ARHGEF15 (Ephexin-5, Vsm-RhoGEF) is a Dbl-family guanine nucleotide exchange factor for Rho-family GTPases, built from a long disordered N-terminal region, a catalytic DH domain at residues 417 to 601 and an adjacent ephexin-type PH domain. It couples Eph receptor tyrosine kinases to actin remodelling: EphA4 associates with it in vascular smooth muscle and, on ephrin-A1 stimulation, tyrosine-phosphorylates it and raises its exchange activity toward RhoA, driving stress-fibre assembly and smooth-muscle contractility; in neurons it binds EphB2 instead, and EphB2-dependent phosphorylation at Tyr-353 triggers UBE3A-mediated ubiquitination and degradation. Exchange activity has been measured directly for RhoA and, in a family-wide screen and in brain tissue, for CDC42 as well, while RAC1 is negative wherever it has been tested. Which of RhoA and CDC42 dominates appears to be set by context and by the phosphorylation state of that same tyrosine rather than by the catalytic site. Expression is largely vascular, in smooth muscle and endothelium, with a separate developmental role in hippocampal dendrites where the protein restrains excitatory synapse formation. Heterozygous loss-of-function variants cause autosomal dominant brain small vessel disease with osteoporosis, through RhoA/ROCK2 inactivation and F-actin disorganisation in vascular cells and impaired Wnt signalling in osteoblasts.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005085 guanyl-nucleotide exchange factor activity | IBA GO_REF:0000033 | ACCEPT | Summary: Exchange activity placed by PAINT at PTN002656129, a Eukaryota-level node of the ephexin family PTHR12845, from five descendant seeds that all carry experimental evidence of their own. Correct and core. Reason: The local PAINT file interpro/panther/PTHR12845/PTHR12845-paint.tsv shows the underlying IBD at taxon:2759 with seeds MGI:MGI:3045246, O94989, Q12774, Q5VV41 and Q8N5V2. Four of the five have a direct assay in their own records: mouse Arhgef15 by IDA from PMID:21029865, human ARHGEF15 by IDA from PMID:12775584, ARHGEF5 by IDA from PMID:15601624 and ARHGEF16 by IDA from PMID:20679435; NGEF carries the term by EXP from PMID:15848799. Placing a Dbl-family exchange activity at the eukaryotic root of a family whose defining feature is the DH domain is a conservative call, not an over-reach. The target's own accession in the WITH/FROM is the expected marker that experimental grounding exists on the target itself. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002656129 · PTHR12845 ancestral node, taxon:2759 (Eukaryota) SUPPORTS TRANSFER IBD for GO:0005085 dated 20250903 in the local PAINT export. MGI:MGI:3045246 · mouse Arhgef15 SUPPORTS TRANSFER Carries GO:0005085 by IDA from PMID:21029865; the ortholog of the target. UniProtKB:O94989 · human ARHGEF15, the target itself SUPPORTS TRANSFER Carries GO:0005085 by IDA from PMID:12775584, so it is one of the descendant evidences the PAINT curator used. Expected, not circular. UniProtKB:Q12774 · ARHGEF5 / Ephexin3 SUPPORTS TRANSFER Carries GO:0005085 by IDA from PMID:15601624. UniProtKB:Q5VV41 · ARHGEF16 / Ephexin4 SUPPORTS TRANSFER Carries GO:0005085 by IDA from PMID:20679435. UniProtKB:Q8N5V2 · NGEF / Ephexin1 SUPPORTS TRANSFER Carries GO:0005085 by EXP from PMID:15848799. Supporting Evidence: PMID:12775584 tyrosine phosphorylation of Vsm-RhoGEF induced by EphA4 upon ephrin-A1 stimulation enhanced the Vsm-RhoGEF activity for RhoA PMID:21029865 We conclude that E5 activates RhoA but not Rac1 or Cdc42 (Fig S2A). |
| GO:0005085 guanyl-nucleotide exchange factor activity | IDA PMID:12775584 EphA4-mediated Rho activation via Vsm-RhoGEF expressed speci... | ACCEPT | Summary: The founding direct assay: EphA4-dependent tyrosine phosphorylation raises exchange activity toward RhoA in vascular smooth muscle cells. Core function, and the substrate is recorded as a has_input extension because GO's molecular-function branch can no longer name it. Reason: GO:0005085 is the only molecular-function term available. GO:0005089, Rho guanyl-nucleotide exchange factor activity, no longer exists: it and the ARF, Ran, Ras and Ral equivalents have all been merged into GO:0005085, confirmed on QuickGO and on OLS4 with zero disagreements, and the substrate names survive only as synonyms of the general term. This is the exact mirror of the GAP side, where GO:0005097, GO:0005099, GO:0005100 and GO:0008060 merged into GO:0005096. The substrate identity is therefore carried in the has_input extension and in core_functions substrates rather than in the term. The cached publication is abstract-only, so the quoted evidence below is from the abstract; the IDA itself rests on the full text and is not second-guessed. Supporting Evidence: PMID:12775584 tyrosine phosphorylation of Vsm-RhoGEF induced by EphA4 upon ephrin-A1 stimulation enhanced the Vsm-RhoGEF activity for RhoA file:human/ARHGEF15/ARHGEF15-uniprot.txt Guanine nucleotide exchange factor (GEF) that activates RhoA, |
