PMID:17344063(https://pmc.ncbi.nlm.nih.gov/articles/PMC1937038/) reports human Cx47 dye transfer and paired-cell currents. Selected original sections were read. Tissue evidence is rhesus; mutant ER retention is separate.
PMID:19056803(https://doi.org/10.1093/brain/awn328) separates surface plaques from channel function: I33M has defective homotypic coupling and altered heterotypic gating. The published Methods cite AF014643, whose sequence differs from AY285161 at historical residues 80 and 249; the actual plasmid sequence is unavailable. Historical I33 maps to current I36. Figure 7 includes a reused WT curve; contemporary WT controls are also reported. Opposing-cell pairing is not heteromeric coassembly. Selected original text and Figure 7 were inspected.
PMID:11160382(https://pmc.ncbi.nlm.nih.gov/articles/PMC3671913/) tests mouse Cx47 in HeLa cells and Xenopus oocytes, untagged or C-terminally EGFP-tagged. Historical 437-residue construct equivalence to current Q8BQU6 remains unresolved. Selected Methods/Results/captions were read. Its neuronal-localization interpretation is not used to override later glial evidence.
PMID:20578039(https://doi.org/10.1002/glia.21032) supplies mouse myelin evidence: antibodies label WT but not knockout tissue, and immunoprecipitation distinguishes specific Cx47 from nonspecific crude Western bands. Cuprizone-associated redistribution does not demonstrate that Cx47 synthesizes myelin. Selected original IHC, immunoprecipitation and discussion passages were read; no figure pixels or supplement were inspected here.
The actual mouse Q8BQU6 experimental records include GO:0005243 and GO:0005921 IDA from PMID:11160382 and GO:0043209 IDA from PMID:20578039. The latter retains a mouse anatomical extension, without copying it into the human row. Official Ensembl maps ENSG00000198835 to ENSMUSG00000043448 as one-to-one orthologs; ENSMUSP00000104421 is the incoming mouse protein. Human-return inferred mouse rows are not used as experimental donor evidence.
The HPA GJC2 record reports HPA069619 plasma membrane and cytosol staining in EFO-21 and SH-SY5Y, and cytosolic staining in U2OS. Overall reliability is supported, while the antibody subassay carries uncertainty. These cell-line observations are not direct human myelin imaging. The retained XML and image descriptions were read, without inspecting image pixels.
The actual PTHR11984 tree has 948 nodes and 497 leaves. Human Q5T442 is AN478/PTN002485539 below AN0/PTN008334405. The three IBDs for GO:0005243, GO:0007267 and GO:0005922 are at AN0, with their original donor groups retained. The target alignment recovers all 439 residues; the four transmembrane intervals are 26–46, 79–99, 217–237 and 266–286. No relevant ancestral negative record was found on this target path. Neither donor count nor experimental target self-inclusion is treated as a failure.
GO:0007267 and GO:0010644 are direct is_a children of GO:0007154. The broad existing communication row is therefore TOO_GENERAL. GO:0005243 describes a pore through each adjacent membrane, supporting direct transport without assuming unapposed hemichannel function. GO:0005922 describes the connexin complex; this does not establish a particular native heteromeric composition.
PMID:15192806(https://pmc.ncbi.nlm.nih.gov/articles/PMC1216059/) supports disease linkage and human tissue transcripts, rather than direct channel or myelin-protein localization assays. PMID:20537300(https://pmc.ncbi.nlm.nih.gov/articles/PMC3032064/) supports a lymphedema association; its channel and lymph-flow mechanisms remain hypotheses. Selected original sections were read for both. PMID:15489334 and PMID:16710414 were assessed from authentic abstracts as sequence-resource background. The Reactome assembly pathway is background; its Cx43-specific trafficking text is not transferred to GJC2.
The current canonical record is Q5T442, 439 residues. Historical constructs and uncertain initiation do not establish a second natural isoform. The sequence and its hash are retained in the source archive; a canonical-only product slot is not added to the review.
The local comparison parses the two original GenBank CDS translations and the cached UniProt sequence. It reproduces the human accession comparisons and residue-number mapping above. The mouse 437-residue construct remains a reported experimental length, with equivalence unresolved; no mouse sequence reconstruction is claimed.