AIGR Deep Research Report: ATG14 Phagosomal Localization and Post-transcriptional Regulation

Gene: ATG14 (human, UniProt Q6ZNE5; mouse ortholog Q8CDJ3)
Focus type: function_assignment
Hypothesis slug: phagosomal-localization-and-posttranscriptional-regulation
Source file: genes/human/ATG14/ATG14-ai-review.yaml

Summary

The seed hypothesis packages two independent GO assignments for human ATG14 that this focused review adjudicates separately, as instructed. On tracing each annotation back to its primary IDA/IMP source, both are over-annotations and neither is supported as a direct function of the ATG14 gene product. This is the inverted-pyramid answer: remove both, retain the correct alternative already present.

The phagocytic-vesicle localization (GO:0045335) originates as a mouse MGI IDA on Q8CDJ3, cited to PMID:22493499(https://pubmed.ncbi.nlm.nih.gov/22493499/). That paper studies Slamf1 recruiting a Beclin-1/Vps34/UVRAG complex to E. coli phagosomes and explicitly reports that Atg14L was excluded from the complex — a negative co-immunoprecipitation, not a positive localization. The annotation therefore inverts the polarity of its own source. It then propagates to human Q6ZNE5 as an IEA by Ensembl orthology.

The post-transcriptional-regulation term (GO:0010608), a human IMP by ParkinsonsUK-UCL cited to PMID:23878393(https://pubmed.ncbi.nlm.nih.gov/23878393/), rests on ATG14 controlling beclin-1 Ser90/Ser93 phosphorylation and BECN1-complex stability — strictly post-translational events. GO:0010608 requires modulation of an RNA-level step (RNA processing, mRNA stability, or translation) after transcript production, and excludes protein-level proteostasis. The same paper already yielded the correct term, GO:0061635 (regulation of protein complex stability). ATG14 has no RNA-binding domain, and the only paper linking it to post-transcriptional machinery (PMID:36097301(https://pubmed.ncbi.nlm.nih.gov/36097301/)) shows ATG14 mRNA is a target of m6A/YTHDF1 translation control, not an effector. The directly demonstrated ATG14 roles in autophagosome–endolysosome fusion, PI3KC3 complex-I scaffolding, complex stability, and mitophagy are undisputed and should be retained.

Executive Judgment

Verdict: Over-annotated (refuted for both sub-claims as stated).

1. GO:0045335 (phagocytic vesicle) — refuted as a positive annotation. The sole IDA source (PMID:22493499) provides no positive localization data placing ATG14 on a phagosome. It reports a negative interaction: Atg14L was excluded from the Slamf1/Beclin-1/Vps34/UVRAG phagosomal complex. The seed's framing — "distinguish exclusion from one interaction from exclusion from every phagosome" — does not rescue the term, because the source supplies no positive phagosomal evidence of any kind. The correct curatorial stance is that the term lacks support, not that localization is affirmatively disproven.

2. GO:0010608 (post-transcriptional regulation of gene expression) — refuted as a term misassignment. The IMP source (PMID:23878393) demonstrates post-translational control (beclin-1 phosphorylation, complex stability), which fails the term's RNA-step requirement. This is best read as a "post-translational → post-transcriptional" conflation. GO:0061635, already annotated from the same paper, correctly captures the biology and should be retained in its place.

Most important caveats. (a) The corroborating statement that the LAP/phagosomal PI3K complex "lacks ATG14" comes from PMID:26098576(https://pubmed.ncbi.nlm.nih.gov/26098576/), retracted in 2024 (PMID:38438803(https://pubmed.ncbi.nlm.nih.gov/38438803/)); it is used only as orientation, and the refutation stands on the primary IDA/IMP sources without it. (b) Full-text figure panels of PMID:23878393 were not programmatically retrievable; the adjudication uses the abstract-level mechanism, which is unambiguously post-translational. (c) Nothing here disputes ATG14's established mitophagy and complex-stability roles.

Key Findings

Finding 1 — The phagocytic-vesicle (GO:0045335) annotation is contradicted by its own IDA source

Mouse Atg14 (Q8CDJ3) carries GO:0045335 (phagocytic vesicle) with evidence IDA / located_in, sourced to PMID:22493499(https://pubmed.ncbi.nlm.nih.gov/22493499/) (Ma et al. 2012, J. Biol. Chem.), per QuickGO annotation retrieval (n=70 annotations for the accession). Human Q6ZNE5 does not independently carry GO:0045335 in QuickGO — it inherits the risk only by orthology.