| GO:0005085 guanyl-nucleotide exchange factor activity | IEA GO_REF:0000120 | ACCEPT | Summary: Domain-derived and ortholog-derived electronic call. The weakest of the seven rows for this term, but it points at the right domain rather than at an unrelated fold that happens to share a name. Reason: InterPro IPR000219 is the DH domain, which is the catalytic module of a Dbl-family GEF, and UniProt delimits it at residues 417 to 601 of this protein. This is not the failure mode where a fold or domain name propagates into GO as an activity the protein does not have: the activity has been measured on the human protein directly. IPR047271 is the Ephexin-like signature and is family-level rather than catalytic, so it adds classification but no independent evidence. The Ensembl protein identifier ENSMUSP00000104311 is recorded unresolved rather than mapped to a guessed accession. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB:Q5FWH6 · mouse Arhgef15 SUPPORTS TRANSFER Carries GO:0005085 by IDA from PMID:21029865. InterPro:IPR000219 · DH domain SUPPORTS TRANSFER The catalytic domain itself, delimited at 417-601 in this protein. InterPro:IPR047271 · Ephexin-like SUPPORTS TRANSFER Family-level signature; classification rather than independent evidence. ensembl:ENSMUSP00000104311 · Ensembl mouse protein identifier UNRESOLVED Namespace not resolved here; recorded as supplied rather than substituted. Supporting Evidence: file:human/ARHGEF15/ARHGEF15-uniprot.txt Guanine nucleotide exchange factor (GEF) that activates RhoA, |
| GO:0005085 guanyl-nucleotide exchange factor activity | IMP PMID:36929019 Mutations in ARHGEF15 cause autosomal dominant hereditary ce... | ACCEPT | Summary: Three BSVD5 variants reduce exchange activity and RhoA activation. Loss-of-function evidence for the same activity the IDA row asserts, on the human protein. Reason: UniProt records decreased guanyl-nucleotide exchange factor activity for R21C and V360M and decreased RhoA activation for all three variants including the C-terminal truncation. The knock-in mouse the paper builds is named Arhgef15-e(V368M), which does not match UniProt's human V360M; aligning the human and mouse proteins shows V360 and V368 are the same residue in the two species' numbering, so this is a numbering difference and not a second variant. All three variants lie outside the DH domain, so the mechanism is not direct disruption of the catalytic site. Supporting Evidence: PMID:36929019 In vitro experiments indicated that ARHGEF15 mutations resulted in RhoA/ROCK2 inactivation-induced F-actin cytoskeleton disorganization in vascular smooth muscle cells and endothelial cells file:human/ARHGEF15/ARHGEF15-bioinformatics/RESULTS.md the alignment shows these are the same residue in the two species' numbering, not two different variants. |
| GO:0005085 guanyl-nucleotide exchange factor activity | ISS GO_REF:0000024 | ACCEPT | Summary: Curator transfer from mouse Arhgef15, whose own row for this term is an IDA from PMID:21029865. A transfer from a measurement, not from another inference. Reason: Querying QuickGO by gene product for UniProtKB:Q5FWH6 confirms GO:0005085 with evidence IDA and reference PMID:21029865, assigned by MGI. The donor assay is a GST pulldown with Ephexin1 as the positive control for Rac1 and Cdc42 in the same experiment, so it is a substrate panel and not a single-target readout. Human and mouse Ephexin5 are close enough that the aligned functional residues are identical: the alpha5 triad L562/Q566/R567 of the mouse protein is L554/Q558/R559 in human. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB:Q5FWH6 · mouse Arhgef15 SUPPORTS TRANSFER Donor row is GO:0005085 IDA from PMID:21029865. Supporting Evidence: PMID:21029865 We conclude that E5 activates RhoA but not Rac1 or Cdc42 (Fig S2A). |
| GO:0005085 guanyl-nucleotide exchange factor activity | TAS Reactome:R-HSA-8980691 | ACCEPT | Summary: Reactome places ARHGEF15 in its confirmed list of RHOA GEFs. Consistent with every other line of evidence, and the citation behind it is a family-wide specificity screen. Reason: The cached Reactome entry lists ARHGEF15 among GEFs shown to bind RHOA and catalyze GDP to GTP exchange on RHOA, citing Fukushima et al. 2016 and Mueller et al. 2020. Mueller et al. 2020 is PMID:32203420, whose Supplementary Table 2 records ARHGEF15 as RhoA positive in a screen of all 145 human RhoGEFs and RhoGAPs. That measurement is in no GOA row for this gene, so this TAS is the only place the human record connects to it. Supporting Evidence: file:human/ARHGEF15/ARHGEF15-bioinformatics/RESULTS.md Cited PMIDs on the row |
