ESPN

Summary: Refine the WH2-derived annotation to actin monomer binding.

Reason: InterPro IPR003124 identifies the WH2 domain at human B1AK53 residues 648–673. The human UniProt DOMAIN section attributes monomer binding to that domain by similarity to mouse Q9ET47. GO:0003785 specifies this monomer-binding scope, whereas the separate GO:0051015 rows describe filament binding supported by human espin 3A experiments (PMID:29572253 Figure 5K). The replacement is a domain-supported refinement, not a newly inspected human WH2-only assay. Cross-linking measured in Figure 5L is a distinct composite activity and remains represented in the core synthesis.

Proposed replacements: actin monomer binding

Summary: Refine cytoskeletal localization to the actin cytoskeleton.

Reason: SL-0090 genuinely maps to cytoskeleton; the human UniProt location is attributed by similarity to mouse Q9ET47. The supported cytoskeletal component is more precisely actin cytoskeleton, already represented by the valid PAINT row 5. Human GFP-espin 3A targets phalloidin-positive microvilli and stereocilia in PMID:29572253 Figures 4 and 5, while the rat small-espin original PMID:9763424 Figures 4–5 places the protein along microvillar actin bundles and actin-rich fibers. These observations justify specificity without asserting localization to other cytoskeletal filament systems.

Proposed replacements: actin cytoskeleton

Summary: Human espin 3A targets actin-rich microvilli in the epithelial-cell assay.

Reason: PMID:29572253 construct and cell Methods identify GFP-tagged human espin 3A expressed in LLC-PK1-CL4 cells. Wild-type protein localizes to apical phalloidin-positive microvilli and elongates them (Figure 4A-C/P); the actin-binding-region truncation fails targeting, while the six-residue deletion separates residual targeting from elongation. This directly supports microvillus localization of the tested human gene product. The heterologous-cell assay does not establish endogenous expression in every human epithelium.

Summary: Retain microvillar localization.

Reason: The current SL-0293 mapping is exactly GO:0005902, and the human UniProt entry links this location to PMID:29572253. The human GFP-espin 3A construct localizes to and lengthens microvilli in LLC-PK1-CL4 cells; Figure 4 and Materials and methods distinguish targeting from the defective elongation of deletion variants. Rat small-espin brush-border localization in PMID:9763424 is complementary ortholog evidence, not a human tissue experiment.

Summary: Retain brush-border localization supported by the rat small-espin donor and independent human microvillar targeting.

Reason: The ISS source Q63618 is rat Espn. PMID:9763424 actually isolates brush borders from rat intestine and localizes the 253-aa small espin to intestinal and renal brush borders, with immunofluorescence and pre/postembedding immunogold labeling along microvillar actin bundles (Methods; Figures 3-4; localization Results). Preimmune controls and biochemical enrichment support the assignment. This is a valid ortholog-source localization assertion, not a direct human tissue experiment. Human GFP-espin 3A targeting to apical microvilli in PMID:29572253 Figure 4 independently supports the conserved localization. Brush border retains the organized absorptive-surface context, while the separate microvillus assertion describes its constituent projections.

Summary: Retain inherited actin-cytoskeletal localization.

Reason: The PTHR24153 target leaf PTN002516521 descends from IBD node PTN001192625; its ungapped alignment exactly matches the 854-residue human sequence. The corresponding IBD record supports GO:0015629 and includes human ESPN, mouse Espn and Espnl, and rat Espn. Human self-evidence in WITH/FROM is legitimate descendant evidence, not circularity. Human espin 3A localization in PMID:29572253 and rat small-espin localization in PMID:9763424 independently substantiate the location. Mouse Espnl targeting to the actin network in PMID:26926603 Figure 4 is donor-specific support and is not mislabeled as ESPN or a human assay. No NOT record occurs on the target path.

Summary: Retain the specific SH3-binding interaction as a contextual, non-core partner-binding activity.

