ABI3BP (also called TARSH, target of Nesh-SH3) is a large secreted glycoprotein that is deposited into the extracellular matrix, where its characterised action is on cells rather than on the mechanical fabric of the matrix. The 1068-residue precursor carries an N-terminal signal peptide, two fibronectin type-III domains near either end, a family-specific C-terminal domain shared with FNDC1, and a very large intrinsically disordered central region that is dense in proline-rich SH3-binding motifs and is the site of extensive alternative splicing. It has no catalytic domain. Its characterised receptor is integrin beta-1: matrix-deposited ABI3BP binds integrin beta-1 and holds it in the activated state, which drives paxillin phosphorylation, sequesters Src and ERK at focal adhesions, and thereby suppresses cyclin-D1 induction and S-phase entry. The consequence is a coupled switch in stem and progenitor cells - proliferation is restrained and differentiation is licensed. Mesenchymal stem cells lacking ABI3BP over-proliferate and fail to undergo osteogenic or adipogenic differentiation; cardiac progenitor cells lacking it expand but fail to express cardiomyocyte markers, and hearts lacking it recover poorly from infarction. Consistent with this brake on growth, ABI3BP expression is lost or epigenetically silenced in lung, thyroid and gallbladder tumours, and restoring it suppresses proliferation, motility and epithelial-mesenchymal transition. In the nervous system secreted ABI3BP acts as a diffusible cue that reduces dendritic complexity of olfactory mitral cells during postnatal refinement. The protein is broadly expressed, most highly in lung, and is recovered from the interstitial matrix of aorta, vein, atrial myocardium and cartilage. It is cleaved by thrombin at Arg-337, and its levels are additionally set post-translationally by proteasomal degradation, giving it a half-life of about one hour. Reports of its relationship to cellular senescence point in both directions in different cell types and are not yet reconciled.
Definition: The activity of a matrix-resident secreted protein that acts on cells by engaging a cell adhesion receptor, such as an integrin, thereby altering the activation state of that receptor and modulating cell adhesion, cytoskeletal organisation, proliferation or differentiation. Whether a protein with this activity also contributes to the mechanical integrity of the extracellular matrix is a separate question, and this term neither asserts nor excludes it.
Justification: GO has no molecular function for the matricellular class of matrix proteins, and this gene shows the concrete cost of that gap. The only extracellular-matrix-specific molecular function in GO is GO:0005201 extracellular matrix structural constituent, whose definition is explicitly mechanical - "The action of a molecule that contributes to the structural integrity of the extracellular matrix" - and whose children are uniformly mechanical too: GO:0030020 conferring tensile strength, GO:0030021 conferring compression resistance, GO:0030023 conferring elasticity, GO:0030197 lubricant activity, plus GO:0008147 structural constituent of bone, GO:0035804 structural constituent of egg coat and GO:0150043 structural constituent of synapse-associated extracellular matrix (child list retrieved from QuickGO). There is no sibling for matrix proteins whose action is to instruct cells rather than to bear load. Searching GO for "matricellular" returns nothing, confirmed both through the OLS search_all_ontologies endpoint restricted to the go ontology and through the QuickGO ontology search, each of which returns an empty result set. The consequence is mechanical over-annotation. A pipeline that knows only that a protein is a non-collagenous core-matrisome glycoprotein has exactly one extracellular-matrix molecular function available and reaches for it. That is how ABI3BP acquired three separate RCA assertions of GO:0005201 from three ECM proteomics datasets, when nothing has ever been measured about its contribution to matrix mechanics and its demonstrated activity is to bind and activate integrin beta-1. The same pressure applies across the matricellular class - thrombospondins, tenascins, SPARC, periostin and the CCN family are all matrix-resident signalling modulators rather than structural elements. GO:0048018 receptor ligand activity is proposed as the parent because its definition, "The activity of a gene product that interacts with a receptor to effect a change in the activity of the receptor", is precisely what ABI3BP does to integrin beta-1, and because its existing children - GO:0008083 growth factor activity, GO:0005125 cytokine activity, GO:0005179 hormone activity, GO:0016015 morphogen activity - are class-level ligand terms of exactly this shape. On what would distinguish the proposed term from those siblings, the honest answer is receptor class and matrix residence, not tethering, and the evidence for this gene does not support a stronger claim. It is tempting to say matricellular proteins act only where they are deposited, but ABI3BP contradicts that: concentrated medium from cells expressing it inhibits mesenchymal stem cell proliferation, and purified secreted TARSH acts on mitral cell dendrites in dissociated culture, so the protein demonstrably works at a distance from its source. What the growth-factor control in PMID:23666637 actually shows is narrower than "not diffusible" - it shows that conditioned medium containing Abi3bp does not raise phospho-ERK1/2 in serum-starved cells, i.e. Abi3bp is not a growth-factor-like ERK agonist. Its output is the opposite sign, and it runs through an adhesion receptor rather than a signalling receptor tyrosine kinase. The proposal therefore rests on the ontology gap rather than on a tethering criterion: there is no molecular function for an extracellular-matrix constituent whose action is to engage cell-surface receptors, and the one ECM-specific molecular function that exists is mechanically scoped. On the differentia itself, two candidates were rejected before settling on receptor class. A negative clause ("without contributing to the mechanical integrity of the matrix") is unmeasurable and, on this review's own position that a structural role for ABI3BP has never been tested, could not be shown to hold for the motivating example. Matrix residence alone is also too weak, and would not separate the term from its siblings: GO:0008083 growth factor activity is satisfied by several matrix-resident ligands (FGFs, HB-EGF, latent TGF-beta complexes), and localization is in any case a poor differentia for a molecular function. Receptor class is the criterion that the siblings genuinely fail - GO:0005125 cytokine activity is defined for a soluble extracellular product, and growth factor activity for stimulation of growth or proliferation, whereas a matricellular protein signals through an adhesion receptor (GO:0004895, whose own definition names integrins) and here restrains proliferation instead. That receptor class is not merely a definitional convenience, it is how production GO already types the receptor ABI3BP engages: in GO-CAM 689e7a5d00003515, human ITGB1 (UniProtKB:P05556) is assigned exactly GO:0004895 cell adhesion receptor activity, with GO:0007229 integrin-mediated signaling pathway and GO:0005925 focal adhesion (gocams/index.tsv). So the proposed differentia lines up with existing curation rather than asserting a new typing. Two things are deliberately kept out of the definition. First, the anti-proliferative sign: it is true of ABI3BP but not of the class, since tenascin-C, periostin and CCN1/CCN2 act through alphaV-beta3, alphaV-beta5 and alpha6-beta1 and are characteristically pro-proliferative. A definition carrying that sign would exclude most of the proteins listed above as the reason the term is needed, and a differentia hedged with "characteristically" is not usable as a differentia anyway. Receptor class alone does the whole exclusion job. Second, the comparison with the sibling terms itself, which is reasoning about the ontology rather than part of the class definition, and belongs here; ProposedOntologyTerm has no comment slot. If the editors want something reasoner-checkable rather than prose, the intended content is GO:0048018 receptor ligand activity and has-target some GO:0004895 cell adhesion receptor activity, with matrix residence as a non-defining characteristic. Whether matricellular proteins are best defined by receptor class, by matrix residence, or as a graded rather than discrete category is a question for the ontology editors, and is raised in suggested_questions for this gene specifically.
