ARHGAP36 is a tissue-regulated signaling protein with a RhoGAP-like domain and an N-terminal protein kinase A inhibitory motif. The N-terminal region binds the PKA catalytic subunit as a pseudosubstrate and also promotes its ubiquitin-dependent lysosomal turnover. This can activate GLI-dependent Hedgehog output downstream of Smoothened, while a Patched1-dependent centrosomal pool in mouse cells regulates PKA and ciliary trafficking. Human isoforms differ in localization and GLI activation: isoform 1 is predominantly perinuclear in the tested expression systems, whereas isoforms 2–5 reach the plasma membrane and can activate GLI. In breast-cell entosis models, ARHGAP36 associates with actin-rich protrusions and junctions and coordinates catenin-dependent adhesion with PKA/RhoA-dependent actomyosin organization. Its RhoGAP-like domain lacks the conventional catalytic arginine, and a cellular screen detected no activity toward RhoA, Rac1 or Cdc42. This domain nevertheless contributes to protein interactions and regulation of GLI activity. Altered expression is associated with developmental abnormalities and several tumor phenotypes.
Curated functional classes representing distinct biological activities. These may be splice variants, cleavage products, or other forms with different functions.
ARHGAP36_PERINUCLEAR_INACTIVEARHGAP36_MEMBRANE_ACTIVE| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005096 GTPase activator activity | IBA GO_REF:0000033 | REMOVE | Summary: Convergent sequence and cellular activity evidence supports loss of conventional catalytic GAP activity in ARHGAP36. This conclusion does not negate its binding, signaling or cytoskeletal roles. Reason: The exact Q6ZRI8 leaf PTN002489201 is inside the clade inheriting GAP IBD PTN000973894. An ancestral GAP placement can be sound while a descendant loses catalysis. Domain-alignment controls map the conventional arginine finger to human T258 (isoform-2 T227); mouse Arhgap36 has the corresponding T246. In the independently re-read PMID:32203420 supplementary screen, human isoform 2 scores negative for RhoA, Rac1 and Cdc42 while positive controls work. This screen is not a purified GTP-hydrolysis assay and includes false-negative examples, so its result is interpreted together with the catalytic substitution and primary domain characterization. These convergent findings justify retaining REMOVE for conventional GAP catalysis, without asserting exclusion of every possible GTPase/condition or requiring that the entire ancestral node was wrongly placed. A subfamily difference, one donor, and an ND root annotation are not negative functional evidence. PKA-mediated RhoA activation and actin organization can remain when catalytic GAP activity is lost. Propagation Review Root cause: PROPAGATION BAD Failure modes: PSEUDO OR SUBACTIVITY LOSS FUNCTIONAL DIVERGENCE Sources checked: MGI:MGI:1196332 · mouse Arhgap6 SUPPORTS SOURCE BUT NOT TARGET Mouse Arhgap6 is an experimentally grounded GAP source with conventional R435. Target-specific catalytic divergence challenges retention, not the validity of a single-source ancestral placement. PANTHER:PTN000973894 · PTHR12635 ancestral node at taxon:6072 (Eumetazoa) SUPPORTS SOURCE BUT NOT TARGET Verified ancestor of human ARHGAP36. A later loss of conventional catalytic activity is compatible with a correct ancestral IBD; actin process and location must be assessed independently. Supporting Evidence: PMID:33999959 However, the GAP-like region in ARHGAP36 lacks the “arginine finger” motif conserved in catalytically active homologs [7, 24], and ARHGAP36 has no effect on the activities of Rac1, Cdc42, and Rho A [25]. PMID:25024229 The replacement of this structural element with a threonine (T246) suggests that catalytic GAP domain function may not be required for Arhgap36-induced Gli activation. PMID:40378841 Arhgap36 functioned not as a Rho GTPase inhibitor but as an inhibitor of PKA signaling by directly binding the catalytic subunit of PKA (PKAC) and targeting it for lysosomal degradation. PMID:27481945 Using database searches, we extracted 66 distinct human RHOGAPs, from which 57 have a common catalytic domain capable of terminating RHO protein signaling by stimulating the slow intrinsic GTP hydrolysis (GTPase) reaction. file:human/ARHGAP36/ARHGAP36-deep-research-affinage.md Its central mechanism is bimodal inhibition of the PKA catalytic subunit (PKAC): a pseudosubstrate motif directly blocks PKAC catalytic activity while ARHGAP36 also targets PKAC for ubiquitin-mediated lysosomal degradation |
