemistry on human cell lines both show cytoplasmic staining of the endogenous protein, using an antibody the same paper validates against exogenously expressed BIG3 with no cross reactivity. The localisation is also mechanistically necessary for the function the same paper establishes, since holding PHB2 out of the nucleus requires BIG3 to be where PHB2 is. Kept as a core location despite being unspecific: ARFGEF3's reported compartment genuinely varies by cell type, and cytoplasm is the one assignment supported in human cells. Deliberately not refined to a more specific compartment here; the granule, lysosomal and mitochondrial pools each rest on non-human or single-paper evidence and are handled separately. Supporting Evidence: PMID:19496786 showed its strong staining in the cytoplasm of two different histological subtypes of breast cancer PMID:19496786 BIG3 trapped PHB2/REA in the cytoplasm and inhibited its nuclear translocation, and caused enhancement of ERα transcriptional activity |
| GO:0005737 cytoplasm | IEA GO_REF:0000044 | ACCEPT | Summary: The electronic duplicate of the EXP row above, derived from UniProt's Cytoplasm subcellular location term. The SL-0086 to GO:0005737 mapping is exact and the underlying UniProt statement is itself experimental. Reason: UniProt's cytoplasm assignment carries ECO:0000269 against PMID:19496786, so this row and the EXP row trace to the same human experiment rather than representing independent support. Duplicate term identifiers across evidence codes are expected and there is no reason to favour one over the other. Checking the mapping rather than assuming it: SL-0086 maps to GO:0005737 exactly, unlike the SL-0244 and SL-0245 mappings on this same gene, which land in a disjoint branch. Accepted as correct, and noted as non-independent so it is not read as corroboration. Supporting Evidence: file:human/ARFGEF3/ARFGEF3-bioinformatics/RESULTS.md | `SL-0086` | Cytoplasm | `GO:0005737` cytoplasm | — | EXACT | |
| GO:0030133 transport vesicle | IEA GO_REF:0000044 | MODIFY | Summary: Right organelle family, wrong branch. The underlying evidence is insulin and glucagon granules, which are regulated secretory granules, whereas GO:0030133 is defined as the constitutive secretory pathway. The two terms are disjoint siblings, so this is not merely a coarse assignment. Reason: The row comes from UniProt's SL-0244, which UniProt itself names Secretory vesicle and defines in regulated-pathway language, as a vesicle that carries cargo such as hormones or neurotransmitters to the cell membrane where it docks and fuses to release its content. The SubCell-to-GO mapping nonetheless sends SL-0244 to GO:0030133, whose GO definition is any of the vesicles of the constitutive secretory pathway. Checked rather than assumed: querying QuickGO's is_a and part_of closure shows GO:0030141 secretory granule is not an ancestor or a descendant of GO:0030133, both instead descending separately from GO:0031410 cytoplasmic vesicle. So the mapping crosses into a disjoint branch and the fix is a replacement, not a refinement. The primary evidence behind UniProt's statement is the mouse work, where immuno-EM puts the protein on insulin granules and where its loss changes granule number, which is squarely the regulated secretory granule branch. Proposing GO:0030141. The defect is in the vocabulary mapping rather than in this gene's curation, and it is applied to 130685 UniProtKB entries, which is the scale at which it would need fixing wherever the regulated reading applies - not a claim that all of those entries are mis-annotated. Proposed replacements: secretory granule Supporting Evidence: PMID:24711543 Furthermore, BIG3 predominantly localized to insulin granules of islet β-cells, as revealed by immuno-EM PMID:25737957 Depletion of BIG3 in alpha-cells leads to elevated glucagon production and secretion. file:human/ARFGEF3/ARFGEF3-bioinformatics/RESULTS.md | `SL-0244` | **Secretory vesicle** | `GO:0030133` transport vesicle | `GO:0030141` secretory granule | **DISJOINT** | |
