EFEMP2

Summary: EFEMP2 contributes to extracellular-matrix organization.

Reason: The RCA source classifies matrisome proteins, and its recovered Supplementary Table I explicitly detects EFEMP2 in decellularized bone-marrow MSC matrices from five of six donors. Detection alone is not a mechanical assay; the structural contribution is supported independently by direct elastic-fiber assembly and LOX–tropoelastin organization studies. Retain this broad structural assertion without assigning LOX catalysis to EFEMP2.

Summary: Fibulin-4 is an extracellular matrix organizer.

Reason: The authenticated historical source identifies fibulin-4 as an extracellular fibulin-family protein; its original body was unavailable for independent reconstruction of every assay. Subsequent operative studies directly establish elastic-fiber assembly and organization of LOX with tropoelastin, supporting the existing structural-constituent assertion. Do not transfer the paper’s fibulin-3 clone length or fibulin-4 electron microscopy to a sequence-authenticated current construct.

Summary: Calcium binding is an accessory property of the EGF-like repeats.

Reason: The InterPro mapping names calcium-binding EGF-like domains, which are retained in the official 443-residue EFEMP2 sequence. Calcium-dependent matrix interactions are compatible with this prediction. Retain the inferred ion-binding property as non-core; calcium dependence of an interaction is not itself a measurement of ion-binding stoichiometry.

Summary: Binding to SGTA supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and SGTA (O43765), with reference-sequence checksums matching the current official record. SGTA directly binds hydrophobic and misfolded clients for transport and quality control, satisfying the current folding-or-transport chaperone definition. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid pooling approach. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: protein-folding chaperone binding

Summary: The SGTB interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The current official SGTB record supports HSC70/HSP70 association and ATPase regulation by similarity. Its IBA adaptor, TRC-complex and ER-targeting annotations make a related client-handling role plausible, but the inspected evidence does not resolve whether SGTB itself binds nascent or unfolded polypeptides for folding or transport, the partner property specified by GO:0051087. This is uncertainty about that narrower partner class, not evidence that SGTB lacks client binding or that phylogenetic inference is invalid. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid pooling approach. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to RHOXF2 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and RHOXF2 (Q9BQY4), with reference-sequence checksums matching the current official record. The verified RHOXF2 partner is a homeobox DNA-binding transcription factor. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report pull down, two hybrid pooling approach. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: The UBQLN1 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. UBQLN1 is a ubiquitin-dependent quality-control adaptor. Binding to this adaptor does not by itself establish binding to an E3 enzyme or a specific chaperone class. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid pooling approach. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: EFEMP2 binds the extracellular matrix protein elastin.

Reason: PMID:16478991 directly reports reciprocal coimmunoprecipitation of human EFEMP2 and human elastin in transfected 293T lysates. Its purified solid-phase assay instead uses mouse fibulin-4 and bovine tropoelastin. Both support the matrix-protein partner class; the human lysate assay should not be described as a purified human binary assay.

Proposed replacements: extracellular matrix binding

Summary: Binding to ATXN1 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ATXN1 (P54253), with reference-sequence checksums matching the current official record. ATXN1 acts as a CBF1/RBPJ-associated transcriptional corepressor. The recorded constructs do not establish a native extracellular transcription complex. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid, two hybrid array. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: transcription corepressor binding

Summary: Binding to ATN1 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ATN1 (P54259), with reference-sequence checksums matching the current official record. ATN1 has experimentally curated transcriptional corepressor activity. The interaction records include overlapping ATN1 fragments; the class refinement does not claim every fragment retains autonomous repression. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: transcription corepressor binding

Summary: The UBQLN1 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The recorded 1–589 UBQLN1 construct is a quality-control adaptor. Its interaction does not establish EFEMP2 ubiquitination, folding, or an informative specific binding class. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: EFEMP2 binds fibrillin-1.

Reason: The original PMID:19349279 Methods and Results describe recombinant human fibulin-4 binding to N-terminal fibrillin-1 constructs by surface-plasmon resonance, with domain/mutation and ligand-competition comparisons. This supports extracellular matrix binding rather than assigning EFEMP2 fibrillin polymerization or an intrinsic catalytic activity.

Proposed replacements: extracellular matrix binding

Summary: EFEMP2 binds tropoelastin during elastic-fiber organization.

Reason: The original human recombinant binding experiments in PMID:19570982 support the elastin partner and show its relationship to LOX-dependent ternary organization. The reported fibulin-4 construct boundaries (26–418) have not been reconciled to the current 443-residue reference sequence; no natural-product equivalence is asserted.

Proposed replacements: extracellular matrix binding

Summary: EFEMP2 recruits LOX to tropoelastin as an enzyme–substrate adaptor.

Reason: The LOX association is supported by a ternary mechanism in which fibulin-4 brings proLOX/LOX and tropoelastin together. PMID:19570982 tests human recombinant proteins and concentration-dependent ternary organization; PMID:19855011 uses conditioned proLOX-containing medium with purified tropoelastin and fibulin-4 reagents. This supports enzyme-substrate adaptor activity rather than EFEMP2 oxidase catalysis. Construct and preparation differences are retained; a completely purified triple is not claimed.

Proposed replacements: enzyme-substrate adaptor activity

Summary: EFEMP2 binds fibrillin-1 during matrix organization.

Reason: PMID:19570982 shows binding of human recombinant fibulin-4 to N-terminal fibrillin-1, which competes with LOX and tropoelastin binding in the tested system. This is extracellular matrix binding; the historical fibulin-4 construct boundaries do not authenticate the current full-length product.

Proposed replacements: extracellular matrix binding

Summary: EFEMP2 binds fibulin-5.

Reason: The original PMID:19570982 binding experiments identify fibulin-5 as an extracellular matrix partner, with affinity and competition dependent on the assay configuration. This supports extracellular matrix binding without assigning fibulin-5’s distinct mechanistic role to EFEMP2.

Proposed replacements: extracellular matrix binding

Summary: EFEMP2 recruits LOX to tropoelastin as an enzyme–substrate adaptor.

