ARHGAP21

Existing Annotations Review

GO Term	Evidence	Action	Reason
GO:0000139 Golgi membrane	IEA GO_REF:0000044	ACCEPT	Summary: Golgi membrane, mapped from the UniProt subcellular-location vocabulary. ARHGAP21 is a peripheral Golgi membrane protein whose Golgi recruitment depends on binding GTP-loaded ARF1. Reason: The IEA mapping agrees with direct evidence: ARHGAP21 is recruited to Golgi membranes by GTP-ARF1, and UniProt's SUBCELLULAR LOCATION states 'Golgi apparatus membrane; Peripheral membrane protein' with the note that localization to the Golgi depends on the GTP-bound ARF1 interaction. This is a core location, more informative than the plain GO:0005794 rows because it records the membrane association. Supporting Evidence: PMID:15793564 Here, we identify ARHGAP10--a novel Rho GTPase-activating protein (Rho-GAP) that is recruited to Golgi membranes through binding to GTP-ARF1.
GO:0005096 GTPase activator activity	IBA GO_REF:0000033	ACCEPT	Summary: GAP activity for Rho-family GTPases, propagated from the PANTHER node PTN008592632. This is the gene's central molecular function. Reason: ARHGAP21's GAP activity is the one thing every study agrees on: both founding papers independently built the R1184A arginine-finger mutant and both saw loss of activity and loss of function (UniProt records the mutagenesis with ECO:0000269 evidence from PubMed:15793564 and PubMed:16184169). What is contested is the substrate -- Cdc42 (Dubois 2005, Hehnly 2010, Barcellos 2013), RhoA and Cdc42 (Sousa 2005), RhoA (Anthony 2011), RhoA and RhoC (Lazarini 2013). GO:0005096 is substrate-agnostic and therefore correct under every one of those readings. GO:0005096 is also already maximal: QuickGO's /complete endpoint shows that GO:0005099 (Ras), GO:0005100 (Rho), GO:0008060 (ARF) and GO:0005097 (Rab) GAP terms have all been MERGED into GO:0005096, which now has zero is_a children. There is no more specific term to move to, so no replacement is proposed and no new term is requested to re-create what GO deliberately merged. The substrate is recorded machine-readably instead, via core_functions[].substrates and a has_input (RO:0002233) extension. The WITH/FROM is the PANTHER ancestral node plus ARHGAP21's own accession (UniProtKB:Q5T5U3). A self-referential source is the expected marker that a PAINT curator judged the function core on the strength of the target's own experimental annotation; it is not circular and not redundant. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN008592632 SUPPORTS TRANSFER PANTHER ancestral node, not a protein. PTHR23175 contains 5182 proteins of which 7 are reviewed; ARHGAP21 sits in subfamily SF16 with its mouse and Xenopus orthologs, and ARHGAP23 in SF5. UniProtKB:Q5T5U3 SUPPORTS TRANSFER The target's own accession. ARHGAP21 carries its own IDA to this term from PMID:15793564, so it is legitimately among the descendant evidences used to place the ancestral assertion. Supporting Evidence: PMID:15793564 We show that ARHGAP10 functions preferentially as a GAP for Cdc42 and
GO:0005096 GTPase activator activity	IDA PMID:15793564 Golgi-localized GAP for Cdc42 functions downstream of ARF1 t...	ACCEPT	Summary: Direct assay of GAP activity, from the paper that identified the protein as a Golgi-localized Rho-family GAP acting downstream of ARF1. This single row is the experimental anchor of ARHGAP21's entire GO record. Reason: ARHGAP21's GAP activity is the one thing every study agrees on: both founding papers independently built the R1184A arginine-finger mutant and both saw loss of activity and loss of function (UniProt records the mutagenesis with ECO:0000269 evidence from PubMed:15793564 and PubMed:16184169). What is contested is the substrate -- Cdc42 (Dubois 2005, Hehnly 2010, Barcellos 2013), RhoA and Cdc42 (Sousa 2005), RhoA (Anthony 2011), RhoA and RhoC (Lazarini 2013). GO:0005096 is substrate-agnostic and therefore correct under every one of those readings. GO:0005096 is also already maximal: QuickGO's /complete endpoint shows that GO:0005099 (Ras), GO:0005100 (Rho), GO:0008060 (ARF) and GO:0005097 (Rab) GAP terms have all been MERGED into GO:0005096, which now has zero is_a children. There is no more specific term to move to, so no replacement is proposed and no new term is requested to re-create what GO deliberately merged. The substrate is recorded machine-readably instead, via core_functions[].substrates and a has_input (RO:0002233) extension. Note what GO does and does not say here: because the term is substrate-agnostic, GOA is not asserting Cdc42 -- it asserts only that ARHGAP21 is a GAP. The Cdc42-versus-RhoA/RhoC disagreement is therefore invisible in GO rather than resolved by it. Supporting Evidence: PMID:15793564 We show that ARHGAP10 functions preferentially as a GAP for Cdc42 and PMID:16184169 The GAP domain of ARHGAP10 has GAP activity for RhoA and Cdc42.
GO:0005096 GTPase activator activity	TAS Reactome:R-HSA-8981637	ACCEPT	Summary: GAP activity, asserted by Reactome from the reaction 'RHOA GAPs stimulate RHOA GTPase activity'. Reason: Redundant with the IDA and IBA rows but not wrong, and it carries a detail worth noticing: Reactome places ARHGAP21 among the RhoA GAPs, i.e. the pathway database sides with the RhoA half of the contested specificity. That substrate information survives only in the Reactome reaction name and is lost in the mapping to GO, because GO:0005096 cannot carry a substrate. This is a concrete instance of the ontology gap recorded in knowledge_gaps. Supporting Evidence: PMID:23200924 In PC3 cells, ARHGAP21 presented GAP activity for RhoA and RhoC and induced changes in cell morphology.
GO:0005515 protein binding	IPI PMID:15161933 Comprehensive proteomic analysis of interphase and mitotic 1...	MODIFY	Summary: ARHGAP21 recovered as a 14-3-3 (zeta) interactor by peptide-elution affinity chromatography from HeLa extracts. 14-3-3 binding is corroborated across four independent publications and three orthogonal methods in IntAct (anti-tag coIP, proximity-dependent biotin identification, pull down), so this is genuine replication rather than one screen logged several ways. Reason: The partner is a 14-3-3 protein and GO:0071889 '14-3-3 protein binding' is active and exactly describes the interaction, so the uninformative generic term should be replaced rather than removed. The assignment is consistent with ARHGAP21 being heavily phosphorylated -- UniProt lists roughly twenty phosphoserine sites -- and 14-3-3 proteins binding phosphoserine motifs. This is a binding call only: no downstream function of the 14-3-3 interaction has been demonstrated for ARHGAP21, so nothing more is claimed. Proposed replacements: 14-3-3 protein binding Supporting Evidence: PMID:15161933 14-3-3-binding proteins were purified from extracts of interphase and mitotic HeLa cells using specific peptide elution from 14-3-3 zeta affinity columns.