The paper studies Slamf1 recruiting a Beclin-1/Vps34/UVRAG complex to E. coli-containing phagosomes to regulate membrane fusion and NADPH oxidase 2 (NOX2) activity. The decisive sentence is a negative result for Atg14L:

"Because Slamf1 did not interact with Atg14L or Rubicon, which can also form a complex with Vps34 and Beclin-1, we conclude that Slamf1 recruits a subset of Vps34-associated proteins, which is involved in membrane fusion and NOX2 regulation." — PMID:22493499(https://pubmed.ncbi.nlm.nih.gov/22493499/)

This is the only reference behind the IDA. It contains no immunofluorescence, fractionation, or other assay placing Atg14 on a phagocytic vesicle; it reports the opposite — Atg14L was excluded from the phagosome-recruited Vps34 subcomplex. An IDA located_in phagocytic vesicle built on a negative co-IP is a probable curation error and should be removed.

Independent orientation: LC3-associated phagocytosis (LAP) uses a Class III PI3K complex "containing UVRAG but lacking ATG14" (PMID:26098576(https://pubmed.ncbi.nlm.nih.gov/26098576/)). This paper was retracted in 2024 (PMID:38438803(https://pubmed.ncbi.nlm.nih.gov/38438803/)) and is cited here only as field-orientation, not primary evidence. The distinction between the ATG14-containing autophagy-initiation complex and the Rubicon/UVRAG-containing phagosomal complex is a broadly accepted feature of the field and does not depend on the retracted paper.

Finding 2 — The GO:0010608 annotation conflates post-translational with post-transcriptional control

Human ATG14 (Q6ZNE5) carries GO:0010608 (post-transcriptional regulation of gene expression) with evidence IMP / involved_in, sourced to PMID:23878393(https://pubmed.ncbi.nlm.nih.gov/23878393/) (Fogel et al. 2013).

The ontology definitions are decisive:
- GO:0010608 = "Any process that modulates the frequency, rate or extent of gene expression after the production of an RNA transcript."
- GO:0010467 (gene expression) = "the process in which a gene's sequence is converted into a mature gene product (protein or RNA) … production of an RNA transcript and its processing, as well as translation and maturation."

GO:0010608 therefore requires a step acting on RNA processing, mRNA stability, or translation, and excludes post-translational proteostasis (protein modification, complex assembly/stability).

Fogel et al. describe exactly a post-translational mechanism:

"we demonstrated that human Atg14 is critical in controlling an autophagy-dependent phosphorylation of beclin-1. We map these novel phosphorylation sites to serines 90 and 93 and demonstrate that phosphorylation at these sites is necessary for maximal autophagy." — PMID:23878393(https://pubmed.ncbi.nlm.nih.gov/23878393/)

Phosphorylating an existing beclin-1 protein and stabilizing the BECN1/Vps34 complex are protein-level events. Critically, the same paper already generated GO:0061635 (regulation of protein complex stability) as an IMP for Q6ZNE5 — that term correctly captures the finding. GO:0010608 appears to be a term-selection error (a likely "post-translational → post-transcriptional" slip), since no RNA-level or translational assay is described.

Finding 3 — ATG14 is a target, not an effector, of post-transcriptional regulation, and has no RNA-binding domain

A directed literature search for ATG14 acting on an RNA/translation step returned a single relevant primary paper — PMID:36097301(https://pubmed.ncbi.nlm.nih.gov/36097301/) (Huang et al. 2022) — reporting the opposite direction of causality:

"YTHDF1 promoted the translation of autophagy-related genes ATG14, thus boosting autophagy." — PMID:36097301(https://pubmed.ncbi.nlm.nih.gov/36097301/)

Here ATG14 mRNA is the substrate of m6A/YTHDF1 post-transcriptional machinery; ATG14 protein is not the regulator of any RNA step. This is precisely the inverse of what GO:0010608 would assert for ATG14.

Structurally, InterPro for Q6ZNE5 returns only the ATG14/VPS38 autophagy family (IPR018791; Pfam PF10186, "Vacuolar sorting 38 and autophagy-related subunit 14"; PANTHER PTHR13664). There is no RNA-binding, nucleic-acid-binding, or ribonucleoprotein domain, so there is no structural basis for a direct post-transcriptional effector activity.

