JIP4 (YDR475C) — curation notes

Journal for the AI GO-annotation review of Saccharomyces cerevisiae JIP4 (systematic
name YDR475C; UniProt Q03361; SGD:S000002883). This is an understudied ("dark") gene:
UniProt names it "Uncharacterized protein JIP4" (AltName "Jumonji-interacting protein 4"),
and SGD classifies it as a protein of unknown function. The deliverable therefore centres
on an honest knowledge_gaps section grounded in domain composition, orthology, and the
(sparse) literature — no invented function.

Summary of what is machine-known (UniProt Q03361)

• 876 aa protein, MW ~98.7 kDa. Entry version 148 (2026). PE 1: Evidence at protein level
  — but the only experimental evidence is large-scale phosphoproteomics (see below);
  there is no functional assay.
• Highly intrinsically disordered: MobiDB-lite predicts 7 disordered regions
  (37–67, 112–155, 226–254, 330–353, 490–513, 661–728, 750–876) covering a large fraction
  of the protein. Multiple low-complexity / compositionally biased stretches, including
  runs of basic+acidic residues, polar residues, and notably several Arg/Ser (RS-type)
  repeats — e.g. ...RSVSSARRSRSRSRSRSICTRR... (~aa 350–370) and
  SKSRSSSKSRIRDKSKPSSP (~aa 675–695), and a C-terminal ...KSRSPSSFRKEDE...RGGLFGFGRL.
  These RS/RD-rich low-complexity tracts are the reason PANTHER places it in the
  SR-repetitive-matrix family (below). [file:yeast/JIP4/JIP4-uniprot.txt]
• No recognizable folded/catalytic domain is annotated in UniProt: no PWI domain, no
  RRM, no enzymatic motif — only disorder + compositional bias. This is important: the human
  family members that carry the splicing function have a folded PWI nucleic-acid-binding
  domain; JIP4 does not have one annotated. [file:yeast/JIP4/JIP4-uniprot.txt]
• Phosphoprotein (KW: Phosphoprotein). Eight phosphoserines mapped by MS:
• Ser-48, Ser-51, Ser-510, Ser-552, Ser-775 [PMID:19779198 — Cdk1 substrate global analysis]
• Ser-48, Ser-360, Ser-552 [PMID:18407956 — multidimensional chromatography phosphoproteome]
• Ser-577 [PMID:15665377 — pheromone-signaling phosphoproteome]
  Several of these sites (from PMID:19779198, "Global analysis of Cdk1 substrate
  phosphorylation sites") indicate JIP4 is among the many proteins phosphorylated in a
  Cdk1-dependent manner; this is a large-scale substrate catalog, not a dedicated study of
  JIP4. [file:yeast/JIP4/JIP4-uniprot.txt]
• Cross-references indicate protein interactions exist in high-throughput datasets:
  BioGRID 87 interactions, IntAct 6, FunCoup 117 — i.e. it appears in interactome data but
  with no curated functional interpretation. [file:yeast/JIP4/JIP4-uniprot.txt]

Orthology / family (PANTHER PTHR23148)

PANTHER assigns JIP4 to family PTHR23148 ("Serine/arginine repetitive matrix"),
subfamily PTHR23148:SF0 = "SERINE/ARGININE REPETITIVE MATRIX PROTEIN 1" (SRRM1 / SRm160).
The reviewed members of SF0 are essentially the metazoan SRRM1 orthologs plus the two
yeasts [file:yeast/JIP4/interpro/panther/PTHR23148/PTHR23148-entries.csv... i.e.
interpro/panther/PTHR23148/PTHR23148-entries.csv]:

The human/metazoan SRRM1 (SRm160) is a nuclear-matrix SR-related splicing coactivator:
it promotes constitutive and ESE-dependent splicing by bridging sequence-specific SR
proteins to snRNP components, is a component of the spliceosome / exon junction complex,
binds RNA/DNA with low sequence specificity, and stimulates mRNA 3'-end cleavage.
[PANTHER family description, file:yeast/JIP4/interpro/panther/PTHR23148/PTHR23148-metadata.yaml;
corroborated by SRRM1/SRm160 literature — UniProt Q8IYB3.]

