PRG-1 (worm, UniProt P90786) — Adjudication of "Nuclear Localization and piRNA Processing"

Focus: function_assignment · Hypothesis slug: nuclear-localization-and-pirna-processing
Source file: genes/worm/prg-1/prg-1-ai-review.yaml · Selector: free-text
Organism: Caenorhabditis elegans (NCBITaxon:6239) · Date: 2026-09-21

Summary

The seed hypothesis bundles two claims that must be adjudicated independently, and when they are, they resolve in opposite directions. PRG-1 is the C. elegans Piwi Argonaute that binds 21U-RNAs (piRNAs). The hypothesis asserts (1) that PRG-1 localizes to the nucleus (GO:0005634) and (2) that it participates in piRNA processing (GO:0034587), and it explicitly asks the curator to distinguish perinuclear germ granules from intranuclear protein and to test whether PRG-1 contributes to precursor processing versus merely stabilizing mature piRNAs. The evidence answers both prompts cleanly.

The piRNA-processing clause is supported as non-catalytic participation. PRG-1 loads and stabilizes the untrimmed 21U pre-piRNA precursor and provides the platform on which 3′-end maturation occurs. The catalytic trimming step is performed by the ribonuclease PARN-1, with HENN-1 adding a protective 2′-O-methyl group — PRG-1 itself does not catalyze cleavage or trimming. Because GO conventions permit non-catalytic participation, GO:0034587 is defensible, but the experimentally grounded, worm-specific term GO:0034585 "21U-RNA metabolic process" (IMP) is the better-supported and more precise choice for this protein.

The nuclear-localization clause is over-annotated. GO:0005634 "nucleus" on PRG-1 rests only on a phylogenetic IBA inference propagated across the PIWI clade (GO_REF:0000033), with no experimental C. elegans backing. This is independently contradicted by UniProtKB's own curated subcellular-location comment for P90786, which lists Cytoplasm only (no Nucleus location or keyword), and by direct experimental IDA evidence placing PRG-1 in perinuclear P granules (GO:0043186), a cytoplasm-facing condensate docked on the outer nuclear envelope. The pathway's genuine nuclear silencing arm is executed downstream by the nuclear Argonautes HRDE-1/NRDE-3 loaded with 22G-RNAs — a pathway consequence, not PRG-1's own localization. The net recommendation is to split the hypothesis: keep the processing/participation claim (preferring the 21U term) and flag/demote the nucleus claim as IBA carry-over.

Key Findings

F001 — The nucleus annotation (GO:0005634) is IBA carry-over, not experimental fact

A live QuickGO annotation pull for P90786 (2026-09) shows that GO:0005634 "nucleus" is supported only by an IBA annotation — Inferred from Biological Ancestor (ECO:0000318; GO_Central; GO_REF:0000033). IBA is a phylogenetic inference: the term was propagated onto PRG-1 because other members of the PIWI Argonaute clade carry a nucleus annotation, not because anyone observed PRG-1 inside the C. elegans nucleoplasm. In the same annotation set, the cellular-component term GO:0043186 "P granule" carries direct experimental IDA (ECO:0000314) from two foundational papers (PMID: 18501605, PMID: 18571452), while the generic cytoplasm term (GO:0005737) is IEA.

The primary literature is unanimous that PRG-1's native compartment is perinuclear. A 2026 study of unloaded-Argonaute degradation states that "the failure to load piRNAs disrupts PRG-1 localization to perinuclear germ granules" (PMID: 41529195), naming the native compartment as perinuclear germ granules. A study of Piwi-mutant sterility observes that "sterile individuals consistently exhibit altered perinuclear germ granules" (PMID: 33658512), again centering PRG-1 function on perinuclear granules rather than the nucleoplasm.

Crucially, the seed hypothesis's own request — distinguish perinuclear germ granules from intranuclear protein — is exactly the test that dissolves the nucleus annotation. P granules dock on the cytoplasmic face of the nuclear envelope over germ-cell nuclei; their perinuclear position is a cytoplasmic-facing localization and is not evidence of intranuclear (nucleoplasmic) PRG-1.