| GO:0005085 guanyl-nucleotide exchange factor activity | TAS Reactome:R-HSA-9013159 | ACCEPT | Summary: Reactome places ARHGEF15 in its confirmed list of CDC42 GEFs as well. This is the only point in the entire GO record where CDC42 is named for this gene, and it is supported by three independent measurements. Reason: CDC42 is the contested half of this protein's substrate range and the evidence now runs three to two in favour. For: PMID:32203420 scores ARHGEF15 Cdc42 positive in a family-wide screen whose library entry for this gene is the human protein at its full 841 residues rather than a rodent cDNA; PMID:23029280 shows VEGF-induced CDC42 activation abolished by siRNA knockdown in human HUVECs; PMID:40138406 measures CDC42 activation in mouse brain and identifies the selectivity switch. Against: PMID:21029865 found no CDC42 activation in mouse neurons, and UniProt reports from PMID:36929019 that the protein does not activate RAC1 or CDC42. The two negatives are not equivalent in weight to the positives, because the same PMID:32203420 screen scores ITSN1, a textbook CDC42 GEF, negative for all three GTPases, so a negative in this assay class means not detected rather than does not act. Keeping the row is the correct call, but it is kept as a substrate claim that GO can only express through the Reactome reaction and the has_input extension, since the CDC42-specific molecular-function term was merged away. Supporting Evidence: PMID:23029280 VEGF-induced activation of Cdc42 was abrogated by Arhgef15 knockdown in human umbilical vascular ECs (HUVECs, Figure 2B and Figure S4). file:human/ARHGEF15/ARHGEF15-deep-research-affinage.md In endothelial cells it acts downstream of VEGF to activate Cdc42 while potentiating inactivation of RhoJ PMID:40138406 Here, we show that Ephexin5 activates both RhoA and Cdc42 in the brain. file:human/ARHGEF15/ARHGEF15-bioinformatics/RESULTS.md A textbook Cdc42 GEF scoring negative for all three is the screen's own false-negative rate made visible |
| GO:0005515 protein binding | IPI PMID:12775584 EphA4-mediated Rho activation via Vsm-RhoGEF expressed speci... | MODIFY | Summary: Bare protein binding for the one interaction in this record that has a mechanism. EPHA4 is the receptor that recruits and activates the exchange factor, so the informative term is ephrin receptor binding. Reason: The interaction is not an isolated binding event: EPHA4 associates with the exchange factor in quiescent cells and, on ephrin-A1 stimulation, phosphorylates it and raises its activity toward RhoA. That is receptor coupling, and GO:0046875 names it. The comparator check supports the term rather than inventing a use for it: NGEF/Ephexin1, the paralog in the same PANTHER family, already carries GO:0046875 for the same relationship to the same receptor class, albeit by Ensembl projection from mouse Ngef rather than experimentally. UniProt records both Eph partners for this protein, EPHA4 experimentally and EPHB2 by similarity, and GO:0046875 covers both classes, so the replacement does not commit to the A-class receptor alone. Proposed replacements: ephrin receptor binding Supporting Evidence: PMID:12775584 Based on the association of Vsm-RhoGEF with EphA4 in quiescent cells PMID:12775584 tyrosine phosphorylation of Vsm-RhoGEF induced by EphA4 upon ephrin-A1 stimulation enhanced the Vsm-RhoGEF activity for RhoA file:human/ARHGEF15/ARHGEF15-uniprot.txt Interacts with EPHB2 (By similarity). Interacts with EPHA4 |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | REMOVE | Summary: Bare protein binding with PIN1 from a proteome-scale two-hybrid screen. Reproducible in IntAct, uninformative as an annotation. Reason: PIN1 is a phospho-dependent prolyl isomerase and ARHGEF15 is phosphorylated on both tyrosine and serine residues, so the pairing is not implausible; but plausibility is not evidence, and nothing downstream of this screen has tested it. GO:0005515 with no extension says nothing about what the protein does. There is no more informative molecular function to modify to, because the screen assigns ARHGEF15 no activity in the interaction, so the policy for a bare GO:0005515 row lands on removal. Removing it says the annotation is uninformative, not that the interaction is false: the IntAct record is untouched and the partner identity is preserved in this document. |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | REMOVE | Summary: Bare protein binding with LASP1 from the same two-hybrid screen. No mechanism attached. Reason: LASP1 is an actin-binding LIM and SH3 protein, so a link to an actin-remodelling exchange factor is thematically consistent, and the independent AP-MS interactome in PMID:32203420 does give ARHGEF15 three actin-associated GOLD-confidence preys, EZR, MSN and RDX. LASP1 is not among them, so the consistency is thematic rather than corroborative and the row remains a bare screen hit with no molecular function to modify to. Removed under the generic-protein-binding policy: the generic term is uninformative, which