GO:0005515 protein binding	IPI PMID:15778465 Targeted proteomic analysis of 14-3-3 sigma, a p53 effector ...	MODIFY	Summary: ARHGAP21 recovered as a 14-3-3 (sigma) interactor by tandem affinity purification coupled to MudPIT. 14-3-3 binding is corroborated across four independent publications and three orthogonal methods in IntAct (anti-tag coIP, proximity-dependent biotin identification, pull down), so this is genuine replication rather than one screen logged several ways. Reason: The partner is a 14-3-3 protein and GO:0071889 '14-3-3 protein binding' is active and exactly describes the interaction, so the uninformative generic term should be replaced rather than removed. The assignment is consistent with ARHGAP21 being heavily phosphorylated -- UniProt lists roughly twenty phosphoserine sites -- and 14-3-3 proteins binding phosphoserine motifs. This is a binding call only: no downstream function of the 14-3-3 interaction has been demonstrated for ARHGAP21, so nothing more is claimed. Proposed replacements: 14-3-3 protein binding Supporting Evidence: PMID:15778465 The majority of identified proteins contained one or several phosphorylatable 14-3-3-binding sites indicating a potential direct interaction with 14-3-3 sigma.
GO:0005515 protein binding	IPI PMID:17347647 Structural basis for ARF1-mediated recruitment of ARHGAP21 t...	MODIFY	Summary: Binding to GTP-bound ARF1, recorded as generic protein binding. The evidence is a 2.1 A crystal structure of ARF1-GTP bound to ARHGAP21's Arf-binding domain -- the strongest single interaction datum on this gene. Reason: ARF1 is a small GTPase, so GO:0031267 'small GTPase binding' is available and far more informative than GO:0005515; it is also the term that the merged GO:0017048 and GO:0017137 now resolve to, so it is where GO expects this class of annotation to sit. The structure defines the interface as a PH domain plus a C-terminal helix engaging the ARF1 switch regions, and IntAct records the ARHGAP21 side of this interaction against the PH domain by x-ray diffraction and cosedimentation. The same region binds ARF6 (UniProt SUBUNIT: 'Interacts with GTP-bound ARF1 and ARF6'). Worth stating explicitly because the symbol invites the opposite reading: this is an ARF-BINDING domain, not an ArfGAP domain, and an independent study classifies ARHGAP21 as an ARF6 effector while contrasting it with the Arf GAPs assayed alongside it. ARHGAP21 has no ArfGAP domain and no ARF-GAP annotation, which is correct. Proposed replacements: small GTPase binding Supporting Evidence: PMID:17347647 We show that ArfBD comprises a PH domain adjoining a C-terminal alpha helix, and that ARF1 interacts with both of these structural motifs through its switch regions and triggers structural rearrangement of the PH domain. PMID:16527809 We also found that the tandem mutation Q37E/S38I prevented the binding of two Arf GAPs, but not the effector ARHGAP10, and blocked the formation of membrane protrusion and actin reorganization.
GO:0005515 protein binding	IPI PMID:17510365 Wilms tumor suppressor WTX negatively regulates WNT/beta-cat...	REMOVE	Summary: ARHGAP21 recovered with beta-catenin in an affinity-purification experiment reported in a paper about the Wilms tumour suppressor WTX. Reason: A single mass-spectrometry co-purification in a study whose subject is a different protein, with no ARHGAP21-specific follow-up anywhere in the literature and no demonstrated function for the association. There is no informative molecular-function term this evidence would support. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does.
GO:0005515 protein binding	IPI PMID:20195357 A comprehensive resource of interacting protein regions for ...	REMOVE	Summary: ARHGAP21 recovered with elongation factor 1-delta in an in vitro virus / display-technology screen built to refine transcription-factor networks. Reason: The weakest interaction row on the gene: IntAct records only 'display technology' and 'tap' for this pair, from a single publication, with no domain mapping on the ARHGAP21 side and no orthogonal method. The screen's subject matter is unrelated to ARHGAP21's biology. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does.
GO:0005515 protein binding	IPI PMID:26496610 A human interactome in three quantitative dimensions organiz...	REMOVE	Summary: ARHGAP21 recovered with AXIN1 in a quantitative affinity-purification interactome built to assign stoichiometries and abundances. Reason: A large-scale interactome row with no ARHGAP21-specific follow-up. The same AXIN1 pairing also appears in the PDZ/PBM holdup survey, so the association is probably a genuine PDZ-motif interaction -- but as explained on those rows, GO has no term for the PDZ-domain side of a PDZ/PBM pair, so there is nothing more informative to modify to. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does.
GO:0005515 protein binding	IPI PMID:26496610 A human interactome in three quantitative dimensions organiz...	REMOVE	Summary: ARHGAP21 recovered with APC in a quantitative affinity-purification interactome built to assign stoichiometries and abundances. Reason: A large-scale interactome row with no ARHGAP21-specific follow-up. The same APC pairing also appears in the PDZ/PBM holdup survey, so the association is probably a genuine PDZ-motif interaction -- but as explained on those rows, GO has no term for the PDZ-domain side of a PDZ/PBM pair, so there is nothing more informative to modify to. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does.
GO:0005515 protein binding	IPI PMID:26496610 A human interactome in three quantitative dimensions organiz...	REMOVE	Summary: ARHGAP21 recovered with MYO18A in a quantitative affinity-purification interactome built to assign stoichiometries and abundances. Reason: A large-scale interactome row with no ARHGAP21-specific follow-up. The same MYO18A pairing also appears in the PDZ/PBM holdup survey, so the association is probably a genuine PDZ-motif interaction -- but as explained on those rows, GO has no term for the PDZ-domain side of a PDZ/PBM pair, so there is nothing more informative to modify to. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does.
GO:0005515 protein binding	IPI PMID:28514442 Architecture of the human interactome defines protein commun...	MODIFY	Summary: ARHGAP21 recovered as a 14-3-3 (zeta) interactor by affinity-purification mass spectrometry (BioPlex). 14-3-3 binding is corroborated across four independent publications and three orthogonal methods in IntAct (anti-tag coIP, proximity-dependent biotin identification, pull down), so this is genuine replication rather than one screen logged several ways. Reason: The partner is a 14-3-3 protein and GO:0071889 '14-3-3 protein binding' is active and exactly describes the interaction, so the uninformative generic term should be replaced rather than removed. The assignment is consistent with ARHGAP21 being heavily phosphorylated -- UniProt lists roughly twenty phosphoserine sites -- and 14-3-3 proteins binding phosphoserine motifs. This is a binding call only: no downstream function of the 14-3-3 interaction has been demonstrated for ARHGAP21, so nothing more is claimed. Proposed replacements: 14-3-3 protein binding Supporting Evidence: PMID:28514442 Here we present BioPlex 2.0 (Biophysical Interactions of ORFeome-derived complexes), which uses robust affinity purification-mass spectrometry methodology to elucidate protein interaction networks and co-complexes nucleated by more than 25% of protein-coding genes from the human genome
