AIGR Gene Hypothesis Deep Research — Paramecium tetraurelia A0BFB4

Focus type: function_assignment
Hypothesis slug: autophagy-ancestry-versus-hmm-family-placement
Target: A0BFB4 (GSPATT00028266001), 381 aa, Paramecium tetraurelia (NCBITaxon:5888)
Source file: genes/PARTE/A0BFB4/A0BFB4-ai-review.yaml

Summary

The seed hypothesis asks whether the four autophagy IBA terms propagated into A0BFB4 through PANTHER family PTHR24348 reflect genuine ATG1/ULK ancestry or a misplacement created by the competing live UniProt/InterPro HMM assignment to the generic family PTHR44167:SF18. The verdict is partially supported, with a clean split: the ancestry is real and the molecular-function terms should be retained, but the propagated autophagy process/location terms are over-annotations that should be removed or flagged uncertain.

The apparent family-placement conflict dissolves on inspection. PTHR24348 is officially named "Serine/threonine-protein kinase UNC-51-RELATED" — the ATG1/ULK autophagy-initiating kinase family (UNC-51 is the C. elegans ATG1 ortholog). An independent Smith-Waterman local alignment of A0BFB4's kinase domain ranks human ULK2 (35.3% identity) and ULK1 (34.5%) as its closest characterized references, above the STE20-family kinases LOK/SLK and above CHEK2. The "competing" best-hit family PTHR44167:SF18 is a functionally uncharacterized "Serine/Threonine Protein Kinase" catch-all with no assigned GO. So the two families are not truly competing: one captures the evolutionary clade, the other is a scoring artifact carrying no function. A0BFB4 genuinely belongs to the ATG1/ULK clade by ancestry, and this ancestry is not a misplacement.

Ancestry, however, does not license the pathway-level terms. A0BFB4 is a bare 381-aa kinase catalytic domain. A comparative-genomics study that explicitly includes Paramecium tetraurelia (PMID: 28123910) shows ciliates do not encode a typical Atg1: every candidate kinase lacks the Atg1-specific C-terminal domain required to build the Atg1 initiation complex, and the complex partners (Atg13, Atg17/FIP200, Atg101) are absent from ciliate genomes. An AlphaFold structural check (AF-A0BFB4-F1) confirms the missing module is genuinely absent — the only confidently folded non-kinase element is N-terminal, while the C-terminus is a short, partly disordered 33-residue tail. Because the machinery that nucleates a phagophore is absent in the ciliate lineage and in this protein specifically, GO:0000045, GO:0000407, GO:0005776, and GO:0010506 rest solely on phylogenetic IBA inference and cannot be executed by the canonical mechanism. The molecular-function terms (protein Ser/Thr kinase activity, ATP binding) are strongly supported and should be kept.

Executive Judgment

Verdict: Partially supported. Family placement is correct (ATG1/ULK ancestry confirmed by two independent lines); MF terms supported; the four IBA-propagated autophagy BP/CC terms are over-annotations (weakly supported) for this ciliate protein.

Reasoning:

1. No family misplacement. The PAINT/TreeGrafter graft is into PTHR24348 = "Ser/Thr kinase UNC-51-RELATED" = the ATG1/ULK family. Top sequence identity is to ULK1/ULK2 (~35%), members of that same family. The graft reflects genuine ancestry.
2. The "competing" HMM is not competing. PTHR44167:SF18 is a generic, functionally uncharacterized kinase family with no specific GO. The PTHR24348-vs-PTHR44167 discrepancy is a best-hit-vs-graft methodology artifact, not evidence of an alternative function.
3. Target-relevant biology weakens the specific autophagy function. Ciliates (including Paramecium tetraurelia) lack a typical Atg1: kinase candidates lack the Atg1 C-terminal domain, and the Atg1-complex partners are absent. A0BFB4 fits this class exactly and, per AlphaFold, has no C-terminal partner-binding module. The process/location terms cannot be executed by the canonical mechanism.

Most important caveats: (a) the Atg1-absence evidence is in-silico and does not fully exclude a divergent/non-canonical role; (b) ~35% kinase-domain identity places the protein in the ATG1/ULK clade but is not proof of one-to-one orthology; (c) there are no experimental data on A0BFB4 itself (localization, substrates, knockdown phenotype).