Parent term: receptor ligand activity
Supporting Evidence:
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005576 extracellular region | IEA GO_REF:0000044 | ACCEPT | Summary: Electronic mapping from a UniProt subcellular-location keyword. This is the only row in the entire GOA record with a WITH/FROM entry, and it resolves cleanly: UniProtKB-SubCell:SL-0243, fetched from https://rest.uniprot.org/locations/SL-0243.json, is "Secreted", defined as "Protein located outside the cell membrane(s)", and its single declared GO mapping is GO:0005576 extracellular region. So the mapping is one-to-one and faithful, and there is no ortholog or paralog transfer anywhere in this record to audit. The underlying claim is correct and is not merely predicted. UniProt marks the location ECO:0000305, a curator inference from the signal peptide at residues 1-21, but secretion has been demonstrated directly for the mouse orthologue, and the protein is repeatedly recovered from extracellular fractions of human tissue by the HDA rows below. Reason: Correct, and the WITH/FROM keyword maps to exactly this term. Kept even though it is the most general location available, because it is the term the source vocabulary supports; the more informative GO:0140149 and GO:0031012 rows sit beneath it. Supporting Evidence: file:human/ABI3BP/ABI3BP-uniprot.txt SUBCELLULAR LOCATION: Secreted file:human/ABI3BP/ABI3BP-uniprot.txt SIGNAL 1..21 PMID:19302145 We also demonstrate that TARSH is a secreted protein. |
| GO:0140149 non-collagenous component of interstitial matrix | TAS PMID:36399478 MatrisomeDB 2.0: 2023 updates to the ECM-protein knowledge d... | ACCEPT | Summary: A GO_Central author statement taken from MatrisomeDB 2.0, which classifies ABI3BP among the non-collagenous glycoprotein constituents of the interstitial matrix. TAS from a database paper is a weak evidence code in the abstract, but here the classification is corroborated from several independent directions: the protein is a secreted, heavily N-glycosylated non-collagenous glycoprotein, with Asn-37 confirmed experimentally and GlyConnect listing 25 N-linked glycans over six sites; it carries no collagen triple-helical repeat; and the HDA rows below recover it specifically from interstitial rather than basement-membrane compartments - aortic extracellular space, varicose vein wall, atrial myocardium - to which osteoarthritic cartilage matrix can be added. Independently of any proteomics, Abi3bp is seen as discrete extracellular deposits around mesenchymal stem cells by antibody staining. This is the most informative cellular-component term in the record, and the only one that distinguishes ABI3BP from any other secreted protein. Reason: Correct and specific. The interstitial rather than basement-membrane assignment is supported by the tissue compartments in which the protein is actually detected. Supporting Evidence: PMID:36399478 The extracellular matrix (ECM) is a complex assembly of proteins that constitutes the scaffold organizing cells, tissues, and organs. PMID:23666637 Abi3bp formed extracellular deposits with expression controlled by Akt1 and ubiquitin-mediated degradation. PMID:41232761 Enriched pathways included extracellular matrix (ECM) organization, ECM-receptor interaction, and phenylalanine metabolism. ABI3BP was notably elevated in OA cartilage. |
| GO:0005201 extracellular matrix structural constituent | RCA PMID:28675934 Characterization of the Extracellular Matrix of Normal and D... | MARK AS OVER ANNOTATED | Summary: This is the central problem with the record. GO:0005201 is defined as "The action of a molecule that contributes to the structural integrity of the extracellular matrix" (definition retrieved from QuickGO), and it is asserted here three times over, by the same BHF-UCL reviewed-computational-analysis pipeline, from three ECM proteomics sources. What those sources establish is that ABI3BP peptides are recovered from ECM-enriched fractions of human tissue. That is a statement about location, and the very same three papers are already cited for the HDA GO:0031012 and GO:0005576 rows, which I accept. Being present in the matrix is not the same as holding the matrix together, and the RCA inference silently crosses that line. This particular citation is the weakest of the three: PMID:28675934 is a protocols paper. It supplies a workflow for solubilising and quantifying ECM proteins and illustrates it on triple-negative breast cancer versus adjacent mammary tissue and on ovarian omental metastasis versus normal omentum. It contains no experiment addressed to ABI3BP, and no experiment addressed to matrix mechanics. No study has asked whether removing ABI3BP alters the mechanical properties of any matrix. The only assembly-relevant observation in the literature is second-hand and concerns a fragment rather than the protein: a partial fragment containing one fibronectin type-III domain promoted cell attachment and could assemble into a matrix in vitro. That fragment result argues for adhesion, and if anything for the cell-instructive reading, not for a load-bearing role. What has actually been measured for the full-length protein is receptor binding, receptor activation and a signalling output: ABI3BP behaves as a matricellular protein. Reason: Not wrong that ABI3BP is a matrix constituent, but wrong about what it does there. Detection in an ECM-enriched proteome supports the co-annotated cellular-component terms, not a molecular function asserting a contribution to structural integrity, and this citation is a methods paper with no ABI3BP-specific and no mechanics-specific experiment. Not removed, because the protein genuinely is matrix-resident and a fragment can assemble into matrix in vitro, so a structural contribution has not been excluded - it has simply never been tested. See proposed_new_terms: GO offers no molecular function for the matricellular class, which is why this term keeps being reached for. Propagation Review Root cause: TERM SCOPING PROBLEM Failure modes: ROLE CONFLATION SOURCE EVIDENCE WEAK Sources checked: PMID:28675934 Β· Naba et al. ECM proteomics protocols paper SUPPORTS SOURCE BUT NOT TARGET Supports the co-annotated HDA cellular-component row from the same dataset; cannot support a molecular function, and contains no ABI3BP-specific or matrix-mechanics experiment. GO:0005201 Β· extracellular matrix structural constituent NOT RELEVANT Term is mechanically scoped ("contributes to the structural integrity of the extracellular matrix") and all its children are mechanical, so it cannot express matrix-resident cell-instructive activity. This is the scoping gap recorded in proposed_new_terms. Supporting Evidence: PMID:28675934 We provide here optimized protocols to solubilize ECM proteins from normal or tumor tissues, digest the proteins into peptides, analyze ECM peptides by mass spectrometry, and interpret the mass spectrometric data. PMID:23666637 Computational screening followed by in vitro assays identified that a partial fragment of Abi3bp, containing one of the two Fibronectin type-III domains found in the full length protein, promoted cell attachment and was capable of assembling into an extracellular matrix PMID:23666637 Upon Abi3bp binding to integrin-Ξ²1 Src associated with paxillin which inhibited proliferation. |