GO:0005515 protein binding	IPI PMID:33961781 Dual proteome-scale networks reveal cell-specific remodeling...	MODIFY	Summary: ARHGAP21 recovered as a 14-3-3 (zeta) interactor by affinity-purification mass spectrometry (BioPlex 3.0). 14-3-3 binding is corroborated across four independent publications and three orthogonal methods in IntAct (anti-tag coIP, proximity-dependent biotin identification, pull down), so this is genuine replication rather than one screen logged several ways. Reason: The partner is a 14-3-3 protein and GO:0071889 '14-3-3 protein binding' is active and exactly describes the interaction, so the uninformative generic term should be replaced rather than removed. The assignment is consistent with ARHGAP21 being heavily phosphorylated -- UniProt lists roughly twenty phosphoserine sites -- and 14-3-3 proteins binding phosphoserine motifs. This is a binding call only: no downstream function of the 14-3-3 interaction has been demonstrated for ARHGAP21, so nothing more is claimed. Proposed replacements: 14-3-3 protein binding Supporting Evidence: PMID:33961781 The first, BioPlex 3.0, results from affinity purification of 10,128 human proteins
GO:0005515 protein binding	IPI PMID:33961781 Dual proteome-scale networks reveal cell-specific remodeling...	REMOVE	Summary: ARHGAP21 recovered with SAPCD1 in a proteome-scale affinity-purification network. Reason: A large-scale interactome row. SAPCD1 is a 148-residue protein of little characterised function and the pairing has no follow-up; it also appears in the PDZ/PBM holdup survey, where GO offers no term for ARHGAP21's side. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does.
GO:0005515 protein binding	IPI PMID:36012204 Differential CFTR-Interactome Proximity Labeling Procedures ...	REMOVE	Summary: ARHGAP21 detected in a proximity-labelling map of the CFTR interactome. Reason: Proximity labelling reports neighbourhood, not binding, and the study's subject is CFTR rather than ARHGAP21. CFTR does carry a C-terminal PDZ-binding motif and the pair is also seen in the holdup survey, so the association is plausible -- but no informative GO term exists for the PDZ-domain side, and nothing here establishes a function. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does.
GO:0005515 protein binding	IPI PMID:36115835 Quantitative fragmentomics allow affinity mapping of interac...	REMOVE	Summary: Affinity of the ARHGAP21 PDZ domain for the AXIN1 PDZ-binding motif, measured by holdup assay in a 65,000-interaction PDZ/PBM survey. Reason: These rows come from a systematic PDZ-domain / PDZ-binding-motif affinity survey. IntAct's recorded participant ranges place the ARHGAP21 side of every holdup measurement at residues 50-159, i.e. the PDZ domain, which is worth checking rather than assuming because the paper's full text never names ARHGAP21. The measurement is real and quantitative, but GO cannot express it: the only non-obsolete PDZ term is GO:0030165, defined as 'Binding to a PDZ domain of a protein' -- the ligand side, for proteins carrying a PBM. ARHGAP21 is on the opposite side of the interaction and GO offers no reciprocal term, so there is nothing more informative to modify to. Recorded as an ONTOLOGY knowledge gap. This partner additionally has an orthogonal anti-tag coIP row from a different publication in IntAct, so the interaction is not a single-method hit. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does. Supporting Evidence: PMID:36115835 Here, we measure the affinities of 65,000 interactions involving PDZ domains and their target PDZ-binding motifs (PBM) within a human interactome region particularly relevant for viral infection and cancer.
GO:0005515 protein binding	IPI PMID:36115835 Quantitative fragmentomics allow affinity mapping of interac...	REMOVE	Summary: Affinity of the ARHGAP21 PDZ domain for the human papillomavirus type 16 E6 PDZ-binding motif, measured by holdup assay in a 65,000-interaction PDZ/PBM survey. Reason: These rows come from a systematic PDZ-domain / PDZ-binding-motif affinity survey. IntAct's recorded participant ranges place the ARHGAP21 side of every holdup measurement at residues 50-159, i.e. the PDZ domain, which is worth checking rather than assuming because the paper's full text never names ARHGAP21. The measurement is real and quantitative, but GO cannot express it: the only non-obsolete PDZ term is GO:0030165, defined as 'Binding to a PDZ domain of a protein' -- the ligand side, for proteins carrying a PBM. ARHGAP21 is on the opposite side of the interaction and GO offers no reciprocal term, so there is nothing more informative to modify to. Recorded as an ONTOLOGY knowledge gap. This partner is a viral oncoprotein whose C-terminal PBM hijacks host PDZ proteins; the row records a host-virus interaction rather than a human cellular function. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does. Supporting Evidence: PMID:36115835 Here, we measure the affinities of 65,000 interactions involving PDZ domains and their target PDZ-binding motifs (PBM) within a human interactome region particularly relevant for viral infection and cancer.
GO:0005515 protein binding	IPI PMID:36115835 Quantitative fragmentomics allow affinity mapping of interac...	REMOVE	Summary: Affinity of the ARHGAP21 PDZ domain for the human papillomavirus type 18 E6 PDZ-binding motif, measured by holdup assay in a 65,000-interaction PDZ/PBM survey. Reason: These rows come from a systematic PDZ-domain / PDZ-binding-motif affinity survey. IntAct's recorded participant ranges place the ARHGAP21 side of every holdup measurement at residues 50-159, i.e. the PDZ domain, which is worth checking rather than assuming because the paper's full text never names ARHGAP21. The measurement is real and quantitative, but GO cannot express it: the only non-obsolete PDZ term is GO:0030165, defined as 'Binding to a PDZ domain of a protein' -- the ligand side, for proteins carrying a PBM. ARHGAP21 is on the opposite side of the interaction and GO offers no reciprocal term, so there is nothing more informative to modify to. Recorded as an ONTOLOGY knowledge gap. This partner is a viral oncoprotein whose C-terminal PBM hijacks host PDZ proteins; the row records a host-virus interaction rather than a human cellular function. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does. Supporting Evidence: PMID:36115835 Here, we measure the affinities of 65,000 interactions involving PDZ domains and their target PDZ-binding motifs (PBM) within a human interactome region particularly relevant for viral infection and cancer.
