ARHGAP4 (originally C1/RGC1, protein p115) is an X-linked, hematopoietically enriched Rho GTPase-activating protein of the SLIT-ROBO RhoGAP (SRGAP) family. It combines an N-terminal F-BAR membrane-deforming module (residues 19-317), a central Rho-GAP catalytic domain (507-695) carrying the canonical arginine finger at R543, and a C-terminal SH3 domain (746-805). Work on the rat protein shows that recombinant ARHGAP4 accelerates GTP hydrolysis on RAC1, CDC42 and RHOA in vitro, and that the F-BAR module targets it to the leading edge of migrating cells and to axon growth cones, where the GAP domain and C-terminus are required for its inhibition of cell migration and axon outgrowth; no equivalent experiment has been done on the human protein. Endogenous rat protein is cytoplasmic, concentrated just beneath the plasma membrane, associated with the Golgi complex and redistributing onto microtubules during mitosis, and the human protein is cytoplasmic by two independent localisation experiments. It is expressed predominantly in hematopoietic tissues (spleen, thymus, leukocytes) and in the developing and adult nervous system, and it is recovered in the Hem-1 (NCKAP1L) leading-edge complexes of neutrophils. Loss of the gene is well tolerated in both mouse and human, so its non-redundant physiological role remains undefined.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005096 GTPase activator activity | IBA GO_REF:0000033 | ACCEPT | Summary: Phylogenetic propagation of GAP activity from the PANTHER node PTN002306152, placed at Opisthokonta (taxon:33154) in family PTHR14166 and seeded by mouse Srgap2 and Srgap1, rat Arhgap4, human SRGAP2 and C. elegans srgp-1. This is the core molecular function of the protein. Reason: The IBD sits on a deep node and its seed list includes ARHGAP4's own rat ortholog, which carries GO:0005096 by IDA from a direct in vitro assay showing stimulation of RAC1, CDC42 and RHOA GTP hydrolysis. Every WITH/FROM token was resolved against its own authority: MGI:MGI:109605 is Srgap2, MGI:MGI:2152936 is Srgap1, RGD:628901 is rat Arhgap4, UniProtKB:O75044 is SRGAP2 and WB:WBGene00006406 is srgp-1. The catalytic arginine finger is intact at R543, verified by aligning UniProt-delimited Rho-GAP domains onto the structurally resolved anchor of ARHGAP1 (Arg-282, deposited residue 49 of PDB 1TX4, whose guanidinium sits 2.73 angstroms from the AlF4- transition-state analogue). That residue is supporting rather than decisive, because residue identity and curated activity come apart in both directions. Four RHOGAP-domain proteins that Amin et al. 2016 report as lacking the arginine finger nonetheless carry GO:0005096 in GOA, OCRL1 by IDA and ARHGAP36, ARAP2 and FAM13B by IBA, IEA or TAS; and ARHGAP11B, which retains the arginine at the aligned column, carries two curated NOT|enables IDA rows. The alignment run here independently confirms the missing arginine for ARHGAP36, DEPDC1B, OCRL1 and INPP5B, the four of Amin's seven whose Rho-GAP domain UniProt delimits; for ARAP2 and FAM13B the residue status is Amin's assignment rather than one recomputed here. GO:0005096 is substrate-agnostic and is where GO now requires this annotation to sit, so no more specific term is available or requested. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002306152 · PTHR14166 ancestral node, taxon:33154 (Opisthokonta) SUPPORTS TRANSFER GAP activity placed at a deep node of the SLIT-ROBO RhoGAP family. RGD:628901 · rat Arhgap4 SUPPORTS TRANSFER Carries GO:0005096 by IDA from PMID:12414125; the ortholog of the target itself. UniProtKB:O75044 · SRGAP2 SUPPORTS TRANSFER MGI:MGI:109605 · mouse Srgap2 SUPPORTS TRANSFER MGI:MGI:2152936 · mouse Srgap1 SUPPORTS TRANSFER WB:WBGene00006406 · C. elegans srgp-1 SUPPORTS TRANSFER Supporting Evidence: PMID:12414125 In vitro, recombinant ARHGAP4 stimulated the GTPase activity of three members of Rho GTPases, Rac1, Cdc42 and RhoA. |
| GO:0005096 GTPase activator activity | IEA GO_REF:0000107 | ACCEPT | Summary: Ensembl Compara projection from rat Arhgap4 (UniProtKB:A0A0G2JVF0). The rat row being projected is GO:0005096 IDA from PMID:12414125, a direct in vitro GAP assay against RAC1, CDC42 and RHOA. This is the only line in the entire human record that traces back to a measurement of the activity. Reason: Ortholog projection from a genuine experimental row, not a chain of inferences: querying QuickGO by gene product for the donor confirms UniProtKB:A0A0G2JVF0 carries GO:0005096 with evidence IDA and reference PMID:12414125, assigned by RGD. Human ARHGAP4 has no experimental molecular-function annotation of its own, so removing or demoting this row would leave the human entry with no evidential connection to the one direct activity measurement that exists for the protein. The Ensembl donor accession is an unreviewed TrEMBL entry, which is a bookkeeping feature of how Compara keys rat proteins and not a weakness of the underlying RGD curation. The token ensembl:ENSRNOP00000069460 is an Ensembl protein identifier and is recorded unresolved rather than mapped to a guessed accession. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB:A0A0G2JVF0 · rat Arhgap4 (Ensembl Compara donor) SUPPORTS TRANSFER Donor row is GO:0005096 IDA, PMID:12414125, assigned by RGD. ensembl:ENSRNOP00000069460 · Ensembl rat protein identifier UNRESOLVED Namespace not resolved here; recorded as-is rather than substituted. Supporting Evidence: PMID:12414125 In vitro, recombinant ARHGAP4 stimulated the GTPase activity of three members of Rho GTPases, Rac1, Cdc42 and RhoA. |