Finding 4 — Both disputed annotations are root manual annotations amplified cross-species by orthology

QuickGO provenance tracing shows how each error propagates:

Each erroneous root is multiplied by electronic/phylogenetic transfer across species. Curation must fix the root (MGI for GO:0045335; ParkinsonsUK-UCL for GO:0010608); correcting only downstream IEA/ISO projections would leave the source error to re-propagate.

Mechanistic Model / Interpretation

Both annotations fail for structurally analogous reasons — each attaches a GO term whose scope or polarity does not match the underlying experiment:

GO:0045335  (phagocytic vesicle, IDA)
   source = PMID:22493499
   experiment: Slamf1 pulls down Beclin-1 / Vps34 / UVRAG at phagosomes
   result for ATG14L: NO interaction  →  EXCLUDED from the complex
   annotation asserts: ATG14 located_in phagocytic vesicle
   ──►  POLARITY ERROR: absence recorded as presence

GO:0010608  (post-transcriptional regulation of gene expression, IMP)
   source = PMID:23878393
   experiment: ATG14 controls beclin-1 Ser90/93 phosphorylation
               + BECN1 complex stability
   level of action: PROTEIN (post-translational)
   term requires:   RNA / translation step (post-transcriptional)
   ──►  SCOPE ERROR: post-translational recorded as post-transcriptional
        (correct term GO:0061635 already exists from the same paper)

Where ATG14 genuinely acts. ATG14 (Barkor/ATG14L) is the autophagy-specific targeting subunit of the Class III PI3K (PI3KC3) complex I, partnering Beclin-1, Vps34 (PIK3C3), and Vps15. Its directly demonstrated functions include promoting membrane tethering and SNARE-mediated autophagosome–endolysosome fusion via STX17–SNAP29–VAMP8 (PMID:25686604(https://pubmed.ncbi.nlm.nih.gov/25686604/)); scaffolding and stabilizing the BECN1/Vps34 complex and enabling beclin-1 phosphorylation (PMID:23878393(https://pubmed.ncbi.nlm.nih.gov/23878393/)); and supporting NRBF2-modulated Atg14L-linked Vps34 activity (PMID:24849286(https://pubmed.ncbi.nlm.nih.gov/24849286/)). These undisputed roles are captured by autophagy-initiation, PI3K-complex, and complex-stability terms — not by phagocytic-vesicle localization or post-transcriptional regulation.

The key curatorial distinction is between the ATG14-containing PI3KC3 complex I (canonical autophagy initiation) and the Rubicon/UVRAG-containing complex operating on phagosomes during LAP. ATG14 belongs to the former; PMID:22493499 is a direct experimental demonstration of that partition.

Evidence Base

GO Curation Implications (leads — require curator verification)

Both removals should occur at the root annotation so the derived cross-species copies stop being generated. Avoid "protein binding" as a substitute; GO:0061635 is the informative, evidence-matched term.

Mechanistic Scope

• Direct molecular functions tested: (a) whether ATG14 physically resides on phagosomal membranes, and (b) whether ATG14 acts at an RNA/translational step of gene expression.
• What the data show: ATG14 is a subunit of the autophagy-specific Class III PI3K complex I. Its beclin-1 regulation is post-translational (phosphorylation, complex stabilization). The phagosomal PI3P/NOX2 activity in PMID:22493499 is driven by a different, ATG14-lacking (UVRAG-containing) complex.
• Downstream vs direct: Phagolysosomal maturation, NOX2 activity, and "gene-expression"-flavored autophagy phenotypes are downstream/indirect; they are not attributable to ATG14 localizing to phagosomes or to an ATG14 RNA-level activity.

Conflicts and Alternatives

• Curation carry-over / polarity error, not a real organism difference: the phagosomal localization belongs to the UVRAG/Rubicon complex, and the IDA reads a negative co-IP as positive localization.
• Scope/lexical conflation: the IMP "post-transcriptional" label most plausibly conflates "post-translational" with "post-transcriptional."
• Paralog/complex confusion: LAP and Slamf1-phagosome PI3K activity use UVRAG/Rubicon complexes explicitly without ATG14 — the opposite subunit branch.
• Direction reversal: the only literature linking ATG14 to an RNA/translation step (PMID:36097301) makes ATG14 mRNA a target of m6A/YTHDF1, not an effector — reinforcing removal of GO:0010608.
• Retraction caveat: PMID:26098576 (ATG14-independent phagosomal PI3K) is retracted and used only as corroborating context; the adjudication rests on the primary IDA/IMP sources and ontology definitions.

