The genuine normal fetch seeded 67 annotations and 33 references for human CTCF (HGNC:13723; UniProt P49711). All original source fields and the two official product identities are retained for reconciliation. The successful fetch and publication-cache run are recorded separately from the genuine Falcon timeout and perplexity-lite quota failure; no synthetic provider output is used.
Existing main source caches are selected byte-for-byte. Fresh normal variants and the pending PTHR16515 PAINT proposal remain separate evidence snapshots. Cache equality does not establish phylogenetic placement.
The official displayed protein has 727 residues. Applying VSP_045350 reproduces the official 399-residue second product by deletion of residues 1–328. Its sequence therefore lacks the first two zinc fingers and part of the third; equivalent DNA recognition, recruitment or regulatory activity is not inferred. The two independent current FASTAs match the reconstructed sequences exactly.
Source-specific annotation, primary-paper and propagation review is ongoing. Supported generic protein-binding annotations will follow the project's standing KEEP_AS_NON_CORE policy when no evidence-backed refinement is established.
The complete review retains the 67 genuine source assertions and both official product identities. Two compact core activities are supported: sequence-specific chromatin-loop anchoring and RNAPII DNA-binding transcription-factor activity. CTCF supplies a structural anchor and locus-specific regulatory contribution; cohesin supplies the extrusion machinery. Human MYC repression, APP activation and MHC-II/apolipoprotein promoter experiments establish context-dependent transcriptional effects, rather than an activator or repressor label inferred from zinc fingers alone. The older human experiments and the separate chicken-fragment/quail-host experiments remain distinct. PMID:8649389(https://pubmed.ncbi.nlm.nih.gov/8649389/), PMID:9407128(https://pubmed.ncbi.nlm.nih.gov/9407128/), PMID:18347100(https://pmc.ncbi.nlm.nih.gov/articles/PMC2292219/), PMID:19322193(https://doi.org/10.1038/emboj.2009.81).
The direct human fragment assays demonstrate motif-specific recognition and methylation-sensitive binding to mouse H19/Igf2 imprinting-control DNA and chicken insulator DNA. They do not establish all human germline imprinting steps. The independent neonatal mouse liver experiments connect inherited allele-specific CTCF occupancy, looping and enhancer restriction. The current 399-residue product is not any of these engineered finger fragments: it deletes the first 328 residues of the 727-residue sequence, including two complete fingers and part of a third. Sequence reconstruction is exact, whereas equivalent function remains unresolved. PMID:17827499(https://pubmed.ncbi.nlm.nih.gov/17827499/), PMID:16815976(https://pmc.ncbi.nlm.nih.gov/articles/PMC1484419/).
Human contact assays and CTCF-site inversions support architectural participation. The CCAT1-L paper's correction changes a supplementary antisense sequence and does not withdraw its results. Its contact assays are distinct from the proposed RNA allostery explanation. The original PMID:18654629 annotation source studies nucleosome positioning and cites looping as prior work; its looping citation is marked MISCITED while the biological looping assertion is retained with direct independent support. PMID:24662484(https://doi.org/10.1038/cr.2014.35), corrigendum, PMID:18654629(https://pmc.ncbi.nlm.nih.gov/articles/PMC2453330/), PMID:26276636(https://pmc.ncbi.nlm.nih.gov/articles/PMC4642453/).
The mouse Jpx sources support conserved RNA binding and promoter-specific chromatin recognition, together with RNA-sensitive anchor occupancy. The early study directly measures RNA/protein binding and Xist promoter recognition; its proposed P1 loop is not itself measured. The later study measures motif-dependent P2 contacts and genome-wide contact changes. Its competition assay mixes RNA and DNA before CTCF, so it does not directly time eviction of protein already bound to DNA. RNA does not universally stabilize CTCF loops, and neither source establishes activity of the current short human product. PMID:23791181(https://pmc.ncbi.nlm.nih.gov/articles/PMC3777401/), PMID:34856126(https://pubmed.ncbi.nlm.nih.gov/34856126/).
