ARHGEF19 (WGEF, "weak-similarity GEF"; also called Ephexin-2) is a Dbl-family guanine nucleotide exchange factor built from a tandem DH-PH catalytic module followed by a C-terminal SH3 domain, preceded by a long disordered region that keeps the enzyme autoinhibited. Its established activity is nucleotide exchange on RhoA: in side-by-side pulldowns the human protein raises RhoA-GTP but not Rac1-GTP or Cdc42-GTP, and it co-precipitates strongly with RhoA, weakly with Rac1 and not at all with Cdc42. Exchange is held off by two independent brakes typical of the Ephexin subfamily, an N-terminal inhibitory helix anchored by a conserved tyrosine (Y295 in human) whose phosphorylation releases the DH domain, and a C-terminal SH3-mediated mode; Dishevelled binding relieves the autoinhibited state. Through Dishevelled and Daam1, ARHGEF19 supplies the RhoA activation step of the Wnt/planar-cell-polarity branch, which drives convergent extension during vertebrate gastrulation, kidney tubule morphogenesis, and the actin remodelling and polarised migration that epidermal keratinocytes need to close a wound, where its transcription is directly driven by GRHL3. A separate, RhoA-independent activity has been reported in carcinoma cells, where the DH and PH domains engage BRAF or HRAS and raise MEK/ERK phosphorylation. The protein is expressed most strongly in intestine, liver, heart and kidney, and in skin.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005085 guanyl-nucleotide exchange factor activity | IBA GO_REF:0000033 | ACCEPT | Summary: Correct, and this is already the most specific molecular-function term the ontology offers. The ancestral node sits at the eukaryote root of PTHR12845, every listed donor is a Rho-family GEF, and the human protein has direct experimental exchange evidence of its own that GO has never captured. Reason: The transfer is sound on every axis. All five donor accessions are Ephexin-family RhoGEFs (ARHGEF15, ARHGEF5, ARHGEF16, NGEF, plus mouse Arhgef15), and the target carries an intact DH-PH module: Pfam PF00621 scores 137.0 bits over residues 382-558, inside the 134.6-166.9 range spanned by five Dbl-family GEFs that have both a substrate co-crystal and measured exchange activity, and 24 of the 25 DH positions that contact RhoA in both 1X86 (LARG:RhoA) and 1LB1 (Dbs:RhoA) are present. The bioinformatics is deliberately not treated as proof of activity - identity at that surface is 36% for ARHGEF19 against LARG and also 36% for Tiam1, a GEF for a different GTPase, so it cannot assign a substrate - but it removes any reason to suspect a pseudo-GEF. The activity itself is measured directly in human cells (PMID:18256687, Fig. 2A). The only defect here is one of evidence code, not of term: a gene with a published human IDA is carrying an IBA. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002656129 · PTHR12845 ancestral node, taxon:2759 (Eukaryota) SUPPORTS TRANSFER The PAINT table places this IBD at the eukaryote root while seeding it entirely from mammalian ephexins. The scope is broad relative to the seeds, but nucleotide exchange by a DH-PH module is genuinely ancient, so the breadth is not a defect here. UniProtKB:O94989 · ARHGEF15 / Ephexin-5 (human) SUPPORTS TRANSFER UniProtKB:Q12774 · ARHGEF5 / Ephexin-3 / TIM (human) SUPPORTS TRANSFER UniProtKB:Q5VV41 · ARHGEF16 / Ephexin-4 (human) SUPPORTS TRANSFER UniProtKB:Q8N5V2 · NGEF / Ephexin-1 (human) SUPPORTS TRANSFER MGI:MGI:3045246 · Arhgef15 (mouse) SUPPORTS TRANSFER Supporting Evidence: PMID:18256687 Expression of hWGEF and XWGEF increased the level of active RhoA, whereas expression of hWGEFΔGEF did not file:human/ARHGEF19/ARHGEF19-bioinformatics/RESULTS.md 24 of the 25 RhoA-contacting positions are present |
| GO:0005085 guanyl-nucleotide exchange factor activity | IEA GO_REF:0000002 | ACCEPT | Summary: Correct domain-to-term mapping. IPR000219 is the DH domain and IPR047271 is the Ephexin-like family; both are genuinely present. Reason: The DH domain called by InterPro is confirmed independently by Pfam PF00621 at the gathering threshold (137.0 bits, residues 382-558) and by structural mapping of the exchange surface. This is the weakest evidence code on the row but it is not a false mapping. Worth noting for anyone auditing IEA quality on this protein: the same domain signature also fires on isoform 2 (Q8IW93-2) at 65.5 bits even though that isoform deletes residues 484-783 and so loses half the DH domain plus all of PH and SH3, retaining only 5 of the 25 GTPase-contact positions. A domain-to-term mapping cannot see that, which is an argument for keeping the IEA subordinate to experimental evidence rather than for removing it. Supporting Evidence: file:human/ARHGEF19/ARHGEF19-bioinformatics/RESULTS.md still clears the Pfam gathering threshold |