Reason: PMID:12598619 supplies positive experimental evidence for the rat Q63618 donor. Rat Purkinje-cell espin variants bind a GST fusion containing the 186-aa C-terminal SH3-containing region of human IRSp53; GST controls and deletion of the espin N-terminal proline-rich segment support domain-directed association (Methods; Figure 7C/D; interaction Results). Mouse IRSp53 and Neuro-2a colocalization experiments are separate from this biochemical assay. The current human long sequence contains a corresponding proline-rich region, supporting the curator's gene-level ISS transfer, while the 288-aa product lacks the long N-terminal region. Retain the specific binding as a contextual interaction rather than making it the principal actin-bundling function. The tested rat PC products are not natural human isoform equivalents, and neither a human ESPN binding assay nor a binding constant for these constructs was reported.

Summary: Retain inherited microvillar actin-bundle assembly.

Reason: Human ESPN leaf PTN002516521 lies below the GO:0030034 IBD at PTN002792299. The sole listed human descendant is valid experimental grounding rather than circular support. The independently read PMID:29572253 experiments show human espin 3A-dependent microvillus elongation, loss of this response with the disease-associated deletion, and impaired F-actin bundling in Figure 5L. Together with direct filament binding, this establishes the protein’s structural work in bundle assembly rather than a necessity-only phenotype. The human target MSA is exact and no target-path NOT is present.

Summary: Human espin contributes directly to microvillar actin-bundle assembly through its actin-cross-linking activity.

Reason: PMID:29572253 combines human espin 3A-driven microvillar elongation in LLC-PK1-CL4 cells (Figure 4) with purified-protein F-actin binding and bundling assays (Figure 5K/L). The six-residue deletion retains measurable actin binding but strongly impairs bundling and microvillar elongation; the charge-pattern replacement provides a further control. The protein therefore does structural work in building the parallel actin bundle, rather than merely being required for a downstream phenotype. Retain the process as core, with the experimentally tested human 3A construct and heterologous host explicit.

Summary: Refine the structural location to actin filament bundles.

Reason: GO:0031941 denotes the actin polymer itself. The inspected evidence instead locates espin along bundles and demonstrates filament cross-linking: rat small-espin immunogold labeling lies along microvillar actin bundles (PMID:9763424 Figure 4), and human espin 3A targets actin-rich projections and bundles purified F-actin (PMID:29572253 Figures 4–5). GO:0032432 captures membership in the bundle assembly without describing espin as a constituent of the actin polymer. PMID:8799813 remains abstract-only; its positive rat-fragment report is preserved as source provenance rather than treated as a fully inspected experiment.

Proposed replacements: actin filament bundle

Summary: Human espin 3A targets stereocilia in mouse cochlear hair-cell explants.

Reason: PMID:29572253 Methods express GFP-tagged human espin constructs in P1-P2 mouse organ-of-Corti explants. Wild-type human espin 3A localizes along stereocilia of inner and outer hair cells and promotes elongation; the truncation fails targeting, whereas the six-residue deletion retains stereociliary localization without elongation (Figure 5A-J). This is positive localization of a human protein in a mouse tissue host, not an endogenous human hair-cell localization experiment.

Summary: Retain stereociliary localization.

Reason: The SL-0302 entry maps directly to GO:0032420. PMID:29572253 Figure 5 and its cochlear-explant Methods show human GFP-espin 3A targeting stereocilia in postnatal mouse hair cells. Human protein identity is distinct from the mouse tissue host. This is positive localization evidence for the experimentally expressed reagent without asserting a particular distribution of every natural human product.

Summary: Retain the phylogenetically supported stereocilium-tip location.

Reason: The GO:0032426 IBD is at PTN008395939, above the human ESPN leaf PTN002516521; this is distinct from the two other asserted nodes. Donor identifiers resolve to mouse Espn (MGI:1861630/Q9ET47) and Espnl (MGI:2685402/Q3UYR4). The target’s 854-residue MSA matches exactly and no target-path NOT is present. PMID:26754646 Results and Figure 8A show endogenous mouse espin-1 at stereocilium tips in P0 inner and outer cochlear hair cells, including control and Myo3a/Myo3b double-mutant mice. Methods specify the espin-1 peptide antibody and confocal imaging. This result does not require universal myosin-III dependence. PMID:26926603 Results and Figure 3 distinguish pan-espin shaft labeling from espin-1-enriched tips; its PB538 antibody also cross-reacts with Espnl, a limitation acknowledged by the authors. Separate Espnl C-terminal antibodies and tagged mouse Espnl support that paralog’s tip localization. These are mouse and isoform-sensitive observations, not direct human hair-cell imaging or proof that the short human product shares the same tip enrichment.