| GO:0031012 extracellular matrix | HDA PMID:28675934 Characterization of the Extracellular Matrix of Normal and D... | ACCEPT | Summary: High-throughput identification of ABI3BP in ECM-enriched fractions prepared from human breast and omental tissue by the protocol this paper describes. This is what the dataset genuinely supports, and it agrees with the two independent tissue datasets below, with the curator classification behind GO:0140149, and with the direct observation of extracellular Abi3bp deposits by immunofluorescence in mesenchymal stem cell cultures. Reason: Correct compartment, reproduced across three independent proteomics datasets and supported by non-proteomic imaging evidence. Contrast with the GO:0005201 row from this same paper, where the same data were pushed into a molecular function claim they do not support. Supporting Evidence: PMID:28675934 We illustrate this protocol with ECMs obtained from two pairs of tissues, which differ in ECM content and cellularity: triple-negative breast cancer and adjacent mammary tissue, and omental metastasis from high-grade serous ovarian cancer and normal omentum. PMID:23666637 myc-tagged Abi3bp showed strong ECM staining with a myc-antibody |
| GO:0005201 extracellular matrix structural constituent | RCA PMID:27559042 Glycoproteomics Reveals Decorin Peptides With Anti-Myostatin... | MARK AS OVER ANNOTATED | Summary: The second of three identical RCA assertions of this molecular function, here from atrial glycoproteomics of human atrial fibrillation tissue. The reasoning is the same as for the PMID:28675934 row and fails in the same way: the experiment localises ABI3BP to the atrial extracellular compartment, which is exactly what the accompanying HDA GO:0031012 row from this same paper records, and adds nothing about structural integrity. Worth noting what this study was actually about. It is a glycoproteomic survey whose reported finding concerns decorin peptides with anti-myostatin activity; ABI3BP is one entry in the accompanying protein inventory. Three repetitions of one inference over three inventories do not amount to three independent pieces of evidence for a molecular function. Reason: Same category error as the other two GO:0005201 rows, and it adds no independent weight: the protein's appearance in a third ECM inventory strengthens the cellular-component call, not the molecular-function call. The demonstrated activity is receptor engagement. Propagation Review Root cause: TERM SCOPING PROBLEM Failure modes: ROLE CONFLATION SOURCE EVIDENCE WEAK Sources checked: PMID:27559042 Β· atrial glycoproteomics inventory (subject is decorin peptides) SUPPORTS SOURCE BUT NOT TARGET Supports the co-annotated HDA GO:0031012 row from the same dataset. ABI3BP is an entry in the accompanying inventory, not a subject of the study. GO:0005201 Β· extracellular matrix structural constituent CIRCULAR OR REDUNDANT Third repetition of one inference over three proteomics inventories; adds no independent support for the molecular function beyond the first. Supporting Evidence: PMID:27559042 Glycoproteomics Reveals Decorin Peptides With Anti-Myostatin Activity in Human Atrial Fibrillation. PMID:23666637 immunoprecipitation of endogenous Abi3bp from MSC-GFP-Akt1 cells resulted in the co-precipitation of integrin-Ξ²1 |
| GO:0005201 extracellular matrix structural constituent | RCA PMID:20551380 Proteomics characterization of extracellular space component... | MARK AS OVER ANNOTATED | Summary: The third identical RCA assertion, from proteomic characterisation of the extracellular space of the human aorta. The aortic study is a careful piece of work - stepwise biochemical fractionation designed specifically to separate loosely bound extracellular proteins from cellular and plasma contaminants - and it is good evidence for where ABI3BP is. It is not evidence for what ABI3BP does, and the same dataset already supports two cellular-component rows in this record. This is also the row where the mismatch is most visible: from this one paper the pipeline produced a located_in GO:0005576, a colocalizes_with GO:0031012, and this enables GO:0005201. A single fractionation experiment cannot license three different kinds of claim at three different confidence levels. Reason: Same category error. A fractionation-based inventory of the aortic extracellular space supports localisation, and this paper is already used for two localisation rows here; it does not establish that ABI3BP contributes to the structural integrity of that matrix. Propagation Review Root cause: TERM SCOPING PROBLEM Failure modes: ROLE CONFLATION SOURCE EVIDENCE WEAK Sources checked: PMID:20551380 Β· human aorta extracellular-space fractionation inventory SUPPORTS SOURCE BUT NOT TARGET The same single fractionation experiment was used to produce located_in GO:0005576, colocalizes_with GO:0031012 and enables GO:0005201 for this gene. It supports the first two; it cannot license a molecular function. GO:0005201 Β· extracellular matrix structural constituent CIRCULAR OR REDUNDANT Second repetition of the same inference from a second inventory; no independent support for the molecular function. Supporting Evidence: PMID:20551380 The main advantage of our methodology is the stepwise approach, which results in the isolation of proteins and protein degradation products loosely bound to the ECM in the NaCl fraction, the depletion of cellular proteins in the SDS fraction, and the effective solubilization of ECM proteins in the guanidine fraction. PMID:23666637 In the absence of Abi3bp, the integrin-Ξ²1 is maintained in a non-active state and the lack of phosphorylated paxillin prevents sequestration of Src and ERK at the plasma membrane, leaving these kinases to activate cyclin-d1 and drive proliferation. |