Key Findings

Finding 1 — The autophagy IBA rests on genuine ATG1/UNC-51 ancestry, not a misplacement

The four autophagy IBA terms enter A0BFB4 via a PAINT/TreeGrafter graft into PANTHER family PTHR24348, officially "Serine/threonine-protein kinase UNC-51-RELATED." UNC-51 is the C. elegans ATG1 ortholog; this family is the ATG1/ULK clade of autophagy-initiating kinases. The seed hypothesis feared that the live UniProt/InterPro HMM best-hit — PTHR44167:SF18 (match 70–348, reported score 7.6e-61) — represented a competing, contradictory placement. It does not: PTHR44167 carries the current InterPro name "Serine/Threonine Protein Kinase," a functionally uncharacterized catch-all family with no specific GO or function assigned. A best-single-HMM assignment to a generic family is a scoring artifact, not evidence of a distinct biological function.

An independent sequence-identity check corroborates the ATG1/ULK reading. A Smith-Waterman local alignment (BLOSUM62, gap -11/-1) of A0BFB4's kinase domain against a panel of characterized human kinases ranked:

Two independent lines — the PANTHER tree graft and top-hit sequence identity — agree on ATG1/ULK ancestry. The PTHR44167 assignment is a generic best-HMM artifact and does not refute the autophagy-family placement. This reconciles the discrepancy in the seed hypothesis in favor of "confirmed ancestry, not misplacement."

Finding 2 — A0BFB4 is an intact, catalytically active single-domain Ser/Thr kinase (MF terms strongly supported)

A0BFB4 (381 aa) carries only protein-kinase signatures: Pfam PF00069, InterPro IPR000719 (protein kinase domain), IPR017441 (ATP-binding site), IPR008271 (Ser/Thr active site); SMART S_TKc; PROSITE PS00107/PS00108/PS50011. All canonical catalytic residues are present and intact:

• Glycine-rich P-loop: GEGTFG
• VAIK β3 lysine: CKV
• HRD catalytic loop: HRDIK
• DFG Mg²⁺-binding motif: present
• APE motif: present

This architecture predicts a catalytically active (non-pseudo) kinase and directly supports GO:0004674 (protein serine/threonine kinase activity) and GO:0005524 (ATP binding). Critically, the protein is a bare catalytic domain: it lacks the ULK/ATG1 C-terminal (CTD/MIT) module that binds ATG13/FIP200/ATG101 and targets the kinase to the phagophore. The molecular-function annotations are robust; the CC and BP autophagy terms depend on machinery the protein does not carry.

Finding 3 — Comparative genomics: Paramecium and all ciliates lack a typical Atg1

Aslan, Küçükoğlu & Arslanyolu (2017, PMID: 28123910; open access PMC5244887) surveyed ATG proteins across five ciliate genomes, explicitly including Paramecium tetraurelia (plus Tetrahymena thermophila, Ichthyophthirius multifiliis, Oxytricha trifallax, Stylonychia lemnae). Two abstract statements directly govern this curation:

"ciliate genomes do not encode typical Atg1 since all the candidate sequences lack an Atg1-specific C-terminal domain which is essential for Atg1 complex formation"

"Consistent with the absence of Atg1, ciliates also lack other members of the Atg1 complex"

The full text adds that "these kinase domain containing proteins should not be considered as Atg1 candidates without further experimental analyses, since they all lack the important C-terminal domains," and that "Other components of this kinase complex (Atg1, Atg13, Atg17, Atg29 and Atg31) are also absent in ciliates." Canonical Atg1/ULK1 has three regions (N-terminal kinase domain, central Pro/Ser-rich region, C-terminal domain); the C-terminal domain binds Atg13/Atg17(FIP200)/Atg101 and is essential for the complex that induces autophagy. A0BFB4 is exactly the kinase-domain-only class the authors say should NOT be counted as Atg1 without experiments. The authors further propose that ciliate autophagosome formation may proceed via Atg6 (Beclin/Vps30) in an Atg1-independent manner — meaning even where autophagy occurs in ciliates, this kinase need not be the initiator.

Finding 4 — AlphaFold confirms no C-terminal ATG1 partner-binding domain

The AlphaFold DB model AF-A0BFB4-F1 (v6) provides a residue-level structural test. Segmenting by the HMM kinase match (residues 70–348):

The only confidently structured non-kinase element is N-terminal, whereas the ATG1/ULK1 partner-binding module (recruiting ATG13/ATG17-FIP200/ATG101) is a C-terminal domain. A0BFB4 has no such C-terminal domain — its C-terminus is a 33-residue low-confidence tail — and the N-terminal folded region matches no InterPro signature, so its function is unresolved. This independently clinches the over-annotation call: the protein structurally lacks the module that would assemble the Atg1 initiation complex.