GO:0005515 protein binding	IPI PMID:36115835 Quantitative fragmentomics allow affinity mapping of interac...	REMOVE	Summary: Affinity of the ARHGAP21 PDZ domain for the CFTR PDZ-binding motif, measured by holdup assay in a 65,000-interaction PDZ/PBM survey. Reason: These rows come from a systematic PDZ-domain / PDZ-binding-motif affinity survey. IntAct's recorded participant ranges place the ARHGAP21 side of every holdup measurement at residues 50-159, i.e. the PDZ domain, which is worth checking rather than assuming because the paper's full text never names ARHGAP21. The measurement is real and quantitative, but GO cannot express it: the only non-obsolete PDZ term is GO:0030165, defined as 'Binding to a PDZ domain of a protein' -- the ligand side, for proteins carrying a PBM. ARHGAP21 is on the opposite side of the interaction and GO offers no reciprocal term, so there is nothing more informative to modify to. Recorded as an ONTOLOGY knowledge gap. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does. Supporting Evidence: PMID:36115835 Here, we measure the affinities of 65,000 interactions involving PDZ domains and their target PDZ-binding motifs (PBM) within a human interactome region particularly relevant for viral infection and cancer.
GO:0005515 protein binding	IPI PMID:36115835 Quantitative fragmentomics allow affinity mapping of interac...	REMOVE	Summary: Affinity of the ARHGAP21 PDZ domain for the APC PDZ-binding motif, measured by holdup assay in a 65,000-interaction PDZ/PBM survey. Reason: These rows come from a systematic PDZ-domain / PDZ-binding-motif affinity survey. IntAct's recorded participant ranges place the ARHGAP21 side of every holdup measurement at residues 50-159, i.e. the PDZ domain, which is worth checking rather than assuming because the paper's full text never names ARHGAP21. The measurement is real and quantitative, but GO cannot express it: the only non-obsolete PDZ term is GO:0030165, defined as 'Binding to a PDZ domain of a protein' -- the ligand side, for proteins carrying a PBM. ARHGAP21 is on the opposite side of the interaction and GO offers no reciprocal term, so there is nothing more informative to modify to. Recorded as an ONTOLOGY knowledge gap. This partner additionally has an orthogonal anti-tag coIP row from a different publication in IntAct, so the interaction is not a single-method hit. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does. Supporting Evidence: PMID:36115835 Here, we measure the affinities of 65,000 interactions involving PDZ domains and their target PDZ-binding motifs (PBM) within a human interactome region particularly relevant for viral infection and cancer.
GO:0005515 protein binding	IPI PMID:36115835 Quantitative fragmentomics allow affinity mapping of interac...	REMOVE	Summary: Affinity of the ARHGAP21 PDZ domain for the beta-catenin (CTNNB1) PDZ-binding motif, measured by holdup assay in a 65,000-interaction PDZ/PBM survey. Reason: These rows come from a systematic PDZ-domain / PDZ-binding-motif affinity survey. IntAct's recorded participant ranges place the ARHGAP21 side of every holdup measurement at residues 50-159, i.e. the PDZ domain, which is worth checking rather than assuming because the paper's full text never names ARHGAP21. The measurement is real and quantitative, but GO cannot express it: the only non-obsolete PDZ term is GO:0030165, defined as 'Binding to a PDZ domain of a protein' -- the ligand side, for proteins carrying a PBM. ARHGAP21 is on the opposite side of the interaction and GO offers no reciprocal term, so there is nothing more informative to modify to. Recorded as an ONTOLOGY knowledge gap. This partner additionally has an orthogonal anti-tag coIP row from a different publication in IntAct, so the interaction is not a single-method hit. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does. Supporting Evidence: PMID:36115835 Here, we measure the affinities of 65,000 interactions involving PDZ domains and their target PDZ-binding motifs (PBM) within a human interactome region particularly relevant for viral infection and cancer.
GO:0005515 protein binding	IPI PMID:36115835 Quantitative fragmentomics allow affinity mapping of interac...	REMOVE	Summary: Affinity of the ARHGAP21 PDZ domain for the PTEN PDZ-binding motif, measured by holdup assay in a 65,000-interaction PDZ/PBM survey. Reason: These rows come from a systematic PDZ-domain / PDZ-binding-motif affinity survey. IntAct's recorded participant ranges place the ARHGAP21 side of every holdup measurement at residues 50-159, i.e. the PDZ domain, which is worth checking rather than assuming because the paper's full text never names ARHGAP21. The measurement is real and quantitative, but GO cannot express it: the only non-obsolete PDZ term is GO:0030165, defined as 'Binding to a PDZ domain of a protein' -- the ligand side, for proteins carrying a PBM. ARHGAP21 is on the opposite side of the interaction and GO offers no reciprocal term, so there is nothing more informative to modify to. Recorded as an ONTOLOGY knowledge gap. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does. Supporting Evidence: PMID:36115835 Here, we measure the affinities of 65,000 interactions involving PDZ domains and their target PDZ-binding motifs (PBM) within a human interactome region particularly relevant for viral infection and cancer.
GO:0005515 protein binding	IPI PMID:36115835 Quantitative fragmentomics allow affinity mapping of interac...	REMOVE	Summary: Affinity of the ARHGAP21 PDZ domain for the RPS6KA1 PDZ-binding motif, measured by holdup assay in a 65,000-interaction PDZ/PBM survey. Reason: These rows come from a systematic PDZ-domain / PDZ-binding-motif affinity survey. IntAct's recorded participant ranges place the ARHGAP21 side of every holdup measurement at residues 50-159, i.e. the PDZ domain, which is worth checking rather than assuming because the paper's full text never names ARHGAP21. The measurement is real and quantitative, but GO cannot express it: the only non-obsolete PDZ term is GO:0030165, defined as 'Binding to a PDZ domain of a protein' -- the ligand side, for proteins carrying a PBM. ARHGAP21 is on the opposite side of the interaction and GO offers no reciprocal term, so there is nothing more informative to modify to. Recorded as an ONTOLOGY knowledge gap. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does. Supporting Evidence: PMID:36115835 Here, we measure the affinities of 65,000 interactions involving PDZ domains and their target PDZ-binding motifs (PBM) within a human interactome region particularly relevant for viral infection and cancer.