| GO:0005085 guanyl-nucleotide exchange factor activity | TAS Reactome:R-HSA-8980691 | ACCEPT | Summary: Correct. Reactome places ARHGEF19 in its RHOA GEF set on the strength of the 2004 characterisation paper, which is the same paper GO itself never annotated. Reason: The Reactome reaction "RHOA GEFs activate RHOA" lists its evidence gene by gene, and for this protein it cites Wang et al. 2004, which is PMID:15485661. That paper has produced zero GO annotations in any species, so Reactome's secondary assertion is currently the only route by which its content reaches the GO record at all. Accepting it is right; it would be better still for GO to annotate the primary paper directly. One caution for anyone reading the quotes below: the activity sentence from PMID:15485661 also names Cdc42 and Rac1, and that half of it is superseded - see the finding_review on PMID:15485661, which marks it OVERTURNED by the side-by-side retest in PMID:18256687. Only the RhoA clause and the domain architecture are load-bearing here, and neither was overturned; the quote is left whole rather than trimmed because truncating it at "RhoA" would misrepresent what the sentence says. Supporting Evidence: PMID:15485661 we report a new member of RhoGEF, WGEF, which has the classical structure of DH-PH domain and a C-terminal SH3 domain PMID:15485661 WGEF was shown to activate RhoA, Cdc42, and Rac1 by pulldown assay |
| GO:0005515 protein binding | IPI PMID:18256687 WGEF activates Rho in the Wnt-PCP pathway and controls conve... | MODIFY | Summary: The interaction is real and well characterised, but "protein binding" throws away everything informative about it. The partner is RhoA, this protein's exchange substrate, and the binding was mapped to the DH-PH module. Reason: PMID:18256687 did not merely detect an interaction; it measured specificity across the three candidate GTPases with GST fusions. hWGEF co-precipitated strongly with RhoA, weakly with Rac1 and not at all with Cdc42, and a construct lacking most of DH-PH lost the interaction, placing the binding site in the catalytic module. GO:0031267 small GTPase binding records that the partner is a small GTPase, which GO:0005515 does not. The term one would actually want, GO:0017048 Rho GTPase binding, is obsolete and merged into GO:0031267 - confirmed against EBI OLS4 (is_obsolete true, term_replaced_by GO:0031267), the GO API (no record returned) and this repository's own cache/go/terms.csv (no row), with QuickGO silently resolving the merge and returning the replacement's live record, which is why it was not relied on. GO:0031267 is therefore the most specific available term, not a compromise. Proposed replacements: small GTPase binding Supporting Evidence: PMID:18256687 We found that hWGEF and XWGEF strongly co-precipitated with RhoA at a level comparable to Ephexin, whereas hWGEFΔGEF did not PMID:18256687 hWGEF but not XWGEF showed a very weak interaction with Rac-1, and neither WGEF bound to Cdc42 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MODIFY | Summary: A binary interactome hit with PAK5, a serine/threonine protein kinase. The partner's identity supports a more informative term, but the interaction has had no functional follow-up and should not be treated as core. Reason: Q9P286 is PAK5 (PAK7), a p21-activated serine/threonine kinase, so GO:0019901 protein kinase binding is directly supported by what was detected and is strictly more informative than GO:0005515. UniProt records the pair with NbExp=3, so this is not a single-assay artefact. The caveat is real and is why this stays non-core: the observation comes from a proteome-scale binary screen, no paper has followed it up, and a RhoGEF binding a Rho-effector-family kinase is plausible but unexplained. Refining the term rather than removing the row is the right call - the assay result stands, only its interpretation is thin. Proposed replacements: protein kinase binding Supporting Evidence: PMID:32296183 With approximately 53,000 protein-protein interactions, HuRI has approximately four times as many such interactions as there are high-quality curated interactions from small-scale studies. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-205039 | MARK AS OVER ANNOTATED | Summary: The compartment is not false, but the route by which ARHGEF19 acquired it is. This Reactome reaction produces RAC1-GTP, and human WGEF does not activate Rac1. Reason: R-HSA-205039 is "p75NTR indirectly activates RAC and Cdc42 via a guanyl-nucleotide exchange factor". Its catalyst is a generic set, "guanyl-nucleotide exchange factor activity of GEFs [cytosol]", and its output is RAC1:GTP. ARHGEF19 reaches the reaction only by membership of that undifferentiated GEF set, and the one experiment that tested the question found that human WGEF does not raise Rac1-GTP above control while Ephexin-1 in the same panel does. The cellular-component term is a compartment label attached to the set, not a localisation measurement on this protein, and it inherits a substrate assignment the protein does not support. Removing it would overstate the case, since cytosol is where an unengaged GEF plausibly sits, but a curator should know this row carries no evidence about ARHGEF19 specifically. This is why the three cytosol rows do not all carry the same action: the other two come from RhoA reactions that ARHGEF19 genuinely belongs to, and only this one is grounded in a reaction whose output the protein cannot produce. Supporting Evidence: PMID:18256687 we find that hWGEF and XWGEF did not activate Rac or Cdc42 above control levels, whereas Ephexin did |
| GO:0005829 cytosol | TAS Reactome:R-HSA-419166 | KEEP AS NON CORE | Summary: Reactome compartment convention for a RhoA-GEF reaction, not a localisation experiment on this protein. Keep, but not as a core location. Reason: "GEFs activate RhoA,B,C" is the right pathway for this protein, and Reactome assigns GEF catalytic activity to the cytosol as a modelling convention. No human localisation experiment on ARHGEF19 worth annotating has been published. The only positive localisation data are Xenopus, where WGEF sits preferentially at the plasma membrane and colocalises with cortical actin, plus a qualitative human observation of broad distribution in proliferating SH-SY5Y cells that concentrates into neurites on differentiation. Since a RhoGEF has to reach membrane-anchored RhoA to work, cytosol is at best half the story; it is retained as non-core rather than removed because a soluble pool is entirely plausible. The Affinage deep-research record for this gene grounds its localisation in plasma membrane and cilium; both are declined here, the first because the only evidence is amphibian and the second because the ciliary phenotype is in dog and mouse cell lines whose result the same paper's Xenopus arm did not reproduce. Supporting Evidence: PMID:18256687 These results suggest that XWGEF is associated with the plasma membrane in the Xenopus embryo. file:human/ARHGEF19/ARHGEF19-deep-research-affinage.md GO:0005886 plasma membrane, GO:0005929 cilium |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8980691 | KEEP AS NON CORE | Summary: Same compartment convention as the R-HSA-419166 row, attached to the RHOA-GEF reaction. Keep as non-core. Reason: This is the cellular-component half of the same Reactome reaction whose molecular-function assertion is accepted above. The reaction assignment is appropriate for ARHGEF19; the cytosol label is a property of how Reactome models GEF catalysis rather than an observation, so it should not become a core location. |
| GO:0032956 regulation of actin cytoskeleton organization | IBA GO_REF:0000033 | ACCEPT | Summary: Correct term and sound transfer, and the with/from set is worth reading closely: one of the four supporting entities is ARHGEF19 itself, which marks that experimental grounding for this term exists on the target, and the IBA then adds a claim the IGI does not make. Reason: Actin remodelling is the downstream consequence of RhoA activation and is directly observed for this protein in three systems: forced expression of human WGEF rearranges the actin cytoskeleton, Xenopus WGEF colocalises with cortical actin and its depletion blocks convergent extension, and knockdown of RhoGEF19 in keratinocytes causes defects in actin polymerization that re-expression rescues. The transfer from ARHGEF15 and ARHGEF5 is legitimate. On the self-reference in the with/from set: the PAINT table shows UniProtKB:Q8IW93 among the leaves that seeded PTN002656129 for this term, which is exactly how the mechanism is meant to work - the gene's own IGI was one of the descendant evidences the curator used to place the IBD, so it reappears as a source of the IBA the gene later receives. It is a marker that experimental grounding exists on the target, not a duplicate of it, and the IBA says something the IGI cannot: that regulation of the actin cytoskeleton is an ancestral property of PTHR12845 reaching back to the eukaryote root, rather than a keratinocyte-specific role established once in mouse skin. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002656129 · PTHR12845 ancestral node, taxon:2759 (Eukaryota) SUPPORTS TRANSFER UniProtKB:O94989 · ARHGEF15 / Ephexin-5 (human) SUPPORTS TRANSFER UniProtKB:Q12774 · ARHGEF5 / Ephexin-3 / TIM (human) SUPPORTS TRANSFER UniProtKB:Q8IW93 · ARHGEF19 (the target itself) SUPPORTS TRANSFER The target's own accession appears among the leaves that seeded PTN002656129 for this term, which is the expected shape rather than a defect: the gene's own IGI from PMID:20643356 is one of the descendant evidences the curator used to place the IBD, so it legitimately reappears as a source of the IBA the gene later receives. Read as a marker that experimental grounding exists on the target itself, the IBA adds what the IGI cannot - that the function is inherited rather than lineage-specific. Supporting Evidence: PMID:15485661 forced expression of WGEF resulted in marked rearrangement of the actin cytoskeleton file:interpro/panther/PTHR12845/PTHR12845-paint.tsv PTHR12845 PTN002656129 GO:0032956 P IBD false UniProtKB:O94989|UniProtKB:Q12774|UniProtKB:Q8IW93 taxon:2759 |
| GO:0032956 regulation of actin cytoskeleton organization | IGI PMID:20643356 Epidermal wound repair is regulated by the planar cell polar... | ACCEPT | Summary: Well supported. GRHL3 (Q8TE85) drives RhoGEF19 transcription, and knocking down either one produces the same actin and polarity defects in keratinocytes, with rescue by RhoGEF19 re-expression. Reason: This is the strongest experimental row on the gene and the genetic-interaction qualifier is used correctly: GRHL3 was identified as a direct transcriptional activator of RhoGEF19 by phylogenetic analysis and chromatin immunoprecipitation, and the two knockdowns phenocopy. The rescue experiment is what turns a correlation into a genetic ordering. Note that acts_upstream_of_or_within, rather than involved_in, is the appropriate qualifier here because the readout is a whole-cell actin and polarity phenotype rather than a demonstration that the protein acts within the actin-organising machinery itself. Worth recording that the keratinocytes are human: the abstract says only "keratinocytes", but the full text identifies the line as HaCaT and describes the leading-edge polarity analysis as being done on human cells, so this row rests on human-cell evidence and not on a mouse assay read across. Supporting Evidence: PMID:20643356 Knockdown of Grhl3 or RhoGEF19 in keratinocytes induced defects in actin polymerization, cellular polarity, and wound healing, and re-expression of RhoGEF19 rescued these defects in Grhl3-kd cells PMID:20643356 RhoGEF19-kd cells formed only rudimentary stress fibers that showed irregular organization and were not directed towards the scratched area PMID:20643356 We analyzed the polarity of leading edge cells 6 hours following scratch wounding of human Grhl3-kd, RhoGEF19-kd, and Scr control keratinocytes |
| GO:0042060 wound healing | IBA GO_REF:0000033 | ACCEPT | Summary: Correct, but GOA has the wrong evidence code on it. The donor is mouse Arhgef19 itself, so the orthologue transfer is sound - and the same paper also knocked RhoGEF19 down in a human keratinocyte line and watched the scratch fail to close, so the human annotation should not be inferential at all. Reason: The single gene-product donor resolves to mouse Arhgef19, the direct orthologue, whose experimental annotation comes from PMID:20643356 - the same paper that supports the human IGI two rows above. Mouse and human ARHGEF19 are 1:1 orthologues and the mouse experiment was done in keratinocytes, a cell type the human gene is also expressed in (UniProt records tissue-enhanced expression in skin). This is exactly the case IBA exists for, and the PAINT table shows PTN002656130 seeded by that single gene product and propagated across taxon:117571 (Euteleostomi). The single seed is recorded as a fact about the shape of the ancestral call and explicitly not as a weakness, since donor count is not a proxy for evidential strength