| GO:0005096 GTPase activator activity | TAS PMID:8570618 An X chromosome-linked gene encoding a protein with characte... | ACCEPT | Summary: The 1996 cloning paper for human C1/p115, which identified a GAP-homologous region by sequence similarity and observed that overexpression disrupted stress fibres. The conclusion drawn was that p115 may down-regulate Rho-like GTPases in hematopoietic cells. Reason: The term is correct and is independently supported by the rat IDA and by the IBA, so the row should stand. Its evidential weight, however, is lower than its TAS code suggests: this paper reports no GAP assay. The authors inferred GAP function from two homology regions plus a cellular phenotype, so the row records a correct author statement rather than a measurement. Nothing here is a reason to modify or remove it, but a curator reading only this row should not treat it as a second, independent demonstration of catalysis. Supporting Evidence: PMID:8570618 It is characterized by two regions of similarity to motifs present in known proteins: GAP and SH3 homologous regions. PMID:8570618 Its localization in a narrow cytoplasmic region just below the plasma membrane and its inhibitory effect on stress fiber organization indicate that p115 may down regulate rho-like GTPases in hematopoietic cells. |
| GO:0005096 GTPase activator activity | TAS Reactome:R-HSA-9013144 | ACCEPT | Summary: Reactome places ARHGAP4 in the reaction "RAC1 GAPs stimulate RAC1 GTPase activity", which is a substrate-specific claim that the GO term itself cannot carry. Reason: This row is the clearest illustration of the ontology limitation affecting this gene. Reactome commits to RAC1 as the substrate; GO:0005096 is substrate-agnostic, so the commitment survives only in the Reactome reaction identifier and is invisible to anything reading the GO term. The substrate-specific GAP terms that would once have carried it are gone. Resolving them across three independent services (OLS4, the GO API at api.geneontology.org, and this repository's own cache/go/terms.csv) shows GO:0005100 Rho, GO:0005099 and GO:0005098 Ras, GO:0008060 ARF, GO:0005097 Rab and GO:0046582 Rap GAP activity all merged into GO:0005096, whose old names survive only as synonyms, and GO:0017048 Rho GTPase binding merged into GO:0031267 so the binding route loses Rho specificity too. GO:0005096 has zero is_a children. It is not childless, since QuickGO reports GO:1902773 GTPase activator complex as a child via capable_of, but a complex capable of the activity is not a more specific activity. QuickGO is not used as an authority for the merge status because it silently resolves merged ids to the surviving term flagged as not obsolete, which it did for seven of the nine ids queried. The merge was deliberate, so no replacement term is proposed and no new term is requested; the substrate is recorded as an RO:0002233 has_input extension and in core_functions substrates instead. |
| GO:0005515 protein binding | IPI PMID:16417406 Hem-1 complexes are essential for Rac activation, actin poly... | REMOVE | Summary: Interaction with NCKAP1L (Hem-1), a scaffold of the neutrophil leading-edge complexes that include the WAVE2 complex. UniProt curates this as the only SUBUNIT statement for this gene, with experimental evidence. Reason: GO:0005515 states nothing about what ARHGAP4 does. Repository policy routes generic protein binding to MODIFY when the cited paper supports a more informative molecular function and otherwise to REMOVE, and this row falls on the second branch. That there is nothing to modify to is established by walking the branch rather than by a text search, since GO search is token-based and a failed search is not evidence of absence. A complete enumeration of the branch, unbounded in depth and deduplicated, returns 105 distinct descendants of GO:0044877 protein-containing complex binding, reached over 106 parent-child edges with every descendant within four edges of the root, among them exactly one actin-machinery term, GO:0071933 Arp2/3 complex binding, and no term for binding the WAVE/SCAR or Hem-1 complex. GO:0031209 SCAR complex exists, but as a cellular component rather than as something to bind. The paper assigns ARHGAP4 no function within the