| GO:0030658 transport vesicle membrane | IEA GO_REF:0000044 | MODIFY | Summary: The membrane counterpart of the row above and wrong in the same way: SL-0245 Secretory vesicle membrane is mapped to the transport-vesicle branch rather than the secretory-granule branch. Corrected to GO:0030667. Reason: GO:0030658 is a child of GO:0030133 and GO:0030667 is a child of GO:0030141, so the two membrane terms inherit exactly the same disjointness as their parents, confirmed against QuickGO's closure. The replacement therefore follows from the GO:0030133 row and needs no separate biological argument. One caveat recorded rather than hidden: UniProt flags the membrane localisation itself as ECO:0000305, a curator inference, and the accompanying single-pass membrane topology rests on a predicted transmembrane helix at residues 1492-1512 with ECO:0000255 rather than on any measurement. So the membrane assignment is weaker than the granule assignment it sits under. Kept rather than removed because the granule association is well evidenced by immuno-EM and a peripheral or integral membrane association is the natural reading of it, but it is explicitly not core and the topology is flagged as a question for experimental resolution. Proposed replacements: secretory granule membrane Supporting Evidence: PMID:24711543 BIG3 is predominantly localized to insulin- and clathrin-positive trans-Golgi network (TGN) compartments. file:human/ARFGEF3/ARFGEF3-bioinformatics/RESULTS.md | `SL-0245` | **Secretory vesicle membrane** | `GO:0030658` transport vesicle membrane | `GO:0030667` secretory granule membrane | **DISJOINT** | |
| GO:0140311 protein sequestering activity | IMP PMID:19496786 Activation of an estrogen/estrogen receptor signaling by BIG... | NEW | Summary: ARFGEF3's best-evidenced molecular function, and GOA records nothing like it. BIG3 binds PHB2 and holds it away from the compartment where PHB2 acts, which is the GO:0140311 definition almost word for word. Reason: GO:0140311 is the exact-fit child of GO:0140313: the same definition with "a specific molecule" narrowed to "a protein". The sequestered species here is PHB2, a protein, so the child adds no claim beyond the one the ligand's identity already fixes - which is precisely why the child is right here and the parent is right on the GO:0051018 row, where the child would have asserted an unproven subunit specificity. GO defines it as binding a protein to prevent it from interacting with other partners or to inhibit its localisation to the area of the cell where it is active. Both clauses are satisfied and each is separately demonstrated. Depleting BIG3 by siRNA in human breast cancer cell lines releases PHB2 into the nucleus on oestradiol stimulation, and re-expressing BIG3 holds it in the cytoplasm even with stimulation, which is the localisation clause. The partner-occlusion clause is established by the karyopherin work, which shows BIG3 blocks the surface through which PHB2 engages its import receptors, so this is not a generic tether. Targeted disruption of the interface with the ERAP peptide, and independently with xanthohumol, releases PHB2 and reverses the downstream effect, which is stronger than knockdown or overexpression alone. The interface is mapped to residues 86-434 by co-immunoprecipitation and to Q165, D169 and Q173 by alanine scanning. Evidence code is IMP rather than IDA because the demonstrations are knockdown, re-expression and targeted disruption in human cells rather than a direct assay on purified components. The experiments are human throughout, so an experimental code is appropriate here. Supporting Evidence: PMID:19496786 BIG3 trapped PHB2/REA in the cytoplasm and inhibited its nuclear translocation, and caused enhancement of ERα transcriptional activity PMID:26052702 These data indicated that BIG3 may block the KPNAs (KPNA1, KPNA5, and KPNA6) binding region(s) of PHB2, thereby leading to inhibition of KPNAs-mediated PHB2 nuclear translocation in the presence of E2 in breast cancer cells. PMID:24051437 Intrinsic PHB2 released from BIG3 by ERAP directly binds to both nuclear- and membrane-associated ERα |
