DSC2 is a desmosomal cadherin with one integrated adhesive and plaque-anchoring role. Its calcium-binding ectodomain engages cadherins across adjacent cells, while its alternative cytoplasmic tails differ in plaque associations. The final core uses GO:0098641, a child of both cadherin binding and cell-cell adhesion mediator activity, at the desmosome. Calcium binding, the adhesive process and tail-dependent anchoring are components of this mechanism; the intestinal Rap1, traction and wound-repair observations do not establish a second catalytic activity.
The official DSC2 record is Q02487, with 901-residue 2A/DGII and 847-residue 2B/DGIII products. The short product has the specified alternative terminal sequence and lacks the final 54 residues of 2A. Existing product identifiers and sequence notes are preserved. Human DSC2a/b tail constructs associate with plakophilin-2 in rat-heart lysate assays; the longer tail additionally supports plakoglobin and desmoplakin-fragment association in the reported experiments. Neither those lysate assays nor human constructs expressed in monkey/rodent hosts are described as purified human binary measurements. See PMID:21062920(https://doi.org/10.1093/cvr/cvq353) and the historical splicing study PMID:2037591(https://pubmed.ncbi.nlm.nih.gov/2037591/).
Historical DSC3 naming in the DSC2 record does not identify the current DSC3 gene, Q14574. Experimental DSC3 reagents are not relabeled from this alias alone. Disease truncations, proteolytic maturation products and experimental fragments are not new natural splice products.
The human ectodomain study PMID:27298358(https://doi.org/10.1073/pnas.1606272113) supports strong desmoglein–desmocollin adhesion. Its weak DSC2 self-association and EC1-containing chimera crystal contact do not establish an exclusively homophilic native junction. Conversely, positive self-binding in the single-molecule experiments PMID:24610950(https://pubmed.ncbi.nlm.nih.gov/24610950/) and PMID:29999492(https://pubmed.ncbi.nlm.nih.gov/29999492/) is not erased by a negative bulk bead assay. Protein preparation, calcium, force and membrane context can distinguish these experiments. The exact construct species in the inspected single-molecule Methods is not inferred merely from a human expression host.
The 2022 surface-protein network PMID:35922511(https://doi.org/10.1038/s41586-022-05028-x) contains DSC2–DSG2, but the 29,000 nM affinity in Supplementary Table 3 and the Figure 2 source-data network is explicitly a previously published result attributed to the 2016 study. It is not a new 2022 high-affinity validation. The library DSC2 sequence matches canonical residues 1–693; this shared ectodomain does not resolve the two different natural tails. IntAct EBI-6901957, also IM-20978-13, instead records the 2011 anti-tag co-IP association and its mutant/sufficient-region context. Reference ROGIDs authenticate the curated reference sequences, not every experimental insert. Reproducible checks and the public workbook/record locators are linked in the source-evidence sidecar.
The complete 2024 correction was inspected. It addresses a missing CLC4M figure label, a concentration-unit correction in Figure 2c and four missing single-cell references. It does not identify a change to the DSC2–DSG2 affinity record used here. The preserved original cache is unchanged.
Measured and modeled adhesive pools support desmosome refinement. Broader plasma-membrane observations from HPA and the 2013 cell-surface study remain nonexclusive contextual observations. HPA identifies supported plasma-membrane staining in HaCaT, SuSa and U-2 OS and additional ER staining in HaCaT/SuSa; these are metadata/caption observations, not newly interpreted pixels. A membrane prediction is refined to the complementary plasma-membrane and ER pools without asserting that every molecule is an assembled desmosome component.
The Reactome paths explicitly place Q02487 residues 136–901 in desmocollin/desmosome networks, including a cornified-envelope component. Membership is not evidence that DSC2 executes filaggrin cleavage, phosphorylation or envelope cross-linking. Intestinal adhesion and timely DSG2 recruitment are experimentally supported; the specific positive protein-localization refinement is GO:0150107. Wound healing remains a contextual process. The tension sensor in the 2021 study is DSG2, not DSC2.
Nine decisions remain UNDECIDED because the exact source assertion cannot be resolved: the transferred adherens-junction pool, two unrecovered exosome target tables, specific His–Purkinje binding/adhesion, heart-rate regulation, ventricular action-potential regulation and two p38 source/transfer claims. The genuine primary text that was recovered, including clinical and original keratinocyte experiments, is documented separately from inaccessible supplements. No negative annotation, wrong-gene accusation or absence claim follows from these limits. The urinary proteomic target was independently found in the author laboratory database, but its one-peptide entry does not establish a unique natural product.