complex, and describes the complexes containing the additional proteins as biochemically separable from the WAVE2 complex proper, so neither Arp2/3 complex binding nor SCAR complex membership would be defensible either. Removal does not mean the interaction is false or the curation unsound. UniProt records it as this protein's only SUBUNIT statement, with experimental evidence, and the partner identity is exactly the content GO:0005515 discards; it is preserved in this review and in the description rather than in an annotation that says nothing. The cached record for this reference is abstract-only and does not name ARHGAP4, which is expected for a proteomic inventory whose component list is in the full text; that is not a reason to doubt the curator, and no part of this decision rests on it. Supporting Evidence: PMID:16417406 A subset of these leading edge complexes are biochemically separable from the WAVE2 complex and contain a diverse set of potential polarity-regulating proteins. |
| GO:0005737 cytoplasm | IBA GO_REF:0000033 | ACCEPT | Summary: Propagation from PANTHER node PTN002680572 at Euteleostomi (taxon:117571). Its only seed is the ARHGAP4 accession itself. Reason: A self-referential WITH/FROM is the expected marker that the PAINT curator judged the assignment core on the strength of an experimental annotation on the target itself, and ARHGAP4 does carry cytoplasm by IDA from a fluorescent-fusion localisation study. It is neither circular nor redundant, and is explicitly not classified as CIRCULAR_OR_REDUNDANT here. The node placement is also informative in what it excludes the gene from. The sibling node PTN002680580, at the same taxonomic level, carries GO:0043197 dendritic spine and GO:0098978 glutamatergic synapse, and ARHGAP4 does not inherit it. The same curator therefore placed ARHGAP4 outside the synaptic compartment of this family while keeping it inside the general cytoplasmic one. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002680572 · PTHR14166 ancestral node, taxon:117571 (Euteleostomi) SUPPORTS TRANSFER UniProtKB:P98171 · ARHGAP4 (the target itself) SUPPORTS TRANSFER Expected marker of experimental grounding on the target, not circularity. |
| GO:0005737 cytoplasm | IDA GO_REF:0000054 | ACCEPT | Summary: Direct localisation of a fluorescently tagged ARHGAP4 fusion in living cells (LIFEdb). Reason: A direct observation in human cells, concordant with the HPA immunofluorescence cytosol row, with the curated UniProt Cytoplasm subcellular-location statement and with the rat data. This is one of only two experimental annotations of any kind on the human protein, both of them localisation. |
| GO:0005737 cytoplasm | IEA GO_REF:0000044 | ACCEPT | Summary: Automatic transfer of the UniProt subcellular-location keyword SL-0086 Cytoplasm. Reason: A faithful restatement of the curated UniProt SUBCELLULAR LOCATION line, which itself rests on the 1996 localisation work, and concordant with two independent experimental rows on the same term. Redundant with them rather than adding to them, but not incorrect. |
| GO:0005737 cytoplasm | TAS PMID:8570618 An X chromosome-linked gene encoding a protein with characte... | ACCEPT | Summary: Author statement from the 1996 cloning paper, which localised p115 to a narrow cytoplasmic region just below the plasma membrane. Reason: The term is correct but loses the informative part of the observation. The paper describes what GO defines as the cell cortex, and GO:0005938 cell cortex is an active term, so a more specific replacement was considered and then dropped on evidence. A comparator check asks what seven well-curated RhoGAPs actually carry: none of ARHGAP1, ARHGAP35, ARHGAP17, ARHGAP21, ARHGAP24, SRGAP2 or DLC1 is annotated to GO:0005938, while four of the seven carry GO:0005737. A zero-of-seven proposal is idiosyncratic rather than a gap in the record, so the row stands as annotated and the sub-plasma-membrane localisation is recorded as a knowledge gap instead. What would change this: a direct cortical-localisation experiment on human ARHGAP4, or GO:0005938 appearing on comparable RhoGAPs. Supporting Evidence: PMID:8570618 Its localization in a narrow cytoplasmic region just below the plasma membrane and its inhibitory effect on stress fiber organization indicate that p115 may down regulate rho-like GTPases in hematopoietic cells. Knowledge gap: Whether human ARHGAP4 is concentrated at the cell cortex is stated only as a 1996 author observation and has never been re-examined, so the compartment in which the GAP acts is not established at better than whole-cytoplasm resolution. OPEN BIOLOGYCURATION CC_DARK Resolve: Immunofluorescence or live-cell imaging of endogenous ARHGAP4 in a hematopoietic cell type, with cortical, Golgi and microtubule markers. "Its localization in a narrow cytoplasmic region just below the plasma membrane and its inhibitory effect on stress fiber organization indicate that p115 may down regulate rho-like GTPases in hematopoietic cells." — PMID:8570618 "We have found that the RHOGAP domain itself is nonselective and in some cases rather inefficient under cell-free conditions." — PMID:27481945 |
