AIRE was fetched fresh for the human Proteostasis PN batch. Falcon deep research was attempted with just deep-research-falcon human AIRE --fallback perplexity-lite; Falcon timed out after 600 seconds and the perplexity-lite fallback failed with a Perplexity API 401 quota error. No provider-generated deep research file was created, so the GOA review was completed from the seeded human GOA, cached publications, UniProt, the PN projection report, and the manual fallback summary in AIRE-deep-research-manual.md.
Core biology: AIRE is best supported as a nuclear chromatin-associated transcriptional regulator for thymic self-tolerance. It promotes central tolerance and prevents autoimmunity PMID:26084028. Mechanistically, AIRE binds histone H3 through PHD1 and preferentially recognizes H3K4me0 [PMID:18292755 "AIRE selectively interacts with histone H3 through its first plant homeodomain (PHD) finger"; PMID:18292755 "preferentially binds to non-methylated H3K4 (H3K4me0)"], with structural support for the H3-binding interface PMID:19446523. AIRE also binds A/T-rich DNA motifs as oligomers PMID:11533054 and activates transcription in reporter/promoter contexts [PMID:11274163 "AIRE can activate the interferon beta minimal promoter in a transfection assay"; PMID:18292755 "in vivo AIRE binds to and activates promoters containing low levels of H3K4me3"].
PN projection: the PN report projects GO:0061630 ubiquitin protein ligase activity for AIRE from Ubiquitin Proteasome System|E3 ubiquitin and UBL ligases|RING variant|PHD|other [file:projects/PROTEOSTASIS/reports/pn_projection/pn_projected_annotations.tsv]. I did not add this as a NEW annotation. The projection is a domain/family-bucket inference and the current AIRE evidence supports PHD histone-reader/chromatin/transcription functions, not catalytic ubiquitin transfer. AIRE's PHD domain is experimentally characterized as a histone-recognition zinc finger rather than an E3 ligase catalytic module PMID:19446523. I added a suggested question/experiment to test whether any direct E3 activity exists, but the current conservative curation decision is no GO ubiquitin ligase annotation.
Annotation decisions: broad protein binding entries were marked over-annotated because the useful biology is captured by histone binding, chromatin binding, DAXX/PRKDC/transcription-complex interactions, or oligomerization rather than generic binding. Automated translation regulator annotations were removed as unsupported. Broad immune/humoral immune response terms were marked over-annotated; specific tolerance, negative selection, chemokine/thymocyte migration, and thymus epithelial terms were kept according to evidence strength and core-vs-non-core status.
This entry supersedes the functional judgments in the June PN notes and historical manual research where they differ. Those records remain intact as session provenance. In particular, their statements that intrinsic E3 activity was never demonstrated, that AIRE has no humoral role, and that transcriptional function disproves translation regulation were too strong. The original recombinant DNA-binding experiments remain positive evidence.
Identity and baseline: HGNC:360, approved symbol AIRE, UniProt O43918; APECED, PGA1 and APS1 searches found no competing review directory or open PR. All seven gene files were byte-checked against main 68e57e5afc25714356f4b5febb50d864e6b849ba before authoring. All 55 original annotation source objects, four alternative products and 24 original reference identifiers/titles are preserved. Only annotation reviews and authored synthesis/reference assessments changed; no NEW annotation was added. GOA, UniProt, historical manual research and PN notes remain unchanged.
The cached interpro/panther/PTHR46386/PTHR46386-paint.tsv was read. PTN005329562 grounds nuclear, transcription, DNA, chromatin and histone annotations; PTN002931439 grounds the immune/developmental group. Source entities use the ancestral PTN nodes rather than extant donor lists. Human descendant experiments contributing to the IBD are expected grounding, not circularity. Tree/MSA node placement was not independently reconstructed. Exact ARBA predicates were not inspected and remain UNRESOLVED even where independent human biology supports the annotation.
