AIGR Focused Review — rad3 (SCHPO, Q02099) nucleolus GO:0005730

Gene: rad3 (SPBC216.05) · Organism: Schizosaccharomyces pombe 972 / ATCC 24843 (NCBITaxon:284812) · UniProt: Q02099
Focus: existing GO annotation decision — existing_annotations[36] of genes/SCHPO/rad3/rad3-ai-review.yaml
Hypothesis slug: existing-go-0005730-mark-as-over-annotated
Annotation under review: nucleolus (GO:0005730), IDA, qualifier is_active_in, reference PMID:18180284
Seed review action: MARK_AS_OVER_ANNOTATED

Summary

The seed hypothesis is SUPPORTED, and on closer inspection the annotation is even weaker than "over-annotated" implies — it is a candidate for outright REMOVAL or reference correction. The rad3→nucleolus (GO:0005730) association rests on a single, uncorroborated IDA annotation whose only cited primary source, PMID:18180284 (Bailis et al. 2008, Mol. Cell. Biol., PMC2258774), does not localize Rad3 to the nucleolus — or anywhere. Full-text mining of that paper (~97,000 characters covering Abstract, Methods, Results, Discussion and References) recovered zero occurrences of "nucleol-", "rDNA", "ribosom-", or "GFP", and the only Rad3 mentions (4 total) are genetic/pathway references (Rad3–Cds1 checkpoint, Δrad3 mutant, Rad3 required for Cds1 activation, and one citation). The proteins actually imaged in that study are Rad22-YFP, Rhp51, Mcm4, Rad52, Crb2 and phospho-H2A. Rad3 itself was never subjected to a localization assay. This strongly indicates the annotation is a mis-attributed reference or a curation error, not a genuine minor or context-specific pool.

Independent lines of evidence corroborate the over-annotation call. QuickGO returns exactly one annotation linking Q02099 to GO:0005730 (IDA/ECO:0000314, assigned by PomBase, qualifier is_active_in); no second, independent primary source exists. The genome-wide YFP ORFeome localization screen (PMID:16823372) places Rad3 in the cytosol/diffuse signal, not the nucleolus. The intensively studied human ortholog ATR (Q13535) is not curated to the nucleolus; its curated compartments are nucleoplasm, chromosome, PML body, nuclear envelope and sites of DNA damage. Where ATR-family kinases do engage the nucleolus in mammals, the activity is demonstrably stress-induced and context-specific (TopBP1 overexpression, RNA Pol I inhibition), not a constitutive core compartment.

Rad3's established, multiply corroborated sites of action are the ATR–ATRIP (Rad3–Rad26) complex, chromatin/stalled replication forks, and telomeres — each supported by multiple primary references with IDA evidence. The is_active_in qualifier on the nucleolus term is particularly problematic because it asserts functional activity in a compartment where neither the cited paper nor any other primary source demonstrates Rad3 presence, let alone function. Recommended lead: REMOVE or correct the reference; MARK_AS_OVER_ANNOTATED is fully supported as a conservative floor, pending a curator's inspection of the journal PDF and supplemental figures to exclude a reference typo or overlooked panel.

Executive Judgment

Verdict: SUPPORTED — and the underlying evidence is weaker than the seed states. The over-annotation call is justified; the evidence in fact points toward outright removal / reference correction, subject to a curator's check of the published PDF and supplemental figures.

The decisive result is the full-text mining of the cited reference (Finding 2). Because the annotation carries evidence code IDA/ECO:0000314 ("inferred from direct assay"), it requires a demonstrated experimental localization of Rad3 in the cited paper. No such assay exists in PMID:18180284. The annotation therefore fails its own evidence-code requirement. The strongest residual caveat is that retrieval was of the PMC HTML article, so a small chance remains that a separately hosted supplemental figure contains Rad3 data — although the main text and figure legends never mention the nucleolus and give no indication Rad3 was imaged.

Key Findings

Finding 1 — The nucleolus annotation is single-source and non-core; the over-annotation call is justified

Rad3's curated cellular-component (CC) annotations span several compartments, but they are not of equal evidential weight. Assembling the PomBase curated set for SPBC216.05 / Q02099:

The nucleolus term is the only CC annotation resting on a single IDA, and its source paper (Bailis et al. 2008) is primarily about the MCM complex and replication-fork protection; its abstract does not mention the nucleolus. The genome-wide YFP ORFeome study (PMID:16823372) that systematically localized S. pombe proteins assigned Rad3 to a cytosolic/diffuse signal — an orthogonal, high-throughput dataset that fails to reproduce a nucleolar pool. The deeply studied human ortholog ATR (Q13535) carries curated CC terms for nucleoplasm, chromosome, PML body, nuclear envelope and sites of DNA damage, but no nucleolus term, despite decades of study.