and a node seeded by one well-characterised MOD gene can be entirely sound. This one is: the seed is an experimental IMP in the direct orthologue. What is wrong here is the evidence code, and it is the sharpest single instance of this gene's coverage problem. The same paper that produced the mouse IMP also did the experiment in human cells: shRNA against RhoGEF19 in the human keratinocyte line HaCaT left the scratch 20% closed where controls closed completely, activated RhoA fell in those cells, and re-expressing RhoGEF19 in Grhl3-knockdown HaCaT restored closure from 10% to 73%. That is a human knockdown-with-rescue for this exact term, in the paper GOA already cites for the mouse orthologue, and the human gene carries only an IBA. An earlier revision of this review asserted that no human wound-healing experiment on ARHGEF19 had been published; that was wrong, and it was wrong because the cached record was abstract-only and the abstract says "keratinocytes" without naming a species. The full text names it. Propagation Review Root cause: NO FAILURE CORE Sources checked: MGI:MGI:1925912 · Arhgef19 (mouse) - the direct orthologue SUPPORTS TRANSFER Resolved at MGI: MGI:1925912 is mouse Arhgef19, not a paralogue. Its IMP to GO:0042060 derives from PMID:20643356. PANTHER:PTN002656130 · PTHR12845 ancestral node, taxon:117571 (Euteleostomi) SUPPORTS TRANSFER The node carries exactly one seed, MGI:MGI:1925912, and spans Euteleostomi. Recorded as a fact about the shape of the ancestral call and not as a weakness: donor count is not a proxy for evidential strength, and no failure_modes value fits either, because the seed is an experimental IMP in the 1:1 orthologue and is therefore neither inferred, statement-level, stale nor otherwise too weak for confident propagation, which is what SOURCE_EVIDENCE_WEAK would assert. Supporting Evidence: PMID:20643356 we identified RhoGEF19, a homolog of a RhoA activator involved in PCP signaling in Xenopus, as a direct target of GRHL3 PMID:20643356 we utilized the shRNA lentivirus to target the RhoGEF19 transcript in the human keratinocyte cell line, HaCAT PMID:20643356 unlike the Scr cells, which migrated to completely close the scratch within 24 hours PMID:20643356 re-expression of RhoGEF19 substantially rescued the wound healing phenotype observed with the Grhl3-kd cells (10% closure versus 73%) file:interpro/panther/PTHR12845/PTHR12845-paint.tsv PTHR12845 PTN002656130 GO:0042060 P IBD false MGI:MGI:1925912 taxon:117571 |
| GO:0051056 regulation of small GTPase mediated signal transduction | TAS Reactome:R-HSA-9012999 | MODIFY | Summary: Right idea, needlessly generic. Both the Reactome source and the experimental data are Rho-specific and directional, so a Rho-specific positive-regulation term is available and better. Reason: The cited Reactome pathway is R-HSA-9012999 "RHO GTPase cycle", and the reaction through which ARHGEF19 enters it is "RHOA GEFs activate RHOA" - both Rho-specific, and both describing activation rather than regulation of unspecified sign. Independently, the experiments are directional in the same way: overexpression of human WGEF raises RhoA-GTP, and RNAi against it lowers Wnt-1-stimulated RhoA activation in MCF-7 cells. GO:0035025 positive regulation of Rho protein signal transduction is current and non-obsolete in OLS4, the GO API and QuickGO alike, and captures both the family and the sign. Unlike the molecular-function branch, where the GTPase-specific GEF terms have been merged away, the Rho-specific biological-process terms survive, so this refinement is actually available. Proposed replacements: positive regulation of Rho protein signal transduction Supporting Evidence: PMID:18256687 depletion of WGEF with the aid of RNA interference (RNAi) attenuates Rho activation in response to Wnt-1 |
