This superseding audit preserves the published 98 machine assertions in their original order. The prior review contained one additional authored membrane-grommet proposal; that whole object is retained in the source audit as withdrawn provenance. The genuine fresh normal seed has 104 assertions: 92 map one-to-one to the retained historical rows, five historical aggregates split into ten fully scoped assertions, and two genuine new source tuples are appended. All 12 current additions have ordinary reviewed actions, giving 110 machine assertions. The historical protein-transport keyword tuple remains present although absent from the current projection. The five historical aggregates are explicitly retained for provenance and are not counted as extra current GOA assertions. No donor, qualifier, processed partner or isoform is retrofitted onto an old unscoped object.
The old product field was absent. Two authentic current UniProt descriptors are appended as descriptors only. Disease-associated truncations, activated 1–154 constructs and tagged proteins are not silently equated with current alternative product 2. All 41 original reference identities and selected publication/provider bytes remain unchanged. The GOA update and prior-byte archive are explicitly documented below. The original Falcon report is retained byte-exact, but its claims are not a substitute for primary evidence. New Falcon and fallback provider failures are recorded as failures, without a fabricated report.
The performed core work is membrane deformation in dynamic ESCRT-III assemblies. Human WT/C-terminal-tag CHMP2B cellular tubes and N-terminal-tag/L4D/F5D controls, plus purified activated human 1–154 after CHMP4B, support membrane bending. PMID:21926173 The defined membrane systems and activation/tagging conditions matter; the 2021 study supplies stiffness/countercondition context and a full-length 1–222 numbering discrepancy relative to current 213-aa Q9UQN3. Bending does not establish universal autonomous membrane breakage or restriction of a membrane fenestration. The old grommet MF therefore gives way to one supported bending MF proposal and one coherent core. No NEW process is proposed. The N-terminal membrane-interacting region is distinct from the C-terminal VPS4-binding MIM; CHMP2B does not hydrolyze ATP. The inherited SMYD2/K6 timing detail is omitted because a verified supporting primary experiment was not established.
Endosomal sorting, cytokinetic abscission and cooperative nuclear-envelope reassembly are retained. Endogenous human CHMP2B at the reforming envelope is positive evidence for a broader envelope replacement of the NPC-specific term; the current source part_of qualifier is preserved as an archive field, not reissued as a claim of stable complex incorporation. CHMP2B-alone depletion has little sealing effect in the tested HeLa system, whereas combined A/B depletion is stronger. Accessible Vietri kinetochore panels name CHMP4B, and a different panel names CHMP2A; incomplete main-body access leaves CHMP2B kinetochore assertions unresolved rather than proving a curation error. The Olmos author manuscript supplies the full Methods absent from the selected cache.
Human2026 CAL51 imaging and toxin recovery establish a later repair role after the initial annexin plug. They do not establish CHMP2B as the initial calcium-dependent seal. Full17984323 mutant-clearance experiments support contextual autophagy, but the inspected immuno-EM labels p62; they do not directly resolve CHMP2B on autophagosome, lysosome or amphisome membranes. Those exact locations remain unresolved. Endosomal fusion phenotypes likewise do not isolate intrinsic CHMP2B bilayer fusion; human mutant cells can retain intraluminal cargo sorting while losing endosome–lysosome delivery and normal Rab7 access. The source-specific boundaries are recorded without rejecting experiments from titles or abstract-only access.
Positive binding assertions retain their assay scope. CHMP3 and CHMP4B associations include exact curated records and original human SPR/co-IP experiments. RAP1 is TERF2IP/Q9NYB0, with target-specific BiFC and pull-down evidence; it is not RIF1. Both SARS processed partners, nsp1 and nsp9, have exact Y2H plus positive LUMIER records. Viral protein association does not imply virion binding or a native infection role. VPS4 binding is refined to ATPase binding with the actual C-terminal peptide/structural scope. No partner's catalytic activity is transferred to CHMP2B. E-cadherin BioID is proximity evidence; exact target binding remains unresolved. Urinary-exosome detection has a real four-peptide author-table row, while the separate prostatic-exosome target supplement remains unresolved.
The current PAINT source node is PTN000049603. Its historical placement/alignment was not reconstructed; current descendant records are not a replacement for that phylogenetic judgment. Short donor lists and target self-reference are not treated as failures. Exact ARBA outputs and the Snf7-family InterPro mapping are recorded separately from target experiments. Reactome's nine cytosol assertions are retained as own-source event/member provenance, corroborated by human target imaging, without assigning another component's reaction to CHMP2B.
Source-access corrections are explicit: several protected full-flag caches lack Results/Methods;23051622 remains the exact-main abstract while its genuine fresh full variant was read separately. A mistakenly addressed unrelated PMC XML and an incorrect publisher route were identity-checked and excluded; neither is biological evidence. The correct17984323 XML was then recovered and verified. No failed download or challenge page is counted as a supplementary table.
Strict normal validation exposed the difference between two authentic GOA snapshots. The historical file lacks two genuine new source tuples, whereas the new file lacks only the old keyword-derived protein-transport tuple at historical index13. The schema's retired metadata records that historical tuple without deleting or rewriting its term, evidence, reference, or KEEP_AS_NON_CORE review. The exact fresh normal GOA is the current file; the entire old GOA is preserved as CHMP2B-goa-prior-20261010.tsv. There is no fabricated combined GOA, manual source-row edit, validation-policy change or implication that retirement falsifies protein-transport biology. All 104 current assertions remain uniquely represented, while the historical-only tuple and five old aggregate assertions remain transparent provenance.
The exact Reactome participant sets for rows91/92/94/96 describe nuclear-envelope assemblies. They establish membership rather than a cytosolic pool within those events; independent target imaging supports the retained cytosol annotation. Newly split current rows describe their single partner or relation rather than repeating the historical aggregate rationale. The genuine prostatic-exosome and cadherin target supplements were not recovered. Their independent source consultation supports the unresolved decisions without treating missing access as negative detection. Author-manuscript and final-publication screen counts are not conflated.
This appendix preserves the complete incoming notes above and the incoming history without alteration. The current source is external head 013b90157d7517a1de22280f403e173f9886a177. It is not a ROOT-published version of the earlier unpublished follow-up; that earlier proposal is retained only as a scientific comparison. No biological source or author identity is inferred from this distinction.
The incoming changes are retained: membrane fission is non-core, the five superseded historical aggregates are explicitly retired, the endogenous CHMP2B panels are corrected, and the 2021 stiffness study is absent from affirmative core support. Alongside the dropped keyword tuple at index 13, there are six retired machine objects and 104 uniquely covered active current assertions, plus one authored MF proposal. Original source fields and the two genuine GOA snapshots remain intact; retirement records representation history rather than refuting biology.
Removed 30 authored-file self-quotations while retaining bare provenance links, and omitted the authored artifact's self-VERIFIED assessment. All primary anchors are exact source text, bounded across the current YAML and the full preserved notes. The incoming 19-word 21926173 notes quotation is retained; the two three-word YAML fragments keep its aggregate at 25 words. No new quotation is copied into this appendix or the authored artifact. Current prose and access locators receive spacing/attribution corrections; historical notes remain exact. No source cache, product, reference identity or existing history is changed.