Mechanistic Model / Interpretation

   ANCESTRY (real)                          EXECUTABLE FUNCTION (not supported)
   ───────────────                          ──────────────────────────────────
   A0BFB4 kinase domain                      Autophagy initiation requires:
     │  ~35% id to ULK1/ULK2                   • Atg1 C-terminal domain  ── ABSENT
     │  grafts into PTHR24348                  • Atg13 partner           ── ABSENT in ciliates
     ▼  (UNC-51/ATG1 family)                   • Atg17/FIP200            ── ABSENT in ciliates
   ATG1/ULK-family Ser/Thr kinase             • Atg101                  ── ABSENT in ciliates
     │                                                 │
     │  intact catalytic motifs                        ▼
     ▼  (P-loop, VAIK, HRD, DFG, APE)          No phagophore-nucleation complex
   ACTIVE kinase  ✔ MF terms                   ✘ BP/CC autophagy terms cannot execute
     │
     │  bare catalytic domain
     ▼
   No C-terminal partner-binding module (AlphaFold: C-term = 33-aa disordered tail)

How the discrepancy is reconciled. The seed hypothesis framed PTHR24348 (autophagy) vs PTHR44167:SF18 (generic kinase) as competing placements. They are not truly competing: PTHR24348 captures the evolutionary clade (ATG1/ULK), and PTHR44167 is a scoring catch-all with no function. So the family-placement conflict dissolves — A0BFB4 really is an ATG1/ULK-family kinase by ancestry and by top sequence identity.

Why ancestry does not license the process/location terms. Membership in the ATG1/ULK clade by kinase-domain homology is necessary but not sufficient to perform autophagosome initiation. That function is delivered by the C-terminal scaffolding domain and its binding partners, all of which are absent in ciliates generally and in A0BFB4 specifically (by architecture and by AlphaFold structure). IBA propagation transfers terms based on the phylogenetic node, but the transfer overreaches because it ignores the loss of the accessory module in the ciliate lineage. This is a textbook case where evolutionary placement is correct but functional propagation is too strong.

The seed guidance cautions that "family labels, missing animal accessory domains, absent target experiments and short donor lists alone are not evidence of loss." The over-annotation call here does not rest on any single such heuristic. It rests on convergent, target-relevant evidence: (1) a dedicated ciliate-wide comparative-genomics study that names Paramecium tetraurelia and states the C-terminal domain and entire Atg1 complex are absent; (2) an AlphaFold structure showing the missing module is genuinely absent, not merely HMM-undetected; and (3) the mechanistic requirement that the Atg1 complex — not a lone kinase domain — nucleates the phagophore. Together these exceed a "missing accessory domain" argument.

Evidence Base / Evidence Matrix

GO Curation Implications (leads — require curator verification)

Overall, the MF annotations should be retained — and are more informative than a bare "protein binding" placeholder — while the four IBA-propagated autophagy BP/CC terms should be treated as over-annotations pending experimental support, and are candidates for removal or an uncertainty qualifier. If the framework prefers not to delete IBA terms, the minimum defensible action is a caveat noting lineage-specific loss of the accessory module, citing PMID 28123910 and the AlphaFold structure. These are leads, not final GO actions.

Mechanistic Scope

• Directly testable / supported: A0BFB4 is an intact, likely catalytically active protein serine/threonine kinase (ATP binding + phosphotransfer). This is the immediate molecular activity and is well supported by domain architecture.
• Downstream / inferred only: "autophagosome assembly," "phagophore assembly site," "autophagosome," and "regulation of autophagy" are pathway-level roles that, in yeast/mammals, depend on ATG1/ULK1 assembling with Atg13–Atg17(FIP200)–Atg101 via its C-terminal domain. A0BFB4 lacks that domain and ciliates lack those partners; hence these are inferred pathway consequences of the ancestral family, not established activities of this gene product.
• Localization (phagophore/autophagosome) is a consequence of targeting via the missing module — not supported.

No Paramecium-specific experiment (substrate, localization, mutant phenotype) exists for A0BFB4, so all process/location claims are inference only.

Conflicts and Alternatives

• Database carry-over / method artifact. The autophagy terms are IBA propagations from the ATG1/ULK ancestral node; the live HMM lands on a generic family. Both are automated outputs; neither is experimental.
• Paralog / expansion context. Paramecium tetraurelia underwent successive whole-genome duplications, producing large kinase paralog sets; A0BFB4 may be one of several divergent ATG1-like kinase-domain paralogs, further weakening one-to-one ATG1 orthology.
• Alternative pathway. Aslan et al. propose ciliate autophagosome formation may proceed via Atg6/Beclin (Vps30) in an Atg1-independent manner — consistent with Tetrahymena programmed nuclear death being VPS34/ATG8-driven (PMID: 24280724) rather than ATG1-driven.
• Not a competing function. The generic PTHR44167 framing is not supported by the current family name; there is no positive evidence for a distinct alternative function (e.g., DNA-damage checkpoint) to replace autophagy.
• Genuine tension. Ciliates DO have autophagy machinery (programmed nuclear death; PMID 24280724, 21624459) — the pathway is not absent wholesale. But the specific Atg1-complex-mediated initiation step that these GO terms describe is the part shown to be absent.