does not deny the interaction. |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | REMOVE | Summary: Bare protein binding with CEP55, the most-reported interactor in this record, but from screens only. Reason: UniProt lists NbExp=9 for this pair, which counts assays across studies rather than independent biological validation, and the pair recurs across three interactome papers that share methodology. No functional experiment connects CEP55, a centrosomal and midbody protein, to exchange-factor activity in any system this gene has been studied in, so there is nothing more informative to modify to. Removed under the generic-protein-binding policy: the generic term is uninformative, which does not deny the interaction. |
| GO:0005515 protein binding | IPI PMID:26871637 Widespread Expansion of Protein Interaction Capabilities by ... | REMOVE | Summary: The CEP55 pairing again, from the isoform-resolved two-hybrid screen. Reason: Same partner (CEP55, UniProtKB:Q53EZ4) and same assay platform as the CEP55 row from PMID:25416956; recurrence across screens from one platform is not independent replication. Removed under the generic-protein-binding policy: the generic term is uninformative, which does not deny the interaction. |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | REMOVE | Summary: Bare protein binding with an isoform of PRKG1 from the HuRI binary interactome. Reason: PRKG1 is the cGMP-dependent protein kinase of vascular smooth muscle, which is the tissue this exchange factor was discovered in, and it opposes RhoA/ROCK signalling there, so this is the screen hit most worth following up. That is a reason to test it, not a molecular function to modify the term to: it remains a single two-hybrid observation on one isoform with no functional test. Removed under the generic-protein-binding policy: the generic term is uninformative, which does not deny the interaction. The partner is also named explicitly in suggested_questions, which sets out why this particular screen hit is the one worth testing. |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | REMOVE | Summary: The CEP55 pairing a third time, from HuRI. Reason: Third report of the ARHGEF15-CEP55 pair (UniProtKB:Q53EZ4) from the same class of assay, after PMID:25416956 and PMID:26871637. Three reports of one pairing from one platform is repetition, not replication, and none of them attaches a function to ARHGEF15. Removed under the generic-protein-binding policy: the generic term is uninformative, which does not deny the interaction. |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | REMOVE | Summary: Bare protein binding with GORASP2 from HuRI. No mechanism attached. Reason: GORASP2 is a Golgi stacking protein with no reported connection to Rho signalling or to any system in which ARHGEF15 has been characterised. A reproducible screen hit with no biology attached and no molecular function to modify to. Removed under the generic-protein-binding policy: the generic term is uninformative, which does not deny the interaction. |
| GO:0005737 cytoplasm | ISS GO_REF:0000024 | KEEP AS NON CORE | Summary: Curator transfer of cytoplasmic localisation from rat Arhgef15, whose own row is an IDA from PMID:12775584. Correct but the least informative of the location terms in this record. Reason: QuickGO confirms UniProtKB:D3ZPJ8 carries GO:0005737 by IDA from PMID:12775584. Worth noting for provenance: that is the same paper that is the human IDA source, so this rat row is not independent corroboration of the human record but the same experiments curated against the rat protein. The term is true and unobjectionable; it is non-core because the informative location for a receptor-coupled exchange factor is the membrane, which GO:0005886 already carries by IDA. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:D3ZPJ8 · rat Arhgef15 SUPPORTS TRANSFER Donor row is GO:0005737 IDA from PMID:12775584, the same paper behind the human IDA rows. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-8980691 | KEEP AS NON CORE | Summary: Reactome's compartment assignment for the RHOA GEF reaction. A modelling convention rather than a measurement. Reason: Reactome assigns cytosol to the GEF participants of this reaction as a class, and the same cytosol row appears on NGEF, ARHGEF5 and ARHGEF19 from the identical Reactome reaction. It is not wrong, since the exchange reaction occurs on the cytosolic face, but it carries no gene-specific information and the experimentally determined location for this protein is the plasma membrane and cortical actin. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-9013159 | KEEP AS NON CORE | Summary: The same compartment assignment from the CDC42 GEF reaction. Reason: Duplicate of the cytosol row carried by Reactome's other GEF reaction for this gene, R-HSA-8980691 'RHOA GEFs activate RHOA'; the same reasoning applies, namely that the compartment assignment is inherited from the reaction rather than measured on this protein. |
