Reviewed all 45 annotation rows and all 34 starting reference entries (27 PMIDs, four GO_REF records and three file references), using the review YAML rather than changing its machine-sourced fields. The updated annotation-reviewer rule was applied: generic protein binding is uninformative rather than over-annotation. Six such annotations were removed without rejecting the underlying associations or assigning a partner's molecular activity to AGO1.
AGO1 guide binding and catalytic activity are directly established [PMID:16081530, "AGO1 is physically associated with miRNAs, transacting siRNAs, and transgene-derived siRNAs"] [PMID:16081530, "We also show that AGO1 has Slicer activity."]. Target-mRNA binding is a direct step in this mechanism, and was changed from non-core to core. The leaf RNA interactome study explicitly identifies AGO1 PMID:32344669. AGO1's participation in secondary-siRNA generation was also recognized as core pathway activity, without assigning it Dicer or RDR6 activity [PMID:15469823; PMID:34192539].
The protein-binding studies establish associations with AMP1, PDS5C/E, RICE proteins, ATRM2, SDE3 and viral P0 [PMID:23622241; PMID:38446745; PMID:28463111; PMID:29941559; PMID:22940249; PMID:17869109]. These observations do not establish distinct partner-specific molecular activities for AGO1. In particular, the AMP1 study explicitly notes that co-immunoprecipitation could reflect shared membrane proximity rather than a common complex. ER association and translational repression are retained in the functional synthesis, separately from this binding claim.
The broad electronic virus-defense annotation was refined to RNAi-mediated antiviral immunity using direct AGO1 viral-siRNA loading and virus-hypersensitivity evidence [PMID:17158744; PMID:11910010]. The PVX annotation remains non-core and context-qualified: suppressor-deficient PVX susceptibility does not imply AGO1 restricts wild-type PVX like AGO2 and AGO5 PMID:26023161. PAMP-triggered bacterial immunity is a separate endogenous-miRNA output PMID:20164210.
The prior rejection of the cytoplasmic ribonucleoprotein-granule annotation relied on an SGS3-focused abstract. The published full article was retrieved from a PDF copy, title and DOI checked against PMID:36640363 (10.1016/j.celrep.2022.111985). Results on printed page 10 and Figure 6B/C on printed page 11 explicitly show AGO1 recruitment into SGS3 condensates in tobacco cells and AGO1-SGS3 BiFC foci in Arabidopsis protoplasts. The paper states: "AGO1 protein was incorporated into SGS3 condensates" [PMID:36640363, Results]. The annotation was changed to ACCEPT. The cached publication remains abstract-only; no full-text excerpt was inserted manually into that generated cache.
Independent corroboration was fetched and cached as PMID:38384836. This study reports AGO1 condensates under heat stress and colocalization with siRNA-body and stress-granule components. That primary paper now supports the row with a cache-verifiable snippet. Heat-associated localization is not evidence that every condensate has enhanced silencing activity: the study reports only moderate changes in guide loading and cleavage signatures.
All 27 starting PMIDs were already cached. The cached full-text flag was checked but was not treated as proof that every result or supplement was present. For example, the extraction of PMID:15829601 includes auxin/root findings but omits the far-red genetic experiment. The indexed primary article explicitly states "phyA is epistatic to ago1 in far-red light". The supported non-core far-red response was retained. The browser's direct page-open attempt encountered a challenge page; the quoted primary-source text was visible through the search index, so no figure number was inferred.
The original specific assays in PMID:35137215 and PMID:16998468 were not exposed in their cached abstracts. The clear AGO1 miRNA-binding function is retained with independent direct support and explicit source-curator deference. Likewise, the exact source-table entries for PMID:27729395 and PMID:28887381, and the AGO1 localization assays in PMID:17442570, remain to be directly inspected. General target-RNA binding and nuclear/cytoplasmic localization remain supported by independent primary observations. This is not a claim that those uninspected source experiments were independently verified.
Targeted CLI refreshes were limited to the initially abstract-only PMID:36640363, PMID:34192539 and PMID:38446745. All remained abstract-only because no PMC records were available. No existing full-text caches were force-refreshed or downgraded. PMID:38384836 was fetched successfully with full text. Existing research/UniProt/PANTHER context was consulted, but invented paraphrases attributed as quotations to those files were removed from the curated YAML.
QuickGO was queried for authored refinements and core terms, including RNA endonuclease activity (GO:0004521), cytoplasmic ribonucleoprotein granule (GO:0036464), mRNA regulatory element binding translation repressor activity (GO:0000900), and endoplasmic reticulum (GO:0005783). Definitions were inspected before using them.
Two evidence-backed missing annotations were added after the 45 source rows: ER localization (IDA, PMID:23622241) and mRNA regulatory element binding translation repressor activity (IC, integrated from the original ribosome-stalling annotation, direct AGO1 RNA binding and ER-associated repression). The latter is explicitly an inference from RNA-mediated repression, not a molecular activity assigned from a protein-binding result.
Research refresh outcome: Falcon timed out after 600 seconds; automatic perplexity-lite fallback failed with HTTP 401 insufficient_quota. The existing Falcon report was preserved. The re-review therefore rests on the primary-source audit described above; no manual text was written under a provider-generated filename. Final validation passed with two advisory warnings: the context-dependent virus-defense actions and preference for primary rather than generated-research snippets. All 45 source annotation identities are preserved exactly; the two additional rows are explicitly NEW.
Generalized the author-proposed translational-repressor MF and matching core function to QuickGO-verified GO:0030371. This preserves the established guide-dependent repression while avoiding a stronger implication of independent cis-element recognition. PMID:34192539 Removed a duplicated summary sentence from its reason. All GOA-sourced rows are unchanged.
Core function 1 keeps GO:0090624 (cleaving miRNA-paired mRNA), matching the only slicer row (PMID:16081530 IDA, MODIFY to GO:0090624). A parallel branch had widened it to the parent GO:0004521 on the grounds that AGO1 also loads ta-siRNAs and transgene siRNAs PMID:16081530. Loading is not cleavage, though, and no cached paper assays AGO1 cleaving an siRNA-paired target. Open question: if a paper showing AGO1-catalysed TAS or transgene-target slicing is found, add GO:0070551 to that row's replacements and widen the core function to GO:0004521.