| GO:0005576 extracellular region | HDA PMID:27068509 Extracellular matrix remodelling in response to venous hyper... | ACCEPT | Summary: Identification of ABI3BP in the extracellular protein complement of human saphenous vein wall by ECM proteomics, comparing varicose with non-varicose vein. Independent tissue, independent cohort, same compartment. Note only that GO:0031012 extracellular matrix, which the same pipeline used for the breast, atrial and aortic datasets, would be the more informative term for an ECM-enriched extract than the root-level GO:0005576. Reason: Correct compartment from an independent human tissue dataset. Under-specific relative to GO:0031012, but not wrong. Supporting Evidence: PMID:27068509 Table 1ECM proteins with differential abundance between NSV and VSV based on in-solution digests |
| GO:0031012 extracellular matrix | HDA PMID:27559042 Glycoproteomics Reveals Decorin Peptides With Anti-Myostatin... | ACCEPT | Summary: ABI3BP recovered from the extracellular matrix of human atrial myocardium. This is the correctly scoped claim from this dataset, and it is consistent with the interstitial matrix assignment in GO:0140149 and with cardiac expression recorded in UniProt. It is also the compartment in which the protein's best-characterised biology occurs: cardiac progenitor cells respond to matrix-deposited ABI3BP through integrin beta-1. Reason: Correct, and mechanistically apt - the atrial extracellular matrix is where cardiac progenitor cells encounter this protein. Supporting Evidence: file:human/ABI3BP/ABI3BP-uniprot.txt TISSUE SPECIFICITY: Expressed in brain, heart, lung, liver, pancreas PMID:25296984 Abi3bp controlled CPC differentiation via integrin-Ξ²1, protein kinase C-ΞΆ, and v-akt murine thymoma viral oncogene homolog. |
| GO:0005576 extracellular region | HDA PMID:20551380 Proteomics characterization of extracellular space component... | ACCEPT | Summary: ABI3BP identified in the extracellular space of the human aorta, from the stepwise fractionation study described above. Correct, and the third independent human tissue in which the protein is recovered extracellularly. The same observation applies as for the varicose vein row: GO:0031012 would be more informative than the root-level term, and this same paper does support a GO:0031012 row - just with a weaker qualifier. Reason: Correct compartment from an independent human tissue dataset; under-specific but not wrong. Supporting Evidence: PMID:20551380 In this study, we combined biochemical fractionation with proteomics to comprehensively characterize the protein composition of the vascular extracellular environment. |
| GO:0031012 extracellular matrix | HDA PMID:20551380 Proteomics characterization of extracellular space component... | ACCEPT | Summary: The same aortic dataset again, this time with the term GO:0031012 but the qualifier colocalizes_with. The compartment call is right and I accept it, but the qualifier is inconsistent with the rest of the record and with the experiment. colocalizes_with exists for evidence, typically imaging, that places a protein in the neighbourhood of a structure without resolving whether it is genuinely part of it. Here the evidence is biochemical fractionation of the extracellular space, which is the same method that produced the located_in GO:0005576 row from this identical paper, and the same method that produced located_in GO:0031012 rows from the breast and atrial datasets. One experiment should not yield located_in for the parent term and colocalizes_with for the child. Accepting rather than downgrading, because the biology is not in doubt and the schema offers no action for correcting a qualifier alone. Raised in suggested_questions instead. Reason: Correct compartment. The colocalizes_with qualifier is the anomaly, not the term: it is unsupported by the fractionation method used and disagrees with the located_in qualifier this same paper carries for the parent term GO:0005576. Flagged for GOA rather than acted on, since no review action targets qualifiers. Supporting Evidence: PMID:20551380 Previous attempts to utilize the analytical power of proteomics to study ECM components and associated proteins have been limited by the fact that without prefractionation cellular and plasma proteins in mammalian tissues overwhelm |
| GO:0005178 integrin binding | ISS PMID:23666637 Abi3bp is a multifunctional autocrine/paracrine factor that ... | NEW | Summary: Proposed new annotation. This is the molecular function the gene actually has, and its absence is why three RCA rows of GO:0005201 were reached for instead. Mouse Abi3bp (MGI:MGI:2444583) binds integrin beta-1, shown three ways in one paper: co-precipitation of integrin beta-1 with myc-tagged Abi3bp from conditioned medium, co-precipitation with endogenous Abi3bp from unmodified extracts, and a blocking-antibody panel in which alpha-4 and alpha-5 antibodies did nothing while the beta-1 antibody raised phospho-ERK1/2 about 3.5-fold. The authors also excluded the alternative that Abi3bp works as a diffusible growth factor before turning to integrins, which is what makes the receptor assignment specific rather than a default. Independent confirmation comes from a second lineage: in cardiac progenitor cells, integrin-beta-1 blocking antibodies completely abolished the differentiation rescue produced by re-expressing Abi3bp. Coded ISS rather than IPI because every binding experiment was performed on the mouse protein; the species of the myc-tagged construct expressed in HEK293 is given only in the supplementary methods, which are not in the publication cache, so no human direct-binding claim is made here. Human ABI3BP is 1068 residues with the same two fibronectin type-III domains and the same C-terminal TARSH domain, and human cells respond to ABI3BP levels with the expected proliferative phenotype, so the transfer is well founded. Reason: Fills the molecular-function vacuum with the activity that has actually been measured, and replaces the inference that produced three GO:0005201 rows. GO:0098640 integrin binding involved in cell-matrix adhesion was considered and rejected: the demonstrated consequence of binding is receptor activation and downstream signalling, not adhesion of the cell to the matrix, so the plain GO:0005178 is the honest choice. Supporting Evidence: PMID:23666637 immunoprecipitation of endogenous Abi3bp from MSC-GFP-Akt1 cells resulted in the co-precipitation of integrin-Ξ²1 PMID:23666637 incubation with the Ξ²1 blocking antibody increased phospho-ERK1/2 levels ~3.5-fold. PMID:23666637 Abi3bp did not affect growth-factor signaling; equivalent p-ERK1/2 levels were observed in serum starved MSC-GFP-Akt1 cells stimulated with conditioned media prepared from HEK293 cells expressing either a control vector or the myc-tagged Abi3bp vector (Figure 6C) PMID:25296984 Integrin-Ξ²1 blocking antibodies completely abrogated the positive effects of re-expression of Abi3bp (Figure 5B) |