| GO:0005829 cytosol | IDA GO_REF:0000052 | ACCEPT | Summary: Human Protein Atlas immunofluorescence on the endogenous protein. Reason: A direct observation on human cells and the only antibody-based localisation in the record. Concordant with the LIFEdb fusion-protein row and with UniProt. Together these two rows are the entirety of the experimental evidence on human ARHGAP4, which is itself the central finding of this review. |
| GO:0005829 cytosol | TAS Reactome:R-HSA-9013144 | ACCEPT | Summary: The compartment Reactome assigns to the RAC1-GAP reaction that ARHGAP4 participates in. Reason: Concordant with the HPA immunofluorescence row on the same term, so the annotation stands. It should not, however, be counted as a second independent observation, since it follows from how Reactome compartmentalises a reaction rather than from an experiment locating ARHGAP4. |
| GO:0005874 microtubule | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Ensembl Compara projection of a rat IDA row (GO:0005874, PMID:12414125). In that experiment the endogenous protein was principally at the Golgi complex and redistributed onto microtubules under particular conditions, such as mitosis. Reason: The projection is sound and the underlying rat evidence is a direct observation, but the localisation is conditional rather than constitutive, so it should not be read as a resting-state compartment for the protein. This row also exposes a curation gap worth recording: the same figure of the same paper reports Golgi localisation in its title, and no GO:0005794 annotation exists for Arhgap4 in rat or for ARHGAP4 in human, so only half of one experiment reached the ontology. No Golgi term is proposed here, because the human protein has no Golgi evidence of its own and the correct place to fix this is an IDA row on the rat gene, which would then project to human by the same route as this one. Propagation Review Root cause: NO FAILURE NON CORE Failure modes: GRANULARITY MISMATCH Sources checked: UniProtKB:A0A0G2JVF0 · rat Arhgap4 (Ensembl Compara donor) SUPPORTS TRANSFER Donor row is GO:0005874 IDA, PMID:12414125. The projection is faithful; what is lost is that the observation is conditional. Supporting Evidence: PMID:12414125 Subcellularly, endogenous ARHGAP4 expression localized to the Golgi complex and could redistribute to the microtubules, for example during mitosis. |
| GO:0007010 cytoskeleton organization | TAS PMID:8570618 An X chromosome-linked gene encoding a protein with characte... | MODIFY | Summary: Author statement from the 1996 paper, whose cytoskeletal observation was that overexpressed p115 inhibited stress fibre organisation. Reason: The annotation asserts that ARHGAP4 organises the cytoskeleton, whereas the cited observation is that it negatively regulates a specific cytoskeletal structure. A RhoGAP switches Rho signalling off, so a regulation term is the correct shape. A comparator check supports the change rather than merely my reading of it: of seven well-curated RhoGAPs, not one carries GO:0007010, so ARHGAP4 is the outlier here, while GO:0051497 is carried by DLC1 with IDA evidence from the same kind of stress-fibre experiment and GO:0032956 regulation of actin cytoskeleton organization is carried by ARHGAP35 and ARHGAP17. GO:0051497 is proposed because it matches the observation exactly; GO:0032956 would be the conservative alternative if a reviewer prefers not to commit to stress fibres specifically from an overexpression phenotype. The evidence code should stay TAS, since this remains an author statement. Proposed replacements: negative regulation of stress fiber assembly Supporting Evidence: PMID:8570618 Its localization in a narrow cytoplasmic region just below the plasma membrane and its inhibitory effect on stress fiber organization indicate that p115 may down regulate rho-like GTPases in hematopoietic cells. |
| GO:0007165 signal transduction | IEA GO_REF:0000002 | KEEP AS NON CORE | Summary: InterPro2GO mapping from IPR000198, the Rho GTPase-activating protein domain. Reason: The term is correct but maximally uninformative, and it discards the one thing the matched domain specifies, namely that the signalling regulated is Rho signalling. Replacing it with GO:0035023 was considered and dropped: all seven RhoGAP comparators carry GO:0007165 by IEA from the same mapping, and only one carries GO:0035023, so rewriting a single gene's row would be idiosyncratic and would not survive the next InterPro2GO refresh. The right place to fix this is the IPR000198 mapping itself, not this annotation. Retained as non-core because GO:0051056 already states the same thing more precisely. |