GO:0005515 protein binding	IPI PMID:36115835 Quantitative fragmentomics allow affinity mapping of interac...	REMOVE	Summary: Affinity of the ARHGAP21 PDZ domain for the SAPCD1 PDZ-binding motif, measured by holdup assay in a 65,000-interaction PDZ/PBM survey. Reason: These rows come from a systematic PDZ-domain / PDZ-binding-motif affinity survey. IntAct's recorded participant ranges place the ARHGAP21 side of every holdup measurement at residues 50-159, i.e. the PDZ domain, which is worth checking rather than assuming because the paper's full text never names ARHGAP21. The measurement is real and quantitative, but GO cannot express it: the only non-obsolete PDZ term is GO:0030165, defined as 'Binding to a PDZ domain of a protein' -- the ligand side, for proteins carrying a PBM. ARHGAP21 is on the opposite side of the interaction and GO offers no reciprocal term, so there is nothing more informative to modify to. Recorded as an ONTOLOGY knowledge gap. This partner additionally has an orthogonal anti-tag coIP row from a different publication in IntAct, so the interaction is not a single-method hit. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does. Supporting Evidence: PMID:36115835 Here, we measure the affinities of 65,000 interactions involving PDZ domains and their target PDZ-binding motifs (PBM) within a human interactome region particularly relevant for viral infection and cancer.
GO:0005515 protein binding	IPI PMID:36115835 Quantitative fragmentomics allow affinity mapping of interac...	REMOVE	Summary: Affinity of the ARHGAP21 PDZ domain for the MYO18A PDZ-binding motif, measured by holdup assay in a 65,000-interaction PDZ/PBM survey. Reason: These rows come from a systematic PDZ-domain / PDZ-binding-motif affinity survey. IntAct's recorded participant ranges place the ARHGAP21 side of every holdup measurement at residues 50-159, i.e. the PDZ domain, which is worth checking rather than assuming because the paper's full text never names ARHGAP21. The measurement is real and quantitative, but GO cannot express it: the only non-obsolete PDZ term is GO:0030165, defined as 'Binding to a PDZ domain of a protein' -- the ligand side, for proteins carrying a PBM. ARHGAP21 is on the opposite side of the interaction and GO offers no reciprocal term, so there is nothing more informative to modify to. Recorded as an ONTOLOGY knowledge gap. This partner additionally has an orthogonal anti-tag coIP row from a different publication in IntAct, so the interaction is not a single-method hit. GO:0005515 carries no functional information. Removal does not mean the reported interaction is false or that the underlying experiment is unsound -- it means the annotation does not say what ARHGAP21 does. Supporting Evidence: PMID:36115835 Here, we measure the affinities of 65,000 interactions involving PDZ domains and their target PDZ-binding motifs (PBM) within a human interactome region particularly relevant for viral infection and cancer.
GO:0005515 protein binding	IPI PMID:36931259 A central chaperone-like role for 14-3-3 proteins in human c...	MODIFY	Summary: ARHGAP21 recovered as a 14-3-3 (sigma) interactor by systematic paralog-resolved 14-3-3 interactome mapping. 14-3-3 binding is corroborated across four independent publications and three orthogonal methods in IntAct (anti-tag coIP, proximity-dependent biotin identification, pull down), so this is genuine replication rather than one screen logged several ways. Reason: The partner is a 14-3-3 protein and GO:0071889 '14-3-3 protein binding' is active and exactly describes the interaction, so the uninformative generic term should be replaced rather than removed. The assignment is consistent with ARHGAP21 being heavily phosphorylated -- UniProt lists roughly twenty phosphoserine sites -- and 14-3-3 proteins binding phosphoserine motifs. This is a binding call only: no downstream function of the 14-3-3 interaction has been demonstrated for ARHGAP21, so nothing more is claimed. Proposed replacements: 14-3-3 protein binding Supporting Evidence: PMID:36931259 Here, we map the interactomes of all human 14-3-3 paralogs and systematically characterize the effect of disrupting these interactions on client localization.
GO:0005515 protein binding	IPI PMID:36931259 A central chaperone-like role for 14-3-3 proteins in human c...	MODIFY	Summary: ARHGAP21 recovered as a 14-3-3 (zeta) interactor by systematic paralog-resolved 14-3-3 interactome mapping. 14-3-3 binding is corroborated across four independent publications and three orthogonal methods in IntAct (anti-tag coIP, proximity-dependent biotin identification, pull down), so this is genuine replication rather than one screen logged several ways. Reason: The partner is a 14-3-3 protein and GO:0071889 '14-3-3 protein binding' is active and exactly describes the interaction, so the uninformative generic term should be replaced rather than removed. The assignment is consistent with ARHGAP21 being heavily phosphorylated -- UniProt lists roughly twenty phosphoserine sites -- and 14-3-3 proteins binding phosphoserine motifs. This is a binding call only: no downstream function of the 14-3-3 interaction has been demonstrated for ARHGAP21, so nothing more is claimed. Proposed replacements: 14-3-3 protein binding Supporting Evidence: PMID:36931259 Here, we map the interactomes of all human 14-3-3 paralogs and systematically characterize the effect of disrupting these interactions on client localization.
GO:0005794 Golgi apparatus	IBA GO_REF:0000033	ACCEPT	Summary: Golgi apparatus, propagated from PANTHER node PTN002754039 with the is_active_in qualifier. ARHGAP21 acts at the Golgi, where ARF1 recruits it. Reason: Correct and core, and corroborated on the target itself by an IDA. is_active_in is the right qualifier: the Golgi is where ARHGAP21's GAP activity is exercised, not merely where the protein can be found. The WITH/FROM is the PANTHER ancestral node plus ARHGAP21's own accession (UniProtKB:Q5T5U3). A self-referential source is the expected marker that a PAINT curator judged the function core on the strength of the target's own experimental annotation; it is not circular and not redundant. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002754039 SUPPORTS TRANSFER PANTHER ancestral node carrying the Golgi, actin-cytoskeleton, cell-junction and microtubule-transport terms for this clade. UniProtKB:Q5T5U3 SUPPORTS TRANSFER The target's own accession; ARHGAP21 holds its own IDA to GO:0005794 from PMID:20525016. Supporting Evidence: PMID:17347647 ARHGAP21 is recruited to the Golgi by binding to another small GTPase, ARF1.
GO:0005794 Golgi apparatus	IDA PMID:20525016 Cdc42 regulates microtubule-dependent Golgi positioning.	ACCEPT	Summary: Golgi localisation observed directly in human cells. Reason: Direct localisation evidence in the paper that also established ARHGAP21's role in repositioning the Golgi after microtubule depolymerisation. Core location, consistent with the ARF1-dependent recruitment mechanism. The quote is this paper's own imaging result; an earlier draft cited an introduction sentence that merely restated prior work with reference callouts, which is not IDA evidence. Supporting Evidence: PMID:20525016 The GFP-tagged ARHGAP21 fragment was localized to the Golgi apparatus as expected (Figure 1B).