| GO:0007015 actin filament organization | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: Retain a contextual role in actin organization, supported by target actomyosin and F-actin spatial-distribution experiments. Reason: GO:0007015 includes spatial arrangement of actin structures and does not require a particular RhoGAP catalytic route or direction of RhoA regulation. The exact target inherits the positive IBD at PTN000973894. Public primary Fig.8 and supplementary S7-S8 of PMID:41644816 show an actin-associated pool and altered polarized F-actin/pMLC2 after target depletion, with a PKA/catenin mechanism described in the abstract. This is work by ARHGAP36 in the regulatory mechanism, not merely a requirement inferred from cell death. Retain KEEP_AS_NON_CORE for this entosis/breast-cell context. The main text remains unavailable, so do not infer uninspected assay details or purified actin binding. The accompanying primary-evidence note identifies the publicly accessible main figures and supplementary files inspected separately from the embargoed article text. The PMID abstract independently supports the functional actomyosin context; neither source is treated as a direct actin-binding assay. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN000973894 SUPPORTS TRANSFER Actual positive ancestral node above Q6ZRI8 leaf PTN002489201; target cytoskeletal evidence supports retained contextual function despite catalytic divergence. MGI:MGI:1196332 SUPPORTS TRANSFER Experimentally grounded mouse Arhgap6 source informs ancestral placement. The process/location assertion does not require an identical catalytic mechanism in every descendant. Supporting Evidence: file:human/ARHGAP36/ARHGAP36-primary-evidence.md S8A compares control and ARHGAP36-knockdown MDA-MB-436-10 cells, with altered polarized F-actin and pMLC2 distribution. PMID:41644816 ARHGAP36 engages both adherens junction and actomyosin to drive cell-in-cell formation |
| GO:0007165 signal transduction | IEA GO_REF:0000002 | MODIFY | Summary: Retain the existing refinement of broad signaling to positive Hedgehog-pathway regulation through PKA/GLI output, with the distinct centrosomal step explicitly qualified. Reason: The InterPro domain is genuinely present and ARHGAP36 is a signaling regulator. Pseudosubstrate inhibition of PKA and GLI-reporter experiments support the proposed GO:0045880 refinement even though ARHGAP36 can bypass Smoothened itself. The same branch contains the experimentally curated mouse ortholog. The centrosomal PKA/Inversin/Smoothened-trafficking step can have an opposing sign, so the replacement describes the demonstrated GLI/output effect rather than every local signaling event. Retain the original source term and this established MODIFY judgment; no additional process term is manufactured from the newer entosis phenotype. Proposed replacements: positive regulation of smoothened signaling pathway Supporting Evidence: PMID:25024229 Arhgap36 acts in a Smoothened (Smo)-independent manner to inhibit Gli repressor formation and to promote the activation of full-length Gli proteins. PMID:42405753 Arhgap36 depletes the levels of PKA and its catalytic subunit PKAC, both strongly activating Hh signaling and making signal transduction less dependent on regulation via Smoothened. |
| GO:0015629 actin cytoskeleton | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: Retain an actin-cytoskeleton-associated pool demonstrated in the newer target imaging experiments. Reason: The GO component includes actin-associated proteins and complexes. Public Fig.8A shows ARHGAP36-EGFP with F-actin at protrusions and tails; supplementary S6A/S7C show junctional localization and S8A includes target-depletion-controlled imaging. Other plasma-membrane, vesicle, centrosome and cilium pools do not exclude this context. The actual target is below the actin-cytoskeleton IBD; catalytic-arginine loss does not establish localization loss. Preserve the original is_active_in qualifier and IBA evidence while using the target experiments as corroboration. Colocalization is not proof of direct actin binding. The accompanying primary-evidence note identifies the publicly accessible main figures and supplementary files inspected separately from the embargoed article text. The PMID abstract independently supports the functional actomyosin context; neither source is treated as a direct actin-binding assay. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN000973894 SUPPORTS TRANSFER Actual positive ancestral node above Q6ZRI8 leaf PTN002489201; target cytoskeletal evidence supports retained contextual function despite catalytic divergence. MGI:MGI:1196332 SUPPORTS TRANSFER Experimentally grounded mouse Arhgap6 source informs ancestral placement. The process/location assertion does not require an identical catalytic mechanism in every descendant. Supporting Evidence: file:human/ARHGAP36/ARHGAP36-primary-evidence.md The public main Figure 8A shows ARHGAP36-EGFP with F-actin at pseudopodia and tail regions; PMID:41644816 ARHGAP36 engages both adherens junction and actomyosin to drive cell-in-cell formation |
| GO:0004862 cAMP-dependent protein kinase inhibitor activity | IDA Q6ZRI8-2 PMID:27713425 Bimodal antagonism of PKA signalling by ARHGAP36. | NEW | Summary: Retain the existing proposed PKA-inhibitor activity, directly measured using the human isoform-2 inhibitory region with recombinant kinase and corroborated in cells. Reason: PMID:27713425 establishes each half of the term's definition, binds to and stops or reduces the activity of a cAMP-dependent protein kinase, on the human protein. Binding: the interaction site maps to a 25-residue N-terminal sequence requiring R153 and R154, which docks into the acidic active-site cleft of PKAC, and a synthesised peptide of that sequence binds purified PKAC with high nanomolar affinity by isothermal titration calorimetry. Inhibition: an in vitro kinase assay with recombinant PKAC showed the peptide strongly reducing the phosphorylated fraction of a PKA substrate, with