F002 — PRG-1 participates in piRNA processing by loading/scaffolding precursors; PARN-1 catalyzes trimming

The mechanistic core of the processing clause is well established across three primary studies. Untrimmed pre-piRNAs — 21U precursors bearing 3′ extensions — are stably bound by PRG-1, and 3′-end maturation is carried out by the nuclease PARN-1, with HENN-1 methylation protecting the mature 3′ end. The foundational trimming study reports that "these longer piRNAs are stable and associate with the Piwi protein PRG-1 but fail to robustly recruit downstream silencing factors" (PMID: 26919432), showing that untrimmed precursors load onto PRG-1 while a separate activity performs the maturating trim. A later study confirms that "long isoforms of untrimmed piRNAs are preferentially modified in parn-1 mutant animals" (PMID: 34469728), establishing PARN-1 as the 3′-trimming nuclease acting on PRG-1-associated precursors. A cross-species synthesis states that "the ribonuclease PARN-1 and its orthologs mediate piRNA 3′ trimming in worms, insects, and mammals" (PMID: 38244197), assigning the catalytic trimming step to PARN-1, not PRG-1.

On the annotation side, QuickGO shows GO:0034587 "piRNA processing" for P90786 is IBA + IEA only, whereas the worm-specific term GO:0034585 "21U-RNA metabolic process" carries experimental IMP (ECO:0000315) from PMID: 18571452. PRG-1 is required for the stable accumulation of mature 21U-RNAs and provides the platform on which 3′ trimming occurs, but it does not itself catalyze the trimming/cleavage step. This directly answers the seed hypothesis's fork: PRG-1 binds the precursor (pre-piRNA), so its role exceeds passive stabilization of mature piRNAs — yet it is non-catalytic, so "participation" (not molecular-function catalysis) is the correct framing.

F003 — UniProt's curated location for P90786 is Cytoplasm only, corroborating the IBA-carry-over conclusion

Independent of the GO annotation set, the UniProtKB record for P90786 (824 aa) lists a subcellular-location comment of "Cytoplasm" (ECO:0000256) with a keyword set of {Cytoplasm, Developmental protein, Hydrolase, RNA-binding, RNA-mediated gene silencing}. There is no "Nucleus" subcellular-location annotation and no Nucleus keyword. The domain architecture is a PAZ domain (~220–331) and a PIWI domain (~499–810) with an N-terminal disordered region (~1–24). Because UniProt curators arrive at their location comment through a pipeline distinct from GO_Central phylogenetic propagation, their convergence on "Cytoplasm only" is an orthogonal line of evidence that the GO:0005634 nucleus annotation is a PIWI-clade IBA artifact rather than a reflection of PRG-1's actual localization.

Mechanistic Model / Interpretation

  21U pre-piRNA precursor (Pol II, capped, 3'-extended)
      │
      │  PETISCO / TOFU factors  ── 5' processing (largely upstream, PRG-1-independent)
      ▼
      ┌───────────────────────┐
      │  PRG-1 (Piwi)         │   ◄── LOADS + STABILIZES untrimmed precursor
      │  binds pre-piRNA      │       (scaffold / platform; NON-catalytic)
      └───────────┬───────────┘
  │  PARN-1 ribonuclease  ── CATALYZES 3' trimming
  ▼
      ┌───────────────────────┐
      │  Mature 21U-RNA on    │
      │  PRG-1  (piRISC)      │
      └───────────┬───────────┘
  │  HENN-1  ── 2'-O-methylation (protects 3' end)
  ▼
    Target scanning → 22G-RNA amplification (RdRP EGO-1)
    → NUCLEAR silencing by HRDE-1 / NRDE-3  ◄── downstream, NOT PRG-1

  LOCALIZATION:  Perinuclear P granules (cytoplasmic face of nuclear pores) — NOT nucleoplasm