The MGI Aire annotation graph, showing database update 2026-09-08 when inspected, confirms the term-specific donor references. Mouse Q9Z0E3 and the seeded Ensembl protein identifier are kept distinct from the human target. The provenance trace is:
| Assertion | Mouse source | Evidence assessment |
|---|---|---|
| Humoral immune response | IMP J:77873, PMID:11854172 | Primary abstract verifies whole-animal knockout, circulating autoantibodies and immune-challenge effects; full assay not recovered. |
| Thymic epithelial morphogenesis/nuclear body | J:141379, PMID:19015306 | Primary abstract/figure material supports epithelial differentiation and localization. |
| Chemokine production/thymocyte migration | IMP J:157503, PMID:19923453 | Primary abstract/figures support chemokine changes, gain-of-expression effects and altered migration. |
| Peripheral T-cell tolerance | IDA J:208220 | Bibliographic mapping not recovered from the dynamic MGI reference page. Independent PMID:23993652 has cell-population/transgenic-antigen scope. |
| Negative thymic selection | IGI J:196474 | Original source record is present; bibliographic mapping not recovered. Conserved mechanism has independent human/established literature support. |
PMID:11854172(https://pubmed.ncbi.nlm.nih.gov/11854172/) is a real experimental humoral donor, not an invented antibody inference. Its accessible abstract cannot resolve direct participation. The primary preprint PMID:38260362 reports human B-cell expression/interaction and mouse B-cell functional experiments connecting AIRE to AID-dependent antibody diversification. The original PMID record is a preprint. A final check recovered the current revised eLife version 2, DOI:10.7554/eLife.110530.2, dated 2026-09-21, and its public assessment rating the evidence convincing. The reviewed current status supersedes the initial publication-quality caveat. Both broad humoral rows are therefore KEEP_AS_NON_CORE on independent positive B-cell evidence; the original mouse IMP and InterPro mapping mechanisms remain unresolved. No NEW or additional core is asserted.
PMID:23993652(https://pmc.ncbi.nlm.nih.gov/articles/PMC3804105/) was inspected through primary full Results: human lymph-node AIRE-positive cells are observed, while mouse tolerance experiments use an Aire-promoter-directed antigen transgene. The measured entity performing antigen presentation is the extrathymic cell population. These experiments do not by themselves isolate the endogenous AIRE-protein-dependent step. Peripheral tolerance transfers remain UNDECIDED rather than asserting absence of the function. Central tolerance and negative thymic selection remain core; AIRE performs transcriptional work needed for self-antigen presentation.
PMID:19015306(https://pubmed.ncbi.nlm.nih.gov/19015306/) supports the existing contextual epithelial-development process; PMID:19923453(https://pubmed.ncbi.nlm.nih.gov/19923453/) supports existing chemokine/migration regulation. These are retained non-core. Their results are not converted into intrinsic structural, chemokine or motility activities of AIRE.
Local OAK GO definitions/parents were inspected. GO:0000977 is regulatory-region sequence-specific binding; GO:0043565 captures sequence selectivity without the regulatory-region qualification. GO:0006955 includes calibrated immune response and is not restricted to direct effector chemistry. The broad immune-response rows are consequently retained. GO:0045182 specifically concerns polypeptide synthesis at the ribosome. The retrieved InterPro-family description concerns transcription, but the exact translation mapping's experimental basis remains unavailable; both that MF and its GO:0006417 logical consequence are UNDECIDED. Distinct transcription and pre-mRNA-processing evidence is not proof that a second function is impossible. No GO-CAM index entry for human O43918/mouse Q9Z0E3 was found. No new process annotation or ancestor/descendant redundancy was introduced.
PMID:14734522(https://pmc.ncbi.nlm.nih.gov/articles/PMC2211764/) reports actual positive reconstituted assays using GST-PHD1 and full-length preparations, E1/E2 and ubiquitin. Thus the historical claim that the candidate ligase function was only a PHD-family projection is incorrect. PMID:15649886(https://pubmed.ncbi.nlm.nih.gov/15649886/) reports no detectable intrinsic PHD1 E3 activity or cognate E2 interaction. The conflict is recorded explicitly; neither an unspecified contaminant nor an experimental error is asserted. No NEW E3 term is added. PMID:15150263(https://kyushu-u.elsevierpure.com/en/publications/subcellular-expression-of-autoimmune-regulator-is-organized-in-a-/) describes localization/nuclear-matrix and proteasome-dependent redistribution in the author-institution abstract, which does not independently assign ubiquitin-transfer chemistry to AIRE.
Generic binding to NCK1, PRKDC and DAXX is removed as uninformative without rejecting interactions. PMID:17474147's stated PLCG1 validation rate is not silently transferred to the AIRE/NCK1 pair. The histone H3 partner in PMID:18292755 permits a specific histone-binding replacement. No unsupported adaptor function is manufactured.
Fresh genuine research was attempted with the default Falcon provider and perplexity-lite fallback, preserving the historical manual artifact and directing any new provider output to /tmp/AIRE-fresh-research/AIRE/. Both provider invocations failed during uvx deep-research-client[cyberian]==0.2.7rc1 dependency resolution because PyPI DNS was unavailable; neither reached provider execution, and no report was created. This is manual primary-source review, not fabricated provider output. Logs: /tmp/AIRE-fresh-research.log.