| GO:0005886 plasma membrane | IDA GO_REF:0000052 | ACCEPT | Summary: Immunofluorescence localisation to the plasma membrane, independently corroborated by a systematic RhoGEF localisation screen. The functionally meaningful location for this protein. Reason: A Dbl-family GEF acts on prenylated GTPases at the membrane, and this protein pairs its DH domain with an ephexin-type PH domain, the canonical membrane-targeting module. PMID:32203420 determined the localisation independently with a Citrine fusion and recorded enrichment on membrane ruffles and peripheral actin, confirmed by a cytochalasin D assay, while noting that plasma membrane localisation in addition cannot be ruled out; the same table lists the Cell Atlas call for this gene as plasma membrane at Supported confidence, which is the evidence stream behind this GOA row. No cortical actin or ruffle term is proposed here, because that construct is overexpressed and the authors hedge their own conclusion. Supporting Evidence: file:human/ARHGEF15/ARHGEF15-bioinformatics/RESULTS.md Their Cell Atlas column for this gene reads `Plasma membrane` |
| GO:0030425 dendrite | IEA GO_REF:0000120 | KEEP AS NON CORE | Summary: Dendritic localisation projected from mouse, where it is an IDA from PMID:21029865 and is restricted to the developing hippocampus. Real, but mouse-derived and developmental. Reason: Mouse Arhgef15 carries GO:0030425 by IDA from PMID:21029865, so the projection has a genuine experimental basis. It is non-core for the human protein because no human measurement supports it and the human expression evidence points elsewhere: UniProt records tissue specificity in coronary artery vascular smooth muscle, HPA reports tissue enhancement in adipose tissue and breast, and PMID:23029280 finds expression in mouse retina to be endothelial with none detectable in neurons. UniProt's own note that the protein is expressed exclusively in dendrites of the developing hippocampus is a By-similarity statement, not a human observation. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:Q5FWH6 · mouse Arhgef15 SUPPORTS TRANSFER Donor row is GO:0030425 IDA from PMID:21029865. UniProtKB-SubCell:SL-0283 · UniProt subcellular location keyword, dendrite SUPPORTS TRANSFER Keyword restatement of the same By-similarity SUBCELLULAR LOCATION line. ensembl:ENSMUSP00000104311 · Ensembl mouse protein identifier UNRESOLVED Namespace not resolved here; recorded as supplied rather than substituted. |
| GO:0030425 dendrite | ISS GO_REF:0000024 | KEEP AS NON CORE | Summary: Curator transfer of the same mouse dendritic localisation. Duplicate of the GO:0030425 IEA row from GO_REF:0000120, reached by a different route. Reason: Same donor (mouse Arhgef15, UniProtKB:Q5FWH6) and the same underlying mouse IDA from PMID:21029865 as the GO:0030425 IEA row from GO_REF:0000120, differing only in being a curator transfer rather than an automatic one. Kept because the mouse evidence is sound, non-core because the human relevance is unestablished. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:Q5FWH6 · mouse Arhgef15 SUPPORTS TRANSFER Donor row is GO:0030425 IDA from PMID:21029865. |
| GO:0032956 regulation of actin cytoskeleton organization | IBA GO_REF:0000033 | ACCEPT | Summary: Actin-cytoskeleton regulation placed at the same Eukaryota-level ephexin node as the exchange activity. This is what the exchange activity is for, and it is core. Reason: The local PAINT export records the IBD for GO:0032956 at PTN002656129, taxon:2759, seeded by O94989, Q12774 and Q8IW93. The human seed is the target's own experimental row, GO:0032956 IMP from PMID:36929019; ARHGEF5 carries the term by IMP from PMID:14662653 and ARHGEF19 by IGI from PMID:20643356. Every downstream readout reported for this gene in any system is cytoskeletal: stress-fibre assembly in smooth muscle, actin polymerisation and motility in endothelium, dendritic spine morphology in neuron, F-actin disorganisation in the disease state. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002656129 · PTHR12845 ancestral node, taxon:2759 (Eukaryota) SUPPORTS TRANSFER IBD for GO:0032956 dated 20260528 in the local PAINT export. UniProtKB:O94989 · human ARHGEF15, the target itself SUPPORTS TRANSFER Carries GO:0032956 by IMP from PMID:36929019, so it is one of the descendant evidences the PAINT curator used. Expected, not circular. UniProtKB:Q12774 · ARHGEF5 / Ephexin3 SUPPORTS TRANSFER Carries GO:0032956 by IMP from PMID:14662653. UniProtKB:Q8IW93 · ARHGEF19 / Ephexin2 SUPPORTS TRANSFER Carries GO:0032956 by IGI from PMID:20643356. Supporting Evidence: PMID:12775584 The requirement of Vsm-RhoGEF for ephrin-A1-induced assembly of actin stress fibers in VSMCs was shown by the overexpression of a dominant-negative form of VSM-RhoGEF and by the depletion of Vsm-RhoGEF using RNA interference. |
