EFNB1 encodes one displayed 346-residue ephrin-B1 precursor. The official record annotates the mature chain at residues 28–346 after signal-peptide removal, a C-terminal fragment at 218–346 and an intracellular domain at 260–346. The latter two boundaries are curator inferences; they are not separate splice isoforms or experimentally sequenced endogenous cleavage starts. The official predicted transmembrane span is 238–258. These product states are distinguished from engineered assay constructs.
The evidence supports an integrated extracellular Eph receptor-binding and signaling function at the plasma membrane. Ligand–receptor engagement can mediate adhesion, while cellular behavior depends on the receptor and presentation context. The historical HEK2 receptor in PMID:8798744 is EPHB3/P54753. Its positive soluble-ligand result is retained rather than replaced by a universal membrane-only rule based on another family study.
PMID:35922511 tests human EFNB1 residues 1–240 with fusion tags. Its original workbooks provide reciprocal raw binding cells and retained source-network entries for EPHA2, EPHB2 and EPHA4. The EPHA2 and EPHA4 SPR values are measurements made by that source; the EPHB2 affinity in the combined table is explicitly a published literature value. No native membrane affinity or intracellular-fragment activity is inferred. PMID:39433908 does not correct those EFNB1 entries.
PMID:22279592 supports human EFNB1–ERBB2 association in endogenous and tagged-construct cell experiments. Receptor-class binding is distinguished from receptor agonism, EFNB1 kinase activity or a purified direct interface. The ANKRD46 and mouse PICK1 interactions remain positive non-core assertions under the specific ClinGen project instruction. The task-specific policy retains supported generic binding when no evidence-backed narrower class is established; lack of native physiology does not by itself make an association false.
PMID:16930449 supports nuclear/cytoplasmic localization of engineered human intracellular fragments in COS cells, with separate mouse presenilin-deficient cell experiments. The membrane remnant and released ICD have different experimental behavior. PMID:19515908 provides a mouse cortical and tagged-domain ZHX2-associated signaling context, with unresolved construct species and antibody cross-reactivity. These observations do not establish a separate human transcription-factor molecular function for full-length EFNB1.
All six PAINT assertions have literal ancestral support above the exact human target sequence. Target self-grounding and a single well-characterized donor are not propagation failures. PMID:19915143 supports rat cortical synaptic contexts; the current donor annotation’s hippocampus extension remains distinct from the inspected cortical culture. The 30 Reactome location assertions resolve to EFNB1 itself at plasma membrane, without transferring the activities of other event participants.
The urinary-exosome table from PMID:19056867 records one EFNB1 peptide; its final value 2 identifies the reference. This is positive contextual detection, without a claim about a particular processed chain or an active exosomal signaling mechanism. HPA supplies positive supported plasma-membrane localization for the exact target, with per-location reliability inspected and no independent pixel interpretation.
The final decisions cover 66 imported source contexts: 45 ACCEPT, 9 MODIFY and 12 KEEP_AS_NON_CORE. There are no NEW annotations or newly quoted passages. Reference-only citations retain their primary identifiers, while the source guide and executed evidence replay expose the exact locators and data joins.
The automated deep-research attempts failed with a Falcon authentication error and a fallback quota error. No provider report was invented. Manual literature and source review proceeded with explicit access scopes. Cached full-text flags and existing cache bytes remain unchanged; independently recovered original text is identified separately from abstract-only caches. The public analysis contains original datasets, not republished article bodies.
The PMID:8798744 adhesion assertion is refined from GO:0007155 to its specific
descendant GO:0098609, cell-cell adhesion. Receptor- and ligand-expressing cells
formed aggregates, and soluble ligand inhibited the contact. EFNB1 provides the
receptor-binding ligand in this attachment, rather than merely being required
for an unrelated adhesive mechanism. This remains a source-specific TAS assertion;
the complete abstract is available but full Methods were not recovered. Adhesion
in this assay does not imply the same outcome in every receptor, cell or ligand
presentation context. The updated counts are 44 ACCEPT, 10 MODIFY and 12
KEEP_AS_NON_CORE across the same 66 source contexts, superseding the earlier
count paragraph without changing its historical text.
The single integrated core distills extracellular Eph receptor engagement. It
does not deny reverse signaling through the cytoplasmic tail. The processing and
localization findings in PMID:16930449 and the mouse cortical/tagged-domain context
in PMID:19515908 remain represented at their measured scope. They do not establish
a direct human DNA-binding or transcription-factor activity, and no separate
transcriptional core is assigned from intracellular-fragment localization alone.
The source-measured EFNB1–EPHA2 dissociation constant in PMID:35922511 is about
41 µM, a low-affinity interaction in that soluble recombinant ectodomain assay.
It supports the receptor partner class, without establishing native membrane
affinity, receptor activation or physiological importance. The same source
measured the EPHA4 value of 25 µM; its EPHB2 value of 0.66 nM was taken from
published literature. These values are not interchangeable native-context
measurements or independent repetitions of one assay.
The craniofrontonasal-syndrome sentence follows the official EFNB1 UniProt disease
record. It does not add a cellular-interference mechanism that was not established
by the disease passage inspected for this clarification.
Reference-only primary citations and explicit locators are retained; no new
quotations are introduced. The unchanged dataset archive preserves the original
inputs needed to reproduce the published selections and context counts offline.
It contains data rather than main article bodies or PDFs. Its existing hashes,
source receipts and executed replay remain the evidence for those calculations.