| GO:0033625 positive regulation of integrin activation | ISS PMID:23666637 Abi3bp is a multifunctional autocrine/paracrine factor that ... | NEW | Summary: Proposed new annotation. ABI3BP does not merely bind integrin beta-1, it changes the receptor's state, which is the difference between a binding annotation and a process annotation. The authors state the model directly: without Abi3bp the receptor stays inactive and paxillin is not phosphorylated, and once enough Abi3bp is present the receptor is activated and drives paxillin phosphorylation. The measured correlate is a lower phospho-paxillin to paxillin ratio in knockout cells, restored in the presence of the ligand. Reason: The activation step is what makes ABI3BP a receptor ligand rather than a passive matrix component, and it is the basis for the proposed matricellular molecular-function term. Coded ISS: the experiments are on mouse cells and a mouse knockout. One caveat a curator should weigh, flagged here in the same spirit as the GO:0098640 rejection documented on the GO:0005178 row. GO:0033622 integrin activation is defined as "The aggregation, arrangement and bonding together of an integrin... that lead to the increased affinity of the integrin for its extracellular ligands", and that branch is conventionally applied to inside-out affinity modulation by talin and kindlin from the cytoplasmic face. ABI3BP acts from the other direction: it is itself an extracellular ligand that engages the receptor and, on the authors' model, holds it in the active state. The annotation is defensible on the explicit wording of the paper - "the integrin-Ξ²1 is maintained in a non-active state" in its absence - and the term is not restricted to inside-out signalling by its definition, but an outside-in ligand is not the canonical filler and a curator may reasonably prefer to carry this claim on GO:0005178 alone. Supporting Evidence: PMID:23666637 In the absence of Abi3bp, the integrin-Ξ²1 is maintained in a non-active state and the lack of phosphorylated paxillin prevents sequestration of Src and ERK at the plasma membrane, leaving these kinases to activate cyclin-d1 and drive proliferation. PMID:23666637 Abi3bp knockout MSCs were found to have a significantly lower phospho-paxillin/paxillin ratio for stained foci area |
| GO:1902461 negative regulation of mesenchymal stem cell proliferation | ISS PMID:23666637 Abi3bp is a multifunctional autocrine/paracrine factor that ... | NEW | Summary: Proposed new annotation, and the single largest gap in this record: GOA gives ABI3BP no biological process at all, yet a knockout mouse shows increased mesenchymal stem cell number and proliferation in bone marrow, lung and liver, and both knockout and stable knockdown increase proliferation in culture through cyclin-D1, ERK1/2 and Src with more cells entering S phase. GO has a term at exactly this granularity, so there is no need to fall back on the generic GO:0008285. The direction is worth stating plainly because one paper reports the opposite. Knockdown in mouse embryonic fibroblasts inhibited proliferation (PMID:19338757). That is a different cell type and a different assay, and the mesenchymal stem cell result is the one supported by both a germline knockout and independent shRNA lines, by in vivo cell counts in three organs, and by the concordant direction in human thyroid, lung and gallbladder cells where restoring ABI3BP arrests growth. The fibroblast discrepancy is recorded in suggested_questions rather than suppressed. Reason: Precisely scoped term, supported by a germline knockout with an in vivo phenotype in three organs plus independent knockdown lines, and concordant with human gain-of-function data. Coded ISS because the perturbation experiments are mouse. Supporting Evidence: PMID:23666637 In vivo, Abi3bp knockout increased MSC number and proliferation in bone marrow, lung, and liver. PMID:23666637 Knockout or stable knockdown of Abi3bp increased MSC and Akt-MSC proliferation, promoting S-phase entry via cyclin-d1, ERK1/2, and Src. PMID:18559958 Re-expression of ABI3BP in thyroid cells resulted in a decrease in transforming activity, cell growth, cell viability, migration, invasion, and tumor growth in nude mice. file:human/ABI3BP/ABI3BP-deep-research-affinage.md ABI3BP is a secreted extracellular matrix protein that acts as a brake on stem and progenitor cell proliferation while licensing their differentiation |
| GO:0070373 negative regulation of ERK1 and ERK2 cascade | ISS PMID:23666637 Abi3bp is a multifunctional autocrine/paracrine factor that ... | NEW | Summary: Proposed new annotation. This is the signalling step through which the proliferation brake operates, and it was dissected rather than inferred. Ligand-engaged integrin beta-1 raises phospho-paxillin, which sequesters Src at the plasma membrane; Src is the kinase that activates Ras-MEK-ERK, so sequestering it lowers phospho-ERK1/2 and prevents cyclin-D1 induction. Removing Abi3bp reverses each step: less phospho-paxillin, more free Src, higher Ras activity, higher phospho-ERK. The blocking-antibody experiment gives the same answer from the receptor side - blocking integrin beta-1 raised phospho-ERK1/2 about 3.5-fold. Knockdown of paxillin, but not of vinculin, raised phospho-ERK, which is the control that makes paxillin the relevant intermediate rather than focal adhesions generally. Reason: Mechanistically specific, supported by perturbation at three points in the chain (ligand, receptor, adaptor) with an internal specificity control. Coded ISS: mouse cells throughout. Supporting Evidence: PMID:23666637 incubation with the Ξ²1 blocking antibody increased phospho-ERK1/2 levels ~3.5-fold. PMID:23666637 This effect was specific to paxillin, knockdown of another focal adhesion protein, vinculin, had no effect on p-ERK (Figure 5D). PMID:23666637 Ras activity was found to be significantly increased by Abi3bp knockdown (Figure 5E). |