| GO:0032956 regulation of actin cytoskeleton organization | IMP PMID:36929019 Mutations in ARHGEF15 cause autosomal dominant hereditary ce... | ACCEPT | Summary: The human experimental row behind the IBA: BSVD5 variants disorganise F-actin in vascular smooth muscle and endothelial cells. Core. Reason: Loss-of-function variants in the human gene produce a cytoskeletal phenotype in the two cell types the protein is expressed in, and the mechanism the paper reports runs through RhoA/ROCK2, which is the pathway the exchange activity feeds. UniProt records for each of the three variants that they cause disorganization of actin cytoskeleton. Supporting Evidence: PMID:36929019 In vitro experiments indicated that ARHGEF15 mutations resulted in RhoA/ROCK2 inactivation-induced F-actin cytoskeleton disorganization in vascular smooth muscle cells and endothelial cells |
| GO:0035025 positive regulation of Rho protein signal transduction | IMP PMID:36929019 Mutations in ARHGEF15 cause autosomal dominant hereditary ce... | ACCEPT | Summary: The one place in this record where GO can still name the GTPase family, because the biological-process branch was never flattened the way the molecular-function branch was. Core. Reason: This row does the work that GO:0005085 can no longer do, and it is worth being precise about how much it does. GO:0032488 Cdc42 protein signal transduction is an is_a descendant of GO:0007266 Rho protein signal transduction, verified on QuickGO and OLS4, so GO:0035025 is satisfied by activating either RhoA or CDC42 and does not distinguish them. It is nevertheless the most specific signed term available: GO:0032489 regulation of Cdc42 protein signal transduction has zero children on both services, so no positive regulation of Cdc42 term exists to be more specific with. Supporting Evidence: PMID:36929019 In vitro experiments indicated that ARHGEF15 mutations resulted in RhoA/ROCK2 inactivation-induced F-actin cytoskeleton disorganization in vascular smooth muscle cells and endothelial cells file:human/ARHGEF15/ARHGEF15-bioinformatics/RESULTS.md There is therefore no *positive* regulation of |
| GO:0051056 regulation of small GTPase mediated signal transduction | TAS Reactome:R-HSA-9012999 | KEEP AS NON CORE | Summary: Reactome's RHO GTPase cycle membership. True, and subsumed by the more specific GO:0035025 row that the human experimental evidence supports. Reason: GO:0035025 is a descendant of GO:0051056, so this row adds no information the record does not already carry more precisely. It is not removed, because a Reactome TAS is a pathway-membership statement rather than a claim about specificity, and it is correct as such; it is marked non-core so the general term is not mistaken for the gene's distinctive process. |
| GO:0051496 positive regulation of stress fiber assembly | ISS GO_REF:0000024 | ACCEPT | Summary: Transfer from rat Arhgef15, whose row is an IDA from PMID:12775584. The specific cytoskeletal readout in the system this protein was discovered in, supported for the human protein by the same paper. Reason: QuickGO confirms UniProtKB:D3ZPJ8 carries GO:0051496 by IDA from PMID:12775584. Unlike the cytoplasm row from the same donor, this one is informative: stress-fibre assembly is the RhoA-dependent readout that defines what the exchange activity does in vascular smooth muscle, and the human abstract states the requirement directly from two opposite perturbations, a dominant-negative construct and RNA interference. Accepted rather than demoted because the transfer restates, in more specific terms, a result that was obtained in the same experiments as the human IDA. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB:D3ZPJ8 · rat Arhgef15 SUPPORTS TRANSFER Donor row is GO:0051496 IDA from PMID:12775584. Supporting Evidence: PMID:12775584 The requirement of Vsm-RhoGEF for ephrin-A1-induced assembly of actin stress fibers in VSMCs was shown by the overexpression of a dominant-negative form of VSM-RhoGEF and by the depletion of Vsm-RhoGEF using RNA interference. |
| GO:0098794 postsynapse | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Ensembl Compara projection of a mouse postsynaptic localisation that rests on IDA and IMP from PMID:28185854. Sound in mouse, unestablished in human. Reason: The donor rows are real experiments, not further inferences: QuickGO shows mouse Arhgef15 with GO:0098794 by both IDA and IMP from PMID:28185854, a paper that manipulates Ephexin5 levels in both directions and images spine outgrowth. Non-core for human for the same reason as the dendrite rows: the neuronal biology is entirely rodent, and the human expression and disease evidence is vascular and skeletal. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:Q5FWH6 · mouse Arhgef15 SUPPORTS TRANSFER Donor rows are GO:0098794 IDA and IMP from PMID:28185854. ensembl:ENSMUSP00000104311 · Ensembl mouse protein identifier UNRESOLVED Namespace not resolved here; recorded as supplied rather than substituted. Supporting Evidence: PMID:28185854 We found that reducing Ephexin5 levels increased spine outgrowth, and increasing Ephexin5 levels decreased spine outgrowth in a GEF-dependent manner |