Reason: The LOX association is supported by a ternary mechanism in which fibulin-4 brings proLOX/LOX and tropoelastin together. PMID:19570982 tests human recombinant proteins and concentration-dependent ternary organization; PMID:19855011 uses conditioned proLOX-containing medium with purified tropoelastin and fibulin-4 reagents. This supports enzyme-substrate adaptor activity rather than EFEMP2 oxidase catalysis. Construct and preparation differences are retained; a completely purified triple is not claimed.

Proposed replacements: enzyme-substrate adaptor activity

Summary: The paper supports EFEMP2 binding, but its tested partner is PITX2.

Reason: The imported source names LOX (UniProtKB:P28300), whereas the actual PMID:22919265 interaction experiments concern PITX2, including an EFEMP2 construct lacking 36 N-terminal residues and reciprocal coimmunoprecipitation in COS-7 cells. Preserve the imported partner provenance and the paper’s positive generic-binding result, but do not refine a LOX-tagged assertion to transcription-factor binding. Independent LOX interaction evidence exists in PMID:19570982 and PMID:19855011; the mismatch does not show that LOX binding is biologically false.

Summary: Human EFEMP2 binds the mouse Hoxa1 transcription-factor construct.

Reason: PMID:23088713 uses mouse Hoxa1 against a human ORFeome. Its target tables list EFEMP2/O95967 in both screen configurations and mark co-purification positive, while BiFC is not determined. The Methods specify tagged Hoxa1 and human ORF products in mammalian lysate pull-downs. This supports the DNA-binding transcription-factor partner class while preserving the cross-species and overexpression context. Human EFEMP2 is the tested prey, so this is a measured cross-species association rather than a transfer from mouse Efemp2. The DNA-binding transcription-factor class describes the assayed mouse Hoxa1 partner; it does not assert an endogenous human HOXA1 complex, physiological co-localization or binding to every member of that class. Unlike the PITX2/LOX citation mismatch, the imported P09022 partner matches the protein used here.

Proposed replacements: DNA-binding transcription factor binding

Summary: Binding to SGTA supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and SGTA (O43765), with reference-sequence checksums matching the current official record. SGTA is a client-binding co-chaperone involved in folding quality control and protein transport. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: protein-folding chaperone binding

Summary: Binding to STK16 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and STK16 (O75716), with reference-sequence checksums matching the current official record. STK16 is an active protein serine/threonine kinase. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: protein kinase binding

Summary: Binding to PTGER3 isoform 12 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and PTGER3 isoform 12 (P43115-12), with reference-sequence checksums matching the current official record. The official P43115-12 sequence alters the C-terminal tail while retaining the prostaglandin-receptor body. The term asserts receptor binding, not receptor agonism. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: signaling receptor binding

Summary: Binding to MEOX2 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and MEOX2 (P50222), with reference-sequence checksums matching the current official record. MEOX2 is a sequence-specific homeobox transcription factor. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: Binding to HDAC4 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and HDAC4 (P56524), with reference-sequence checksums matching the current official record. HDAC4 is the verified histone-deacetylase-class partner; the interaction does not give EFEMP2 deacetylase activity. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: histone deacetylase binding

Summary: Binding to CREB5 isoform 3 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and CREB5 isoform 3 (Q02930-3), with reference-sequence checksums matching the current official record. The official Q02930-3 edit removes the first 33 canonical residues. The assayed 142–475 partner region retains its bZIP DNA-binding region. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: The LCE3C interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. A late cornified-envelope protein is not thereby an actin, microtubule or intermediate-filament component. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The LCE2D interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. Cornified-envelope membership does not establish a cytoskeletal-protein partner class. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The NUFIP2 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The verified NUFIP2 RNA-associated scaffold is not itself a DNA-binding transcription factor or an enzyme on the evidence inspected. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The C5orf24 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. No informative molecular partner class was established for this poorly characterized protein. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The RBPMS interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The reference RBPMS partner is an RNA-binding protein; coactivator behavior attributed to a different isoform should not be transferred to the assayed reference record. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The SGTB interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The current official SGTB record supports HSC70/HSP70 association and ATPase regulation by similarity. Its IBA adaptor, TRC-complex and ER-targeting annotations make a related client-handling role plausible, but the inspected evidence does not resolve whether SGTB itself binds nascent or unfolded polypeptides for folding or transport, the partner property specified by GO:0051087. This is uncertainty about that narrower partner class, not evidence that SGTB lacks client binding or that phylogenetic inference is invalid. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The UBQLN1 isoform 2 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The Q9UMX0-2 sequence deletes canonical 418–445 within the STI1-repeat region. Its quality-control adaptor identity does not establish a more informative enzyme or chaperone binding class. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The PRPF31 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The curated PRPF31 pair includes a mutation reported not to affect interaction. Spliceosomal association does not assign EFEMP2 RNA splicing or a defined enzymatic binding activity. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid bait and prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The AMMECR1 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The exact AMMECR1 association is recovered from the alternative-splicing interaction study, but a specific functional binding class is not established. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Recombinant fibulin-4 binds extracellular-matrix partners in the cited study.

Reason: The complete PMID:27339457 abstract reports binding of recombinant fibulin-4 to LTBP1, LTBP3 and LTBP4, together with LOX/LOXL1 interactions and mutation-dependent differences. These LTBP interactions support extracellular matrix binding (GO:0050840) within the mixed imported partner set; the refinement does not assert that every listed partner belongs to the same class. The current UniProt SUBUNIT account also attributes TGFB1 and LOXL2 comparisons to this paper, but the full target-specific Methods and binding tables were not recovered. Preserve all seven imported partners and their source identity without inventing separate records, a mature-TGFB1 interface, or a native seven-protein complex.

Proposed replacements: extracellular matrix binding

Summary: The SGTB interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The mutation-tested SGTB pair remains positive. The current official SGTB record supports HSC70/HSP70 association and ATPase regulation by similarity. Its IBA adaptor, TRC-complex and ER-targeting annotations make a related client-handling role plausible, but the inspected evidence does not resolve whether SGTB itself binds nascent or unfolded polypeptides for folding or transport, the partner property specified by GO:0051087. This is uncertainty about that narrower partner class, not evidence that SGTB lacks client binding or that phylogenetic inference is invalid. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The CYSRT1 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The cysteine-rich tail protein lacks an established informative partner class in the inspected record. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The PRR32 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The proline-rich protein’s identity is verified, but its molecular class remains insufficiently resolved for a narrower binding assertion. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: EFEMP2 binds the experimentally characterized transporter SLC16A5/MCT6.