| GO:0060071 Wnt signaling pathway, planar cell polarity pathway | NAS PMID:24431302 Wnt signaling in midbrain dopaminergic neuron development an... | ACCEPT | Summary: The right term on evidence that cannot be checked. It rests on an author statement in a Parkinson's-disease Wnt review whose cached abstract does not mention planar cell polarity or ARHGEF19 at all, while the primary paper that established the fact - complete with human-cell loss of function - is cited by GO exactly once for this gene, for GO:0005515 protein binding with RhoA, and not for this term at all. Reason: There is no doubt the term belongs: WGEF binds Dishevelled through the Dvl PDZ domain and binds Daam1, Frizzled-7 drives Dvl-WGEF colocalisation at the membrane, WGEF overexpression rescues convergent extension suppressed by dominant-negative Wnt-11, depletion blocks convergent extension and is rescued by RhoA or Rho-kinase, and RNAi in human MCF-7 cells attenuates Wnt-1-induced RhoA activation. All of that is PMID:18256687, which GO cites exactly once for this gene - as the reference for the GO:0005515 IPI with RhoA reviewed above - and not for this term. What GO has instead for the pathway is a non-traceable author statement in PMID:24431302, a review of Wnt signalling in midbrain dopaminergic neurons; its abstract, which is all that is cached, does not mention ARHGEF19 or planar cell polarity, so the statement the annotation rests on is in full text I could not access and I do not second-guess the ParkinsonsUK-UCL curator who read it. The annotation is accepted because the term is correct and removing it would leave the pathway unrepresented. The defect is that the evidence code understates what is known, and this schema has no way to propose an evidence-code upgrade on a term already present in GOA, so it is recorded as a curation knowledge gap instead. Supporting Evidence: PMID:18256687 These results indicate that Dvl interacts with WGEF through the PDZ domain. PMID:18256687 Furthermore, WGEF co-immunoprecipitated with N-Daam-1 but not with the C-terminal portion, C-Daam-1 |
| GO:0070374 positive regulation of ERK1 and ERK2 cascade | IDA PMID:29164615 ARHGEF19 interacts with BRAF to activate MAPK signaling duri... | NEW | Summary: Proposed new annotation. Three independent human cancer studies report that ARHGEF19 raises MEK/ERK phosphorylation, and two of them map the effect to the DH and PH domains binding BRAF or HRAS rather than to RhoA. Reason: This is a second activity, explicitly RhoA-independent, that GO does not represent at all. In non-small-cell lung cancer, ARHGEF19 overexpression promotes MEK1/2 phosphorylation, ARHGEF19 co-immunoprecipitates with BRAF, and deletion of either the DH or the PH domain abolishes the effect; in small-cell lung cancer the same two domains are reported to engage HRAS and activate MAPK/ERK in cells and xenografts; in breast cancer ARHGEF19 activates the MAPK pathway and promotes growth. All three are human. Three caveats argue for non-core rather than core status: the lung-cancer papers share an institution, the breast-cancer paper is in a low-visibility journal, and no structural or direct-binding work explains how a DH-PH module could engage a kinase or a Ras GTPase, so the interaction may be indirect. Two of the three nonetheless commit to the same specific domain requirement, which is a mechanistic claim rather than a correlation, and none of these papers has produced a single GO annotation in any species. On the evidence code: IDA is chosen for the arm this row rests on - overexpression of ARHGEF19 raising MEK1/2 phosphorylation, with the BRAF co-immunoprecipitation alongside it. Each of the three papers also has a knockdown arm, and a curator taking this forward may prefer to split the row and record those as IMP; the underlying claim is the same either way. All three sources are cached abstract-only, so no figure-level detail is asserted. Supporting Evidence: PMID:29164615 ARHGEF19 interacted with BRAF and facilitated the phosphorylation of its downstream kinase MEK1/2; both the Dbl homology (DH) and Pleckstrin homology (PH) domains of ARHGEF19 were indispensable for the phosphorylation of MEK1/2 PMID:32993957 ARHGEF19-DH and -PD domain interacts with HRAS and activates the MAPK/ERK pathway in SCLC cells and SCLC xenografts. PMID:34813497 we found that ARHGEF19 could activate the MAPK pathway in breast cancer cells |
Loading supporting content…
Download this section (compressed HTML)Q: Should the 2004 report that WGEF activates Cdc42 and Rac1 (PMID:15485661) be treated as superseded by the 2008 side-by-side retest with an internal positive control (PMID:18256687), or as evidence of genuinely broader specificity under different conditions - for example the autoinhibited versus the released conformation?
Suggested experts: Dawid IB, Sondek J
Q: With the GTPase-family-specific GEF activity terms merged into GO:0005085, is an annotation extension or a GO-CAM has_input the intended way to record that ARHGEF19 is a RhoA-specific exchange factor, and if so should the roughly seventy human Dbl-family GEFs be revisited systematically?