Where mammalian ATR/ATM signaling has been documented at the nucleolus, it is explicitly inducible and stress-dependent: TopBP1 overexpression triggers ATR-dependent shut-down of rRNA transcription and nucleolar segregation (PMID:25916852), and Pol I transcription inhibition (CX-5461) activates non-canonical ATM/ATR signaling within nucleoli (PMID:27391441). These are context-specific responses, not evidence of a constitutive nucleolar compartment for the checkpoint kinase. Taken together, the nucleolus annotation is single-source, non-core, and inconsistent with orthogonal localization data and ortholog curation — MARK_AS_OVER_ANNOTATED is well supported.

Finding 2 — The cited reference PMID:18180284 contains no Rad3 localization and never mentions the nucleolus; the annotation is likely mis-attributed

This is the decisive finding. Full text of PMID:18180284 (Bailis et al. 2008, PMC2258774; ~97,000 characters including Abstract, Methods, Results, Discussion and References) was mined directly:

• nucleol → 0 occurrences
• rDNA → 0 occurrences
• ribosom → 0 occurrences
• GFP → 0 occurrences
• Rad3 → 4 occurrences, all as genetic/pathway references (the Rad3–Cds1 checkpoint; the Δrad3 mutant; Rad3 required for Cds1 activation; one bibliographic citation). None occurs within 60 characters of "foci", "focus", "localiz", "nucle", "stain", "YFP" or "GFP".

The proteins actually imaged in the paper are Rad22-YFP, Rhp51, Mcm4, Rad52, Crb2 and phospho-H2A. Rad3 itself is never subjected to a localization assay. The PomBase annotation detail (id 433558) records IDA / ECO:0000314, low-throughput, assigned_by PomBase on 2015-11-25, with no extension and only the is_active_in qualifier. Because IDA/ECO:0000314 demands a demonstrated experimental localization in the cited paper, and no such assay for Rad3 exists in PMID:18180284, the annotation fails its own evidence-code requirement. The two most parsimonious explanations are (1) a mis-attributed/typo'd reference (the wrong PMID was entered), or (2) a curation slip in which text about a different imaged protein was mapped to Rad3. Either way, the citation does not support the term.

Finding 3 — QuickGO confirms a single, uncorroborated is_active_in nucleolus annotation

An independent QuickGO annotation search (geneProductId=Q02099, goId=GO:0005730) returns numberOfHits = 1: IDA / ECO:0000314, reference PMID:18180284, assignedBy PomBase, qualifier is_active_in. There is no second or independent annotation from any other database or reference. Targeted PubMed searches for an alternative primary paper reporting S. pombe Rad3 nucleolar localization returned none. The is_active_in qualifier is especially problematic: it asserts that Rad3 is functionally active in the nucleolus, a claim demanding functional evidence there. The cited paper provides neither localization nor function, so an is_active_in nucleolus assertion is unsupported on its own terms.

Mechanistic Model / Interpretation

The entity under evaluation is Rad3, the S. pombe ATR-family PIKK Ser/Thr checkpoint kinase (InterPro domains: PI3/4-kinase catalytic, FAT, FATC, HEAT-ATR, UME). It functions as the Rad3–Rad26 (ATR–ATRIP) heterodimer — the sensor apex of the DNA replication/damage checkpoint — sensing stalled replication forks and DNA damage and activating downstream effector kinases Cds1 and Chk1. The GO term in question, nucleolus (GO:0005730), asserts a cellular location (and, via is_active_in, a functional site) that must be judged against where Rad3 is actually demonstrated to reside and act.

   Rad3 established sites of action (multiply corroborated)
   ┌───────────────────────────────────────────────────────────┐
   │  Rad3–Rad26 (ATR–ATRIP) complex  ── PMID:10559981 (IDA)     │
   │  chromatin / stalled forks       ── PMID:17531813,21945095  │
   │  telomeric repeat region         ── PMID:12196391,20140190  │
   │  nuclear chromosome              ── PMID:8843195            │
   └───────────────────────────────────────────────────────────┘
         │  activates
         ▼
      Cds1 / Chk1 effector kinases
         │
         ▼
   Downstream: Clp1/Flp1 released FROM the nucleolus on stress
   (PMID:18385517, 22918952) — a SUBSTRATE relocalization event,
   NOT evidence that Rad3 itself resides in the nucleolus.