a PKI peptide achieving a similar extent at equal concentration in the same assay, and a cellular FRET biosensor reproducing the effect for the full-length protein. The evidence code is IDA and the constructs are stated in the methods to be human ARHGAP36 Q6ZRI8-2, so this is a human experiment on a named isoform rather than an inference from an ortholog. The comparator check supports the term rather than merely permitting it. Its human annotations are carried by the PKI pseudosubstrate family, PKIA, PKIB and PKIG, and by the PKA regulatory subunits PRKAR1A, PRKAR1B, PRKAR2A and PRKAR2B, which is exactly the class of entity that occupies the PKAC active-site cleft and blocks catalysis. Eccles et al. make that comparison themselves and in the same experiment. Recorded as this specific inhibitor term rather than as protein binding, which would record that ARHGAP36 touches PKAC while discarding the only thing known about what it does to it. The separate ubiquitin-dependent lysosomal degradation of PKAC described in the same paper is a different assertion about a different mechanism and is not proposed here. The isoform field records what was tested, not a claim that the activity is unique to isoform 2. It is noted because PMID:33999959 reports that isoform 1, which is the UniProt canonical displayed sequence, is perinuclear and does not activate Gli, so the isoform a curator works from is not an incidental detail on this gene. Supporting Evidence: PMID:27713425 These experiments demonstrate the direct interaction between ARHGAP36 and PKAC, mediated by a pseudosubstrate motif on ARHGAP36. PMID:27713425 First, we performed an in vitro kinase assay with recombinant PKAC. Addition of 36i peptide strongly reduced the fraction of phosphorylated PKA substrate (Fig. 3a). PMID:27713425 The above experiments collectively demonstrate that ARHGAP36 is a pseudosubstrate inhibitor of PKAC with potency comparable to PKI. PMID:27713425 The full-length sequence refers to human ARHGAP36 UniProt ID: Q6ZRI8-2. |
| GO:0005886 plasma membrane | IDA Q6ZRI8-2 PMID:27713425 Bimodal antagonism of PKA signalling by ARHGAP36. | NEW | Summary: Retain the existing proposed plasma-membrane localization of human isoform 2. It is compatible with the additional cytoskeletal pool. Reason: PMID:27713425 imaged human ARHGAP36 Q6ZRI8-2 in MDCK cells and found it concentrated at the plasma membrane while PKAC, expressed alone, was largely cytosolic; on coexpression PKAC was recruited to ARHGAP36. PMID:33999959 reports the same for four of the five isoforms and ties the localisation to function: the variants that reach the plasma membrane are the ones that activate Gli, while isoform 1 is perinuclear and does not. So this is not an incidental compartment but the one in which the protein's characterised activity takes place. Proposed at the plasma membrane rather than at the centrosome, although the centrosomal pool is real and better tied to a mechanism, because the mother-centriole work in PMID:30598432 was done in mouse cells, and an IDA on the human gene would assert a human experiment that was not performed. The centrosome is a good candidate for a separate ortholog-based annotation and is raised in suggested_questions instead. The isoform field records what was tested. It matters more than usual here because isoform 1 is the displayed canonical sequence in UniProt and is precisely the variant reported not to reach the membrane. One caveat belongs on this row and not only in the knowledge gaps: both supporting observations use tagged, overexpressed constructs, and PMID:27713425's imaging is in MDCK cells, which are canine. IDA is still the right code, because the gene product assayed is the human protein and its identity is stated in the methods -- that is what separates this row from the mouse centrosomal localisation, which is declined above for the opposite reason. But a compartment seen only for an overexpressed construct in a heterologous line is weaker than one seen for the endogenous protein in a cell that expresses it, the later entosis study adds human-cell target-depletion-controlled imaging, and a curator should weigh the row accordingly. The experiment that would settle it is listed in suggested_experiments. Supporting Evidence: PMID:27713425 ARHGAP36 expressed alone was concentrated at the plasma membrane, while PKAC appeared largely cytosolic. PMID:33999959 isoform 1 adopts a perinuclear distribution and does not affect Gli function, while the other four variants localize to the plasma membrane and can activate Gli transcription factors |
Loading supporting content…
Download this section (compressed HTML)Q: Does the Rho-GAP domain of ARHGAP36, which has lost its arginine finger but retains a GTPase-contacting surface about as conserved as mouse Arhgap6's, bind a Rho-family GTPase productively, and if so which one and with what consequence? UniProt records an interaction with the active form of RAC1 via this domain from PubMed:35986704, but the cached record is abstract-only and the experiment cannot be inspected here. OCRL is the precedent for binding GTP-bound RAC1 without stimulating hydrolysis; establishing whether ARHGAP36 does the same would let an informative interaction term replace the GTPase activator activity being removed.