| GO:0042308 negative regulation of protein import into nucleus | IMP PMID:26052702 BIG3 Inhibits the Estrogen-Dependent Nuclear Translocation o... | NEW | Summary: The process counterpart of the sequestering activity: BIG3 stops PHB2 entering the nucleus. Demonstrated in human cells by knockdown, re-expression and karyopherin epistasis, and entirely absent from GOA. Reason: ARFGEF3 does the work of this process rather than merely being required for it: it is the entity that binds the cargo and occludes the receptor-binding surface, and the import that is prevented is measured directly as PHB2 nuclear signal. Knocking BIG3 down increases the proportion of cells with nuclear PHB2 after oestradiol, re-expressing it abolishes that increase, and knocking down KPNA1, KPNA5 or KPNA6 individually blocks the translocation even in BIG3-depleted cells, which identifies the step BIG3 acts on rather than merely correlating with it. Chosen in preference to any oestrogen-signalling term: BIG3's substrate is PHB2, and the effect on ERalpha is two steps downstream through a protein BIG3 was tested against and found not to bind. Annotating the ERalpha process would conflate ARFGEF3's role with its downstream consequence. Human cells throughout, so IMP. Supporting Evidence: PMID:26052702 Here, we report that BIG3 blocks the estrogen (E2)-dependent nuclear import of PHB2 via the karyopherin alpha (KPNA) family in breast cancer cells. PMID:19496786 BIG3 trapped PHB2/REA in the cytoplasm and inhibited its nuclear translocation, and caused enhancement of ERα transcriptional activity PMID:19496786 we investigated the possibility of a direct interaction between BIG3 and ERα, but failed to indicate their interaction (data not shown) |
| GO:0008157 protein phosphatase 1 binding | IPI PMID:28555617 A-kinase anchoring protein BIG3 coordinates oestrogen signal... | NEW | Summary: Endogenous BIG3 binds the PP1 catalytic subunit PP1C-alpha in human breast cancer cells through a canonical RVxF docking motif. Missing from GOA. Reason: Chosen in preference to the uninformative GO:0005515 because the partner and the binding determinant are both identified: the interaction is detected between endogenous proteins in two human ERalpha-positive lines, and deleting the canonical RVxF motif at residues 1228-1232 abolishes it. RVxF is the standard docking motif by which PP1 regulatory subunits engage the PP1 catalytic subunit, so the structural basis is the recognised one rather than an inference from co-precipitation. Annotated as IPI with the human PPP1CA accession, and the experiment really was human protein with human protein in human cells. Supporting Evidence: PMID:28555617 We detected an endogenous interaction between BIG3 and PP1Cα in the ERα-positive breast cancer cell lines MCF-7 and KPL-3C, which highly express both proteins |
| GO:0004865 protein serine/threonine phosphatase inhibitor activity | IMP PMID:28555617 A-kinase anchoring protein BIG3 coordinates oestrogen signal... | NEW | Summary: BIG3 does not merely bind PP1C-alpha, it inhibits its catalytic activity, and the inhibition depends on the docking motif. A second molecular function absent from GOA. Reason: GO:0004865 rather than its parent GO:0004864 for the same reason the sequestering row uses GO:0140311: PPP1CA is a serine/threonine phosphatase by definition (UniProt "Serine/threonine-protein phosphatase PP1-alpha catalytic subunit", EC 3.1.3.16), so naming the substrate class adds no claim the inhibited enzyme's identity did not already fix. Binding and inhibiting are separate claims and the paper separates them. Titrating wild-type BIG3 into cells reduces endogenous PP1C-alpha phosphatase activity in a dose-dependent manner against a defined substrate, while the RVxF-deleted mutant at equivalent expression does not, so the inhibition tracks the docking interaction rather than expression level. Removing endogenous BIG3 by siRNA in two human breast cancer lines relieves the inhibition, which is the reciprocal test. Evidence code IMP rather than IDA because the readouts are transfection and knockdown in cells rather than an assay on purified proteins. This is also what makes the AKAP description mechanistically coherent: BIG3 holds a kinase and a phosphatase on the same scaffold and gates the phosphatase, with PKA phosphorylation of BIG3 relieving the gate. Supporting Evidence: PMID:28555617 This showed that the introduction of WT-BIG3 inhibited endogenous PP1Cα activity in a dose-dependent manner, while ΔPP1Cα-BIG3 did not |
