Focused Curation Report — PIK3R4 / VPS15 (Q99570)

Hypothesis (function_assignment / slug protein-phosphorylation-complex-contribution):
Human PIK3R4/VPS15 directly contributes to protein phosphorylation (GO:0006468),
potentially as a noncatalytic cofactor/scaffold rather than the phosphotransferase.

Source annotation under review: GO:0006468 "protein phosphorylation" (BP),
evidence code NAS, reference PMID:8999962.

Executive Judgment

Verdict: Weakly supported / over-annotated as written — recommend the GO:0006468
(NAS) annotation be treated as non-core and removed (or, at most, generalized), pending curator verification.

The hypothesis correctly separates two claims that the original annotation conflates:

1. A direct VPS15 protein-kinase molecular function (GO:0004672) — REFUTED.
   Modern cryo-EM of the human class III PI3K complex explicitly designates VPS15 a
   pseudokinase (PMID:39913640, 2025); the seed further notes nucleotide HPLC/MS
   showing GTP (not ATP) binding and degenerate phosphotransferase geometry.
   UniProt Q99570 keeps a legacy "Protein kinase" domain (aa 26–324), an active-site
   "proton acceptor" (D148) and ATP-binding sites annotated by similarity, plus the
   keywords "Serine/threonine-protein kinase" and "Transferase" — yet lists no
   catalytic-activity reaction, and describes the protein's FUNCTION solely as a
   "Regulatory subunit." The catalytic residues are present as a fold relic, not a
   working active site.

2. A broad BP "contribution to protein phosphorylation" via the complex — UNRESOLVED,
   and not established for VPS15 specifically by the cited evidence.
   The protein-substrate phosphorylation in PMID:8999962 (Table I) is a property of the
   co-purified GST‑p150·VPS34 complex, with Mn²⁺ preference and no individual
   autophosphorylation. Attribution to VPS15 is not made by the authors; the more
   parsimonious catalyst is VPS34 (or an associated kinase). In the same paper, p150/VPS15
   appears as a substrate (Fig 4, in‑vivo labeling) — the opposite of being the kinase.
   The annotation is NAS, i.e. author assertion, not a traceable experimental
   attribution to VPS15. Subsequent direct biochemistry firms this direction: VPS15 is a
   ULK substrate (PMID:34121209, six ULK-dependent sites, major Ser861) whose
   phosphorylation regulates VPS34 activity, and a GTP-binding conformational regulator of
   VPS34 (PMID:40537377) — i.e. a phospho-target/regulator, not a protein phosphotransferase.

Most important caveat (from the seed, and respected here): a pseudokinase can still
participate noncatalytically. Loss of catalysis ≠ loss of all participation. So the
correct curation posture is not "VPS15 has nothing to do with phosphorylation," but
"the specific evidence does not establish a direct VPS15 phosphotransferase function,
and a bare BP 'protein phosphorylation' term mis-states VPS15's actual regulatory/scaffold role."

Evidence Matrix

Computed Provenance — VPS15 Kinase-Domain Catalytic-Motif Audit

I retrieved UniProt Q99570 (this run) and audited the kinase domain (aa 26–324) for the
canonical Ser/Thr-kinase catalytic elements, benchmarked against canonical PKA (P17612).
Executed code + output; table saved as vps15_catalytic_motif_audit.csv.

Benchmark: canonical PKA has an intact P-loop (GTGSFG@51) and DFG@185.
Interpretation (conservative): two core nucleotide-positioning/catalytic elements
(Gly-rich loop; DFG glycine) are degenerate in VPS15, consistent with impaired canonical
ATP-dependent phosphotransfer and the pseudokinase designation. The retained K53 and
D148 are the relics that drive UniProt's "by similarity" ATP-binding/active-site and the
PANTHER family-root kinase IBA — i.e. the legacy kinase labels are homology carry-over, not
demonstrated catalysis. This sequence audit is corroborative and secondary to the direct
cryo-EM structural evidence (PMID:39913640).

GO Decision Table (leads — require curator verification)

GO Curation Implications (leads — require curator verification)

• GO:0006468 "protein phosphorylation" (BP), NAS, PMID:8999962 — lead: REMOVE or down-rank to non-core.
  The cited evidence demonstrates a complex activity, not a VPS15 molecular function, and
  modern structure shows VPS15 is a pseudokinase. A bare BP "protein phosphorylation" on a
  scaffold subunit mis-attributes a co-subunit's chemistry.
• GO:0004672 / GO:0004674 "(serine/threonine) protein kinase activity" (MF) — if present via IBA/ISS, lead: REMOVE.
  Refuted by pseudokinase evidence; keep only as evolutionary/fold context, not as active MF.
• Do NOT downgrade to a bare "protein binding" term. More informative supported terms exist:
• CC: GO:0034271/GO:0034272 (phosphatidylinositol 3-kinase complex, class III, type I/II) — supported.
• BP: GO:0000045 (autophagosome assembly), GO:0016192 (vesicle-mediated transport) — supported by function/structure.
• MF: consider enzyme regulator / kinase activator style term for its stimulation of VPS34 lipid kinase (PMID:8999962), and note GTP-binding (from PMID:39913640) if curators accept it.

