The frozen catalog's next unchecked entry after CHAMP1 is CHAT (HGNC:1912; reviewed UniProt P28329). The authenticated source baseline is 2f68dd87cb96a240d03d5debc6c7f131bb7b2808. CHAT/CHOACTase/P28329 had no existing gene folder, queue entry or overlap among all 202 open PRs with complete, same-head file inventories. There is no inherited notes prefix or old provider tree.
The genuine normal intake produced 117 source assertions, three alternative products and ten original reference identities. All are preserved exactly. The 102 interaction assertions explicitly name P28329-3; they are not silently assigned to the larger product. Existing source caches and PANTHER metadata were selected from authenticated main byte-for-byte. The fresh variants remain archived separately. The official PTHR22589 name is CARNITINE O-ACYLTRANSFERASE; the family metadata is retained as fetched rather than relabeled from the CHAT symbol.
The normal Falcon research attempt timed out after 90 seconds; the genuine perplexity-lite fallback failed with an account/quota response. No research report was fabricated. Manual primary reading, exact interaction and pathway records, and independent bounded annotation/core peers provide the evidence described below.
CHAT transfers an acetyl group from acetyl-CoA to choline. Human structures and purified-enzyme kinetics support one choline O-acetyltransferase core and the existing acetylcholine-biosynthesis/neuromuscular-transmission annotations. The broad acyltransferase mapping is refined to GO:0004102. This is intrinsic performed enzyme work, not an inference from disease severity alone. The human disease study includes endplate physiology and direct recombinant kinetic measurements PMID:11172068; the structural paper reports binary complexes and a modeled ternary complex, not an experimentally solved covalent enzyme intermediate PMID:17144655.
The unchanged products are M/83 kDa (P28329-1), S/74 kDa (P28329-2) and R/70 kDa (P28329-3). Older papers call the larger and common smaller forms 82/69 kDa. Translation starts and transcript identity must be distinguished: a transcript can yield more than one protein. A tagged 70-kDa construct can migrate near 74 kDa; it is not thereby the distinct S product. The separately expressed 74-kDa enzyme's comparison is reported as data not shown in PMID:11172068.
Nuclear localization is retained as non-core, rather than rejected because the principal enzyme role is cytosolic. Human recombinant enzyme in insect cells changes location with expression time PMID:10861222; human-cell localization/shuttling and native human neuronal observations independently corroborate a nuclear pool [PMID:10383456; PMID:12637523; PMID:16797789]. The physiological nuclear mechanism is unresolved. Human 69-kDa ChAT-GFP expression in SH-SY5Y cells increases SLC5A7 expression, requiring nuclear targeting and catalytic competence, whereas HEK293 lacks that response PMID:21163949. Neither direct DNA binding, a coactivator interface, histone acetylation nor another catalytic specificity was demonstrated. No new transcription process or molecular function is manufactured from this endpoint.
All 102 exact human P28329-3/partner tuples resolve to 306 positive IntAct records from PMID:32814053: array, pooled and validated yeast two-hybrid stages. The full original institutional PDF was recovered and selected Results, automated-screen Methods and data-deposition text were read. These stages share one study; no individual CHAT pair is claimed to have the general mammalian DULIP validation. The exact 13 partner-fragment features and all partner isoform identifiers remain visible in the source extract and reasons. Missing feature fields are not presented as verified full-length constructs.
The screen recovers DNA-binding transcription-factor partners, HSPD1/HSP60 and other biologically noisy partner classes with the same yeast two-hybrid evidence profile. Partner identity alone is not enough to refine those generic rows to DNA-binding transcription factor binding or protein-folding chaperone binding. Exact WT1-6 and GATA2-2 associations remain generic with their product limits; no canonical function is mechanically transferred across their sequence differences. Supported generic interactions are non-core under the explicit published ClinGen project instruction: immutable policy. Normal binding warnings are expected and do not justify deleting correct interaction evidence.
Both broad pathways genuinely contain CHAT reaction 264622, whose catalyst 264613 consumes choline plus acetyl-CoA and yields acetylcholine plus CoA in the cytosol. The phosphatidylcholine pathway includes a choline/ACh side branch, and the release cycle includes synthesis alongside uptake, vesicle loading, docking and release. This is real target membership, not a wrong-gene citation. However, the inspected CHAT step neither forms phosphatidylcholine nor moves transmitter. The source-scoped phosphatidylcholine pathway projection is removed, the neurotransmitter-transport projection is marked over-annotated, and the existing biosynthetic and synaptic roles are retained. No universal claim that CHAT cannot influence lipid metabolism or transport regulation is made. The sources share a curated reaction and are not three independent experimental replications.