| GO:0007266 Rho protein signal transduction | TAS PMID:8570618 An X chromosome-linked gene encoding a protein with characte... | KEEP AS NON CORE | Summary: Author statement from the 1996 paper that p115 may down-regulate Rho-like GTPases in hematopoietic cells. Reason: The term names the pathway ARHGAP4 acts on rather than what it does to it, and the obvious reading is that a GAP regulates Rho signal transduction instead of participating in it. I checked that reading rather than acting on it, and it does not hold up as a curation convention: two of seven well-curated RhoGAPs carry GO:0007266 (ARHGAP1 by TAS, ARHGAP35 by IBA), which is as many as carry GO:0035024, so placing a RhoGAP on the transduction term is established practice and not an error to correct. No replacement is proposed. It is marked non-core because the informative statement about this gene is the GAP activity itself, recorded on GO:0005096, and because the supporting author statement is explicitly hedged. What would change this: the GO consortium moving RhoGAPs off GO:0007266 onto regulation terms, which would show up as the comparator counts inverting. Supporting Evidence: PMID:8570618 Its localization in a narrow cytoplasmic region just below the plasma membrane and its inhibitory effect on stress fiber organization indicate that p115 may down regulate rho-like GTPases in hematopoietic cells. |
| GO:0007399 nervous system development | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: Propagation from PANTHER node PTN008351815, seeded by mouse Srgap2, rat Arhgap4, human SRGAP2 and SRGAP2C. The seed list does include the lineage of the target itself. Reason: The term is broad and the ARHGAP4-lineage contribution to the node is weak in kind rather than absent. The rat Arhgap4 GO:0007399 row is IEP, that is, evidence from an expression pattern, which establishes that the gene is expressed throughout the developing and adult nervous system but not that the protein does any of the work of nervous system development. The rest of the seed list is the SRGAP branch, where the neuronal role is genuinely established. Retained because the family-level judgement is reasonable and the term is general enough to be safe, but not core: the Arhgap4 knockout mouse returns zero significant IMPC genotype-to-phenotype associations across 413 statistical tests, and human deletions removing the whole gene have not produced a neurological phenotype attributed to it. Propagation Review Root cause: NO FAILURE NON CORE Failure modes: SOURCE EVIDENCE WEAK Sources checked: PANTHER:PTN008351815 · PTHR14166 ancestral node, no taxon scope recorded SUPPORTS TRANSFER RGD:628901 · rat Arhgap4 SUPPORTS SOURCE BUT NOT TARGET Its GO:0007399 row is IEP, an expression pattern, which does not establish participation. UniProtKB:O75044 · SRGAP2 SUPPORTS TRANSFER UniProtKB:P0DJJ0 · SRGAP2C SUPPORTS TRANSFER Hominin-specific partial duplicate of SRGAP2; a derived family member, not a deep signal. MGI:MGI:109605 · mouse Srgap2 SUPPORTS TRANSFER Supporting Evidence: PMID:12414125 ARHGAP4 mRNA expression was observed in multiple tissues with marked expression throughout the developing and adult nervous systems. |
| GO:0007399 nervous system development | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Ensembl Compara projection of the rat Arhgap4 GO:0007399 row, which is IEP from PMID:12414125. Reason: The projection is mechanically correct, but the row it projects is expression-pattern evidence. Marked expression throughout the nervous system is a fact about where the gene is transcribed, and the annotation it produces asserts involvement in development, which the experiment does not test. Retained because it is not wrong and because the SRGAP-family context makes a neural role plausible, but it carries less weight than its involved_in qualifier suggests and duplicates the IBA row above. Propagation Review Root cause: SOURCE WEAK OR INFERRED Failure modes: SOURCE EVIDENCE WEAK Sources checked: UniProtKB:A0A0G2JVF0 · rat Arhgap4 (Ensembl Compara donor) SOURCE WEAK OR INFERRED Donor row is GO:0007399 IEP; an expression pattern is not participation evidence. Supporting Evidence: PMID:12414125 ARHGAP4 mRNA expression was observed in multiple tissues with marked expression throughout the developing and adult nervous systems. |