Reason: The source-linked human EFEMP2–SLC16A5 two-hybrid records are positive and match the O15375 reference sequence. The primary PMID:16174808 abstract explicitly reports human MCT6-mediated uptake of bumetanide and other drugs in Xenopus oocytes, while distinguishing its substrate specificity from other MCT family members. This supports the transmembrane-transporter partner class despite uncertainty about its endogenous substrate. Refine to transporter binding without assigning EFEMP2 transport activity, modulation of MCT6, or a native extracellular complex. Full Methods and the historical transporter insert sequence were not independently recovered from that paper.

Proposed replacements: transmembrane transporter binding

Summary: Binding to TLX3 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and TLX3 (O43711), with reference-sequence checksums matching the current official record. TLX3 is a homeobox DNA-binding transcription factor. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: Binding to SGTA supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and SGTA (O43765), with reference-sequence checksums matching the current official record. SGTA binds misfolded/hydrophobic client proteins for chaperone-mediated quality control and transport. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: protein-folding chaperone binding

Summary: Binding to PROP1 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and PROP1 (O75360), with reference-sequence checksums matching the current official record. PROP1 is a homeobox DNA-binding transcription factor. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: Binding to STK16 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and STK16 (O75716), with reference-sequence checksums matching the current official record. The independently identified STK16 partner is a protein kinase. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: protein kinase binding

Summary: The C3 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The assayed complement precursor has several processed functional products. The record does not identify which region mediates binding, so a processed cytokine-like or receptor-ligand class is not assumed. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The RBP3 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. RBP3 is an interphotoreceptor retinoid carrier; its peptidase-like fold does not establish protease activity. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to ITGB5 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ITGB5 (P18084), with reference-sequence checksums matching the current official record. ITGB5 is the beta subunit of an integrin receptor. Binding to this integrin component supports integrin binding without asserting integrin activation or identifying the physiological ligand interface. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: integrin binding

Summary: The IGFBP6 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. IGFBP6 binds growth factors but is not thereby a growth factor. The pair does not establish EFEMP2 growth-factor binding. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to OTX1 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and OTX1 (P32242), with reference-sequence checksums matching the current official record. OTX1 is a sequence-specific homeobox transcription factor. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: The HPCAL1 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. HPCAL1 is a calcium-sensor protein, not a protein kinase or a calcium ion. No supported more informative binding class was identified. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to CXCL5 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and CXCL5 (P42830), with reference-sequence checksums matching the current official record. The verified CXCL5 partner belongs to the C-X-C chemokine class. No EFEMP2 chemotactic activity is implied. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: C-X-C chemokine binding

Summary: Binding to BAG6 isoform 2 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and BAG6 isoform 2 (P46379-2), with reference-sequence checksums matching the current official record. BAG6 is an ATP-independent client-binding chaperone. P46379-2 deletes six residues (canonical 185–190), rather than removing the protein’s chaperone scaffold; the assay does not establish isoform-specific activity. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: protein-folding chaperone binding

Summary: Binding to MXI1 isoform 3 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and MXI1 isoform 3 (P50539-3), with reference-sequence checksums matching the current official record. MXI1 is a DNA-binding transcriptional repressor. The official P50539-3 N-terminal alteration and recorded 62–295 fragment preserve its bHLH region. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: The MRPL12 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. MRPL12 is a mitochondrial ribosomal protein; the interaction does not establish an EFEMP2 translation or ribosome-assembly activity. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to LBX1 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and LBX1 (P52954), with reference-sequence checksums matching the current official record. LBX1 is a homeobox DNA-binding transcription factor. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: The KRTAP12-2 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The inspected record describes a keratin-associated interfilamentous-matrix protein, without the component assignment needed to equate it with a keratin filament or another cytoskeletal polymer. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The KRTAP10-8 interaction is retained without assigning a cytoskeletal partner class.

Reason: The source-linked human EFEMP2–KRTAP10-8 records are positive two-hybrid interactions, and the P60410 reference-sequence checksum matches. The keratin-filament annotation traces to the IPR002494 family mapping. The corresponding InterPro and UniProt descriptions place this protein in the hair interfilamentous matrix surrounding keratin intermediate filaments. This does not adequately resolve the protein-component-of-a-cytoskeleton requirement of GO:0008092 for this partner. Retain the supported generic interaction as non-core; do not deny the interaction or require the partner itself to polymerize into a filament.

Summary: Binding to PITX1 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and PITX1 (P78337), with reference-sequence checksums matching the current official record. PITX1 is a homeobox DNA-binding transcription factor, distinct from the PITX2 construct in PMID:22919265. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: Binding to CREB5 isoform 3 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and CREB5 isoform 3 (Q02930-3), with reference-sequence checksums matching the current official record. The Q02930-3 partner is sequence-verified, and its assayed 142–475 region retains the bZIP DNA-binding region after the documented N-terminal deletion. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: Binding to TLE5 isoform 2 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and TLE5 isoform 2 (Q08117-2), with reference-sequence checksums matching the current official record. TLE5 is a transcriptional corepressor. Its official isoform-2 N-terminal replacement retains the corepressor protein body; no isoform-specific regulatory effect is inferred. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: transcription corepressor binding

Summary: The INCA1 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The partner’s inhibition of a cyclin-dependent kinase is not itself protein kinase activity; the binding hit cannot be refined to kinase binding on that basis. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to RAB40B supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and RAB40B (Q12829), with reference-sequence checksums matching the current official record. RAB40B is a small monomeric GTPase. Its additional ubiquitin-ligase-complex role need not be transferred to interpret this binding record. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: small GTPase binding

Summary: Binding to KLF1 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and KLF1 (Q13351), with reference-sequence checksums matching the current official record. KLF1 is a sequence-specific DNA-binding transcription factor. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: The IL16 isoform 2 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The Q14005-2 reference sequence is verified, but the assayed 1–453 region is N-terminal and excludes the C-terminal mature cytokine region. Cytokine binding would overstate the tested construct. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to ZNF345 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ZNF345 (Q14585), with reference-sequence checksums matching the current official record. ZNF345 carries current phylogenetic DNA-binding transcription-factor and cis-regulatory-region annotations, corroborated by an experimentally curated sequence-specific DNA-binding annotation. The partner-class inference is distinct from direct EFEMP2 transcriptional activity. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: Binding to CLPP supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and CLPP (Q16740), with reference-sequence checksums matching the current official record. CLPP is the proteolytic subunit of the mitochondrial Clp protease. A bound protease class is sufficient; no catalytic-site recognition is asserted. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: protease binding