Normal publication caches remain unchanged even where authentic external primary text was available. PMID:11172068 full PMC and PMID:21163949 full normal text support the direct human assays; PMID:32814053 full institutional PDF supports screen methods. Selected indexed primary text supplies PMID:10383456/12637523 localization context. The PMID:10861222/17144655/16797789 caches remain abstract-only and their uninspected detailed protocols are not claimed. Failed PDF/challenge responses are excluded. The PMID:12637523 detailed isolated-nuclei enzyme-assay Methods were not independently recovered; only its reported substrate-supported nuclear activity is described. Optional full-text-availability flags are omitted in favor of these explicit cache/access distinctions.
PAINT PTN000502016 is recorded as the proximate source with historical reconstruction unresolved. Independent human evidence corroborates the retained assertions; neither donor number nor human self-reference is a defect. The authenticated GO-CAM index has 0 target matches. No NEW annotation is proposed, and no comparator-absence argument is used.
The 117 decisions are 11 ACCEPT, 103 KEEP_AS_NON_CORE, 1 MODIFY, 1 MARK_AS_OVER_ANNOTATED and 1 REMOVE. The three products and ten original identifier/title pairs are exact; six additional references document five genuine supplemental publications and the manual factual evidence extract. Authored quotations are counted cumulatively across the whole new YAML and notes, including repetitions, with a maximum of 25 words per source. This is authoring discipline, not a repository validator rule. Notes and the evidence artifact add no quotation excerpts. Genuine normal validation/status/render receipts and final finite source guards accompany the sealed packet; campaign completion still requires the normal review and merge gates.
This section supersedes the earlier blanket statement that partner identity cannot justify a binding-class refinement, and the earlier OVER disposition for neurotransmitter transport. The complete notes prefix from external commit 37a3235b179be3162f42206408d0b55719e74d18 is retained above, together with both published history records. Its genuine phosphatidylcholine-projection correction is retained. All 117 original assertion objects, source partners and qualifiers, three products, reference identities and immutable cache bytes remain exact.
Binding specificity and physiological confidence answer different questions. GO:0140297 concerns binding to a DNA-binding transcription factor; GO:0051087 concerns binding to a protein-folding chaperone. The definitions do not require CHAT to regulate transcription or folding. Exact positive CHAT-3 pair records plus a verified partner class support these narrower binding terms, with the same assay limitations as retained generic pairs. None becomes a core role, an endogenous neuronal interaction, a mitochondrial encounter, or the partner's molecular activity. The seven prior refinements are restored on this basis, rather than claiming better physiological evidence for those seven hits.
The original full PMID:32814053 Results and STAR Methods exclude autoactivating bait strains, select HIS3/URA3 growth and lacZ positives, and retest pooled hits pairwise. These are real controls, but do not establish zero false positives. Three related curated Y2H stages are one study, not three independent biological replications. The separate DULIP subset is not attributed to individual CHAT pairs without target-level evidence. A heterogeneous partner repertoire and lack of established endogenous coexpression limit physiological interpretation. They do not demonstrate that a particular positive pair is false simply because the partner is a transcription factor or chaperone. These limits apply equally to generic and class-specific associations.
The same-class audit also corrects an overly conservative omission. ASCL4 has positive curated ISA transcription-factor and IBA sequence-specific DNA-binding annotations; tentative function prose alone is not a reason to dismiss those inferences. ZNF829's exact isoform-3 interaction region 82–513 is the complete 432-residue displayed sequence, reconstructed from the official VSP_057606 feature, including its KRAB domain and all ten annotated zinc fingers. Its curated IBA transcription-factor class is therefore used with the class-inference limit explicit. Rows 59 and 39 receive the same GO:0140297 refinement. No historical PAINT tree was reconstructed. THAP3 remains generic because the inspected official evidence establishes a regulatory complex, not THAP3 as its sequence-specific DNA-binding subunit. The THAP1 partner is not substituted for it. WT1-6, GATA2-2, HOXB1 Q4VB01, BAHD1 and TSC22D4 retain the stated exact-product/class uncertainties; none is asserted to lack the proposed class merely because it was not resolved here.
