The normal seed identifies human ATF6 (UniProt P18850; 670 amino acids), with 88 existing annotations and no seeded alternative-product records. The annotation list is still unreviewed. Cleavage of the membrane precursor into an active N-terminal transcription factor is protein processing; it must not be represented as an invented splice isoform.
The normal falcon research request with perplexity-lite fallback ended unsuccessfully and created no research file. These are manual notes. The normal Seed22 publication output is being checked and imported separately; existing publication caches must remain unchanged. No duplicate gene-wide fetch was started while that import was pending.
ATF6 is an ER membrane bZIP transcription factor. Its precursor and released transcription factor have different locations. ER stress promotes trafficking to the Golgi, proteolytic release of the cytosolic N-terminal region, and nuclear transcriptional activation. The source UniProt record cites the human studies that distinguish these events:
The official GO:0036500 record defines the ATF6 branch of the unfolded protein response as signaling from ATF6 activation through regulation of a downstream process such as transcription. This is consistent with the mechanistic frame above. The official GO:0001228 record identifies RNA polymerase II-specific DNA-binding transcription activator activity. These are candidate core terms, subject to full evidence review and validation.
Several distinctions need explicit assessment before choosing actions:
No new GO assertion, final action, core-function entry or full-text verification is claimed at this stage.
All 26 PMID abstracts associated with the seeded review have now been read, using the existing caches and the exact staged normal-fetch output. This does not imply that every paper's full text or every reported interaction has been inspected. The existing cache for PMID:23661758 supplies its title, Networks of bZIP protein-protein interactions diversified over a billion years of evolution; the seed's missing-title placeholder can be corrected in the authored review without editing that cache.
PMID:16236796(https://pubmed.ncbi.nlm.nih.gov/16236796/): the actual Methods name FLAG-ATF6 and its 1–373 construct. The Figure 10 Results explicitly describe transfected ATF6 in the ER before treatment, in the Golgi after 30 minutes of DTT, and in the nucleus at 120 minutes. CREB4 remains distinct from ATF6 in that comparison. Thus the ATF6 localization rows are supported by experiments on ATF6 within this CREB4-focused paper, rather than merely by its background discussion. Construct/assay Methods and the relevant Results were read; no image-level inspection is claimed.
PMID:19543265(https://pubmed.ncbi.nlm.nih.gov/19543265/): the Results and Methods distinguish CRTC2-associated-protein discovery from endogenous and tagged CRTC2 in human HEK293T cells, interaction/co-IP and promoter experiments in primary mouse hepatocytes, and mouse liver experiments. The N-terminal ATF6 region associates with CRTC2; the ER-luminal fragment does not. CRTC2 recruitment to Xbp1 regulatory DNA depends on ATF6. The mouse metabolic outcome should not become a new human gluconeogenesis annotation. Detailed Results and Methods were read; supplemental raw interaction tables were not.
PMID:26029869(https://pubmed.ncbi.nlm.nih.gov/26029869/): patient-derived fibroblasts carrying homozygous Arg324Cys show reduced induction of ATF6 target genes. Controlled HEK293 experiments use a DHFR-ATF6(1–373) fusion and the corresponding mutant. This supports impaired transcriptional activation as the disease mechanism. Human foveal hypoplasia and cone dysfunction must be distinguished from the mouse phenotype: young knockout mice retain retinal function, whereas older animals develop rod and cone impairment. Genetic/clinical Results, cellular functional Results, and relevant Methods were read. These findings do not make ATF6 a phototransduction enzyme.
PMID:22682248(https://pmc.ncbi.nlm.nih.gov/articles/3372931/): the existing cache reports available full text but contains Introduction and Discussion without Results or Methods. The indexed official PMC page supplies Figure 5, binding Results and Methods. The purified luminal fragments depicted in Figure 5B/C are mouse ATF6; human ATF6 GAL4 fusion constructs were also tested with Thbs4. The text names HEK293 cells for the latter assay, whereas its caption names COS cells. Do not silently resolve that discrepancy or describe all assays as purified human binding. The cardiac co-IP is from mouse tissue. The cached availability flag remains unchanged; quote only text actually present in the cache when using a cached-reference supporting snippet.
The apparent ATF6B repression in PMID:14973138 and its absence in the knockout comparison in PMID:17765680 come from different experimental contexts. Neither abstract supports a universal assertion that ATF6B always inhibits ATF6.
A search for ATF6/P18850 in the local GO-CAM index found no matching entry. No inference about all external GO-CAM content follows from that local result.
All 88 original annotations have been reviewed: 64 ACCEPT, 15 KEEP_AS_NON_CORE, seven UNDECIDED and two MODIFY. The source objects, qualifiers, partner accessions and evidence codes are unchanged; there are no NEW annotations or invented alternative-product records. The single core is nuclear DNA-binding transcription activation in the ATF6-mediated unfolded protein response. ER/Golgi locations describe the precursor activation route.
Protein folding is refined to the ATF6 signaling response, and direct ERAD participation is refined to positive regulation of ERAD pathway. The official definition/parents were read: regulation of degradation is distinct from executing the proteolysis or retrotranslocation steps. The existing positive regulation of ATF6-response annotation is retained with the concrete human ERSE reporter evidence in PMID:9837962; it does not assert feedback regulation or self-cleavage.