GO:0005829 cytosol	TAS Reactome:R-HSA-8981637	KEEP AS NON CORE	Summary: Cytosol, asserted by Reactome as the compartment of the 'RHOA GAPs stimulate RHOA GTPase activity' reaction. Reason: A Reactome compartment label is a property of the reaction as modelled, not a measurement of where ARHGAP21 was observed, so it should not be read as primary localisation evidence. It is nonetheless not wrong: ARHGAP21 is a peripheral membrane protein and necessarily has a soluble cytosolic pool, and independent studies report nuclear and cytoplasmic signal. But the functionally meaningful pools are the Golgi membrane and the cell junction, which are separately annotated, so cytosol is retained as non-core rather than accepted as a core location. Supporting Evidence: PMID:23200924 ARHGAP21 is located in the nucleus and cytoplasm of both cell lines and its depletion resulted in decreased proliferation and increased migration of PC3 cells but not LNCaP cells.
GO:0005856 cytoskeleton	IEA GO_REF:0000044	ACCEPT	Summary: Cytoskeleton, mapped from the UniProt subcellular-location vocabulary. Reason: Broad, but correct and consistent with UniProt's 'Cytoplasm, cytoskeleton' location and with the more specific GO:0015629 actin cytoskeleton rows the gene already carries. An IEA being more general than the experimental annotations alongside it is not a defect.
GO:0007030 Golgi organization	IBA GO_REF:0000033	ACCEPT	Summary: Golgi organization, propagated from PANTHER node PTN002754039 and independently supported by knockdown experiments on the human protein. Reason: Core biological process, and the propagation lands on exactly the term the target's own IMP supports. The WITH/FROM is the PANTHER ancestral node plus ARHGAP21's own accession (UniProtKB:Q5T5U3). A self-referential source is the expected marker that a PAINT curator judged the function core on the strength of the target's own experimental annotation; it is not circular and not redundant. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002754039 SUPPORTS TRANSFER Ancestral node for the Golgi/cytoskeleton term set. UniProtKB:Q5T5U3 SUPPORTS TRANSFER The target's own accession; ARHGAP21 holds its own IMP to GO:0007030 from PMID:20525016.
GO:0007030 Golgi organization	IMP PMID:20525016 Cdc42 regulates microtubule-dependent Golgi positioning.	ACCEPT	Summary: Knockdown of ARHGAP21 impairs the ability of dispersed Golgi membranes to reposition at the centrosome, establishing a role in Golgi organization. Reason: A clean loss-of-function experiment in human cells with a structural readout. Core process, and mechanistically coherent with the ARF1/Cdc42 axis the same paper works out. Supporting Evidence: PMID:20525016 Here we show that reduced expression of the Cdc42-specific GTPase-activating protein, ARHGAP21, inhibits the ability of dispersed Golgi membranes to reposition at the centrosome following nocodazole treatment and washout.
GO:0007165 signal transduction	IEA GO_REF:0000002	MODIFY	Summary: Signal transduction, mapped electronically from the InterPro RhoGAP domain signature IPR000198. Reason: The mapping is sound in direction but pitched several levels too high: GO:0007165 is one of the broadest terms in the biological-process branch, while the domain it derives from is specifically a Rho-family GAP domain. GO:0051056 'regulation of small GTPase mediated signal transduction' is exactly what a RhoGAP domain licenses, and the gene already carries it from Reactome, which shows the term is considered apt here. Duplicate terms from different evidence are acceptable, so this is a specificity upgrade rather than a removal. Proposed replacements: regulation of small GTPase mediated signal transduction
GO:0015629 actin cytoskeleton	IBA GO_REF:0000033	ACCEPT	Summary: Actin cytoskeleton, propagated from PANTHER node PTN002754039 with is_active_in. Reason: ARHGAP21 controls Arp2/3-dependent F-actin dynamics through its GAP activity, so the actin cytoskeleton is genuinely where it acts rather than merely where it is found. Corroborated on the target by an independent HPA immunofluorescence IDA. The WITH/FROM is the PANTHER ancestral node plus ARHGAP21's own accession (UniProtKB:Q5T5U3). A self-referential source is the expected marker that a PAINT curator judged the function core on the strength of the target's own experimental annotation; it is not circular and not redundant. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002754039 SUPPORTS TRANSFER Ancestral node for the Golgi/cytoskeleton term set. UniProtKB:Q5T5U3 SUPPORTS TRANSFER The target's own accession; ARHGAP21 holds an independent HPA IDA to GO:0015629. Supporting Evidence: PMID:15793564 regulates the Arp2/3 complex and F-actin dynamics at the Golgi through the control of Cdc42 activity.
GO:0015629 actin cytoskeleton	IDA GO_REF:0000052	ACCEPT	Summary: Actin cytoskeleton localisation from HPA immunofluorescence. Reason: Independent imaging evidence that agrees with the IBA row and with the protein's established role in F-actin regulation.
GO:0030054 cell junction	IBA GO_REF:0000033	ACCEPT	Summary: Cell junction, propagated from PANTHER node PTN002754039 with is_active_in. Reason: Well supported on the human protein: ARHGAP21 is transiently recruited to cell-cell junctions during adhesion formation, where it lowers Cdc42 activity, and it binds alpha-catenin there. is_active_in is appropriate because this is where the GAP activity is exercised. The WITH/FROM is the PANTHER ancestral node plus ARHGAP21's own accession (UniProtKB:Q5T5U3). A self-referential source is the expected marker that a PAINT curator judged the function core on the strength of the target's own experimental annotation; it is not circular and not redundant. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002754039 SUPPORTS TRANSFER Ancestral node for the Golgi/cytoskeleton term set. UniProtKB:Q5T5U3 SUPPORTS TRANSFER The target's own accession; ARHGAP21 holds an independent HPA IDA to GO:0030054. Supporting Evidence: PMID:23235160 ARHGAP21 interacts with Cdc42, and decreased Cdc42 activity coincides with the appearance of ARHGAP21 at the cell-cell junctions.
GO:0030054 cell junction	IDA GO_REF:0000052	ACCEPT	Summary: Cell junction localisation from HPA immunofluorescence. Reason: Independent imaging evidence consistent with the junction biology established by the alpha-catenin and Cdc42 studies.
GO:0030659 cytoplasmic vesicle membrane	IEA GO_REF:0000044	ACCEPT	Summary: Cytoplasmic vesicle membrane, mapped from the UniProt subcellular-location vocabulary. Reason: Consistent with UniProt's 'Cytoplasmic vesicle membrane; Peripheral membrane protein' location and with ARHGAP21's role in retrograde and post-Golgi trafficking. A peripheral association with transport-vesicle membranes follows naturally from ARF1-dependent recruitment. Non-controversial IEA.