| GO:0051018 protein kinase A binding | IPI PMID:28555617 A-kinase anchoring protein BIG3 coordinates oestrogen signal... | NEW | Summary: Endogenous BIG3 interacts with PKA in human breast cancer cells, which together with the PP1 binding is what makes it an A-kinase anchoring protein. Missing from GOA. Reason: Annotated at the parent term rather than at GO:0034237 protein kinase A regulatory subunit binding, deliberately. What was measured is an interaction between endogenous BIG3 and PKA; the assignment to the RII regulatory subunit comes from in silico identification of two candidate RII-binding domains at residues 318-338 and 1225-1245 and was not itself demonstrated. Annotating the child would assert a subunit specificity the data do not establish, so the claim is pitched at the level the evidence supports and the predicted specificity is recorded as a question instead. No supporting_entities are given for the same reason: naming a specific PKA subunit accession would encode the unproven half of the claim. Anchoring both a kinase and a phosphatase places ARFGEF3 in the same functional class as its paralogues BIG1 and BIG2, which carry AKAP sequences too, so this is the part of the family inheritance that survived the loss of catalysis. Supporting Evidence: PMID:28555617 We found that endogenous BIG3 interacts with PKA in breast cancer cells PMID:28555617 BIG3 contains two potential RII-binding domains (RIIBDs) that bind to the RII subunit of PKA |
| GO:1904410 negative regulation of secretory granule organization | ISS PMID:24711543 BIG3 inhibits insulin granule biogenesis and insulin secreti... | NEW | Summary: The best-characterised physiological role of the protein: it restrains biogenesis of hormone-containing secretory granules, so losing it increases hormone content and stimulated release. Established in mouse, hence ISS. Reason: Chosen at this level rather than as an insulin-specific term because the same effect is seen for two different hormones in two different islet cell types. Knocking BIG3 down in beta cells raises insulin content, granule number and stimulated secretion and accelerates proinsulin processing; depleting it from alpha cells raises glucagon production and secretion, and knockout animals release more glucagon under hypoglycaemic challenge. An insulin-only term would understate that. Chosen as biogenesis-level regulation rather than exocytosis because capacitance measurements localise the change to granule number rather than to the release kinetics of individual granules. That measurement is what decides between two terms that are not interchangeable: GO:1903306 negative regulation of regulated secretory pathway was considered and rejected, because it targets the release step the capacitance data specifically exonerate, and QuickGO's is_a and part_of closure confirms it is neither an ancestor nor a descendant of GO:1904410 - the two are disjoint, so choosing the wrong one asserts a different claim rather than a coarser one. GO:1904410's target GO:0033363 is defined as the assembly and arrangement of a secretory granule, which is exactly what changes when BIG3 is removed. No regulation-of-secretory-granule-maturation term exists, so the accelerated proinsulin and chromogranin A processing seen on knockdown is not separately annotated. Evidence code is ISS with the mouse accession, not IMP: every one of these experiments was performed in mouse cells or mice, and the human protein has never been tested in this pathway. The one human reagent involved is exogenous human BIG3 rescuing the knockdown in a mouse cell line, which is a cross-species rescue and is exactly what ISS is for. Supporting Evidence: PMID:24711543 these results demonstrate that BIG3 negatively modulates insulin granule biogenesis and insulin secretion and participates in the regulation of systemic glucose homeostasis PMID:24711543 BIG3-deficient insulin-secreting cells display increased insulin content and granule number and elevated insulin secretion upon stimulation. PMID:25737957 Depletion of BIG3 in alpha-cells leads to elevated glucagon production and secretion. PMID:25737957 Consistently, BIG3-knockout (BKO) mice display increased glucagon release under hypoglycemic conditions. |
Loading supporting content…
Download this section (compressed HTML)Q: Does purified ARFGEF3, or its isolated Sec7 domain, bind ARF1 or ARF6 at all, and does it inhibit exchange catalysed by BIG1 or BIG2 in trans? This distinguishes a genuine competitive dead GEF - the model its characterisers proposed but never tested - from a protein whose Sec7 region has simply become a structural element.