| GO:0098978 glutamatergic synapse | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Companion projection to the postsynapse row, from the same mouse IDA and IMP evidence in PMID:28185854. Reason: Same donor (mouse Arhgef15, UniProtKB:Q5FWH6) and the same underlying mouse IDA and IMP from PMID:28185854 as the GO:0098794 postsynapse row. Kept because the mouse evidence is direct, non-core because nothing places the human protein at a synapse. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:Q5FWH6 · mouse Arhgef15 SUPPORTS TRANSFER Donor rows are GO:0098978 IDA and IMP from PMID:28185854. ensembl:ENSMUSP00000104311 · Ensembl mouse protein identifier UNRESOLVED Namespace not resolved here; recorded as supplied rather than substituted. |
| GO:0150052 regulation of postsynapse assembly | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Projection of the mouse spine-outgrowth phenotype. The paper behind it shows regulation in both directions, which is why the unsigned term is the right one. Reason: PMID:28185854 reports that lowering Ephexin5 increases spine outgrowth while raising it decreases outgrowth, and also that the protein is required for new spine formation at specific sites, which the authors call a dual role. An unsigned regulation term is therefore more accurate than either signed alternative, and the curator's choice is correct. Non-core for human on the species grounds applied to all the neuronal rows. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:Q5FWH6 · mouse Arhgef15 SUPPORTS TRANSFER Donor rows are GO:0150052 IDA and IMP from PMID:28185854. ensembl:ENSMUSP00000104311 · Ensembl mouse protein identifier UNRESOLVED Namespace not resolved here; recorded as supplied rather than substituted. Supporting Evidence: PMID:28185854 We found that reducing Ephexin5 levels increased spine outgrowth, and increasing Ephexin5 levels decreased spine outgrowth in a GEF-dependent manner |
| GO:2000297 negative regulation of synapse maturation | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Projection of the developmental brake on excitatory synapse formation described in PMID:21029865, where mouse Arhgef15 carries this term by IMP. Reason: The mouse donor row is GO:2000297 IMP from PMID:21029865, the paper that established the EphB2 to Ephexin5 to UBE3A axis restraining excitatory synapse number. The mechanism is well supported in rodent and the human protein retains the machinery it depends on: the EphB2 phosphosite Y361 of mouse maps to human Y353, which UniProt already annotates by similarity. Retention of a phosphosite is not evidence the pathway operates in human, so the row is kept and marked non-core rather than promoted. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:Q5FWH6 · mouse Arhgef15 SUPPORTS TRANSFER Donor row is GO:2000297 IMP from PMID:21029865. ensembl:ENSMUSP00000104311 · Ensembl mouse protein identifier UNRESOLVED Namespace not resolved here; recorded as supplied rather than substituted. Supporting Evidence: PMID:21029865 EphB mediates phosphorylation of Ephexin5 at tyrosine-361 |
| GO:2000297 negative regulation of synapse maturation | ISS GO_REF:0000024 | KEEP AS NON CORE | Summary: Curator transfer of the mouse GO:2000297 IMP from PMID:21029865, duplicating the GO_REF:0000107 Ensembl projection of the same result by a second route. Reason: Same donor (mouse Arhgef15, UniProtKB:Q5FWH6) and the same underlying mouse IMP from PMID:21029865 as the GO:2000297 IEA row from GO_REF:0000107, differing only in being a curator transfer rather than an Ensembl Compara projection. Kept because the rodent mechanism is well supported; non-core because retention of the EphB2 phosphosite is not evidence the pathway operates in human. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:Q5FWH6 · mouse Arhgef15 SUPPORTS TRANSFER Donor row is GO:2000297 IMP from PMID:21029865. |
Loading supporting content…
Download this section (compressed HTML)Q: Does human ARHGEF15 activate CDC42 in vascular smooth muscle, where its RhoA activity was established, or is the CDC42 arm restricted to endothelium and neurons? The three CDC42-positive results come from endothelial cells, a family-wide screen and mouse brain, while both CDC42-negative results come from systems where the RhoA arm dominates. If selectivity is set by the phosphorylation state of Y353 as PMID:40138406 reports, the answer should depend on receptor input rather than on cell type as such, and the two literatures would stop looking contradictory.