Both Reactome source rows explicitly assert involved_in. The official GO:0006656 definition describes phosphatidylcholine formation, whereas GO:0006836 describes directed neurotransmitter movement. The inspected CHAT catalyst 264613 in reaction R-HSA-264622 transfers acetyl to choline and forms acetylcholine in the cytosol. Both parent pathways truly contain that reaction, but the PC-forming and neurotransmitter-moving steps are separate events. Rows 110 and 111 therefore remove the unsupported GO process projections, without deleting or denying genuine target pathway membership. Existing acetylcholine biosynthesis and neuromuscular transmission remain. No new process term is needed, and no universal absence of metabolic or transport regulation is claimed. In particular, the independently observed SLC5A7 expression response is not recast as CHAT transporting a neurotransmitter.
Reference findings now include direct recombinant mutant kinetics from PMID:11172068 and the nuclear/catalytic mutant boundary for SLC5A7 expression from PMID:21163949. The generic structural anchor is replaced with a short catalytic-mechanism anchor from the actual PMID:17144655 abstract; its ternary mechanism remains a model, not a newly solved covalent intermediate. Existing access distinctions remain: genuine external full text and selected indexed primary evidence are not confused with normal cache completeness. The manual artifact's VERIFIED status records checked extraction/provenance, not an independent experiment or physiological verification of every screen interaction. Meaningful file-support paragraphs are explicitly authored paraphrases and match the added artifact fields verbatim.
The resulting 117 decisions are 11 ACCEPT, 94 KEEP_AS_NON_CORE, 10 MODIFY and 2 REMOVE, with no NEW annotation or core change. The nine screen refinements comprise eight DNA-binding transcription-factor partners and HSPD1; the tenth MODIFY is catalytic specificity. All primary excerpts in the full current YAML and complete campaign notes are counted cumulatively, including repetitions, within 25 words per source. This appendix adds no primary quotations. Prior notes and histories remain provenance and are superseded only where stated above. Normal validation, derived status and rendering are performed after these final text changes; source and trailing-whitespace guards accompany the finite packet.
Preserve the substantive existing human reaction, isoform and direct-interaction analysis. Human enzyme activity (PMID:10861222), substrate-bound structures (PMID:17144655) and congenital-myasthenia enzyme assays (PMID:11172068) support acetyl-CoA + choline → acetylcholine + CoA. The ternary arrangement in the structural study is modeled from binary structures; it is not a solved native enzyme–transporter complex.
Current-session binding policy supersedes historical notes requesting retention of all bare GO:0005515 observations. Ninety-three formerly non-core bare-binding rows are now REMOVE solely as uninformative under the explicit policy; their machine partner/source payloads and durable source-evidence artifact remain intact. Nine measured interactions retain precise partner-class refinements (transcription-factor or chaperone binding). This does not make CHAT a transcription factor or chaperone. Existing phosphatidylcholine-biosynthesis and neurotransmitter-transport removals remain grounded in reaction/role mismatch. The 82-kDa nuclear isoform is retained; cytoplasmic predominance is not exclusion. Prior source-evidence JSON remains untouched and its old policy text is historical, not current authorization.
Methods and limits: reviewed all 117 original annotations, preserving every machine-sourced term, qualifier, reference, partner and evidence field. No new annotation was added. Genuine Falcon and fallback provider attempts were made in the foundation batch and failed with authentication/quota errors; no provider report was fabricated. Normal PMID fetching and targeted public-source recovery were used instead, with access limitations stated per claim. Logs are retained under .cache/module-audit-logs/acetylcholine-foundation-fetch/; cached source files were not hand-edited. This review follows the current bare-protein-binding and root-organelle policies.
Actions: ACCEPT 11, KEEP_AS_NON_CORE 1, MODIFY 10, REMOVE 95. Focused strict/schema, ontology-term and source-quote checks, record-section backfill and HTML rendering are recorded in the final audit report; remaining biological uncertainties above are not validation failures.
Normalized human-facing citation identifiers and spacing, preserving every machine annotation field and existing supporting quotation. Removed exactly repeated summary prefixes from reasons where a separate substantive explanation remains. All annotation actions remain unchanged for this gene.