GO:0051056 regulation of small GTPase mediated signal transduction	TAS Reactome:R-HSA-9012999	ACCEPT	Summary: Regulation of small GTPase mediated signal transduction, asserted by Reactome from the RHO GTPase cycle. Reason: The correct process-level description of what a Rho-family GAP does, and pitched at the right level given that the substrate is contested. It is the term this review also proposes as the replacement for the over-broad GO:0007165 InterPro mapping.
GO:0051683 establishment of Golgi localization	IBA GO_REF:0000033	ACCEPT	Summary: Establishment of Golgi localization, propagated from PANTHER node PTN002754039. Reason: Matches the target's own IDA for the same term, so the propagation lands at exactly the right specificity rather than above its evidence. The WITH/FROM is the PANTHER ancestral node plus ARHGAP21's own accession (UniProtKB:Q5T5U3). A self-referential source is the expected marker that a PAINT curator judged the function core on the strength of the target's own experimental annotation; it is not circular and not redundant. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002754039 SUPPORTS TRANSFER Ancestral node for the Golgi/cytoskeleton term set. UniProtKB:Q5T5U3 SUPPORTS TRANSFER The target's own accession; ARHGAP21 holds its own IDA to GO:0051683 from PMID:20525016.
GO:0051683 establishment of Golgi localization	IDA PMID:20525016 Cdc42 regulates microtubule-dependent Golgi positioning.	ACCEPT	Summary: Establishment of Golgi localization, from the nocodazole washout assay in which ARHGAP21-dependent Cdc42 signalling is required for dispersed Golgi membranes to re-cluster at the centrosome. Reason: Direct, well-controlled evidence in human cells, and the term is precisely what the assay measures. Supporting Evidence: PMID:20525016 Together, our results reveal that dynein and microtubule-dependent Golgi positioning is regulated by ARF1-, coatomer-, and ARHGAP21-dependent Cdc42 signaling.
GO:0051684 maintenance of Golgi location	IMP PMID:20525016 Cdc42 regulates microtubule-dependent Golgi positioning.	ACCEPT	Summary: Maintenance of Golgi location, from knockdown experiments in the same study. Reason: Distinct from establishment of Golgi localization and separately supported -- and the distinction turns on which experiment is cited. The nocodazole washout assay measures RE-establishment of position after dispersal, so it belongs on GO:0051683; the evidence for maintenance is the steady-state observation that Golgi membranes are already partially dispersed in unperturbed cells when ARHGAP21 is knocked down. This row is quoted with the latter. Retained as a core process. Supporting Evidence: PMID:20525016 Golgi membranes were partially dispersed in cells when ARHGAP21 levels were reduced by RNA interference but not in control cells (Figures 1A and S1B).
GO:0070161 anchoring junction	IEA GO_REF:0000044	ACCEPT	Summary: Anchoring junction, mapped from the UniProt subcellular-location vocabulary. Reason: Correct and better-specified than the plain cell-junction rows: the junction ARHGAP21 is documented at is the adherens junction, where it binds alpha-catenin, and adherens junction is a kind of anchoring junction. Supporting Evidence: PMID:16184169 This protein colocalized with alpha-catenin at cell-cell junctions and was recruited at L. monocytogenes entry sites.
GO:0072384 organelle transport along microtubule	IBA GO_REF:0000033	KEEP AS NON CORE	Summary: Organelle transport along microtubule, propagated from PANTHER node PTN002754039. Reason: Supported on the target by its own IMP, so the propagation is sound. Marked non-core rather than core because ARHGAP21 does not itself move cargo: it acts upstream, restraining Cdc42 so that dynein-dependent motility of Golgi membranes proceeds normally. The transport is executed by dynein and microtubules; ARHGAP21 regulates it. The WITH/FROM is the PANTHER ancestral node plus ARHGAP21's own accession (UniProtKB:Q5T5U3). A self-referential source is the expected marker that a PAINT curator judged the function core on the strength of the target's own experimental annotation; it is not circular and not redundant. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002754039 SUPPORTS TRANSFER Ancestral node for the Golgi/cytoskeleton term set. UniProtKB:Q5T5U3 SUPPORTS TRANSFER The target's own accession; ARHGAP21 holds its own IMP to GO:0072384 from PMID:20525016.
GO:0072384 organelle transport along microtubule	IMP PMID:20525016 Cdc42 regulates microtubule-dependent Golgi positioning.	KEEP AS NON CORE	Summary: Organelle transport along microtubule, from the permeabilised-cell Golgi capture and motility assay. Reason: The experiment is sound and the term describes the process affected. Kept as non-core for the same reason as the IBA row: ARHGAP21 is an upstream regulator of dynein-dependent Golgi motility rather than a component of the transport machinery, and the motility readout was blocked by an inhibitory dynein antibody. Supporting Evidence: PMID:20525016 Together, our results reveal that dynein and microtubule-dependent Golgi positioning is regulated by ARF1-, coatomer-, and ARHGAP21-dependent Cdc42 signaling.
GO:0045294 alpha-catenin binding	IPI PMID:16184169 ARHGAP10 is necessary for alpha-catenin recruitment at adher...	NEW	Summary: ARHGAP21 was identified as a direct partner of alpha-catenin in a two-hybrid screen using alpha-catenin as bait, colocalises with it at cell-cell junctions, and UniProt maps the interaction to residues 1592-1861 with ECO:0000269 evidence from this paper. Reason: This is the single largest curation gap on the gene. Querying QuickGO by reference shows PMID:16184169 produced two annotations in all of GOA, both on alpha-catenin (CTNNA1, P35221) and neither on ARHGAP21 -- even though the paper is titled for ARHGAP21's requirement in alpha-catenin recruitment, and even though UniProt records the interaction in both FUNCTION and SUBUNIT and maps a binding region for it. GO:0045294 'alpha-catenin binding' is active and defined as 'Binding to catenin complex alpha subunit', which is exactly this interaction, so the gap is not an ontology limitation but an uncurated paper. IPI is the correct code: both proteins are human and the assay is a direct interaction assay. Supporting Evidence: PMID:16184169 Using a two-hybrid screen, we identified a new partner of alpha-catenin: ARHGAP10. PMID:16184169 This protein colocalized with alpha-catenin at cell-cell junctions and was recruited at L. monocytogenes entry sites.