Two conclusions follow. For the process (BP): PRG-1's contribution is precursor loading, scaffolding, and stabilization — not enzymatic maturation. The trimming enzyme is PARN-1; the methyltransferase is HENN-1. "Participates in piRNA processing" does not require catalysis, so GO:0034587 is annotatable, but the phenotype-grounded worm term GO:0034585 (IMP) is more informative. For the location (CC): PRG-1 resides in perinuclear P granules. "Perinuclear" describes granules docked on the cytoplasmic face of the nuclear envelope; it is not intranuclear. The nuclear transcriptional silencing that piRNAs ultimately trigger is executed downstream by HRDE-1/NRDE-3, so the pathway's nuclear activity must not be back-attributed to PRG-1 as a localization.

GO decision table

Avoid "protein binding" as a recommendation; the informative MF is 21U-RNA/piRNA binding (GO:0034583/0034584).

Evidence Base

Evidence Matrix

Narrative on the evidence

The processing evidence forms a coherent three-paper mechanistic chain (PMID: 26919432, PMID: 34469728, PMID: 38244197) that consistently places PRG-1 as the carrier/scaffold of the precursor and PARN-1 as the trimming nuclease. This is the strongest part of the evidence base and cleanly answers whether PRG-1 contributes to processing (it does — via precursor loading and stabilization) versus catalysis (it does not).

The localization evidence uniformly points away from the nucleoplasm. PMID: 41529195 and PMID: 33658512 both name perinuclear germ granules; UniProt lists cytoplasm only; the only experimentally supported CC GO term is P granule (IDA). The lone support for nucleus is a phylogenetic IBA inference. PMID: 41414669 adds that PRG-1 localization is robust even when N-terminal RG motifs are mutated, further underscoring that granule localization is the stable, defining feature of the protein and that downstream silencing (WAGO-siRNA production) is genetically separable from loading/localization.

Mechanistic Scope

Direct PRG-1 activity: sequence-specific binding of 21U-RNAs/pre-piRNAs (MF), assembling piRISC in perinuclear P granules (CC). PRG-1 loads the untrimmed precursor and stabilizes it, licensing 3′ maturation.
Adjacent, non-PRG-1 catalysis: 5′ processing by PETISCO/PICS + TOFU factors (upstream); 3′ trimming by PARN-1; 2′-O-methylation by HENN-1.
Downstream (not PRG-1 localization/activity): target licensing → 22G-RNA amplification (RdRP EGO-1) → nuclear silencing by HRDE-1/NRDE-3. Transposon silencing, fertility loss, transgenerational sterility, and longevity are loss-of-function phenotypes, not molecular functions of PRG-1.

Conflicts and Alternatives

• Paralog/clade carry-over. Some PIWI-clade proteins in other taxa are nuclear (e.g., mouse MIWI2/PIWIL4; Drosophila Piwi). The IBA "nucleus" call most plausibly propagates that clade property onto PRG-1, which is not experimentally nuclear in C. elegans — a classic over-annotation risk. (Used here as orientation for the carry-over argument; those cross-organism primary abstracts were not retrieved in this run and are not cited as retrieved evidence.)
• "Processing" vs "stabilizing only." The seed's fork resolves: PRG-1 binds the precursor (pre-piRNA), not only mature piRNA, so its role exceeds passive stabilization — but it does not catalyze trimming (PARN-1 does).
• Biogenesis is partitioned. PETISCO/TOFU (5′) act largely upstream/independent of PRG-1, so "piRNA processing" as a whole is not PRG-1-centric.