PTHR24025 target PTN002513172 was joined to the exact 901-residue sequence and its ancestral path. Literal PTN001931974 calcium-binding/cell-cell-adhesion and PTN002771418 desmosome assertions were inspected. Target self-evidence is legitimate; no donor-count heuristic was used. The current ARBA cornified-envelope rule matches the target. The current general-junction rule fails a current-version replay, but that does not prove the historical annotation invalid; independent target experiments support the refined junction location.
The 58 imported source objects, including identifiers, evidence codes and flags, are preserved. Actions are 14 ACCEPT, 26 MODIFY, 9 KEEP_AS_NON_CORE and 9 UNDECIDED; there are no NEW annotations. Supported generic binding would be kept non-core under the scoped project instruction; here the identified measured pairs permit informative classes. The original cache flags are retained and distinguish cached full text from separately recovered original papers. Genuine Falcon and fallback provider attempts failed with access/quota responses; no provider report was authored to replace them. Manual original, database and ontology work supplies the evidence described here.
New quotations appear once each in the YAML: 10 words from PMID:21220045, 9 from PMID:24610950, 8 from PMID:31967937, 13 from PMID:33596089 and the ROOT-allocated 8 from PMID:27298358. No publication quotations are duplicated in these notes or the source guide. Each source remains below the 25-word aggregate limit.
The existing main-branch publication caches for PMID:29999492 and PMID:15775979 are retained byte-for-byte instead of the later fetch variants. The entire preserved PMID:15775979 HTML extract was inspected: its full-text flag is true, but it contains only the abstract, Introduction and selected Results/Discussion sections, omitting the operative Figure 2/3 Results and Methods. The corresponding reference flag follows that cache metadata; the unresolved adherens-junction decision still reflects incomplete target-specific evidence. No annotation source, action, replacement, product, core or quotation changes in this reconciliation.
The PKP2 interaction retains the broad cytoskeletal-protein binding refinement. This class includes non-filament-forming structural components; it is not synonymous with binding an assembled intermediate filament. The measured partner is rat plakophilin-2, while the DSC2 tails are human. Partner annotations and documented plaque/intermediate-filament interactions support the component class, with ortholog-based localization provenance kept distinct from the target lysate-association assay. The public source guide links the literal partner records and ontology definition.
The broad membrane refinement now explicitly includes desmosome alongside the independently broader plasma-membrane and additional ER observations. HPA labels the ICC/IF record supported and names ER in HaCaT and SuSa. The observation does not establish permanent ER residence or an ER-specific function; trafficking or antibody-background explanations are hypotheses, not measurements that negate the localization. The selected merged image was inspected, but no new marker-colocalization or antibody-specificity conclusion is drawn from it.
Both DSG2-binding rows now use the adhesion-specific term with their assays distinguished. The 2011 co-IP establishes association; independent human ectodomain experiments establish adhesive function for the same pair. The existing single core already describes calcium-binding ectodomains and natural-tail-dependent plaque coupling as one integrated adhesive mechanism; no extra core or annotation is added merely to mirror accepted binding rows.
Current reference notes and source navigation describe the actual reading scope without internal agent labels. Earlier note entries remain an append-only historical record; the biological conclusions and the unchanged experimental limitations above are the current account.
Full-text access is distinguished from cache format: PMID:27298358 has a separately recovered complete original despite its preserved abstract-only cache, so its reference access flag is false. PMID:2037591 remains access-limited, while PMID:21062920 has available full text. No publication cache bytes were changed.
The earlier localization account incorrectly generalized HPA's overall supported gene score to the additional ER observation. The same saved public page and the downloadable location record distinguish supported plasma membrane from uncertain ER. HaCaT and SuSa retain the reported ER staining, but it is no longer proposed as an established replacement compartment. Desmosome and the independently supported broader plasma-membrane pool remain. This correction follows the per-location source assessment, not absence from the GOA import or an assumed transit mechanism. The previous notes are preserved as historical provenance.
The PKP2 partner classification remains supported by explicitly traced component annotations: rat ISO and human ISS GO:0005882 assertions trace to mouse Q9CQ73's IDA assertion from PMID:16917092. The donor-specific experiment was not independently recovered; the transfers are not presented as direct rat or human microscopy. A protein can contribute to cytoskeletal structure without itself forming the filament. The calcium-dependent adhesion and tail-mediated plaque-coupling mechanism remains integrated in the existing core.
Round 4 of projects/INVOLVED_IN_PRECOMPOSITION.md (2026-11-07): the core-function molecular function now uses the genus GO:0045296 cadherin binding instead of the precomposed GO:0098641, and the MODIFY rows target the genus; the adhesion context is carried by GO:0098609 cell-cell adhesion in the core functions. The earlier statement above that the final core uses GO:0098641 is superseded.