The normal GOA publication fetch found all 14 original PMIDs cached. Standard fetch-pmid attempts for 12 additional required records returned DNS failures and produced no caches: PMID:11854172, PMID:14734522, PMID:15150263, PMID:15649436, PMID:15649886, PMID:19015306, PMID:19923453, PMID:22310661, PMID:23993652, PMID:38260362, PMID:38480882 and PMID:39169234. This list includes historical notes, current YAML and the primary sources used for donor/mechanism adjudication. Local full-text flags reflect actual caches: only the existing PMID:18292755 and PMID:19446523 caches contain full experimental sections; external reading does not set a local full-text flag. All original source records remain immutable. The review is DRAFT until normal cache recovery and outstanding validation advisories are resolved. Final source-preservation, quote, history and rendered-output checks are recorded in the handoff manifest.
Independent review: root read all 55 decisions, the integrated core, questions and new reference assessments and accepted the principal biology. The requested current-publication-status check recovered the eLife revision above; root independently read the saved current eLife record and actual B-cell Results/Methods and accepted the final two-row humoral delta on experimental participation, with historical donor limitations retained. The bounded annotation_aars1 consultation confirmed the original human DNA-binding experiments.
Final checks: the full targeted validator completed its ontology and reference checks; references reported only the 12 unavailable normal caches. Two knowledge-gap strings were converted to the schema-required gap_statement objects without changing their content. Final schema/term/GOA/best-practice validation passed with the sole advisory that the historical manual research is not cited by annotation reviews; that artifact contains superseded conclusions and was not used as primary support. History validation and HTML rendering passed. A separate case-sensitive, whitespace-normalized check matched all 58 cached quotations. The 55 source objects, four isoforms, original 24 reference identity/title pairs and four immutable provenance/source files remain preserved; all authored files pass trailing-whitespace and YAML-alias checks.
The seven local gene files were rechecked byte-for-byte against the reviewed head 6912ab4f1c14f8b4fbaa3d3ffd398f80683f0dd3 before editing. This bounded session addresses review 5329239748 and its full comment. It preserves all 55 seeded annotation assertions, their actions, all 36 reference identities/titles, the four alternative products, and the single integrated core. There are no NEW rows or new molecular-function/process assertions. Action totals remain 26 ACCEPT, 15 KEEP_AS_NON_CORE, 6 MODIFY, 5 UNDECIDED and 3 REMOVE.
The normal fetch-pmid source4 run 36294925088 generated all twelve previously missing AIRE records. Its fixed artifact was independently verified against the source run/head, request inventory, ZIP digest and per-file raw-output hashes; root imported exact bytes with exclusive creates and no overwrites. The import receipt is tmp/source4-canonical-import-receipt.json (SHA256 3c32b81b66e9234a64b687be4203965fb8d960bcffdbcfd2f6dc9b335428698b). These are machine-generated publication records, not authored replacements. Prior local DNS failures remain historical facts, but no longer represent current cache availability.
| Recovered PMID | Local cache now | Inspected evidence and boundary |
|---|---|---|
| 11854172 | Abstract only | Mouse Aire knockout, immune challenge and circulating autoantibodies; no newly accessible full humoral assay. |
| 14734522 | Full text | Human cDNA and recombinant PHD/full-length preparations, E1/E2/ubiquitin and GST/PHD2 controls; the positive E3 experiment remains in explicit conflict with 15649886. |
| 15150263 | Abstract only | Nuclear-matrix association and proteasome-dependent redistribution, not intrinsic ubiquitin-transfer chemistry. |
| 15649436 | Abstract only | Human recombinant-domain motif binding; native promoter selection is not established. |
| 15649886 | Abstract only | Human PHD1 structure and reported negative E3/E2-binding results; full preparation-by-preparation comparison remains incomplete. |
| 19015306 | Full text | Aire-locus GFP/disruption, nuclear staining, altered mouse mTEC morphology/distribution and reduced involucrin/Hassall-like structures (Figures 1–3). |
| 19923453 | Full text | Mouse thymic expression/protein measurements and adenoviral Aire/conditioned-medium/chemokine-blocking experiments in the 1C6 epithelial system. |
| 22310661 | Abstract only | Cooperative histone/DNA-PK recruitment; the existing external reading remains distinct from local full-text availability. |
| 23993652 | Full text | Human lymph-node observation and mouse AdBDC antigen-presentation/tolerance experiments; this still distinguishes cell-population function from endogenous AIRE-protein participation. |
| 38260362 | Full text of original preprint | The recovered record is bioRxiv DOI 10.1101/2024.01.10.574926, PMC10802573. It is not the revised eLife v2. The separate v2 route and judgment above remain explicit. |
| 38480882 | Full text | Mouse F1-hybrid cis-regulatory, Z-DNA mapping and promoter-poising experiments; no purified AIRE-Z-DNA recognition or negative replication of the early human EMSA was inferred. |
| 39169234 | Full text | Human WT/domain-mutant AIRE in 4D6 cells (Figures 1–2); selected mouse-domain experiments retain their construct boundaries. The six existing abstract quotations are now locally checkable. |
Seven local full_text_unavailable flags change to false; the five abstract-only records stay true. Some XML text duplicates parent/child sections or joins headings, which is an extraction limitation rather than absence of full text. The original 23993652 reference title is preserved: the normal cache's superscript CD4⁺ is a rendering difference from CD4+, with the same PMID/DOI and paper. The original preprint's cache is not used to imply that every later eLife revision is now available locally.