Q: Should GO restore a way to state which GTPase a GEF or GAP acts on, given that both the GEF and the GAP molecular-function branches have now been flattened to a single term? GO:0005089 and its four siblings merged into GO:0005085, exactly as GO:0005097, GO:0005099, GO:0005100 and GO:0008060 merged into GO:0005096. The information is not recoverable from the biological-process branch either: Cdc42 protein signal transduction is an is_a descendant of Rho protein signal transduction, and there is no positive regulation of Cdc42 protein signal transduction term at all. For a family of 145 human proteins whose entire functional distinction is substrate choice, that leaves the molecular-function annotation uninformative.
Q: Is the PRKG1 interaction real, and does cGMP-dependent kinase signalling regulate this exchange factor in vascular smooth muscle? Seven bare protein-binding rows in this record, covering five distinct partners, are recommended for removal because the generic term is uninformative, not because the interactions are disputed; the partners are PIN1, LASP1, CEP55, PRKG1 and GORASP2 and they are named here so that removing the annotations does not lose them. PRKG1 is the one whose partner is a central regulator of the tissue the protein was discovered in, and it opposes RhoA/ROCK signalling there. A confirmed interaction would place a known vasodilatory pathway directly onto the exchange factor. It currently rests on one two-hybrid observation against one isoform.
Q: Do the mouse neuronal functions of Ephexin5 operate in human at all? Seven of the thirty-one GOA rows are neuronal or dendritic and every one of them is a projection from mouse. The human expression evidence is vascular and adipose, the human disease is cerebrovascular and skeletal rather than synaptic, and the one human neurological association is a single de novo variant of uncertain significance. The rows are kept because the mouse evidence is strong, but nothing currently distinguishes a conserved function from a rodent-specific one.
Experiment: Measure exchange activity of purified human ARHGEF15 DH-PH against RHOA, RAC1 and CDC42 in a single side-by-side nucleotide-exchange assay, with the L554A/Q558A/R559A triple mutant as the negative control and the phosphorylated and Y353F forms as the selectivity comparison. Every existing substrate call for this protein comes from cell-based effector pulldowns in different laboratories, systems and species, which is why they disagree.
Hypothesis: Human ARHGEF15 catalyses nucleotide exchange on both RHOA and CDC42 but not RAC1, and the balance between the first two is set by phosphorylation of Y353 rather than by the catalytic site.
Type: in vitro guanine nucleotide exchange assay
Experiment: Repeat the PMID:32203420 screen conditions for ARHGEF15 alongside ITSN1 and FGD1. ITSN1 scored negative for all three GTPases in that screen despite being a textbook CDC42 GEF, so the screen's false-negative behaviour is a measurable property that determines how much weight its negative calls can bear, including the RAC1 negative for this gene.
Hypothesis: The negative calls in the PMID:32203420 screen, including RAC1 for ARHGEF15, have a false-negative rate large enough that they cannot be read as evidence of absence.
Type: replication with positive and known-negative controls
Experiment: Test whether ARHGEF15 localises to cortical actin and membrane ruffles at endogenous levels, by endogenous tagging rather than overexpression. PMID:32203420 reports ruffle and peripheral-actin enrichment confirmed by a cytochalasin D assay, and independently gives the protein three ERM-family AP-MS preys, EZR, MSN and RDX; but that localisation used an overexpressed Citrine fusion and the authors explicitly could not exclude plasma membrane in addition, so no new cellular-component annotation is proposed from it here.
Hypothesis: Endogenous ARHGEF15 is enriched at cortical actin and membrane ruffles rather than distributed over the plasma membrane generally.
Type: endogenous tagging and live-cell imaging
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)