| GO:0010764 negative regulation of fibroblast migration | IEA GO_REF:0000107 | MARK AS OVER ANNOTATED | Summary: Ensembl Compara projection of the rat Arhgap4 GO:0010764 IMP row, which comes from a gain-of-function experiment in NIH/3T3 mouse fibroblasts. Reason: The biology is right but the cell type in the term is an artefact of the assay. NIH/3T3 fibroblasts were the reporter cell line for an overexpression experiment; they are not a setting in which ARHGAP4 normally acts. The protein is expressed predominantly in hematopoietic tissues and in the nervous system, so an annotation that commits specifically to fibroblast migration asserts a cell-type specificity the data do not support. The general claim is already captured, and captured better, by the GO:0030336 IBA row, which is why no replacement term is proposed here: moving this row to GO:0030336 would only duplicate that one. Marked over-annotated rather than removed because the underlying rat IMP is a real experiment whose result is not in doubt. Propagation Review Root cause: TERM SCOPING PROBLEM Failure modes: CONTEXT OR TISSUE MISMATCH Sources checked: UniProtKB:A0A0G2JVF0 · rat Arhgap4 (Ensembl Compara donor) SUPPORTS SOURCE BUT NOT TARGET The rat IMP is sound; the fibroblast-specific term generalises an assay cell line into a biological context. Supporting Evidence: PMID:17804252 Our studies show that ARHGAP4 inhibits the migration of NIH/3T3 cells and the outgrowth of hippocampal axons. |
| GO:0030336 negative regulation of cell migration | IBA GO_REF:0000033 | ACCEPT | Summary: Propagation from PANTHER node PTN001021265 at Eumetazoa (taxon:6072), seeded by rat Arhgap4, human SRGAP2 and C. elegans srgp-1. This is the best-supported biological process on the gene. Reason: The seed set spans three phyla and includes the rat ortholog of the target itself, whose GO:0010764 IMP comes from a direct migration assay. It is also the one process on which the independent human literature converges: separate groups report migration and invasion effects of ARHGAP4 in pancreatic, colorectal and mammary epithelial cells, and the ARHGAP4-SEPT2-SEPT9 complex is reported to modulate integrin-mediated focal adhesions and migration. None of that literature produced a GO annotation, so the IBA is carrying the weight of a body of work that is otherwise invisible in GOA. The cell-type-agnostic term is the right level, given that the underlying experiments span fibroblasts, epithelial cells and neurons. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN001021265 · PTHR14166 ancestral node, taxon:6072 (Eumetazoa) SUPPORTS TRANSFER RGD:628901 · rat Arhgap4 SUPPORTS TRANSFER Carries GO:0010764 by IMP from PMID:17804252; the ortholog of the target itself. UniProtKB:O75044 · SRGAP2 SUPPORTS TRANSFER WB:WBGene00006406 · C. elegans srgp-1 SUPPORTS TRANSFER Supporting Evidence: PMID:17804252 Our studies show that ARHGAP4 inhibits the migration of NIH/3T3 cells and the outgrowth of hippocampal axons. PMID:34524873 ARHGAP4, one of the Rho-GAPs, forms a complex with SEPT2 and SEPT9 |
| GO:0030426 growth cone | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Ensembl Compara projection of the rat Arhgap4 GO:0030426 IDA row. The rat observation comes from a structure-function analysis showing that the F-BAR domain targets the protein to the leading edges of migrating cells and to axon growth cones. Reason: A real direct observation, and mechanistically coherent given that F-BAR domains bind curved membranes, which is what a growth cone and a lamellipodial leading edge present. Kept non-core rather than accepted as core because the localisation was established through domain constructs in a heterologous structure-function setting, because the human protein has no localisation data of its own beyond whole-cytoplasm, and because ARHGAP4 is expressed predominantly in hematopoietic tissue where growth cones do not arise. The analogous statement that matters for the hematopoietic compartment, leading-edge localisation in a migrating leukocyte, has no GO term applied and no direct experiment. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:A0A0G2JVF0 · rat Arhgap4 (Ensembl Compara donor) SUPPORTS TRANSFER Donor row is GO:0030426 IDA, PMID:17804252. Supporting Evidence: PMID:17804252 Our structure/function analyses show that the FCH domain appears to be important for spatially localizing ARHGAP4 to the leading edges of migrating NIH/3T3 cells and to axon growth cones. |
| GO:0030517 negative regulation of axon extension | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Ensembl Compara projection of the rat Arhgap4 GO:0030517 IMP row, from an experiment in which ARHGAP4 inhibited the outgrowth of hippocampal axons and in which the GAP domain and C-terminus were required for that inhibition. Reason: This is the best mechanistic experiment in the ARHGAP4 literature, because it ties a phenotype to specific domains rather than merely to the protein. It is nonetheless a gain-of-function result. There is no loss-of-function counterpart in any species. The Arhgap4 knockout mouse returns zero significant IMPC genotype-to-phenotype associations across 413 statistical tests, and humans hemizygous for deletions removing the entire gene have no reported axonal or neurodevelopmental phenotype attributable to ARHGAP4. An overexpressed RhoGAP will suppress neurite outgrowth fairly generically, so the experiment establishes that ARHGAP4 is sufficient to inhibit axon extension rather than that it normally does so. Retained as a real and well-executed finding, marked non-core because sufficiency without necessity does not establish a physiological role. Propagation Review Root cause: NO FAILURE NON CORE Failure modes: SOURCE EVIDENCE WEAK Sources checked: UniProtKB:A0A0G2JVF0 · rat Arhgap4 (Ensembl Compara donor) SUPPORTS TRANSFER Donor row is GO:0030517 IMP, PMID:17804252; a gain-of-function experiment. Supporting Evidence: PMID:17804252 Our analyses also show that the GAP domain and C-terminus are necessary for ARHGAP4-mediated inhibition of cell and axon motility. |