   Nucleolus (GO:0005730) annotation
   ┌───────────────────────────────────────────────────────────┐
   │  Single IDA · PMID:18180284 · qualifier is_active_in        │
   │  → cited paper: 0 mentions of nucleolus/rDNA/ribosom/GFP    │
   │  → Rad3 never imaged; proteins imaged = Rad22, Rhp51, Mcm4… │
   │  → orthogonal ORFeome screen: Rad3 = cytosol/diffuse        │
   │  → human ATR ortholog: NOT curated to nucleolus             │
   └───────────────────────────────────────────────────────────┘

A crucial conceptual distinction is that downstream substrates of the Rad3 pathway do relocalize at the nucleolus. The Cdc14-family phosphatase Clp1/Flp1 is nucleolar-sequestered and is released into the nucleoplasm on replication stress in a Cds1/checkpoint-dependent manner (PMID:18385517, PMID:22918952). This is genuine and interesting nucleolar biology — but it concerns the substrate (Clp1/Flp1), not Rad3 itself. It would be an error to infer Rad3 nucleolar residence from the nucleolar release of one of its pathway's downstream targets. This substrate-versus-kinase conflation is a plausible route by which a nucleolar annotation could have entered Rad3's record. The correct GO representation of Rad3's function localizes it to chromatin, forks and telomeres — the sites where the kinase is directly demonstrated to bind and act.

Evidence Base / Evidence Matrix

(Computed provenance from the iterations was saved as evidence_matrix.csv, rad3_CC_provenance.csv, and GO_decision_table.csv.)

GO Curation Implications

Lead (requires curator verification): remove or correct the reference for GO:0005730 (nucleolus, CC) on rad3.

• Term class: GO:0005730 is a cellular-component term. The question is whether the nucleolus is even a supported localization of Rad3, and whether it is core. The weight of evidence says the cited support is absent and the nucleolus is not core — chromatin, forks, telomeres and the ATR–ATRIP complex are the core CCs.
• Primary recommendation: REMOVE (IDA/ECO:0000314 is not satisfied by PMID:18180284 — the paper neither images Rad3 nor mentions the nucleolus). At minimum, the reference is wrong and must be corrected.
• Fallback (if an inaccessible supplemental figure does show Rad3 in the nucleolus): MARK_AS_OVER_ANNOTATED / KEEP_AS_NON_CORE with a context/stress qualifier — still not a core compartment.
• is_active_in qualifier: This is the strongest form of location assertion (functional activity in the compartment) and is the least defensible here — it should be dropped along with the term.
• What the evidence does support (unchanged): chromatin (GO:0000785), telomeric repeat–containing region, nuclear chromosome, and the ATR–ATRIP kinase complex (GO:1990635) remain well corroborated.
• Avoided fallback: No "protein binding" fallback is invoked; the recommendation is specific.
• Bottom line: The seed action MARK_AS_OVER_ANNOTATED is supported and, if anything, conservative.

Mechanistic Scope

The immediate molecular function of Rad3 is a PIKK Ser/Thr protein kinase acting as the sensor apex of the DNA replication/damage checkpoint, functioning as the Rad3–Rad26 (ATR–ATRIP) complex. Its direct action occurs on chromatin at stalled replication forks and at telomeres, where it phosphorylates downstream effectors (Cds1, Chk1). Direct gene-product activity (the kinase acting on chromatin/forks/telomeres) is not in dispute. What is in dispute is whether Rad3 resides and is active in the nucleolus — and the available primary evidence does not place it there. The nucleolar phenomena in this pathway that are real (Clp1/Flp1 release from the nucleolus; mammalian ATR-dependent rRNA shut-down) are either downstream substrate relocalizations or stress-induced context-specific responses of orthologs, not constitutive Rad3 nucleolar residence. Any nucleolar/rDNA role (as inferred from mammalian ATR) would be a downstream, stress-inducible activity, not the primary function.

Conflicts and Alternatives

1. Substrate–kinase conflation (most likely). The Rad3 pathway has a bona fide nucleolar node — Clp1/Flp1 sequestration and checkpoint-dependent release (PMID:18385517, PMID:22918952). A curation step that linked "nucleolar checkpoint biology" to Rad3 rather than to its substrate would produce exactly this annotation. This alternative reinforces removal.
2. Reference mis-attribution / typo (strongest alternative, now evidenced). Full-text mining of PMID:18180284 found no Rad3 localization experiment and no mention of the nucleolus. The most parsimonious explanation is a mis-attributed reference or curation slip — the annotation may have been intended for a different paper, or the curator conflated Rad3 with another imaged protein. If confirmed, this favours REMOVE.
3. Genuine but context-specific nucleolar biology of ATR-family kinases (PMID:25916852, PMID:27391441). ATR is recruited to rDNA/nucleoli during nucleolar stress and Pol I inhibition, so the term is not biologically absurd — it argues for "non-core/context-specific" rather than "impossible," if any signal survives verification.
4. Systematic vs. targeted disagreement. The ORFeome screen (cytosol) and the single IDA (nucleolus) disagree; both are low-to-medium resolution for a diffuse nuclear kinase. Neither establishes the nucleolus as a preferred compartment.
5. No paralog confusion in S. pombe. Rad3 is the single-copy PIKK-ATR; Tel1 is the ATM ortholog, separately annotated. Cross-organism carry-over is also unlikely, since human ATR is itself not curated to nucleolus.