Suggested experts: Catherine Jackson, Weiping Han
Q: Is the PKA contact really made through the predicted RII-binding domains at residues 318-338 and 1225-1245? The interaction was measured but the subunit specificity was only predicted, which is why this review annotates GO:0051018 rather than its child GO:0034237.
Suggested experts: Toyomasa Katagiri
Q: Are the cytoplasmic, granule, lysosomal and mitochondrial pools of ARFGEF3 the same protein in different cell types, or does the protein have compartment-specific isoforms or modifications? A single comparative study across islet, neuronal and carcinoma cells would settle three currently unannotatable localisations.
Suggested experts: Weiping Han
Q: Is ARFGEF3 an integral membrane protein? UniProt records single-pass topology from a predicted helix at 1492-1512 with ECO:0000255 and marks the membrane localisation ECO:0000305, while PMID:24997568's own analysis found no evidence for any predicted helix being transmembrane. The two computational assessments disagree and neither is a measurement.
Suggested experts: Weiping Han
Q: Should PAINT reconsider what node PTN008950430 donates to Mon2-like members? Yeast MON2 is a WITH/FROM donor for GO:0005085 on this row despite having no annotated SEC7 domain and no IPR000904, and holding the term only by IGI, IPI and ISS. MON2 and ARFGEF3 share IPR015403, so the same node may be donating exchange activity to a second member that does not perform it. Checked on the reciprocal side and it holds; the check is committed as ARFGEF3-bioinformatics/reciprocal_donor_check.py, which selects suspect donors by measurement (a protein donor lacking the term's own InterPro signature) rather than by name, and finds exactly one. MON2 itself RECEIVES GO:0005085 by IBA from PANTHER:PTN008950430 - the same node this ARFGEF3 row comes from - with a WITH/FROM set identical to this row's (19 tokens vs 19), and all of its non-IBA support traces to one reference, PMID:12052896, as ISS plus IGI and IPI. It holds no IDA. That paper is itself careful: it reports Ysl2p/Mon2 as homologous to the Sec7 family and claims only a potential GEF function, not a measured one. So neither of the two members lacking a functional Sec7 catalytic site has ever had exchange activity measured directly, and both inherit the term from the same node.
Suggested experts: Pascale Gaudet
Q: ARFGEF1 and ARFGEF2 carry the identical GO:0005085 IBA row from the identical PANTHER node, and should keep it: their glutamic fingers are intact at E793 and E738 respectively, and both hold their own IDA. Reviews of the three paralogues should diverge on this row for a stated structural reason rather than silently.
Suggested experts: Pascale Gaudet
Experiment: Express and purify the ARFGEF3 Sec7 domain (residues 583-796) alongside the equivalent BIG1 domain as a positive control and a BIG1 E793A mutant as the dead control. Measure ARF1 nucleotide exchange by tryptophan fluorescence or mant-GTP loading, and ARF binding independently by isothermal titration calorimetry or surface plasmon resonance, so that binding and catalysis are separated. Then test whether adding the ARFGEF3 domain slows BIG1-catalysed exchange in trans. The controls matter more than the query here: without the E793A mutant the assay cannot demonstrate that it is capable of reporting zero activity.
Hypothesis: ARFGEF3's Sec7 domain retains ARF binding but cannot catalyse exchange, allowing it to act as a competitive inhibitor of BIG1 and BIG2 at the trans-Golgi network.
Type: biochemical assay
Experiment: Measure ARF1-GTP levels in ARFGEF3-knockout and wild-type islets and MIN6 cells by GGA-GAT domain pull-down, with brefeldin A treatment as the positive control for a reduction in active ARF1. This is the experiment the literature already cites as having been done but which is not locatable in the cited paper, so performing and reporting it would close the review's principal evidential gap.