Summary: The DOK7 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. DOK7 is a kinase activator/adaptor rather than a protein kinase; its positive pair does not justify kinase binding or an EFEMP2 neuromuscular-signaling role. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to ZNF699 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ZNF699 (Q32M78), with reference-sequence checksums matching the current official record. ZNF699 has current phylogenetic DNA-binding transcription-factor/cis-regulatory-region annotations and an intact annotated zinc-finger protein sequence. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: The KRTAP19-5 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The inspected keratin-associated interfilamentous-matrix description is not equivalent to direct identification as an intermediate-filament component. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The KRTAP13-3 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The interfilamentous-matrix/keratin-associated partner is not shown here to be a polymerized keratin or an established cytoskeletal component. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The LCE3E interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The late cornified-envelope protein is not an intermediate filament merely because it is keratinocyte-associated. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The LCE1C interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. Cornified-envelope membership does not establish actin, microtubule or intermediate-filament identity. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The LCE1E interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. No supported narrower partner-class function follows from late cornified-envelope membership. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The LCE1A interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The partner is a cornified-envelope protein, which is not equivalent to a cytoskeletal filament. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The LCE3A interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. Cornified-envelope association supports no specific cytoskeletal or receptor binding class here. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The LCE2C interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The late cornified-envelope partner is not an enzyme or established cytoskeletal filament component. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to ZNF57 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ZNF57 (Q68EA5), with reference-sequence checksums matching the current official record. ZNF57 has current phylogenetic DNA-binding transcription-factor and transcription-regulatory-region annotations. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: Binding to NR1D2-associated Q6NSM0 record supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and NR1D2-associated Q6NSM0 record (Q6NSM0), with reference-sequence checksums matching the current official record. The literal Q6NSM0 reference sequence is an unreviewed NR1D2-associated record retaining the annotated nuclear-receptor DNA-binding domain. This is a bounded partner-class inference, not authentication of a full-length canonical NR1D2 assay or of receptor agonism. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: The CDPF1 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The cysteine-rich-domain protein’s informative biochemical class is unresolved in the inspected evidence. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to ZNF774 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ZNF774 (Q6NX45), with reference-sequence checksums matching the current official record. ZNF774 has current phylogenetic DNA-binding transcription-factor and cis-regulatory-region annotations. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: The KRTAP26-1 interaction is retained without assigning a cytoskeletal partner class.

Reason: The source-linked human EFEMP2–KRTAP26-1 records are positive two-hybrid interactions, and the Q6PEX3 reference-sequence checksum matches. The keratin-filament annotation traces to the IPR007659 family mapping. The corresponding InterPro and UniProt descriptions place this protein in the hair interfilamentous matrix surrounding keratin intermediate filaments. This does not adequately resolve the protein-component-of-a-cytoskeleton requirement of GO:0008092 for this partner. Retain the supported generic interaction as non-core; do not deny the interaction or require the partner itself to polymerize into a filament.

Summary: The SPATA8 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The spermatogenesis-associated name does not establish a specific biochemical binding class or an EFEMP2 reproductive role. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to ADAMTSL4 isoform 3 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ADAMTSL4 isoform 3 (Q6UY14-3), with reference-sequence checksums matching the current official record. ADAMTSL4 is an extracellular matrix glycoprotein, not an ADAMTS protease. The Q6UY14-3 reference and assayed 674–1097 region preserve its C-terminal thrombospondin-repeat matrix-protein region. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: extracellular matrix binding

Summary: The SLC25A48 isoform 3 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The Q6ZT89-3 product is only 157 residues and substantially truncates the canonical mitochondrial-carrier architecture. Transporter activity cannot be transferred from the full-length gene product by name alone. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The NUFIP2 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The RNA-associated scaffold interaction does not itself establish a defined enzyme, DNA-binding transcription-factor or chaperone class. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The SPATA24 isoform 2 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The Q86W54-2 altered-tail product is sequence-verified, but its molecular role remains insufficiently characterized for a narrower term. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The FAM90A1 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The family-associated protein has no established informative molecular partner class in the inspected sources. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The KRTAP11-1 interaction is retained without assigning a cytoskeletal partner class.

Reason: The source-linked human EFEMP2–KRTAP11-1 records are positive two-hybrid interactions, and the Q8IUC1 reference-sequence checksum matches. The keratin-filament annotation traces to the IPR007659 family mapping. The corresponding InterPro and UniProt descriptions place this protein in the hair interfilamentous matrix surrounding keratin intermediate filaments. This does not adequately resolve the protein-component-of-a-cytoskeleton requirement of GO:0008092 for this partner. Retain the supported generic interaction as non-core; do not deny the interaction or require the partner itself to polymerize into a filament.

Summary: Binding to TRIM42 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and TRIM42 (Q8IWZ5), with reference-sequence checksums matching the current official record. TRIM42 carries a current electronic ubiquitin-ligase annotation and retains its annotated RING-type zinc finger. The E3 partner class is inferred, not an independently measured EFEMP2-dependent ubiquitination mechanism. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: ubiquitin protein ligase binding

Summary: The CNNM3 interaction is retained while its transporter partner class remains unresolved.

Reason: The source-linked human EFEMP2–CNNM3 two-hybrid records are positive and match the Q8NE01 reference sequence. UniProt describes a probable transporter by similarity, and a current IBA assigns transmembrane transporter activity. However, primary human CNNM3 experiments demonstrate regulation of the TRPM7 transport complex, with differing effects across assays; the direct ion-translocation role of CNNM3 itself remains unresolved. The TRPM7-independent efflux experiments in PMID:34928937 principally test CNNM2/CNNM4, not CNNM3. Retain the generic interaction as non-core rather than treating the CNNM3 hit as sufficient evidence of binding to an established transmembrane transporter or an assembled TRPM7 complex. This is a source-specific class limitation, not a rejection of phylogenetic evidence generally or of the interaction.