Limitations and Knowledge Gaps

1. Is there any bona-fide intranuclear PRG-1 pool? Checked: QuickGO evidence tiers + all retrieved localization papers → only perinuclear IDA found; no experimental nuclear evidence. Matters because it determines keep/remove for GO:0005634. Resolve with quantitative subcellular fractionation or super-resolution/immuno-EM/APEX imaging distinguishing nucleoplasm from granule.
2. Does PRG-1 loading constitute a required processing step, or merely precede it? Checked: PARN-1/HENN-1 papers show trimming on PRG-1-bound precursors. Resolve with in-vitro reconstitution testing whether trimming requires PRG-1 loading.
3. Term granularity. Whether curators prefer GO:0034585 (21U) vs GO:0034587 (piRNA) is a nomenclature/consistency decision, not a biology gap.
4. Provenance timing. GO evidence tiers reflect a live QuickGO pull (2026-09) and may lag the newest primary literature; a curator should re-pull at curation time. PMIDs 18501605/18571452 were confirmed via QuickGO annotation references rather than re-read in full; their figures should be re-inspected. UniProt location evidence for P90786 is ECO:0000256 (automatic).

Discriminating Tests

• Cell fractionation / immuno-EM of endogenously tagged PRG-1: quantify nucleoplasmic vs perinuclear-granule signal (directly tests GO:0005634).
• In-vitro pre-piRNA trimming ± PRG-1 loading with recombinant PARN-1: tests whether PRG-1 is a required processing scaffold vs a mere carrier.
• Small-RNA-seq in parn-1, henn-1, prg-1, and PETISCO/TOFU mutants: partitions PRG-1's contribution (loading/3′) from 5′ biogenesis.
• Curator audit of the GO_Central IBA reference tree for the PIWI clade to identify which ortholog(s) seeded the nucleus propagation and whether they have experimental nuclear support.

Proposed Follow-up Actions / Curation Leads (verify before applying)

Reference snippets to verify:
- PMID: 41529195 — "the failure to load piRNAs disrupts PRG-1 localization to perinuclear germ granules."
- PMID: 26919432 — untrimmed piRNAs "are stable and associate with the Piwi protein PRG-1 but fail to robustly recruit downstream silencing factors."
- PMID: 34469728 — "long isoforms of untrimmed piRNAs are preferentially modified in parn-1 mutant animals."
- PMID: 38244197 — "The ribonuclease PARN-1 and its orthologs mediate piRNA 3′ trimming in worms, insects, and mammals."
- PMID: 41414669 — RG motifs "are not required for PRG-1 expression, localization, or piRNA loading."

Candidate GO actions:
- Demote/remove GO:0005634 (nucleus, IBA-only) — do not treat as a core direct-localization annotation.
- Retain GO:0043186 (P granule, IDA) as the primary CC and GO:0005737 (cytoplasm).
- Prefer/promote GO:0034585 (21U-RNA metabolic process, IMP); if GO:0034587 (piRNA processing) is kept, frame it as non-catalytic participation (loading/scaffolding) with PARN-1 as the catalytic partner.
- Retain GO:0034583/0034584 (21U/piRNA binding) as the informative MF; do not fall back to "protein binding."

Suggested curator questions:
- Is any experimental nuclear-localization citation on file that would upgrade GO:0005634 beyond IBA?
- Should the review explicitly annotate PRG-1's processing role as precursor loading/scaffolding (non-catalytic) with PARN-1 named as the catalytic partner?

Suggested experiments: fractionation/imaging for a nuclear pool; in-vitro trimming ± PRG-1 loading; comparative small-RNA-seq across the maturation-pathway mutants above.

Bottom line

Split the hypothesis. Keep the piRNA-processing participation (best expressed as the experimentally grounded 21U-RNA metabolic process term, with PRG-1 as non-catalytic scaffold/stabilizer and PARN-1 as the trimming nuclease). Flag/demote the nucleus localization as PIWI-clade IBA carry-over that conflicts with direct IDA P-granule evidence and UniProt's cytoplasm-only curated location. Perinuclear germ granules are cytoplasm-facing — they are not the nucleus.

Artifacts

• OpenScientist final report
• OpenScientist final report
• OpenScientist prg1 current go annotations
• OpenScientist prg1 evidence matrix
• OpenScientist prg1 go decision table