| GO:0051056 regulation of small GTPase mediated signal transduction | TAS Reactome:R-HSA-9012999 | ACCEPT | Summary: The Reactome placement of ARHGAP4 in the RHO GTPase cycle. Reason: Correct and pitched at the right level of generality for a GAP whose measured substrate range spans RAC1, CDC42 and RHOA. Of the process terms on this gene it is the one that states what the protein does without committing to a cell type, a substrate or a downstream phenotype, and it subsumes the uninformative GO:0007165 IEA row. |
| GO:0051963 regulation of synapse assembly | IBA GO_REF:0000033 | MARK AS OVER ANNOTATED | Summary: Propagation from PANTHER node PTN008351815. Unlike every other inherited assertion on this gene, the seed list contains no ARHGAP4-lineage member: it is mouse Srgap2, mouse Srgap3, and the hominin-specific paralogs SRGAP2C and SRGAP2B. Reason: Three independent observations point the same way. First, the evidence base is entirely the SRGAP branch, and two of its four seeds are SRGAP2B and SRGAP2C, partial duplicates of SRGAP2 that arose on the human lineage, so the assertion rests on the most derived members of the family rather than on a deep conserved signal. Second, the same PAINT curator placed the synaptic cellular-component terms GO:0043197 dendritic spine and GO:0098978 glutamatergic synapse on a different node, PTN002680580, which ARHGAP4 does not inherit; the boundary excluding ARHGAP4 from the synaptic compartment of this family was therefore drawn deliberately, which makes inheriting the synaptic process term the anomaly. Third, no experiment in any species links ARHGAP4 or its orthologs to synapse assembly, the Arhgap4 knockout mouse returns zero significant IMPC associations across 413 statistical tests, and ARHGAP4 is expressed predominantly in hematopoietic tissue. This is marked over-annotated rather than removed, because an IBA encodes a considered judgement about where in the tree the function arose and that judgement deserves argument rather than deletion. What would reverse this: any experimental synapse phenotype for ARHGAP4 or a non-SRGAP ortholog, or a PAINT revision that moves this term to a node ARHGAP4 demonstrably descends from on independent evidence. Propagation Review Root cause: PROPAGATION BAD Failure modes: WRONG ORTHOLOG OR PARALOG CONTEXT OR TISSUE MISMATCH Sources checked: PANTHER:PTN008351815 · PTHR14166 ancestral node, no taxon scope recorded SOURCE WEAK OR INFERRED The sibling node PTN002680580, carrying the synaptic CC terms, is not inherited by ARHGAP4. UniProtKB:P0DJJ0 · SRGAP2C SUPPORTS SOURCE BUT NOT TARGET Hominin-specific partial duplicate of SRGAP2. UniProtKB:P0DMP2 · SRGAP2B SUPPORTS SOURCE BUT NOT TARGET Hominin-specific partial duplicate of SRGAP2. MGI:MGI:109605 · mouse Srgap2 SUPPORTS SOURCE BUT NOT TARGET MGI:MGI:2152938 · mouse Srgap3 SUPPORTS SOURCE BUT NOT TARGET |
Loading supporting content…
Download this section (compressed HTML)Q: Which Rho-family GTPase does ARHGAP4 act on in a hematopoietic cell? The in vitro assay was done on rat protein against RAC1, CDC42 and RHOA together, and the RHOGAP domain is non-selective in cell-free conditions across this family, so the cellular answer is undetermined.
Q: ARHGAP4 is recovered in Hem-1 (NCKAP1L) leading-edge complexes in neutrophils, and those complexes are required for RAC activation. Is ARHGAP4 the brake within that machine, and if so does it act on RAC1 directly?
Q: Does ARHGAP4 have a Golgi function? The rat cloning paper reports Golgi localisation in its title, yet no Golgi GO annotation exists for the gene in any species, and the Golgi is not a site where a RhoGAP would obviously act.
Q: Is the T491M variant reported in an X-linked intellectual disability family, which lies outside the Rho-GAP domain (507-695), pathogenic? No functional assay has been performed on it.
Q: Why is complete loss of ARHGAP4 tolerated? Patients hemizygous for deletions removing the gene have no attributed phenotype, and the Arhgap4 knockout mouse returns zero significant IMPC associations across 413 statistical tests, so this is a tested absence rather than an untested one. Which paralogs compensate is open and tractable, and it is not transcriptional, since RhoGAP-family expression is unchanged in patient cells lacking ARHGAP4.