Independent annotation/core consultation covered every source row, all 26 cached abstracts, six pathway summaries and selected primary Results/Methods. Current HPA subcellular tables and HPA071169 ICC assessment support the contextual nucleolar/fibrillar-center rows; no image pixels or processing-state resolution was claimed. PAINT target self-inclusion is legitimate experimental grounding; the tree/MSA was not reconstructed.
The separate primary-source consultation of PMID:16469704(https://www.sciencedirect.com/science/article/pii/S0092867406000043) inspected original indexed Results and Figures 6-7 legends. Reciprocal co-IP uses endogenous human ATF6 in 293T with FLAG-CREBH; p50 reporter activation and human CRP probe capture supply further support. The exact p50 construct species/residue boundaries and purified dimer stoichiometry remain unresolved. The original array Methods for PMID:20102225 include human-human pairs, and PMID:23661758 uses solution FRET across species. Their exact ATF6-XBP1/ATF6B supplemental pair entries were inaccessible. P17861 is XBP1, not ATF1. The unchanged canonical caches retain their original availability flags, even where external or quarantined original text was richer.
Four contextual generic associations are retained as non-core. Unresolved target-specific interactions, apoptosis and autophagy evidence remain UNDECIDED. Established homodimerization is retained for two sources with explicit curator deference; the TAP/MS self-association row remains source-specifically unresolved because the analysis removes bait self-identifications. No assay is rejected merely because its title foregrounds a different gene or because a generic term is uninformative.
The normal publication import is complete, with existing caches preserved. The unsuccessful provider-research attempt produced no report; manual source work supplied the review. Earlier pending statements above document the progression of the evidence assessment. Exact cached excerpts support seven reference findings, and the authored title placeholder for PMID:23661758 is corrected from its existing record.
The folding annotation is refined to positive regulation of protein folding (GO:1903334), preserving the effect of chaperone induction without assigning folding chemistry to ATF6. The two annotations to regulation of the ATF6 pathway are instead refined to ATF6-mediated unfolded protein response (GO:0036500). The human ERSE-output result in PMID:9837962 establishes execution of this pathway; it does not, on its own, separate an additional regulatory mechanism. This is a term-precision judgment, not a claim that pathway components cannot have feedback roles. The original experimental/electronic evidence and sources are preserved.
The five disputed interaction assertions are retained as contextual NON_CORE associations with explicit curator deference. Biological corroboration does not recreate inaccessible source-specific measurements. The ATF6–XBP1 mechanism in PMID:17765680 is from mouse experiments. Exact human zipper-array/FRET pair cells and affinities in PMID:20102225/23661758 remain unread. The processed-product UniProt record separately corroborates homodimerization and ATF6B heterodimerization. The PMID:25609649 Methods temporarily remove bait self-identifications before SAINT and add them back afterward; mere bait recovery cannot establish homomerization. This corrects the incomplete account in the earlier notes without claiming that its exact curated self-association evidence has been reconstructed.
For ATF6–HSPA5, the independent original PMID:15657421, PMC543992 indexed Results, Figure 1 legend and selected experimental passages were read. They describe ATF6/BiP association in HeLa cells, human BiP substrate-pocket mutagenesis and separate experiments using hamster BiP or mouse cells. The ATF6 construct species was not independently re-established in this bounded read. No single species assignment is imposed on every assay. The source-specific PMID:27238082 experiment and erratum remain unread. This independent reading supports the established association only; it does not certify that later paper or add a new annotation. No cache was fetched or rewritten for this supplementary reading.
The original PMID:16469704 Cell paper supplies evidence absent from its cached abstract. Indexed Results and Figure 7C legend identify ATF6p50 association with a wild-type human CRP probe in COS1 nuclear extracts and loss with a mutant probe; selected DNA–protein assay Methods were also read. Results for Figures 6D–F describe reporter activation and cooperation with CREBH. The earlier independent source consultation additionally read Figure 6C reciprocal co-IP of endogenous human ATF6 in 293T cells. These are external text readings, not literal quotations from the normal cache. Complete supplementary construct Methods and image pixels remain unread; the p50 construct species and purified dimer stoichiometry remain unresolved. The cache and its full-text-unavailable flags are preserved.
The current HPA subcellular table still reports supported nucleoli/fibrillar-center locations with HPA071169 in the named cell contexts. Retaining these as NON_CORE does not assign a nucleolar function or resolve cleavage state. One antibody and an unexpected location alone do not contradict the observation. Broad existing transcription, membrane and signaling terms can represent aspects of the same core mechanism; there is no quota requiring their demotion merely because more specific terms also exist.
The single core remains the released nuclear transcriptional activator. ER/Golgi trafficking and proteolysis define how that activity is activated, rather than a second independently established molecular activity. The precursor is the substrate of processing proteases. No additional core, splice isoform, NEW process or NEW binding assertion is proposed. Earlier dated research and pending statements remain historical; a new append-only history event will record completion of this followup.