Limitations and Knowledge Gaps

Discriminating Tests

1. RNAi/knockdown of A0BFB4 in Paramecium + autophagy readout (ATG8 lipidation/puncta; starvation- or autogamy-induced macronuclear autophagy) — directly tests any autophagy role.
2. In-vitro kinase assay on recombinant A0BFB4 to confirm catalytic activity (validates MF) and, with candidate substrates, probe the pathway.
3. Co-IP / proximity labeling for Atg13/FIP200/Atg101 homologs — tests whether an Atg1 complex can form (predicted negative given their absence).
4. GFP localization under starvation/conjugation — tests "phagophore assembly site"/"autophagosome" CC terms.
5. Phosphoproteomics comparing wild-type vs knockdown to identify substrates and infer the real pathway.
6. Phylogenetic placement / Foldseek-DALI with a curated ATG1/ULK vs CAMK-group alignment across ciliates and model eukaryotes, scoring the C-terminal-domain presence/absence character; structural search of the unassigned N-terminal element.

Curation Leads (require curator verification)

• Candidate reference to add: PMID: 28123910 (Aslan, Küçükoğlu & Arslanyolu, PeerJ 2017). Verify snippets: "ciliate genomes do not encode typical Atg1 since all the candidate sequences lack an Atg1-specific C-terminal domain which is essential for Atg1 complex formation" and "Consistent with the absence of Atg1, ciliates also lack other members of the Atg1 complex." Full text (PMC5244887) lists Paramecium tetraurelia among five genomes and states these kinase-domain proteins "should not be considered as Atg1 candidates without further experimental analyses."
• Supporting reference: PMID: 24280724 — verify: "Programmed nuclear death (PND) in the ciliate protozoan Tetrahymena thermophila is a novel type of autophagy" and "the absence of some ATG genes in this organism."
• Structural provenance: AlphaFold AF-A0BFB4-F1 (v6) — C-terminal region (349–381) mean pLDDT 64.2, no C-terminal partner-binding domain; document as evidence the Atg1 accessory module is genuinely absent.
• Candidate action changes: Retain MF (GO:0004674, GO:0005524). Flag/remove or add uncertainty to BP GO:0000045, GO:0010506 and CC GO:0000407, GO:0005776 as ATG1-complex-dependent functions unsupported in ciliates.
• Suggested curator questions: (i) Should IBA autophagy terms be down-weighted when the target lineage demonstrably lacks the required complex? (ii) Is a generic "protein phosphorylation" BP a safer core than "autophagosome assembly"? (iii) Should the MF terms be re-sourced to sequence/architecture evidence (ISS/ISM) rather than remain tied to the same IBA graft?
• Suggested experiments: the six Discriminating Tests above.

Proposed Follow-up Experiments / Actions

1. For the curator (immediate): apply the GO decision table — retain the two MF terms, remove or flag the four autophagy BP/CC terms — and attach PMID 28123910 plus the AlphaFold provenance as justification.
2. For wet-lab resolution: RNAi knockdown + autophagy-flux assay under starvation/autogamy is the single most decisive experiment; a negative result would refute the BP terms outright.
3. For bioinformatic firming-up: run an ML phylogeny of ciliate ATG1-like kinases scoring C-terminal-domain presence, and a Foldseek/HHpred search of the unassigned N-terminal folded element to determine whether it confers any lineage-specific targeting or regulatory role.

Provenance / Artifacts

• Catalytic-motif scan, family reconciliation table, and Smith-Waterman identity table (executed code + outputs).
• AlphaFold AF-A0BFB4-F1 (v6) per-residue pLDDT-by-region computation.
• UniProt/InterPro/PANTHER record retrieval; EuropePMC abstract + full-text extraction confirming Paramecium inclusion and Atg1-absence statements.

Limitations: all computational results are conservative inferences; no experimental data on A0BFB4 exist; the Atg1-absence evidence is in-silico. Local repository *-bioinformatics analyses were withheld from this run.

Artifacts

• OpenScientist final report
• OpenScientist final report
• OpenScientist a0bfb4 alphafold plddt
• OpenScientist a0bfb4 family and alignment