Suggested experts: Xue Zhang, Mohammad Reza Ahmadian
Q: GO has no biological-process term for Hedgehog signalling that is not defined by activation of Smoothened, and the two terms that once covered regulation of hh target transcription factor activity, GO:0007228 and GO:1990787, are obsolete. ARHGAP36 is the clearest case of a regulator that raises Gli-dependent transcription without Smoothened involvement. Is a term for this warranted, and if so should it sit alongside GO:0045880 or under it?
Suggested experts: James K Chen
Q: How do the catenin and PKA-binding routes partition ARHGAP36-dependent actin organization across adhesion and entosis contexts? The existing actin organization/location IBAs are now supported by accessible primary target images; determine which steps are separable from contractility without assuming that GAP catalysis is required.
Suggested experts: Qiang Sun
Q: Does the centrosomal and mother-centriole pool described in PMID:30598432 exist in human cells? That work was done in mouse, so no IDA can be written for the human gene, but the localisation is mechanistically the best characterised one this protein has and a cellular component annotation for it would be worth having.
Q: Which isoforms predominate in each endogenous human context, and how do their localization and inhibitory capacities differ? Isoform1 lacks GLI activation in the tested expression comparisons, which is not proof of global inactivity; Supplementary Fig.S3 of PMID:41644816 measures endogenous isoform expression in breast-cell clones.
Suggested experts: James K Chen, Oliver Rocks
Experiment: Measure GAP activity of the purified ARHGAP36 Rho-GAP domain against a panel of Rho-family GTPases directly, alongside a T258R revertant and mouse Arhgap6 as the positive control. The negative result currently in the literature comes from a systems-scale screen; a targeted assay with the revertant would establish whether the threonine is the whole reason. The ARHGAP11B precedent, where the arginine is retained and the protein is still GAP-dead, is the reason this cannot be assumed.
Hypothesis: The threonine at the arginine-finger position is by itself sufficient to account for the absence of GAP activity, so restoring the arginine restores catalysis.
Type: in vitro GTP hydrolysis assay with a catalytic revertant
Experiment: Test binding, rather than catalysis, of the isolated ARHGAP36 Rho-GAP domain to GTP-loaded RAC1, CDC42 and RHOA by isothermal titration calorimetry or biolayer interferometry, with OCRL as the binds-without-catalysing control and ARHGAP1 as the catalysing control. This is the experiment that would decide whether an interaction term belongs on this gene in place of the GTPase activator activity being removed.
Hypothesis: The Rho-GAP domain of ARHGAP36 binds a GTP-loaded Rho-family GTPase without accelerating its hydrolysis, as OCRL does.
Type: binding assay with matched positive and negative controls
Experiment: Screen substitutions across the 25-residue pseudosubstrate motif for variants that retain isothermal titration calorimetry binding to PKAC and in vitro kinase inhibition while losing lysosomal targeting, scoring degradation by endogenous PKAC immunoblot with bafilomycin rescue. The answer decides whether GO needs one annotation here or two.
Hypothesis: Blocking PKAC catalysis and driving PKAC into the lysosome are separable activities of the same N-terminal motif.
Type: structure-function mutagenesis with two orthogonal readouts
Experiment: Tag endogenous ARHGAP36 in a human cell line that expresses it, such as the neuroblastoma lines in which it and PKAC are present at approximately equimolar levels, and image its localisation through a Hedgehog stimulation timecourse. This would test the mouse result in human cells and would also establish which isoform is actually expressed, which no study has yet reported for an endogenous human source.
Hypothesis: The Patched1-dependent centrosomal pool of ARHGAP36 described in mouse also exists in human cells that express the protein endogenously.
Type: endogenous tagging and live-cell imaging
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)