CAF120 (YNL278W / P53836) — curation notes

Journal of research for the AI GO-annotation review of Saccharomyces cerevisiae CAF120.
CAF120 is an understudied ("dark") gene: it is named as a CCR4-associated factor and
is reported to be associated with the CCR4-NOT complex, but its own molecular function and
biological role are essentially uncharacterized. The primary deliverable of this review is
therefore an honest knowledge_gaps section plus conservatively-reasoned description /
core_functions grounded in domain content, orthology and the sparse literature — never
invented function.

Identity (verified from UniProt record)

Domain / family architecture (read inline from CAF120-uniprot.txt)

CCR4-NOT association — nuanced, do not overstate

Localization

Phenotypes (SGD, background context — not directly annotatable here)

caf120Δ null: decreased competitive fitness in minimal medium; altered free amino-acid profile;
elevated cell-surface metal reductase activity; decreased vegetative & anaerobic growth;
abnormal vacuolar morphology; increased stress sensitivity [SGD locus S000005222, phenotype
summary]. These are diffuse, mostly high-throughput phenotypes and do not pin down a specific
molecular function — consistent with the "dark gene" status.

Annotation-by-annotation reasoning (GOA)

  1. GO:0004672 protein kinase activity — IBA (GO_REF:0000033) — from PANTHER family
    PTN001969686, with reference members = Arabidopsis MAP3K loci AT1G05100, AT3G50310,
    AT4G26890     (WITH/FROM). CAF120 has no protein kinase domain (only the PH domain). This
    is a family over-propagation: a shared accessory PH module pulled a non-kinase yeast protein
    into a plant-MAP3K-dominated PANTHER cluster, inheriting the kinase MF from the kinase-domain
    members. Biologically indefensible → REMOVE.
  2. GO:0007165 signal transduction — IBA (GO_REF:0000033) — same PANTHER family
    (PTN001969686), same Arabidopsis MAP3K reference set. "Signal transduction" is the generic
    BP the plant MAP3Ks carry. No evidence CAF120 acts in a signaling cascade; inherited via the
    same over-propagation. → REMOVE (over-annotation; no gene-specific support).
  3. GO:0005634 nucleus — IEA (GO_REF:0000044, SL-0191) — UniProtKB-SubCell mapping backed by
    PubMed:14562095 (GFP atlas). Reasonable subcellular location. → KEEP_AS_NON_CORE.
  4. GO:0005737 cytoplasm — IEA (GO_REF:0000044, SL-0086) — same GFP-atlas backing. Broad but
    correct location. → KEEP_AS_NON_CORE.
  5. GO:0005935 cellular bud neck — IEA (GO_REF:0000044, SL-0029) — redundant with the IDA
    below but derived from the SubCell mapping. → KEEP_AS_NON_CORE (subsumed by the IDA).
  6. GO:0003674 molecular_function — ND (GO_REF:0000015) — root "unknown MF" placeholder.
    Honest reflection of the dark-gene state; the ND is appropriate given no defensible MF. →
    ACCEPT (it correctly records that MF is unknown).
  7. GO:0008150 biological_process — ND (GO_REF:0000015) — root "unknown BP" placeholder.
    Same rationale. → ACCEPT.
  8. GO:0005935 cellular bud neck — IDA (PMID:25028499) — direct-assay localization to the
    bud neck. Well supported (and consistent with the Huh GFP atlas). → ACCEPT (best-supported
    experimental annotation).

What is KNOWN vs NOT known

KNOWN:
- Subcellular localization: cytoplasm, nucleus, and bud neck (GFP atlas + IDA); relocalizes
under DNA-damage stress.
- Has a divergent PH domain (Skg3/CAF120-like PH family, IPR058155); is a Cdk1 phosphosubstrate.
- Co-purifies with / is associated with the CCR4-NOT machinery (named "CCR4-associated factor";
UniProt SUBUNIT), though not enumerated as a GO/SGD core subunit.
- Has a WGD paralog SKG3 (YHR133C), also uncharacterized.

NOT known (knowledge gaps):
- Molecular function: no demonstrated catalytic, binding, adapter, or scaffold activity. The PH
domain's ligand (phosphoinositide? protein? — divergent family) is unknown.
- Biological role: whether it functionally contributes to CCR4-NOT mRNA deadenylation/turnover
or transcriptional regulation, or acts independently at the bud neck, is unresolved.
- Whether it is a bona fide stable CCR4-NOT core subunit vs a transient/substoichiometric
associated factor.
- Physiological meaning of bud-neck localization (cytokinesis/budding role?) and of the
DNA-damage-induced relocalization.
- Functional relationship to / redundancy with paralog SKG3.

UPDATE (post-merge, from falcon deep research + independent verification): meiotic crossover role

The falcon deep-research report (CAF120-deep-research-falcon.md, 21 citations) surfaced a
specific, previously-missed functional finding, which I then verified independently against the
primary paper (NOT taking falcon's word for it):

Actions taken: added NEW annotation GO:0010520 (regulation of reciprocal meiotic recombination,
IMP, PMID:31351878); added a meiotic core_function; added a RESIDUAL_SUBGAP knowledge gap for the
unknown mechanism; updated description. All supporting_text verbatim-verified against the cached
full text.

Note on falcon's other claims: falcon also cited Goldstrohm 2007 (JBC; PUF/Ccr4 deadenylation)
as testing caf120Δ with NO effect on HO mRNA deadenylation, and Kim 2018 (Genome Res; homomer-
ization) for self-interaction + cytoplasm/bud/bud-neck localization. These are plausible but I
did NOT independently verify the caf120-specific data in those papers, so I did not add them as
annotations (only the crossover finding, which I verified, was incorporated).

Provenance summary of key references