Limitations and Knowledge Gaps

1. Exact content of PMID:18180284 for Rad3. Checked: abstract and complete main text/legends/references via PMC HTML (PMC2258774) — "nucleolus" and "rDNA" absent; Rad3 never imaged. Remaining unknown: separately hosted supplemental figures could not be programmatically retrieved. This matters because it decides REMOVE vs. non-core. Resolve by opening the journal PDF + supplemental files and confirming no Rad3-nucleolus panel exists.
2. Independent S. pombe evidence for nucleolar Rad3. Checked PomBase CC set and ORFeome — none corroborate the nucleolus. A second independent localization (endogenous tag, high resolution, or under replication stress) would resolve whether a minor nucleolar pool is reproducible.
3. Functional relevance at rDNA in fission yeast. No S. pombe assay ties Rad3 catalytic function to rDNA/nucleolar processes. A Rad3-dependent phenotype at rDNA (fork stalling, rDNA stability, Pol I regulation) would upgrade the term from CC-only to functionally meaningful.
4. Negative-evidence weight of ortholog curation. ATR lacking a nucleolus CC term is suggestive but is absence-of-evidence, not evidence-of-absence.

Discriminating Tests

• Re-read PMID:18180284 localization figures/legends (main + supplemental) to confirm whether any Rad3 image panel exists and, if so, its marker, condition and quantitation.
• Endogenous high-resolution imaging of Rad3-GFP/mNeonGreen (native locus) with a nucleolar marker (Gar2/Nop1/fibrillarin) and rDNA FISH, ± replication stress (HU) / rDNA fork-block, to test for a reproducible minor nucleolar pool vs. general nucleoplasmic signal.
• rDNA-directed ChIP / CUT&RUN of Rad3–Rad26 to test physical association with rDNA repeats.
• Curation provenance audit: retrieve PomBase annotation id 433558 history and original curator note to verify the intended reference and figure.
• Functional test: rad3Δ (or kinase-dead) effect on rDNA stability / Pol I regulation to determine whether any nucleolar presence is functional.

Proposed Follow-up Actions / Curation Leads (require curator verification)

• Primary action lead: REMOVE GO:0005730 (nucleolus, IDA, PMID:18180284) — or, at minimum, flag the reference as not supporting the annotation and request re-curation. The seed's MARK_AS_OVER_ANNOTATED is fully supported as a floor.
• Decisive verifiable evidence: Full text of PMC2258774 mined — "nucleolus"/"nucleolar" = 0 occurrences; "rDNA"/"ribosom-" = 0; "Rad3" = 4 occurrences, all genetic/pathway, none localization; imaged proteins = Rad22-YFP, Rhp51, Mcm4, Rad52, Crb2, phospho-H2A (Rad3 absent). Curator: reproduce by searching the article PDF for "nucleol" and for any Rad3 image panel.
• Candidate supporting references / snippets to verify:
• PMID:16823372 (ORFeome) — systematic localization assigns rad3 to cytosol, not nucleolus.
• PMID:17531813 — "Cdc18 persists in a chromatin-bound complex including the checkpoint kinases Rad3 and Rad26." (core = chromatin).
• PMID:18385517 — "Active Cds1p checkpoint kinase is required to release Flp1p into the nucleus" (nucleolar biology here is a substrate event, not Rad3 residence).
• UniProt Q13535 — human ATR curated CC lacks a nucleolus term.
• PMID:25916852 / PMID:27391441 — ATR-family nucleolar activity is stress/context-specific (rationale for "non-core," not "impossible").
• Suggested curator questions: (1) Does PMID:18180284 actually contain any Rad3 localization panel (main or supplemental)? (2) If not, was the intended reference a different paper, and which? (3) Should PomBase annotation id 433558 be retracted or re-referenced?
• Suggested experiment (only if a genuine signal must be resolved): Endogenous Rad3-GFP + nucleolar marker (Gar2/fibrillarin) colocalization ± HU (see Discriminating Tests).

Bottom line

The seed action MARK_AS_OVER_ANNOTATED is supported, and full-text mining of the cited reference (PMC2258774) shows the evidence is weaker still: the paper never mentions the nucleolus and never localizes Rad3, so the annotation is most likely a mis-attributed reference / curation error and a candidate for REMOVE. Rad3's core compartments are the ATR–ATRIP complex, chromatin/stalled forks and telomeres. A curator should confirm against the journal PDF + supplemental figures before finalizing REMOVE vs. non-core.

Artifacts

• OpenScientist final report
• OpenScientist final report