| GO:0045669 positive regulation of osteoblast differentiation | ISS PMID:23666637 Abi3bp is a multifunctional autocrine/paracrine factor that ... | NEW | Summary: Proposed new annotation, the other half of the switch. Mesenchymal stem cells from Abi3bp knockout mice cannot undergo osteogenesis: all three wild-type isolations mineralised under dexamethasone and ascorbic acid, and the knockout isolations were severely deficient. The authors frame the protein as required for the transition out of proliferation into differentiation. A note on term choice, because the obvious candidate is a trap. GO:2000738 positive regulation of stem cell differentiation looks like the natural lineage-agnostic term for "knockout MSCs fail to differentiate", but its target GO:0048863 stem cell differentiation is defined as "The process in which a relatively unspecialized cell acquires specialized features of a stem cell" - differentiation *into* a stem cell. Annotating it would assert that ABI3BP promotes cells becoming stem cells, the opposite of the observed biology. GO:2000741 positive regulation of mesenchymal stem cell differentiation inherits the same inversion through GO:0072497. So the outcome-specific terms are the only correct route, and the two affected lineages are annotated separately here and in the GO:0045600 row. Reason: Supported by a germline knockout losing osteogenic capacity across three independent cell isolations. Coded ISS: mouse knockout cells. Term definitions were checked against QuickGO rather than inferred from labels, which is what ruled out the more convenient GO:2000738 and GO:2000741. Supporting Evidence: PMID:23666637 MSCs from Abi3bp knockout mice displayed severe deficiencies in osteogenic and adipogenic differentiation. PMID:23666637 All of the wild-type MSC isolations were capable of undergoing osteogenic differentiation as evidenced by robust alizarin-red staining. PMID:23666637 Our results support this hypothesis and show that MSCs require Abi3bp to switch from a proliferative to differentiating state. |
| GO:0045600 positive regulation of fat cell differentiation | ISS PMID:23666637 Abi3bp is a multifunctional autocrine/paracrine factor that ... | NEW | Summary: Proposed new annotation, the second lineage affected in the same knockout cells. Adipogenic differentiation is severely deficient in Abi3bp knockout mesenchymal stem cells alongside the osteogenic defect. Annotated separately from GO:0045669 rather than rolled into a stem-cell parent for the term-definition reason set out on that row: the available stem-cell differentiation regulators (GO:2000738, GO:2000741) target differentiation *into* a stem cell, not differentiation *of* one, so they would invert the claim. Recording both lineages is also the more informative statement, since it is the combination - one upstream switch, two divergent fates lost together - that identifies ABI3BP as acting on the proliferation/differentiation decision rather than on any single lineage programme. Reason: Same germline knockout, same experiment, second lineage. Coded ISS: mouse knockout cells. Supporting Evidence: PMID:23666637 MSCs from Abi3bp knockout mice displayed severe deficiencies in osteogenic and adipogenic differentiation. PMID:23666637 In summary, we have identified a novel extracellular matrix protein necessary for the switch from proliferation to differentiation in MSCs. |
| GO:2000727 positive regulation of cardiac muscle cell differentiation | ISS PMID:25296984 Abi3bp regulates cardiac progenitor cell proliferation and d... | NEW | Summary: Proposed new annotation. In vivo ablation of Abi3bp inhibited cardiac progenitor differentiation while expanding the progenitor pool, and this correlated with worse recovery after myocardial infarction; in vitro, knockout or shRNA progenitors under differentiating conditions expressed less of both early and late cardiomyocyte markers. Receptor dependence was demonstrated rather than assumed - integrin beta-1 blocking antibodies abolished the rescue by re-expressed Abi3bp, and blocking integrin beta-1 markedly inhibited Gata4 and Gata6 induction - which ties this process to the same molecular function proposed above. The downstream kinases here are PKC-zeta and Akt rather than the Src-ERK arm used in mesenchymal stem cells, so the receptor is shared but the effector branch is lineage-specific. This term passes the target-definition check that ruled out the stem-cell differentiation regulators used elsewhere in this review: GO:0055007 cardiac muscle cell differentiation is "The process in which a cardiac muscle precursor cell acquires specialized features of a cardiac muscle cell", so promoting it is the right direction for a protein whose loss blocks cardiomyocyte marker expression. Reason: Supported by in vivo genetic ablation with a functional cardiac phenotype, in vitro marker analysis, and antibody-based demonstration that the effect runs through integrin beta-1. Coded ISS: mouse in vivo and mouse cells. Supporting Evidence: PMID:25296984 In vivo, genetic ablation of the Abi3bp gene inhibited CPC differentiation, whereas CPC number and proliferative capacity were increased. This correlated with adverse recovery after myocardial infarction. PMID:25296984 integrin-Ξ²1 blocking antibody markedly inhibited Gata4 and Gata6 expression (Figure 5A) indicating that the effects on CPC differentiation were mediated by integrin-Ξ²1 PMID:25296984 Abi3bp controlled CPC differentiation via integrin-Ξ²1, protein kinase C-ΞΆ, and v-akt murine thymoma viral oncogene homolog. |
| GO:0050774 negative regulation of dendrite morphogenesis | ISS PMID:19302145 Secreted TARSH regulates olfactory mitral cell dendritic com... | NEW | Summary: Proposed new annotation, and a genuinely separate function from the integrin axis above. Secreted TARSH reduces mitral cell dendritic complexity and restricts interneurone branching and outgrowth in dissociated olfactory bulb cultures, and mitral cells overexpressing it change dendritic morphology cell-autonomously. The full text adds a third arm, purified TARSH-GFP applied at 500 ng/ml reproducing the conditioned-medium effect, which rules out an indirect medium component; that detail is in the locally cached full text but the reference validator's own record for this PMID holds only the abstract, so the quotations here are taken from the abstract, which states the loss-of-complexity and overexpression results directly. Two features make this a developmental cue rather than an incidental effect: expression is transient in mitral cells