The BAG-1 source explicitly reports that “CTCF inhibited BAG-1 promoter activity”. Its BORIS-focused title is not a wrong-gene finding. The selected machine cache says full text, but actually contains Introduction/Discussion without Methods/Results; direct recovery returned failures or a challenge page. Positive promoter binding and repression are retained with curator deference, and complete assay verification remains UNVERIFIED. The nucleolar study and historical mapping study remain abstract-only at the reviewer access boundary, with their explicit positive results preserved. PMID:18413740(https://pmc.ncbi.nlm.nih.gov/articles/PMC2733164/), PMID:16595548(https://doi.org/10.1242/jcs.02890), PMID:9591631(https://pubmed.ncbi.nlm.nih.gov/9591631/).
Human centromeric occupancy and CENP-E recruitment are positive results at the tested conditions. The 2015 binding experiments compare bacterial MBP-CENP-E fragments with wheat-germ translated CTCF; they are not a homogeneous two-protein purification. The source reports that “CENP-E binding decreased” after CTCF depletion, with cell-cycle and region controls. The 2026 human HCT116-AID study instead finds unchanged CENP-E recruitment after three days and measures centromere spring/plate maintenance. This difference is preserved, not described as a retraction. Regulation of centromeric sister cohesion remains UNDECIDED because the original statement was prospective and the newer partial-RAD21 comparator preserves sister cohesion. PMID:26321640(https://pmc.ncbi.nlm.nih.gov/articles/PMC4633288/), PMID:18550811(https://pubmed.ncbi.nlm.nih.gov/18550811/), PMID:41486985(https://doi.org/10.1242/jcs.264181).
Endometrial-cell assays support contextual growth regulation. Wild-type expression can suppress MTT/colony endpoints, whereas partial KLE depletion does not significantly change proliferation; the authors state that “there is not a simple linear relationship”. This positive but context-dependent process is NON_CORE. Localization is also nonexclusive: human nucleus/nucleoplasm, mitotic chromosomes and specific nucleolar conditions can coexist. HPA independently reports supported nucleoplasmic staining in A-431, U-251MG and U2OS, without product-specific resolution. PMID:28319062(https://pubmed.ncbi.nlm.nih.gov/28319062/), PMID:16107875(https://doi.org/10.1038/sj.emboj.7600793).
Five generic interactions receive specific partner-class refinements: TFAP4 bHLH factor, OCT4 RNAPII factor, KDM5B enzyme, CENP-E kinesin and TRAF2 ubiquitin ligase. These name the bound partner and do not assign that partner's catalysis, motor function or signaling role to CTCF. The CTCF–CGGBP1 association is positive, but its particular non-DNA-binding coregulator classification remains UNDECIDED; a DNA-binding protein can have another regulatory role, yet a DNase-resistant complex alone does not establish it. PMID:19505873(https://pubmed.ncbi.nlm.nih.gov/19505873/), PMID:19536159(https://pmc.ncbi.nlm.nih.gov/articles/PMC3057664/), PMID:24937458(https://pmc.ncbi.nlm.nih.gov/articles/PMC4079039/), PMID:26321640(https://pmc.ncbi.nlm.nih.gov/articles/PMC4633288/), PMID:32296183(https://pmc.ncbi.nlm.nih.gov/articles/PMC7169983/), PMID:31547883(https://pubmed.ncbi.nlm.nih.gov/31547883/).
Eight supported generic-binding assertions remain NON_CORE under the standing CLINGEN_MENDELIAN curation instructions. This explicit campaign instruction overrides generic removal of positive protein-binding annotations; there is no action quota. RPL31 binding does not establish binding to a complete ribonucleoprotein complex. The 170-residue ZMYM6 isoform-4 backbone and its sufficient 1–156 assay region remain distinct. APP records contain a mutation feature at residues 670–671 rather than a deletion interval. A backbone ROGID authenticates the reference sequence but cannot erase tags, mutations or fragments. Co-isolation and two-hybrid observations do not automatically establish native stoichiometry or physiological activity.