Summary: Binding to LNX1 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and LNX1 (Q8TBB1), with reference-sequence checksums matching the current official record. LNX1 is an E3 ubiquitin-protein ligase. The binding assertion does not identify EFEMP2 as a ubiquitination substrate. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: ubiquitin protein ligase binding

Summary: Binding to RNF138 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and RNF138 (Q8WVD3), with reference-sequence checksums matching the current official record. RNF138 is an E3 ubiquitin-protein ligase. The record supports binding to that partner class, not EFEMP2 ubiquitination or ligase activity. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: ubiquitin protein ligase binding

Summary: The PRPF31 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. PRPF31 is a spliceosome-associated protein. A binary interaction does not itself establish EFEMP2 splicing activity or a specific enzyme-binding term. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The FBXW5 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. FBXW5 is a substrate-recognition component of SCF/DCX ligase complexes, not by itself the ubiquitin-ligase enzyme asserted by the narrower enzyme-binding term. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The ARRDC3 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The arrestin-domain trafficking adaptor is not a signaling receptor merely because it regulates receptor trafficking. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to SPACA9 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and SPACA9 (Q96E40), with reference-sequence checksums matching the current official record. SPACA9 is a structurally identified microtubule-associated cytoskeletal component. This supports cytoskeletal-protein binding without assigning EFEMP2 microtubule binding or ciliary motility. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: cytoskeletal protein binding

Summary: Binding to ZNF837 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ZNF837 (Q96EG3), with reference-sequence checksums matching the current official record. ZNF837 has current phylogenetic DNA-binding transcription-factor and cis-regulatory-region annotations. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: The SGTB interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The current official SGTB record supports HSC70/HSP70 association and ATPase regulation by similarity. Its IBA adaptor, TRC-complex and ER-targeting annotations make a related client-handling role plausible, but the inspected evidence does not resolve whether SGTB itself binds nascent or unfolded polypeptides for folding or transport, the partner property specified by GO:0051087. This is uncertainty about that narrower partner class, not evidence that SGTB lacks client binding or that phylogenetic inference is invalid. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The HHIPL1 isoform 2 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The Q96JK4-2 product alters/truncates the C terminus and lacks the canonical SRCR domain. Neither the HHIP-like name nor this pair establishes a receptor class. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The HSPA12B interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. An atypical Hsp70-family similarity assignment without a demonstrated or reliably annotated client-binding activity is insufficient for a chaperone-binding refinement. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to ZNF558 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ZNF558 (Q96NG5), with reference-sequence checksums matching the current official record. The ZNF558 record combines a direct transcriptional-regulation function with phylogenetic DNA-binding transcription-factor annotations. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: Binding to TRIB3 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and TRIB3 (Q96RU7), with reference-sequence checksums matching the current official record. TRIB3 is a catalytically inactive kinase-like protein with a documented transcriptional corepressor role involving ATF4/DDIT3. Use the latter class rather than protein kinase binding. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: transcription corepressor binding

Summary: Binding to ZNF587 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ZNF587 (Q96SQ5), with reference-sequence checksums matching the current official record. ZNF587 has current phylogenetic DNA-binding transcription-factor and cis-regulatory-region annotations. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: Binding to P2RX7 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and P2RX7 (Q99572), with reference-sequence checksums matching the current official record. P2RX7 is an ATP-gated purinergic receptor. The binding class does not establish receptor agonism or ion-channel regulation by EFEMP2. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: signaling receptor binding

Summary: The MDFI interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. MDFI can inhibit transcription factors through cytoplasmic sequestration; that does not establish the basal-transcription-machinery corepressor class required by GO:0001222. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The FAM110A interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. Centrosomal/cell-cycle association of the partner does not establish a cytoskeletal polymer or enzyme class. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to ZNF670 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ZNF670 (Q9BS34), with reference-sequence checksums matching the current official record. ZNF670 has current phylogenetic DNA-binding transcription-factor and cis-regulatory-region annotations. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: Binding to THAP7 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and THAP7 (Q9BT49), with reference-sequence checksums matching the current official record. THAP7 is a chromatin-associated transcriptional repressor that recruits HDAC3 and nuclear-receptor corepressors. The corepressor class is more appropriate than inferring sequence-specific DNA binding from its THAP name. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: transcription corepressor binding

Summary: The AP1M1 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. AP1M1 is an adaptor-complex subunit, not a transmembrane transporter or an enzyme. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The LCE3D interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The late cornified-envelope partner is not an intermediate filament merely because it is expressed in keratinocytes. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The KRTAP3-2 interaction is retained without assigning a cytoskeletal partner class.

Reason: The source-linked human EFEMP2–KRTAP3-2 records are positive two-hybrid interactions, and the Q9BYR7 reference-sequence checksum matches. The keratin-filament annotation traces to the IPR007659 family mapping. The corresponding InterPro and UniProt descriptions place this protein in the hair interfilamentous matrix surrounding keratin intermediate filaments. This does not adequately resolve the protein-component-of-a-cytoskeleton requirement of GO:0008092 for this partner. Retain the supported generic interaction as non-core; do not deny the interaction or require the partner itself to polymerize into a filament.

Summary: The SPRY4 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. A signaling regulator that inhibits receptor pathways is not itself a receptor or kinase; the pair supports no narrower class here. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The FTO isoform 2 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The Q9C0B1-2 product is 127 residues, with deletion of canonical 1–378 and replacement of 379–413. It lacks the canonical demethylase architecture, so enzyme binding cannot be assigned from the FTO gene name. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to HSD3B7 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and HSD3B7 (Q9H2F3), with reference-sequence checksums matching the current official record. HSD3B7 is a steroid-metabolizing oxidoreductase. The binding observation supports enzyme binding without transferring its catalytic reaction to EFEMP2. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: enzyme binding