precisely during the dendritic refinement window, and it is unchanged when odour-evoked activity is blocked by naris closure or in AC3 knockout mice, so it is genetically programmed rather than activity-driven. Term direction checked the same way as the differentiation terms: GO:0050774's target GO:0048813 dendrite morphogenesis is "The process in which the anatomical structures of a dendrite are generated and organized", so reducing branching and complexity is negative regulation of it. No inversion here. This row was initially withheld on the grounds that the work is single-laboratory and mouse-only. On review that reasoning does not survive contact with the rest of this file: PMID:23666637 is also single-laboratory and mouse-only and carries five NEW rows, and PMID:25296984 is from the same laboratory as PMID:23666637. Applying ISS to one body of mouse work while withholding it from another of the same grade was inconsistent, so the row is added. Reason: Direct gain-of-function evidence on the phenotype from conditioned medium, purified protein and a cell-autonomous overexpression arm, with an expression time course matching the process and a control excluding activity dependence. IDA in mouse, so ISS for human, on the same footing as the other NEW rows. Held out of core_functions deliberately, and this is the one asymmetry worth stating: unlike the integrin axis, no receptor has been identified for the neuronal effect, so there is no molecular function to attach the process to, and it has not been reproduced outside the originating laboratory. Recorded as a real function of the protein rather than as part of its synthesised core. Supporting Evidence: PMID:19302145 In dissociated OB cultures, secreted TARSH promotes the reduction of mitral cell dendritic complexity and restricts dendritic branching and outgrowth of interneurons. Dendritic morphological changes were also observed in mitral cells overexpressing TARSH themselves. PMID:19302145 Through a genome-wide microarray screen, we identified TARSH (Target of NESH SH3) as a transiently expressed molecule in mitral cells during the dendritic refinement period. PMID:19302145 The dynamic TARSH expression is not altered when odor-evoked activity is blocked by naris closure or in AC3 knockout mice. |
Loading supporting contentβ¦
Download this section (compressed HTML)Q: Should ABI3BP have any biological-process annotation at all? At present it has none - not one BP row in the whole GOA record - despite a knockout mouse with proliferation and differentiation phenotypes in two independent progenitor lineages, a defined receptor, a worked-out signalling chain, and loss-of-function data in three human tumour types. The candidates that follow from the published work, and that are proposed as NEW rows here, are negative regulation of mesenchymal stem cell proliferation (GO:1902461), positive regulation of integrin activation (GO:0033625), negative regulation of the ERK1/ERK2 cascade (GO:0070373), positive regulation of osteoblast differentiation (GO:0045669), positive regulation of fat cell differentiation (GO:0045600), positive regulation of cardiac muscle cell differentiation (GO:2000727) and negative regulation of dendrite morphogenesis (GO:0050774). Because every perturbation experiment is mouse-based, all are ISS on the human gene with mouse Abi3bp (MGI:MGI:2444583) as supporting entity. Note that GO:0050774 comes from a different axis of the biology with no identified receptor, which is why it is the one proposed process kept out of core_functions. A curator taking these up should note the term-definition trap documented on the GO:0045669 row: the apparently ideal lineage-agnostic terms GO:2000738 and GO:2000741 both target differentiation *into* a stem cell, so they invert the claim, and the two mesenchymal lineages have to be annotated by outcome instead.
Q: Does ABI3BP act where it is deposited, or does it diffuse? This is the crux of the proposed matricellular term and the evidence currently cuts both ways for the same protein. It forms discrete extracellular deposits and is recovered from insoluble matrix fractions of four tissues, which is matrix-tethered behaviour. Yet concentrated medium from cells expressing it inhibits mesenchymal stem cell proliferation, and purified secreted TARSH reduces mitral cell dendritic complexity in dissociated culture - both of which are action at a distance from the source. Thrombin cleavage at Arg-337 may be the reconciling mechanism, releasing a soluble fragment from a tethered pool, but that has not been tested. Until it is, "matricellular" should be justified for this gene by receptor class and matrix residence rather than by any claim that the protein cannot diffuse.
Q: What receptor mediates the neuronal effect? The dendritic-refinement activity is the one demonstrated function with no receptor attached, which is why it is annotated as a process (GO:0050774) but kept out of core_functions - there is no molecular function to hang it on. Integrin beta-1 is the obvious candidate to test first given the rest of the biology, but no integrin experiment has been done in neurons, and mitral cell dendritic refinement is not an adhesion phenotype in any obvious way.
Q: Does ABI3BP contribute to the structural integrity of any matrix, or only to signalling through it? This is the question that decides whether the three GO:0005201 rows should stand. Nothing published measures a mechanical property in the absence of ABI3BP. The only assembly-relevant datum is a fibronectin type-III fragment that promoted cell attachment and assembled into matrix in vitro, reported second-hand through PMID:23666637 from PMID:18757743, whose full text is not retrievable and whose abstract does not name the protein.
Q: Is the effect on senescence positive or negative? The literature supports both signs with perturbation experiments, and says so explicitly. Knockdown in mouse embryonic fibroblasts increases senescence-associated beta-galactosidase in a p53/p21-dependent way, and re-expression in human thyroid carcinoma lines also triggers senescence through p21 - opposite manipulations reported to give the same outcome. Meanwhile knockdown or knockout reduces senescence in vascular smooth muscle, renal tubular epithelium and chondrocytes. Until this is resolved, whether by cell type or by distinguishing replicative from stress-induced senescence, no senescence term should be asserted, and senescence has deliberately been left out of core_functions.