Mechanistic Scope

Immediate molecular function tested: phosphotransfer onto protein substrates by VPS15 itself.
- Direct VPS15 activity: not demonstrated; fold present, catalysis absent (pseudokinase).
- Complex activity (VPS34-containing): protein-substrate phosphorylation observed in vitro, catalyst unresolved — a distinct reaction, not VPS15's MF.
- Downstream/indirect (not the tested MF): VPS15 scaffolds PI3KC3, activates VPS34 lipid
(not protein) kinase, and is itself a phospho-substrate/regulatory target — regulatory
participation in phosphorylation-linked processes, not a phosphotransferase role.

Conflicts and Alternatives

• Attribution artifact: complex activity assigned to a subunit that is a substrate in the same paper.
• Homology carry-over: UniProt/PANTHER retain kinase/transferase labels by similarity; IBA at
  the family root risks propagating catalytic MF to a leaf that lost catalysis.
• Lipid- vs protein-kinase conflation: the robust, reproducible VPS34-complex chemistry is
  lipid kinase (PtdIns→PtdIns3P); the "protein phosphorylation" claim rests on the weaker Table I assay.
• Not contamination by default: an unexpected Mn²⁺-preferring activity is not proof of
  contamination; it is best logged as unresolved associated activity, per the seed's caution.
• VPS15 is the phospho-TARGET, not the enzyme: PMID:34121209 shows VPS15 is a ULK substrate
  (six ULK-dependent sites; major Ser861), directly realizing the seed's "phosphorylated by another
  kinase" alternative. Its phosphorylation regulates VPS34 lipid-kinase output — a regulatory
  node in a phosphorylation pathway, not a protein phosphotransferase.
• Real noncatalytic moonlighting: PMID:37414850 shows a Vps34-independent nuclear coactivator
  role (Bmal1-Clock), confirming that "pseudokinase" ≠ "no function," but these functions are not
  "protein phosphorylation."

Knowledge Gaps

1. Which subunit is the Table I catalyst? Checked: abstract + seed description; unresolved.
   Matters because it decides whether any protein-phosphorylation term belongs to VPS15.
   Resolve with: recombinant, individually purified catalytically-dead VPS34 vs VPS15 point mutants + substrate assays.
2. Full text of PMID:39913640 nucleotide/geometry data. Checked: abstract (truncated) + seed;
   full methods not retrieved here. Matters for firmness of "GTP-binding, no phosphotransfer."
3. Current GO/UniProt evidence codes on the live record (IBA vs NAS vs ISS). Checked: UniProt JSON
   (kinase keywords, no catalytic reaction); GO evidence-code audit needed for the exact annotation set.

Discriminating Tests

• Subunit-resolved kinase assay: individually purified WT vs kinase-dead (D148A) VPS15 and
  VPS34 with peptide/protein substrates; ± Mn²⁺; test which subunit carries Table I activity.
• Nucleotide-loading assay (HPLC/MS or thermal shift) on isolated human VPS15 pseudokinase
  domain to confirm GTP-over-ATP preference and absence of γ-phosphate transfer.
• Structure-guided catalytic-motif audit (HRD, DFG, VAIK, glycine loop) on Q99570 vs bona fide STKs.
• Orthology/PANTHER review to reclassify the IBA MF at the leaf as pseudokinase.

Curation Leads (require curator verification)

• Candidate action: Remove/retire GO:0006468 (NAS, PMID:8999962) as a VPS15 direct annotation;
  do not assign GO:0004672/GO:0004674 MF. Replace informativeness with PI3KC3 CC terms + autophagy/vesicle BP + a VPS34-lipid-kinase regulator/activator MF.
• Candidate reference snippets to verify:
• PMID:39913640 — "The inactive conformation of the VPS15 pseudokinase stabilizes the inactive conformation…"
• PMID:40442316 — "…extensive contacts between the FIP200 scaffold subunit of ULK1C and the VPS15, ATG14 and BECN1 subunits of PI3KC3-C1."
• PMID:8999962 — "Recombinant p150 associated with PtdIns 3-kinase in vitro in a stable manner, resulting in a 2-fold increase in lipid kinase activity."
• Suggested question for curators: Is the NAS GO:0006468 meant to capture VPS15's own MF, or
  the complex's process participation? If the latter, a co-complex/regulatory annotation is more accurate than "protein phosphorylation."
• Suggested experiment: subunit-resolved catalytically-dead mutant kinase assay (above).

Note: This report deliberately did not re-run the previously completed NVJ analysis; it targets the
Panaretou-1997 vs modern-structure phosphotransfer chemistry, as instructed.

Artifacts

• OpenScientist final report
• OpenScientist final report