The OpenCell and 2025 RPL31 target tables were not recovered; these assertions retain curator deference and independent positive evidence for the same pair, with paper-specific confidence UNVERIFIED. Other screen references distinguish literal curated records from direct original-table inspection. Empty current joins are not evidence of biological absence. Actual read scopes, including external main-text recovery and failed supplements, are recorded in the portable evidence index.
All 67 raw GOA records reconcile exactly with the seeded assertions, including original qualifiers and omitted fields. The actual current PANTHER v19 resolver gives PTHR24404; its 952-node tree and aligned target sequence support the three on-path PTN002812688 IBDs. The protected PTHR16515 entries contain CTCF, whereas the separately fetched current PTHR16515 tree lacks it. These are distinct objects. Selected main family files and the unmerged PAINT proposal remain unchanged; byte equality does not prove ancestry, and target self-evidence is legitimate descendant evidence.
Complete donor queries preserve 943 current records, 60 joins and 143 literal occurrences across 19 queries. Mouse UniProt Q61164 and ENSMUSP00000005841 are two identifiers for one product. All 1,365 branches of two current ARBA rules were evaluated against official predicates and taxonomy with no unknown condition type: the nuclear rule matches three branches and the epigenetic rule one. Actual interaction reconstruction retains 262 full 42-field records, 26 used records and 52 exact backbone ROGID checks. Two Reactome paths reach the 1–727 CTCF entity in nucleoplasm through free input and complex output. No target GO-CAM activity was found in the selected main index; this is a scoped index result, not a claim that no model can exist.
The notes and review use short exact primary anchors. The public sidecar stores quote hashes and review locators rather than repeating their text. Source caches and genuine variants are preserved, and reference access flags describe actual reviewer access rather than overwriting machine cache metadata. The six-paper core consultation, 14-row binding consultation and two ROOT RNA consultations are attributed separately from owner reads. They do not become claims that one reviewer personally read all papers or every figure. The full gene judgments remain the owner's responsibility.
The published evidence index now summarizes source reconciliation, product checks, primary read/access limits, and propagation, interaction and model conclusions with repository paths and public record URLs. The earlier full audit remains unchanged in the frozen research record; its raw external responses, tree nodes, interaction records and branch evaluations are not republished in the gene folder. The curated review is the sole copy of annotation decisions and core functions. Snapshot counts in the summary are reported review results, not a promise that evolving public databases reproduce the same historical data.
The centromeric-cohesion annotation remains UNDECIDED: the 2008 source leaves the function prospective, while the 2026 centromere spring measurements and partial-RAD21 comparator do not resolve the specific cohesion-regulation role. Positive centromere mechanics and mitotic effects are retained. The cell-growth annotation remains non-core: matched-expression wild-type/mutant assays in Ishikawa and partial depletion in KLE are different conditions, so the latter is not a universal falsification of the former. No action was changed.
The description now states a contribution to mitotic centromere organization without foregrounding a universal CENP-E recruitment effect. The original 2015 recruitment evidence and the distinct 2026 result remain in their source-specific assessment. The broad chromosome refinement is explicitly identified as converging on the existing IDA centromere term from the same paper, with both original source tuples retained. A question surfaces the PTHR16515/PTHR24404 resource-version distinction; the authenticated target path and protected prior records remain separate. A descriptive histone-demethylase-binding ontology proposal complements the existing KDM5B enzyme-binding refinement without adding an annotation or implying catalysis by CTCF.
All 67 original source objects, ordinary qualifiers, 40 reference identities, both products, both core units and existing literal quotations are retained. The earlier notes are an exact prefix. Repeated opening summaries were removed from a bounded set of reasons without removing their scientific or access qualifications. No new primary retrieval, source-cache change or provider attempt was needed.