Q: Does full-length ABI3BP actually bind ABI3? The gene is named for this interaction, but it was a yeast two-hybrid hit against the isolated Nesh/ABI3 SH3 domain, and the clone recovered was the 486-residue short isoform - canonical 1068 residues minus the 1-607 deletion plus the 25-residue insertion at position 669 gives exactly the 486 residues the cloning paper reports. Twelve years later the caveat still stood in print, and UniProt still hedges with "Probably interacts with ABI3". GOA carries no annotation for this pair, which is the right outcome; the open question is whether the interaction is real or whether the gene is named after an artefact.
Q: Why is the colocalizes_with qualifier used on the GO:0031012 row from PMID:20551380 when the same paper, the same experiment and the same pipeline produced located_in for the parent term GO:0005576, and located_in GO:0031012 for the breast and atrial datasets? colocalizes_with is for evidence that cannot resolve whether a protein is genuinely within a structure, which does not describe stepwise biochemical fractionation of the extracellular space. There is no review action for correcting a qualifier alone, so it is raised here.
Q: What is the significance of the huge disordered, proline-rich central region? Of 21 PxxP motifs in the 1068-residue sequence, 18 fall inside the MobiDB-lite disordered segment 384-811 - 4.2 per 100 residues against 0.47 per 100 residues elsewhere, a roughly ninefold enrichment. This is the region that alternative splicing rearranges across all four human isoforms and the region reported as an SH3-binding motif cluster in mouse Tarsh. Does it recruit SH3-domain proteins at the cell surface, or is it a spacer that projects the fibronectin type-III domains away from the matrix?
Q: Are the four isoforms functionally different? Isoforms 2, 3 and 4 all lack the first fibronectin type-III domain at residues 116-214, which is the domain implicated in the cell-attachment and matrix-assembly fragment experiment, and isoforms 3 and 4 substitute alternative proline-and-basic-rich N-termini. If integrin beta-1 engagement requires that domain, three of the four isoforms would be signalling-dead - and no annotation in the record is isoform-resolved.
Q: Is thrombin cleavage at Arg-337 a physiological regulatory step? The site is chemically convincing without taking the report on trust: residue 337 of the canonical human sequence is arginine, and the local context is Pro-336/Arg-337/Ser-338, a textbook thrombin recognition motif with proline at P2, arginine at P1 and serine at P1'. It also sits in the disordered linker rather than in either fibronectin type-III domain, so cleavage would separate the two halves of the protein. If it is regulatory rather than incidental, coagulation-linked terms may eventually be warranted.
Experiment: Assemble matrix from wild-type and ABI3BP-null human fibroblasts, or from wild-type matrix immunodepleted of ABI3BP, and measure mechanical properties directly - AFM indentation stiffness, tensile testing of decellularised sheets, and collagen fibril architecture by second-harmonic imaging. In parallel measure integrin beta-1 activation state on cells plated on each matrix using conformation-specific antibodies. The prediction is unchanged stiffness and fibril architecture with a large change in integrin activation. A negative mechanical result together with a positive signalling result would settle the GO:0005201 question and support the proposed matricellular term.
Hypothesis: ABI3BP is a matricellular signalling ligand and contributes nothing measurable to matrix mechanics, so GO:0005201 should give way to an integrin-binding molecular function.
Type: biomechanics and cell biology
Experiment: Express and purify human full-length ABI3BP, each of the four isoforms, and isolated fibronectin type-III domain constructs, then measure binding to recombinant integrin alpha5-beta1 and alpha4-beta1 ectodomains by surface plasmon resonance or biolayer interferometry, with and without manganese activation. This would convert the mouse co-immunoprecipitation result into a direct human affinity measurement suitable for an experimental GO:0005178 annotation, and would test the isoform prediction in the same experiment.
Hypothesis: Human ABI3BP binds human integrin beta-1 directly, and the interaction requires the first fibronectin type-III domain, so isoforms 2, 3 and 4 cannot signal.
Type: biophysics and protein biochemistry
Experiment: Deplete ABI3BP in a single panel of primary human cells - fibroblasts, mesenchymal stem cells, vascular smooth muscle cells, renal tubular epithelium and chondrocytes - with one shared reagent set and one shared readout panel of SA-beta-galactosidase, p16, p21, lamin B1 and SASP cytokines, assayed both under serial passage and under acute stress such as angiotensin II, irradiation or interleukin-1 beta. Running every cell type in one experiment with one protocol is the only way to separate a genuine cell-type dependence from between-laboratory variation, and the outcome determines whether any senescence GO term can be asserted for this gene.
Hypothesis: The opposing senescence phenotypes are cell-type-dependent rather than contradictory, and reflect replicative versus stress-induced senescence.
Type: cell biology
Experiment: Test the interaction in mammalian cells with full-length proteins at near-endogenous levels - co-immunoprecipitation from a cell type expressing both, proximity labelling, and a split-luciferase complementation assay - comparing full-length ABI3BP against the 486-residue isoform 2 that the original two-hybrid screen recovered. A negative full-length result would justify UniProt dropping the "Probably" and would confirm that GOA is right to carry no annotation for this pair.
Hypothesis: Full-length ABI3BP does not bind ABI3, and the gene name records a two-hybrid artefact of the proline-rich central region.
Type: interaction proteomics
Experiment: Treat purified recombinant human ABI3BP with thrombin, confirm the cleavage position by N-terminal sequencing or mass spectrometry of the fragments, and compare intact protein, cleaved protein and an R337A mutant for integrin beta-1 binding, for the ability to hold integrin beta-1 in the active conformation, and for suppression of mesenchymal stem cell proliferation. This distinguishes regulated inactivation from incidental proteolysis.
Hypothesis: Thrombin cleavage at Arg-337 inactivates ABI3BP signalling by separating the two fibronectin type-III domains.
Type: protein biochemistry and cell biology
Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)