Q: Does the internal PDZ-binding motif at 349-359 replace the earlier N-terminal Dishevelled-binding assignment, or do Dishevelled and ARHGEF19 make two contacts, only one of which each study detected?
Suggested experts: Kulkarni K, Dawid IB
Experiment: Repeat the three-GTPase pulldown panel of PMID:18256687 with purified full-length human ARHGEF19 and defined phosphorylation states of Y295, using RhoA, Rac1, Cdc42 and RhoG, to settle the substrate question quantitatively and to test whether the broader specificity reported in 2004 reflects the autoinhibited versus released conformation rather than a discrepancy between laboratories.
Experiment: Reconcile the two Dishevelled-binding maps directly: measure the affinity of the Dvl2 PDZ domain for full-length human ARHGEF19, for the 1-213 fragment and for the 349-359 peptide in the same assay, then test which mutation abolishes Wnt-induced RhoA activation in cells.
Experiment: Determine whether the DH-PH interaction with BRAF is direct, using purified components, and whether a catalytically inactivated DH domain still supports MEK1/2 phosphorylation. This separates a scaffolding function from an exchange-dependent one and would decide whether the MAPK branch deserves a molecular-function term of its own.
Experiment: Localise endogenous human ARHGEF19 in a keratinocyte or epithelial line with and without Wnt/Frizzled stimulation, by knock-in tagging and fractionation, to replace the Reactome-derived cytosol assignment with a measurement.
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: The best-established fact about this protein - that it is a RhoA exchange factor and specifically not a Rac1 or Cdc42 exchange factor - cannot be stated in a GO term.
OPEN ONTOLOGY
What is known: The biology is not in doubt. The human protein was tested against all three GTPases side by side with Ephexin-1 as an internal positive control: it raised RhoA-GTP, it did not raise Rac1-GTP or Cdc42-GTP, and its binding followed the same order. What is missing is vocabulary.
Significance: GO:0005085 guanyl-nucleotide exchange factor activity is now a leaf: the GTPase-family-specific children (GO:0005089 Rho, GO:0005088 Ras, GO:0030676 Rac) are obsolete and merged into it, and GO:0017048 Rho GTPase binding is likewise merged into GO:0031267 small GTPase binding. Verified across three independent services - OLS4 reports is_obsolete with term_replaced_by, the GO API returns no record, and this repository's cache/go/terms.csv has no row - while QuickGO silently resolves each merge and returns the replacement as a live term, so it cannot be read at face value in either direction. The practical consequence is that every one of the roughly seventy human Dbl-family GEFs now carries the identical molecular-function term regardless of which GTPase it acts on, and the substrate distinction that took a decade of biochemistry to establish is invisible in a GAF.
What would resolve it: No new term is proposed here: the merges look deliberate rather than accidental, and re-proposing what was just collapsed would be arguing with an ontology decision rather than filling a hole. The expressive route that already exists is an annotation extension or a GO-CAM activity with has_input(UniProtKB:P61586), and the useful action is to record the substrate that way rather than to reinstate a term.
Provenance (the field's own admissions):
Gap: Almost none of the published experimental work on this gene has reached GO. Eleven of the fourteen primary papers on ARHGEF19 have produced no GO annotation of any gene in any species, and twelve of fourteen have produced no annotation of ARHGEF19 itself.
OPEN CURATION MF_DARK
What is known: This is not a dark gene in the usual sense - it has a measured activity, a mapped regulatory mechanism, identified partners and a developmental role. The literature exists; the annotations do not.
Significance: The single richest paper, PMID:18256687, yielded exactly one ARHGEF19 annotation - GO:0005515 protein binding with RhoA. The same figure panel that supports that IPI sits beside a RhoA activation assay, a Rac1/Cdc42 negative, a Dishevelled domain mapping, a Daam1 co-immunoprecipitation, a human MCF-7 RNAi experiment and an autoinhibition mapping, none of which reached GO. Uncurated alongside it are the first characterisation of the activity (PMID:15485661), the autoinhibition mechanism (PMID:18537266), the 2024 structural and activity study (PMID:38714795), three cancer-pathway papers, the ciliogenesis study and the only published negative result (PMID:20810787). The consequence is that the gene currently reads as electronically inferred when it is in fact experimentally characterised.