Summary: Retain the ortholog-supported stereocilium-tip location.

Reason: The Q9ET47 donor is mouse Espn, with experimental GO:0032426 records for PMID:26754646 and PMID:22264607. The former original was read directly; the latter is recorded as donor provenance without a claim to have inspected its experiments. PMID:26754646 Results and Figure 8A show endogenous mouse espin-1 at stereocilium tips in P0 inner and outer cochlear hair cells, including control and Myo3a/Myo3b double-mutant mice. Methods specify the espin-1 peptide antibody and confocal imaging. This result does not require universal myosin-III dependence. PMID:26926603 Results and Figure 3 distinguish pan-espin shaft labeling from espin-1-enriched tips; its PB538 antibody also cross-reacts with Espnl, a limitation acknowledged by the authors. Separate Espnl C-terminal antibodies and tagged mouse Espnl support that paralog’s tip localization. These are mouse and isoform-sensitive observations, not direct human hair-cell imaging or proof that the short human product shares the same tip enrichment.

Summary: Retain inherited actin-filament binding.

Reason: The GO:0051015 IBD lies at PTN001192625 and the human leaf descends from it with an exact canonical alignment. Its rat Espn and mouse Espnl donors have experimental records; mouse Espnl actin-network targeting in PMID:26926603 Figure 4 does not establish Espnl crosslinking and is not treated as a human ESPN assay. Independent human espin 3A high-speed cosedimentation with preformed rabbit F-actin in PMID:29572253 Figure 5K supports filament binding, with low-speed bundling measured separately in Figure 5L. Rat small-espin binding and bundling in PMID:9763424 independently corroborate the conserved activity. No target-path NOT is present.

Summary: Actin-filament binding is independently established for human espin and supports the retained ortholog annotation.

Reason: The full original PMID:8799813 could not be read; its abstract positively reports rat-fragment F-actin binding, so source access does not justify rejecting the annotation. PMID:29572253 independently assays purified N-terminally His-tagged human espin 3A with polymerized rabbit skeletal-muscle actin: high-speed cosedimentation recovers wild-type espin with F-actin but not without actin (Figure 5K), separately from the low-speed bundling assay. The deletion mutant retains reduced binding. PMID:9763424 also supplies inspected rat small-espin binding controls and measured affinities (Methods; Figure 6), but those rat-reagent constants are not assigned to human ESPN. Retain the specific core binding activity.

Summary: Refine the generic actin-based projection to microvillus and stereocilium.

Reason: The current ARBA00027839 target-matching condition set is zero-based set 6: FunFam 1.25.40.20:FF:000174 AND Metazoa. Human B1AK53 has this FunFam cross-reference and metazoan lineage; no unrelated rule alternative is used as evidence. The rule’s GO:0098858 location is valid but less specific than the independently supported microvillus and stereocilium locations already carried in rows 2/3 and 10/11. Human GFP-espin 3A experiments in PMID:29572253 Figures 4–5 directly support both projection types. The two replacements are sibling location classes rather than an ancestor/descendant pair.

Proposed replacements: microvillus stereocilium

Summary: Refine the projection cluster to brush border and stereocilium bundle.

Reason: The ARBA00044845 target-matching condition set is zero-based set 7: FunFam 1.25.40.20:FF:000174 AND Eukaryota. This is a genuine matching rule, but GO:0098862 is broader than the experimentally identified projection clusters. Rat small-espin localization along intestinal and kidney brush-border microvilli in PMID:9763424 Figures 3–4 supports the existing brush-border assignment, with its ortholog scope preserved. Human espin 3A localizes to the sensory hair-cell stereocilia bundle in the mouse explant experiment of PMID:29572253 Figure 5; the mouse donor originals independently demonstrate endogenous espin in hair bundles. Brush border and stereocilium bundle are separate current subclasses of the imported cluster term, so these refinements retain both supported contexts without inventing a new biological process.

Proposed replacements: brush border stereocilium bundle