GO:0034333 adherens junction assembly	IMP PMID:16184169 ARHGAP10 is necessary for alpha-catenin recruitment at adher...	NEW	Summary: Knockdown of ARHGAP21 impairs recruitment of alpha-catenin to cell-cell contacts, and its overexpression raises alpha-catenin and cortical actin at junctions; an independent knockdown study finds weaker cell-cell adhesions. Reason: ARHGAP21 does part of the work of building the junction rather than merely being required for it: it binds alpha-catenin directly through a mapped region, is itself recruited to the junction during adhesion formation, and determines whether alpha-catenin -- a core structural component linking cadherin to actin -- arrives there. Loss and gain of function move the phenotype in opposite directions, which is what distinguishes participation from being a passive requirement. Two independent human knockdown studies agree. The alternative framing would be a regulation-level term such as GO:1903393, which a curator may prefer if the effect is judged to run entirely through GAP activity; the direct alpha-catenin interaction is why the assembly term is proposed instead. Species caveat, stated because an experimental code asserts the experiment was done in the annotated organism: both supporting papers are abstract-only in the cache, neither abstract names a cell line, and PMID:16184169 is not in PMC or Europe PMC full text and returns no text-mined annotations, so the cell system could not be confirmed from any source available here. IMP is retained because the two-hybrid partner is human alpha-catenin and the annotation target is human ARHGAP21, but a curator with the full text should confirm the cell line before accepting the code. Supporting Evidence: PMID:16184169 In ARHGAP10-knockdown cells, L. monocytogenes entry and alpha-catenin recruitment at cell-cell contacts were impaired. PMID:16184169 Its overexpression disrupted actin cables, enhanced alpha-catenin and cortical actin levels at cell-cell junctions and inhibited L. monocytogenes entry. PMID:23235160 Cells lacking ARHGAP21 expression show weaker cell-cell adhesions, increased cell migration, and a diminished ability to undergo hepatocyte growth factor-induced epithelial-mesenchymal transition (EMT).
GO:0032956 regulation of actin cytoskeleton organization	IMP PMID:15793564 Golgi-localized GAP for Cdc42 functions downstream of ARF1 t...	NEW	Summary: ARHGAP21 controls Arp2/3-complex and F-actin dynamics at the Golgi through its GAP activity toward Cdc42. The direction of the effect is context-dependent: overexpression disrupts actin cables and raises cortical actin at junctions, while antisense knockdown in neonatal mouse pancreatic islets lowers F-actin. Reason: The gene's founding paper is titled for control of the Arp2/3 complex and F-actin dynamics, yet ARHGAP21 carries no actin-related biological process term at all -- only the cellular-component rows for actin cytoskeleton. A regulation term is the right shape, and is right for a second reason: ARHGAP21 does not nucleate or bundle actin itself, it sets the activity level of the Rho GTPase that drives Arp2/3-dependent polymerisation, and the reported sign of the effect differs between systems (overexpression disrupts actin cables in epithelial cells; antisense knockdown reduces F-actin in mouse islets). GO:0032956 is sign-agnostic and so accommodates both, where a directional regulation term would not. Evidence-code caveat: PMID:15793564 is abstract-only in the cache, so the assay detail behind the F-actin phenotype cannot be read here; IMP is retained because the abstract states the control of Arp2/3 and F-actin dynamics as this paper's own result on the human protein, but a curator with the full text should confirm the perturbation used. Note the supporting PMID:25744409 experiment is in MOUSE islets and is cited only to show the direction is not fixed, not as evidence for the human annotation. On the evidence structure: the human data on this row is GAIN-of-function (PMID:16184169, overexpression in epithelial cells); no human LOSS-of-function actin readout is quotable from the cache. UniProt's MISCELLANEOUS line records the human depletion phenotype, but it spans a CC continuation (ARHGAP21-uniprot.txt:274-275), so the only fragment quotable on one physical line truncates at 'induces cell spreading and' and drops the F-actin clause that would make it relevant. Left uncited rather than quoted misleadingly. Supporting Evidence: PMID:15793564 regulates the Arp2/3 complex and F-actin dynamics at the Golgi through the control of Cdc42 activity. PMID:16184169 Its overexpression disrupted actin cables, enhanced alpha-catenin and cortical actin levels at cell-cell junctions and inhibited L. monocytogenes entry. PMID:25744409 F-actin was reduced in AS-islets, as judged by lower phalloidin intensity.
GO:0031005 filamin binding	IPI PMID:41957357 ARHGAP21 enhances metastasis in hepatocellular carcinoma by ...	NEW	Summary: ARHGAP21 binds filamin A directly, with the PDZ domain implicated in the interaction with FLNA's 1-1200 aa fragment, and recruits HSP90-alpha to protect FLNA from ubiquitination and degradation. Reason: GO:0031005 'filamin binding' is active and partner-appropriate, so unlike the ten PDZ holdup rows this interaction CAN be expressed informatively -- which is why it is proposed as a row rather than left as generic binding. It also corrects the darkness framing of the PDZ knowledge gap: the domain does have one reported cellular ligand. The paper hedges the domain mapping ('functions as a potential mediator'), so the row is deliberately scoped to the binding itself, which the paper states directly, rather than to a PDZ-dependence claim. Both proteins are human and the work is in human HCC cell lines, so IPI is the correct code. The downstream FLNA-stabilisation and metastasis phenotypes are not proposed as process terms here: they are effects of the interaction in a cancer context rather than a demonstrated normal function, and the HSP90-alpha arm would need its own evidence. Supporting Evidence: PMID:41957357 Mechanistically, we demonstrated that ARHGAP21 directly binds to FLNA, and the PDZ domain of ARHGAP21 functions as a potential mediator of its binding to the 1-1200 aa fragment of FLNA.

Gene Description

Existing Annotations Review

Core Functions

Rho-family GTPase-activating protein: accelerates GTP hydrolysis on Cdc42, and on RhoA and RhoC in some settings, via the R1184 arginine finger of its RhoGAP domain, thereby switching off Rho-GTPase signalling to the actin cytoskeleton.

ARF1/ARF6 effector: binds the switch regions of GTP-loaded ARF1 through a bipartite ARF-binding domain formed by the PH domain and an adjacent C-terminal helix, which is what recruits ARHGAP21 to Golgi membranes and couples ARF1 signalling to Cdc42 inactivation.

Adherens-junction component: binds alpha-catenin through a region in its C-terminal half and is required for alpha-catenin recruitment to nascent cell-cell contacts, coupling junction assembly to local Cdc42 inactivation and cortical actin remodelling.

References

Suggested Questions for Experts

Suggested Experiments

Knowledge Gaps

Deep Research

Affinage

📚 Additional Documentation

Notes

Bioinformatics Results

📄 View Raw YAML