Experiment: Real-time GAP kinetics of the human ARHGAP4 Rho-GAP domain against a RHO panel (RHOA, RHOB, RHOC, RAC1, RAC2, RAC3, CDC42, RHOG, RHOQ, RHOJ), with the R543A arginine-finger mutant as the negative control, following the design of PMID:27481945, which characterised 14 RHOGAPs this way but not ARHGAP4. Note that an arginine-finger mutant is a poor loss-of-function tool in cells, because it still sequesters the target GTPase.
Hypothesis: The human ARHGAP4 Rho-GAP domain is catalytically active and, like other isolated RHOGAP domains, non-selective in vitro.
Experiment: Rho-family biosensor imaging in a neutrophil-like line after ARHGAP4 knockdown, to determine which GTPase activity rises and where, testing the Hem-1 leading-edge hypothesis directly.
Hypothesis: ARHGAP4 acts as a local brake on RAC1 within the Hem-1 leading-edge complex.
Experiment: Immunofluorescence of endogenous ARHGAP4 in a hematopoietic cell type against cortical, Golgi and microtubule markers, to establish the compartment in which the human protein resides. The sub-plasma-membrane localisation reported in 1996 has never been re-examined.
Hypothesis: Endogenous human ARHGAP4 occupies the cell cortex, as the 1996 report suggests, rather than the bulk cytoplasm.
Experiment: Combinatorial knockout of ARHGAP4 with the RhoGAPs co-expressed in the same hematopoietic lineage, assayed under migration or chemotaxis challenge rather than at baseline.
Hypothesis: The absent single-knockout phenotype reflects redundancy with other RhoGAPs expressed in the same lineage.
Experiment: Reconstitute the ARHGAP4-SEPT2-SEPT9 complex and measure GAP activity with and without the septins. This would convert the strongest interaction result in the literature into a statement about molecular function.
Hypothesis: Septin binding modulates ARHGAP4 catalytic activity rather than only its localisation.
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: ARHGAP4 has no non-redundant physiological function demonstrated in any organism. Every functional result in the literature is a gain-of-function or knockdown experiment in a cell line; no loss-of-function phenotype has been attributed to the gene in vivo.
OPEN BIOLOGY BP_DARK
What is known: The biochemistry is settled in outline: it is a Rho-family GAP with an intact catalytic arginine finger, an F-BAR targeting module and an SH3 domain, expressed predominantly in hematopoietic tissue and the nervous system. What is missing is a phenotype, and the absence is a tested one rather than an untested one: IMPC ran 413 statistical tests on the Arhgap4 knockout and recorded zero significant genotype-to-phenotype associations, whereas Srgap1 and Srgap3 have never been phenotyped there at all.
Significance: Without a loss-of-function anchor, the process annotations on this gene rest on sufficiency (overexpression inhibits migration and axon outgrowth) and on phylogenetic inheritance from the SRGAP branch. Functional redundancy with other RhoGAPs is the hypothesis the field has settled on, and the only direct test of it argues against one version of it. Microarray profiling of patient PBMC lacking ARHGAP4 found no change in RhoGAP family expression, so whatever compensates does not do so by being transcriptionally upregulated. That leaves redundancy at the protein level, which is what a combinatorial knockout would test, and which nobody has done.
What would resolve it: Conditional or compound knockouts in a hematopoietic lineage, tested under challenge rather than at baseline, and in combination with the closest expressed RhoGAP paralogs.
Provenance (the field's own admissions):
Gap: The GO record for ARHGAP4 is not wrong so much as absent. Sixteen of the nineteen primary papers about this gene produced no GO annotation on ARHGAP4 in any species, and the human protein has no experimental molecular-function or biological-process annotation at all.
OPEN CURATION RESIDUAL_SUBGAP
What is known: Three papers account for every non-electronic functional row: PMID:8570618 gives four TAS rows on the human protein; PMID:12414125 gives the rat GO:0005096 IDA, GO:0005874 IDA, GO:0007399 IEP and GO:0035023 NAS; PMID:17804252 gives the rat GO:0010764 IMP, GO:0030517 IMP and GO:0030426 IDA. Everything else on the human entry is IBA, Ensembl projection, InterPro mapping, or Reactome.
Significance: A reviewer who assumes a sparse GO record reflects a sparsely studied gene will reach the wrong conclusion here, and a reviewer who treats the existing rows as over-annotation will delete the little that is correct. The defect is coverage. Concretely, the ARHGAP4-SEPT2-SEPT9 complex, the HDAC2 and p53 interactions, the EMT work, and the Golgi localisation reported in the title of PMID:12414125 are all uncurated.
What would resolve it: Curation of the primary papers, on the rat gene where the experiments were done, so that the existing Ensembl Compara route carries the results to human. The single highest-value addition is a GO:0005794 Golgi apparatus IDA on rat Arhgap4, since the microtubule half of the same observation is already annotated and projected.
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)