Hypothesis: The change in insulin granule number on ARFGEF3 loss is caused by altered ARF1 activation at the trans-Golgi network.
Type: biochemical assay
Experiment: Perform proximity labelling with ARFGEF3 as bait in an islet beta cell line, a hippocampal neuron culture and an ERalpha-positive breast cancer line in parallel, and compare the interactomes. Consistent recovery of PHB2 across all three would argue for one conserved function detected in different compartments; cell-type specific partner sets would argue for genuinely distinct roles and would explain the four reported localisations.
Hypothesis: ARFGEF3's sequestering function is not restricted to PHB2 and not restricted to breast carcinoma.
Type: proteomics
Experiment: Replicate the lysosomal assignment in human neurons, ideally iPSC-derived, using a knockout-validated antibody or an endogenously tagged allele rather than overexpression, with LAMP1 co-staining and with the islet granule markers as a negative control. A single mouse report is currently the only evidence, which is why no lysosome term is proposed in this review.
Hypothesis: ARFGEF3 localises to lysosomes in neurons and this pool regulates inhibitory neurotransmission.
Type: imaging
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: No experiment has directly measured whether ARFGEF3 can load GTP onto any ARF protein. The conclusion that it cannot rests on absence of the catalytic glutamate and on the authors' repeated sequence-level statements. PMID:24997568 asserts a prior GST-GAT pull-down demonstration and its cross-reference resolves to PMID:24711543, but no such assay is locatable in that paper's accessible text.
OPEN BIOLOGY
What is known: Established: the Sec7 catalytic glutamate is absent in human and mouse ARFGEF3 and present in 17 of 17 catalytically verified relatives. Not established: that ARFGEF3 fails a direct exchange assay, or whether its Sec7 domain still binds ARF without catalysing exchange.
Significance: Decides whether GO:0032012 should be restored. A dead Sec7 domain that still binds ARF would make ARFGEF3 a genuine regulator of ARF signalling; one that binds nothing would make the whole ARF connection an artefact of the family name.
Gap: ARFGEF3 is reported in four different compartments in four cell types - cytoplasm in breast carcinoma, TGN and secretory granules in islet cells, lysosomes in hippocampal neurons, mitochondria in osteosarcoma lines - and no study has compared them in one system.
OPEN BIOLOGY
What is known: Each individual report is internally consistent and the osteosarcoma paper explicitly contrasts its result with the breast cancer localisation rather than disputing it. What is missing is whether these are genuinely different pools, a single pool differently detected, or an antibody artefact.
Significance: Only the cytoplasmic assignment is supported in human cells, so three of the four compartments cannot presently be annotated. A comparative study would either license three new CC terms or retire them.
Gap: UniProt's SubCell mapping sends SL-0244 Secretory vesicle to GO:0030133 transport vesicle and SL-0245 to GO:0030658, but the secretory-granule and transport-vesicle branches are disjoint in GO, so the mapping places proteins in the constitutive pathway when the SubCell term's own definition describes regulated exocytosis.
OPEN CURATION
What is known: Verified against QuickGO's is_a and part_of closure that GO:0030141 is neither an ancestor nor a descendant of GO:0030133. Not established: how many of the 130685 entries carrying SL-0244 are actually in the regulated pathway, which would need case-by-case review.
Significance: A single vocabulary-level rule, so a fix propagates to every affected entry at once rather than needing per-gene correction.
Gap: The interpro2go line mapping IPR000904 to GO:0005085 fires on a profile match to the Sec7 fold without testing whether the catalytic glutamate is present, so it assigns exchange activity to Sec7-fold pseudo-GEFs.
OPEN CURATION
What is known: Demonstrated for ARFGEF3, where the domain is genuinely present and the catalytic residue genuinely absent. Not quantified: how many other IPR000904 members lack the glutamic finger.
Significance: Gating the mapping on the catalytic motif rather than on the domain match would fix every Sec7-fold pseudo-GEF in one edit, in the same way that a catalytic-residue gate would help the metalloprotease and serine-hydrolase families.
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)