Limitations and Knowledge Gaps

1. Full-text of PMID:23878393 not programmatically retrievable. Only front matter/abstract was accessible via NCBI efetch (PMC3753860). Why it matters: a hidden RNA-level assay could theoretically justify GO:0010608. Resolution: manual full-text read of figures. Risk is low because the described ATG14 activity (regulating an existing protein's phosphorylation/stability) is post-translational regardless of any transcript readout.
2. Negative-space certainty for phagosome localization. Absence of positive localization in one paper does not logically prove ATG14 is never on any phagocytic vesicle. Why it matters: a curator might consider a generalized term. Resolution: a targeted search found no positive ATG14 phagosome-localization report; the correct stance is that the term lacks support.
3. Database-level provenance/domain evidence. QuickGO/InterPro results are database evidence, appropriate for tracing propagation and domain content but not primary experiment.
4. Retraction of PMID:26098576 removes one corroborating line; the core refutation is unaffected.

Discriminating Tests

• Read PMID:23878393 full text/figures to confirm no assay measures ATG14 acting on ATG mRNA levels, transcript stability, or translation. If none, GO:0010608 removal is definitive.
• Re-examine PMID:22493499 figures for any imaging/fractionation placing Atg14L on phagosomes independent of the Slamf1 co-IP. If none, GO:0045335 removal is definitive.
• Direct localization assay: immunofluorescence / immuno-EM / phagosome fractionation for endogenous ATG14 in macrophages engulfing beads/bacteria, versus UVRAG/Rubicon controls (expected: ATG14 absent).
• ATG14-KO vs Rubicon-KO effect on phagosomal LC3/PI3P (LAP should be ATG14-independent).
• RNA-interaction screen (CLIP/RIP/eCLIP) for ATG14 protein to test for any RNA-binding (expected negative, consistent with absence of an RNA-binding domain).
• qRT-PCR vs Western for BECN1 plus reporter/translation assays upon ATG14 perturbation to formally exclude any RNA/translation-level step.

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

Candidate reference snippets to verify:
- PMID:22493499 — "Because Slamf1 did not interact with Atg14L or Rubicon … we conclude that Slamf1 recruits a subset of Vps34-associated proteins." → supports removing GO:0045335 for ATG14.
- PMID:23878393 — "human Atg14 is critical in controlling an autophagy-dependent phosphorylation of beclin-1 … serines 90 and 93." → supports removing GO:0010608 and retaining GO:0061635; consider GO:0031647.
- PMID:36097301 — "YTHDF1 promoted the translation of autophagy-related genes ATG14." → ATG14 is a target, not an effector, of post-transcriptional control.

Suggested curator questions:
1. Is the GO:0045335 IDA a mis-transfer of the complex's phagosomal localization to ATG14, given the source reports ATG14 exclusion?
2. Is GO:0010608 a lexical conflation of "post-translational" with "post-transcriptional"?
3. Should GO:0061635 be treated as the canonical replacement capturing the PMID:23878393 finding?

Act at the root annotation: GO:0045335 originates from the MGI IDA on mouse Q8CDJ3; GO:0010608 originates from the ParkinsonsUK-UCL IMP on human Q6ZNE5. Each propagates to the other species by Ensembl IEA / GO_Central ISO, so correcting the root removes the derived copies. Consider contacting MGI and the ParkinsonsUK-UCL group.

Do not disturb: the directly demonstrated ATG14 roles in autophagosome–endolysosome fusion (PMID:25686604), PI3KC3 complex-I scaffolding/Vps34 regulation (PMID:24849286), BECN1-complex stability (PMID:23878393 → GO:0061635), and mitophagy remain well supported and should be retained.

Bottom-Line Answer

For human ATG14 (Q6ZNE5), the phagocytic-vesicle localization (GO:0045335) and post-transcriptional-regulation (GO:0010608) assignments are not supported as positive functions and are recommended for removal at their root annotations. The localization IDA (PMID:22493499) documents ATG14's exclusion from the Slamf1/Beclin-1/Vps34/UVRAG phagosomal complex; the GO:0010608 IMP (PMID:23878393) describes post-translational control of beclin-1 phosphorylation/complex stability that fails the term definition. Retain GO:0061635 (regulation of protein complex stability), consider GO:0031647 as a lead, and leave the established mitophagy/complex-stability roles undisturbed.

Artifacts

• OpenScientist final report
• OpenScientist final report