What would resolve it: Direct curation of PMID:18256687 would by itself convert the molecular-function evidence from IBA to IDA, add Dishevelled and Daam1 as partners, and support the planar-cell-polarity term with experiment rather than an author statement. The counts are reproducible from ARHGEF19-bioinformatics/go_coverage_by_reference.py. Even the one paper that did produce annotations is under-curated on the human side: PMID:20643356 knocked RhoGEF19 down in human HaCaT keratinocytes and measured failed wound closure with rescue, yet the human gene carries GO:0042060 only as an IBA projected from the mouse orthologue in the same paper. That single row would convert to a human IMP on evidence already published.
Provenance (the field's own admissions):
Gap: Which region of ARHGEF19 binds Dishevelled is disputed between the two papers that measured it, and neither result has been independently reproduced.
OPEN BIOLOGY
What is known: That Dishevelled binds ARHGEF19 and relieves its autoinhibition is agreed by both. The disagreement is over the site.
Significance: PMID:18256687 mapped binding to the N-terminal 1-213 by deletion analysis, and built its autoinhibition model on that colocation - the Dishevelled-binding region and the inhibitory region being the same region is what makes "binding releases the brake" mechanistically natural. PMID:38714795 instead identifies an internal PDZ-binding motif at residues 349-359 and says explicitly that this contradicts the earlier mapping. The two are not reconcilable as stated, and the second paper's exchange assays were performed on Xenopus rather than human protein because the human constructs were insoluble in E. coli. A further trap for anyone reading the database rather than the paper: UniProt lists PDB 8YR7 under Q8IW93, which looks like a structure of human ARHGEF19, but the authors report that this fusion approach failed to yield the complex, and the WGEF peptide is not usefully resolved.
What would resolve it: Isothermal titration calorimetry or NMR with full-length human ARHGEF19 against the Dvl2 PDZ domain, and a cellular rescue assay comparing the 1-213 deletion with a 349-359 point mutant.
Provenance (the field's own admissions):
Gap: The physiological input that relieves the C-terminal SH3 brake on ARHGEF19 is unidentified, and no Eph receptor partner has ever been found despite the protein being named Ephexin-2.
OPEN BIOLOGY
What is known: The architecture is understood at family level. Ephexins carry two independent autoinhibitory modes: an N-terminal inhibitory helix released by phosphorylation of a conserved tyrosine - Y295 in human ARHGEF19, confirmed present in the current UniProt sequence - and a C-terminal SH3-mediated mode. For Ephexin-4 and SGEF the C-terminal brake is released by PDZ proteins binding a motif at the extreme C-terminus. ARHGEF19 has no such C-terminal motif; it ends in an SH3 domain.
Significance: The family's structural study leaves the activating input for the non-Ephexin-4 members explicitly undefined, and the kinase that phosphorylates Y295 in ARHGEF19 has not been identified either - Src does this for Ephexin-1 downstream of Eph receptors, but the family review is explicit that no Eph receptor interacting with Ephexin-2 has been identified. The protein is therefore named for a receptor connection it has never been shown to have, which is a standing invitation to transfer Eph-pathway annotations to it by homology.
What would resolve it: A phosphoproteomic survey of Y295 occupancy under candidate stimuli, a kinase screen against the isolated N-terminal inhibitory domain, and a pulldown of the ARHGEF19 SH3 domain against a receptor-tail library.
Provenance (the field's own admissions):
Gap: How ARHGEF19 activates MEK/ERK is unexplained, and whether the reported BRAF and HRAS interactions are direct is untested.
OPEN BIOLOGY
What is known: The cellular phenotype is reproduced in three independent human cancer settings, and two of them agree that both the DH and the PH domain are required. Nothing beyond co-immunoprecipitation supports a physical interface.
Significance: A DH-PH module binding a RAF kinase or a Ras GTPase in a manner that does not involve nucleotide exchange would be a genuinely novel use of the fold, and the papers explicitly rule out RhoA as the intermediate. The alternative - that the interaction is indirect and the domain requirement reflects folding or scaffolding rather than contact - is equally consistent with the published data and has not been excluded.
What would resolve it: Recombinant binding assays with purified ARHGEF19 DH-PH against BRAF and HRAS, and a catalytically dead DH mutant to separate scaffolding from exchange.
Provenance (the field's own admissions):
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)