Summary: Binding to ZNF768 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ZNF768 (Q9H5H4), with reference-sequence checksums matching the current official record. ZNF768 has experimentally supported promoter/MIR sequence binding and a current phylogenetic transcription-factor annotation. The assayed 21–540 fragment retains its annotated DNA-binding zinc-finger region. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: The ANAPC11 isoform 2 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The Q9NYG5-2 sequence replaces canonical 38–84, disrupting the annotated RING region 34–77. A full-reference checksum match does not restore the missing catalytic scaffold or establish ubiquitin-ligase activity of this product. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to RBAK supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and RBAK (Q9NYW8), with reference-sequence checksums matching the current official record. RBAK has current phylogenetic DNA-binding transcription-factor annotations and published transcription-regulatory functions. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: Binding to ZNF581 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ZNF581 (Q9P0T4), with reference-sequence checksums matching the current official record. ZNF581 carries current phylogenetic DNA-binding transcription-factor and cis-regulatory-region annotations. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: DNA-binding transcription factor binding

Summary: The Neurotrimin isoform 3 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The Q9P121-3 sequence retains extracellular Ig-like domains but loses the canonical C-terminal GPI-anchor region. Its adhesion-protein identity does not alone establish the signaling-receptor binding definition. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The SLC23A1 isoform 3 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The Q9UHI7-3 product deletes canonical 91–429, removing much of the transporter architecture. A transporter-binding refinement would assume activity not established for the assayed product. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to USP21 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and USP21 (Q9UK80), with reference-sequence checksums matching the current official record. USP21 is a ubiquitin-specific protease; the recorded 185–565 partner fragment contains its catalytic region. This is binding to the protease class, not an EFEMP2 deubiquitination assay. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: ubiquitin-specific protease binding

Summary: The UBQLN1 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. UBQLN1 is a quality-control adaptor. Binding does not establish a ubiquitin-ligase enzyme or an EFEMP2 folding reaction. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: The AMMECR1 isoform 3 interaction is retained without an unsupported class refinement.

Reason: The exact source-linked pair is present in the curated interaction records and the partner reference sequence is verified. The Q9Y4X0-3 sequence deletes canonical 159–195, but neither the intact nor shortened partner has a sufficiently resolved functional class for an informative refinement. Retain the supported generic association as non-core under the project-specific binding policy. The source-linked records report two hybrid array, two hybrid prey pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Summary: Binding to ATXN1 supports a specific partner class.

Reason: The curated pair records identify EFEMP2 and ATXN1 (P54253), with reference-sequence checksums matching the current official record. The partner’s CBF1-associated transcriptional-corepressor function is independently established. Retain the tested construct context without inferring that secreted EFEMP2 participates in native nuclear repression. The refinement describes the bound partner; it does not transfer that partner’s activity to EFEMP2 or demonstrate a native interaction in every compartment. The source-linked records report two hybrid array, two hybrid pooling approach, validated two hybrid. These are assay-context associations; a native extracellular complex is not established by the screen alone.

Proposed replacements: transcription corepressor binding

Summary: Reactome places mature EFEMP2 in the extracellular region.

Reason: The saved R-HSA-1592387 input is DefinedSet 2395327, whose hasMember 2161231 is EFEMP2/O95967 residues 24–443 in GO:0005576. This is an exact target entity and compartment assertion, not an inference from another fibulin’s name.

Summary: EFEMP2 remains extracellular in the modeled elastic-fiber assembly.

Reason: The saved R-HSA-2161282 input complex 2396077 contains 2395337, which contains fibulin DefinedSet 2395327 and its EFEMP2/O95967 member 2161231, residues 24–443. The model explicitly places that chain in the extracellular region; association of other proteins in the reaction does not confer their activities on EFEMP2.

Summary: The narrower basement-membrane placement remains source-limited.

Reason: The literal ortholog source is mouse Efemp2/Q9WVJ9. Its curated location and PMID:17324935 comparative-family abstract were inspected, but the target-specific basement-membrane images/assay were not recovered. The abstract establishes microfibril-associated localization and matrix interactions, which do not alone resolve this narrower compartment. Preserve the positive imported assertion pending that source rather than treating general extracellular localization as proof or disproof.

Summary: The narrower basement-membrane placement remains source-limited.

Reason: The literal ortholog source is mouse Efemp2/Q9WVJ9. Its curated location and PMID:17324935 comparative-family abstract were inspected, but the target-specific basement-membrane images/assay were not recovered. The abstract establishes microfibril-associated localization and matrix interactions, which do not alone resolve this narrower compartment. Preserve the positive imported assertion pending that source rather than treating general extracellular localization as proof or disproof.

Summary: The historical basement-membrane attribution could not be independently resolved.

Reason: PMID:10601734 is bibliographically authenticated but remains abstract-only in the available source set. Its fibulin-family characterization does not expose the target-specific localization experiment needed to distinguish basement membrane from other extracellular matrix. Preserve the source assertion without declaring it false.

Summary: Multimeric fibulin-4 binds heparin in the reported biochemical assay.

Reason: PMID:23782690 directly assays recombinant human fibulin-4 forms and finds stronger heparin binding for multimeric material, with calcium-dependent cell-adhesion behavior. Retain this accessory interaction without assigning a demonstrated endogenous heparan-sulfate receptor mechanism or a separate principal function.

Summary: EFEMP2 is detected in cell-derived extracellular matrix.

Reason: The recovered PMID:28327460 Supplementary Table I row 29 lists EFEMP2 in matrix from five of six bone-marrow MSC donors, with donor-specific peptide counts and precursor-ion abundance. The Methods describe decellularization, matrix enrichment and LC-MS/MS. This supports the ECM location, without equating detection with a target-specific mechanical perturbation assay.

Summary: The extracellular-matrix location is inherited within the fibulin lineage.

Reason: The literal IBD for GO:0031012 is placed at PTN001170088. Current PANTHER PTHR24034 tree and MSA reconstruction places human O95967 below that node and exactly matches its 443-residue sequence. The descendant donor paths, including target evidence, are consistent with the transfer; self-inclusion is not circularity. Independent human matrix proteomics corroborates the location.

Summary: EFEMP2 contributes to aortic development through extracellular-matrix organization.

Reason: The transfer traces to mouse Efemp2/Q9WVJ9 and its developmental knockout studies. PMID:20019329 measures disrupted elastic laminae, smooth-muscle organization and aortic pathology after germline or smooth-muscle loss. These are genuine tissue-development effects, but they remain contextual outcomes of the extracellular organizer rather than a distinct molecular activity.

Summary: EFEMP2 contributes to aortic development through extracellular-matrix organization.