Full targeted validation passed with one intentional advisory: the superseded historical manual research artifact is not cited as annotation evidence. The title check passed with the original 23993652 title preserved; its typographic difference was not hidden by editing either record. All 64 ordinary supporting quotations match their cached sources under case-sensitive whitespace normalization, including all six 39169234 instances; none relies on case folding. The recursive citation census covers current YAML, notes and both historical manual-note artifacts: all 26 cited PMIDs are cached, with no genuine provider report present and no remaining source gate. The DRAFT status is retained for the intentional manual-research advisory; the PR may be ready for normal review. Source-preservation, history, rendering and final file hashes are recorded in the follow-up manifest.
Live PR #3265 was independently confirmed open at 30c18ceaab6457e4c158b720352f384104c60057; all seven gene files matched that exact published tree before editing. Formal review 5329722498 and full comment 5854568649 were read. The formal review is COMMENTED, withdraws the prior blockers, and identifies a real local provenance omission. No source record, immutable manual artifact, previous history or shared project file is changed here.
The checked-in rules/arba/_interpro2go.txt has header version 2025/09/01 20:18:52. Lines 10876–10880 explicitly assign five terms to the AIRE entry IPR008087: DNA binding (GO:0003677), translation regulator activity (GO:0045182), humoral immune response (GO:0006959), nucleus (GO:0005634) and cytoplasm (GO:0005737). The map is recovered, so the current YAML no longer treats retrieval of the mapping itself as pending. These are entry-level assignments; the flat file contains no member-specific experimental assay or primary-evidence citation. That observation does not show that the entry curators had no underlying evidence, nor that its member proteins can never have the mapped function.
The live GO:0045182 definition and its part_of relationship to GO:0006417 were rechecked. The former concerns molecular function regulating polypeptide synthesis at the ribosome; the latter concerns modulation of translation of RNA into protein. The inspected PMID:11274163 abstract reports promoter transactivation and subcellular targeting. PMID:20085707 reports AIRE-associated nuclear transport, chromatin, transcription and pre-mRNA-processing proteins and RNAi experiments. Its abstract does not describe a translation assay; no negative translation experiment is inferred from this omission. The original auditor independently confirmed no additional positive translation evidence had been recovered.
Both electronic translation rows remain UNDECIDED after deliberate reconsideration. The exact mapping chain is now established, but it supplies no result that settles AIRE's translation-regulatory capacity. There is neither a resolved positive mechanism nor demonstrated target divergence or a negative assay here. Thus the uncertainty is biological rather than a promise to retrieve a nonexistent per-member field from this flat mapping. A family-level assignment alone is not proof of over-propagation, and the transcript-level experiments are not used to establish intrinsic inability. The source statuses now identify these as inferred records; the question asks for the primary basis underlying the known mapping.
The humoral IEA remains KEEP_AS_NON_CORE. The exact InterPro entry mapping is now traced, while the independent B-cell evidence remains separate from its historical provenance. The recovered PMID:38260362 original preprint was reread at Results Figure 5: human tonsillar B-cell AIRE–AID association is distinguished from the mouse CH12 domain/depletion functional experiments. The earlier separately inspected eLife version 2 and public assessment remain explicitly versioned; the cached bioRxiv record is not represented as that revision. Nothing in finding the map changes this positive contextual judgment.
The optional core suggestion was assessed against the actual schema: contributes_to_molecular_function is for a complex function that the gene product contributes to but does not independently enable. The existing zinc-binding and self-association assertions describe AIRE's own structural binding/assembly properties, already integrated into its single core description and evidence. They are not recast as accessory contributions solely to duplicate every ACCEPT term in the core table. No new core or molecular function is added. The shared INTERPRO derived harvest remains outside this gene-only follow-up; its next regeneration should read the final reviewed actions rather than rewrite them to match an older report.
All 55 source objects/actions, 36 reference identities, four alternative products, description and core are preserved. The historical failed-API statements remain above as session history, superseded by this dated local-file finding. Full validation, exact-quote/source preservation, recursive PMID/decoded DOI/Reactome census, new scaffolded history and rendering are checked before freeze. DRAFT remains appropriate for the known advisory concerning the preserved, superseded manual research artifact; it is not promoted into primary evidence just to silence that advisory.