Reason: The transfer traces to mouse Efemp2/Q9WVJ9 and its developmental knockout studies. PMID:20019329 measures disrupted elastic laminae, smooth-muscle organization and aortic pathology after germline or smooth-muscle loss. These are genuine tissue-development effects, but they remain contextual outcomes of the extracellular organizer rather than a distinct molecular activity.

Summary: Recombinant fibulin-4 forms dimers and larger assemblies.

Reason: PMID:23782690 separates monomeric, dimeric and multimeric recombinant human fibulin-4 and relates the assemblies to heparin binding and cell adhesion. Retain the measured homodimerization property as accessory to matrix organization; do not infer a universal endogenous stoichiometry or a separate catalytic unit.

Summary: The elastic-fiber assembly role follows the curated fibulin-4 ancestral node.

Reason: The literal GO:0048251 IBD at PTN002778230 is experimentally grounded by mouse Efemp2 and human EFEMP2. The current tree/MSA independently places the exact human sequence below this post-duplication node; no target-specific loss was identified. Human ternary binding and ortholog assembly experiments support the inherited organizer role.

Summary: The electronic elastic-fiber assembly prediction matches direct functional evidence.

Reason: ARBA00089721 maps a specified fibulin-related FunFam condition to GO:0048251. Its rule output was inspected, without claiming that every historical classifier feature was independently recomputed. Direct binding/ternary organization and knockout/reconstitution studies establish EFEMP2’s participation as an extracellular organizer, not an oxidase.

Summary: Fibulin-4 regulates the maturation phase of tropoelastin coacervation.

Reason: PMID:18973305 uses defined elastin polypeptides and recombinant fibulin proteins. Fibulin-4 reduces the rate of coacervate maturation without the same effect on the initial coacervation rate or temperature. Retain its direct participation in assembly while keeping fibulin-5-specific imaging and mutant effects separate.

Summary: The elastic-fiber assembly assertion is consistent with fibulin-4 interaction biology.

Reason: The PMID:27339457 abstract reports fibulin-4 matrix interactions and secretion defects for tested disease-associated variants; the full target-specific assembly experiment was unavailable. The existing process is independently supported by direct LOX–tropoelastin organization and assembly studies, so retain it while deferring to the curator for the inaccessible source’s detailed IMP design.

Summary: Ortholog evidence supports EFEMP2 participation in elastic-fiber assembly.

Reason: The literal mouse Efemp2/Q9WVJ9 donor has direct knockout and biochemical assembly evidence. PMID:16478991 distinguishes human lysate co-IP from purified mouse fibulin-4/bovine tropoelastin assays and mouse knockout phenotypes; together with human ternary organization, this supports the conserved organizer role.

Summary: Aortic smooth-muscle morphogenesis is a conserved matrix-dependent context.

Reason: The mouse donor studies show disrupted smooth-muscle shape, arrangement and contacts with elastic laminae after Efemp2 loss. This supports the transferred tissue-morphogenesis assertion. The evidence does not establish a cell-intrinsic morphogenetic enzyme or transcription-factor role for secreted EFEMP2.

Summary: Aortic smooth-muscle morphogenesis is a conserved matrix-dependent context.

Reason: The mouse donor studies show disrupted smooth-muscle shape, arrangement and contacts with elastic laminae after Efemp2 loss. This supports the transferred tissue-morphogenesis assertion. The evidence does not establish a cell-intrinsic morphogenetic enzyme or transcription-factor role for secreted EFEMP2.

Summary: EFEMP2 was detected in the human parotid-saliva exosome fraction.

Reason: The original supplementary workbook identifies EFEMP2 at physical row 284 as IPI00296058.1, with 3 sequences, 5 spectra and 6.3% sequence coverage. Retain this measured fraction-associated pool as non-core. The table’s functional/location descriptions are annotations, whereas the peptide measurements are the experimental observation; neither establishes a physiological exosome-specific function.

Summary: A non-core extracellular-exosome-associated pool is supported, while this urinary target measurement remains unverified.

Reason: Independent PMID:19199708 supplementary data identify EFEMP2 in a human parotid-saliva exosome fraction with 3 sequences, 5 spectra and 6.3% coverage. This supports the gene-level non-core extracellular-exosome association also represented by the adjacent annotation. PMID:23533145 describes pooled expressed-prostatic-secretion urine, differential ultracentrifugation and LC-MS/MS with FDR below 1%, but its EFEMP2-specific Supplementary Table 2 record was not recovered. Retaining the supported location does not claim that the parotid measurement verifies this separate urinary experiment, establish a physiological exosome-specific function, or imply contamination.

Summary: Fibulin-4 associates with elastic fibers.

Reason: The PMID:27339457 original target-specific localization experiment was unavailable, but the existing location is corroborated by ortholog ultrastructure and human matrix/elastic-fiber organization studies. Retain the curator’s elastic-fiber assignment without claiming that each corroborating study resolves the same microscopy endpoint.

Summary: The elastic-fiber location is consistent with the conserved matrix organizer.

Reason: The literal mouse Efemp2/Q9WVJ9 source and the electronic rule support elastic-fiber association. Comparative localization places fibulin-4 on microfibrils surrounding elastin cores; knockout and matrix-binding studies corroborate the conserved context. The inference does not make EFEMP2 elastin itself or a catalytic cross-linking enzyme.

Summary: The elastic-fiber location is consistent with the conserved matrix organizer.

Reason: The literal mouse Efemp2/Q9WVJ9 source and the electronic rule support elastic-fiber association. Comparative localization places fibulin-4 on microfibrils surrounding elastin cores; knockout and matrix-binding studies corroborate the conserved context. The inference does not make EFEMP2 elastin itself or a catalytic cross-linking enzyme.

Summary: The review’s elastic-fiber assignment agrees with primary evidence.

Reason: PMID:23962539 summarizes fibulin-4 in elastic-fiber assembly and LOX/tropoelastin organization. This is secondary evidence, supported independently by the operative human binding studies and ortholog assembly experiments; it is not treated as an additional primary assay.

Summary: Vascular smooth-muscle development is a conserved tissue-level consequence.

Reason: The mouse donor experiments show altered smooth-muscle maturation/differentiation markers and elastic-lamina architecture after Efemp2 loss. Retain the developmental association as non-core, with the extracellular environment and angiotensin signaling distinguished from a demonstrated autonomous differentiation activity of EFEMP2.

Summary: Vascular smooth-muscle development is a conserved tissue-level consequence.

Reason: The mouse donor experiments show altered smooth-muscle maturation/differentiation markers and elastic-lamina architecture after Efemp2 loss. Retain the developmental association as non-core, with the extracellular environment and angiotensin signaling distinguished from a demonstrated autonomous differentiation activity of EFEMP2.

Summary: The specific non-collagenous basement-membrane location remains unresolved.

Reason: The operative PMID:25885889 Methods define positive immunostaining as cytoplasmic granules, and Results report strong cancer-cell cytoplasmic staining plus perivascular signal in ovarian stroma. These are positive EFEMP2 observations, but they do not resolve the non-collagenous compartment of basement-membrane extracellular matrix named by GO:0140144. The separately inspected ARBA rule also did not establish target-specific localization. The gene-level claim therefore remains unresolved across the two sources; the limitation of this staining assay is not evidence that EFEMP2 cannot occur in that compartment.

Summary: The specific non-collagenous basement-membrane location remains unresolved.

Reason: ARBA00096893 outputs GO:0140144 from domain/taxonomic or FunFam combinations. The saved rule and target record were inspected, but target-specific localization support was not established; the current target feature snapshot does not itself reconstruct the rule application. Broad fibulin/ECM membership and the separately observed cytoplasmic/perivascular staining do not resolve the non-collagenous basement-membrane compartment. Preserve this inference as unresolved, consistently with the same location claim from PMID:25885889, without treating a rule-snapshot mismatch as biological refutation.

Summary: The MatrisomeDB citation does not yet expose the target’s interstitial-matrix evidence.

Reason: The database paper and its proteomics inclusion/search pipeline were inspected. The EFEMP2-specific record linking its measurements to the interstitial rather than another ECM compartment was not recovered. Retain the TAS assertion as unresolved without treating broad ECM detection as proof of this finer location.

Summary: The extracellular-vesicle measurement is source-limited.

Reason: PMID:24769233 is bibliographically authenticated, but the EFEMP2-specific vesicle preparation and target-table record were unavailable. The positive imported HDA assertion is preserved; general extracellular secretion cannot establish the assay’s vesicle association, and no contamination claim is made.

Summary: EFEMP2 regulates collagen fibril organization in defined tissue contexts.

Reason: The mouse donor evidence includes E57K knock-in and knockout studies showing altered collagen fibril dimensions, organization and cross-linking. PMID:26220971 directly documents aortic collagen disorganization and reduced mature cross-links, with largely retained LOX processing and no stimulation of PCOLCE/BMP1 cleavage by fibulin-4 in vitro. Retain the regulatory tissue role without converting proposed scaffolding mechanisms into EFEMP2 catalysis.

Summary: EFEMP2 regulates collagen fibril organization in defined tissue contexts.

Reason: The mouse donor evidence includes E57K knock-in and knockout studies showing altered collagen fibril dimensions, organization and cross-linking. PMID:26220971 directly documents aortic collagen disorganization and reduced mature cross-links, with largely retained LOX processing and no stimulation of PCOLCE/BMP1 cleavage by fibulin-4 in vitro. Retain the regulatory tissue role without converting proposed scaffolding mechanisms into EFEMP2 catalysis.

Summary: Fibulin-4 supports normal collagen fibril assembly in ortholog experiments.

Reason: The literal mouse donor has loss-of-function and E57K evidence for delayed or abnormal collagen fibril assembly in tendon, skin and aorta. The direction supports positive regulation, but tissue-specific diameter changes and incomplete mechanistic resolution preclude a universal direct catalytic interpretation. Preserve this contextual process separately from the principal elastic-fiber organization mechanism.

Summary: Fibulin-4 supports normal collagen fibril assembly in ortholog experiments.

Reason: The literal mouse donor has loss-of-function and E57K evidence for delayed or abnormal collagen fibril assembly in tendon, skin and aorta. The direction supports positive regulation, but tissue-specific diameter changes and incomplete mechanistic resolution preclude a universal direct catalytic interpretation. Preserve this contextual process separately from the principal elastic-fiber organization mechanism.

Summary: EFEMP2 restrains smooth-muscle hyperproliferation in the mouse aortic context.

Reason: The donor knockout studies measure increased proliferative labeling and ACE/angiotensin signaling. In PMID:23636094, angiotensin-pathway inhibition suppresses hyperproliferation/aneurysm phenotypes without repairing the elastic-fiber defect. Retain the conserved regulatory association as non-core and distinguish matrix-dependent signaling from a direct antiproliferative enzyme.

Summary: EFEMP2 restrains smooth-muscle hyperproliferation in the mouse aortic context.

Reason: The donor knockout studies measure increased proliferative labeling and ACE/angiotensin signaling. In PMID:23636094, angiotensin-pathway inhibition suppresses hyperproliferation/aneurysm phenotypes without repairing the elastic-fiber defect. Retain the conserved regulatory association as non-core and distinguish matrix-dependent signaling from a direct antiproliferative enzyme.

Summary: EFEMP2 supports aortic smooth-muscle differentiation in an extracellular-matrix context.

Reason: The mouse Efemp2 donor’s loss alters smooth-muscle differentiation markers and elastic-lamina contacts. The studies explicitly leave direct versus matrix-mediated effects incompletely separated. Retain the positive differentiation association without assigning transcriptional regulation to EFEMP2 itself.

Summary: EFEMP2 supports aortic smooth-muscle differentiation in an extracellular-matrix context.

Reason: The mouse Efemp2 donor’s loss alters smooth-muscle differentiation markers and elastic-lamina contacts. The studies explicitly leave direct versus matrix-mediated effects incompletely separated. Retain the positive differentiation association without assigning transcriptional regulation to EFEMP2 itself.

Summary: Multimeric fibulin-4 promotes smooth-muscle-cell adhesion under the tested conditions.

Reason: PMID:23782690 directly compares recombinant human fibulin-4 assemblies and their cell-adhesion/heparin-binding behavior. The positive smooth-muscle adhesion endpoint is retained as a contextual matrix effect; the responsible endogenous receptor and whether heparan sulfate is the physiological mediator remain unestablished.