This family assignment is the sole basis for the IBA annotations in GOA:
- part_of spliceosomal complex (GO:0005681) — with/from MGI:1858303 (mouse Srrm1),
PANTHER:PTN000567596, UniProtKB:Q8IYB3 (human SRRM1). [file:yeast/JIP4/JIP4-goa.tsv]
- involved_in regulation of mRNA splicing, via spliceosome (GO:0048024) — with/from
FB:FBgn0036340 (Drosophila), PANTHER:PTN000567596. [file:yeast/JIP4/JIP4-goa.tsv]
- (UniProt DR also lists an IBA RNA binding GO:0003723 from GO_Central, not currently in
the GOA TSV row set.) [file:yeast/JIP4/JIP4-uniprot.txt]

Caveat on the IBA splicing propagation (KEY reasoning)

Whether the metazoan SRRM1 splicing function transfers to yeast JIP4 is doubtful, for
biological reasons that are worth stating:

1. JIP4 lacks the PWI domain. The defining folded, nucleic-acid-binding module of the
   SRm160/SRRM1 and RED120 (PWI-motif) splicing proteins is the PWI domain. JIP4's UniProt
   record annotates only disorder and low-complexity bias — no PWI, no RRM. The S. pombe
   family member (Q9USH5) is explicitly a "PWI domain-containing protein" and is only 301 aa,
   suggesting the true PWI-splicing ortholog in fission yeast is a different, shorter protein,
   and that JIP4's placement rests on shared RS/low-complexity composition rather than the
   catalytic/binding domain. [file:yeast/JIP4/JIP4-uniprot.txt; PANTHER entries csv]
2. The budding-yeast splicing machinery is exhaustively characterized. The S. cerevisiae
   spliceosome has been purified and defined in great biochemical/structural detail. The
   established yeast counterparts of the human SR-related nuclear-matrix splicing proteins are
   Cwc21 (YDR482C; ortholog of SRm300/SRRM2) and candidate PWI proteins such as Snu71
   — not JIP4. JIP4 has never been reported as a spliceosome or EJC component in yeast
   spliceosome proteomics. [web: RNA journal / genesdev PWI-motif and Cwc21 characterization —
   background, not gene-specific experimental evidence for JIP4.]
3. Yeast introns are few and simple — there is no minor (U12) spliceosome and essentially
   no exon-junction-complex-style splicing coactivation of the metazoan kind; the specific
   ESE-dependent activation role of SRm160 has no clear yeast equivalent.

Conclusion: the IBA MF/BP splicing terms are plausible-by-homology but unverified and
biologically questionable for yeast. They should be treated as low-confidence
homology-only inferences (mark as non-core / over-annotated candidates), not as established
JIP4 function. The RS/low-complexity composition supports at most a possible RNA/nucleic-
acid-associated or nuclear role, but this is not established.

SGD / literature facts (verified via SGD, BioGRID)

• Protein of unknown function. Previously annotated as two separate ORFs (YDR474C and
  YDR475C) that were merged after corrections to the systematic reference sequence.
  [SGD locus JIP4/YDR475C; web:yeastgenome.org]
• Paralog YOR019W, arising from the whole-genome duplication; YOR019W is likewise
  uncharacterized. jip4∆ is viable; yor019w∆ is viable (double mutant not annotated) —
  consistent with a non-essential, possibly redundant gene. [SGD; web:yeastgenome.org]
• Name "Jumonji-interacting protein 4" (JIP4) is a legacy/heuristic name (likely from an
  interaction dataset with a Jumonji/JmjC-family protein); it does not by itself establish a
  molecular function. No dedicated functional paper on yeast JIP4 was found.
• Appears in high-throughput interactome data (BioGRID 87 / IntAct 6) and in
  phosphoproteomics, but with no curated functional role.

KNOWN vs NOT-KNOWN

KNOWN (evidence-supported):
- It is expressed and made as a protein (PE1).
- It is an in vivo phosphoprotein with ≥8 mapped phosphoserines, including Cdk1-dependent
sites (large-scale MS). [PMID:19779198; PMID:18407956; PMID:15665377]
- It is intrinsically disordered with RS/low-complexity composition (sequence-based).
- It is a member (by low-complexity composition) of the SR-repetitive-matrix PANTHER family
that in metazoa contains SRRM1/SRm160.
- Non-essential; has a WGD paralog (YOR019W).

NOT KNOWN (genuine knowledge gaps):
- Molecular function — no biochemical activity demonstrated; no folded catalytic/binding
domain identified. RNA binding is only an IBA guess.
- Whether it is truly part of the spliceosome / involved in splicing in yeast (IBA only;
biologically questionable, see caveat).
- Subcellular localization — not experimentally established (no CC annotation beyond ND).
RS-rich composition hints at nuclear, but unverified.
- Biological process / pathway — unknown (ND at the BP root).
- Loss-of-function phenotype beyond "viable"; role of the phosphorylation; the identity/
meaning of its interactors; functional relationship to paralog YOR019W.
- What the "Jumonji-interacting" name refers to mechanistically.

Curation plan

• IBA spliceosomal complex (CC) and regulation of mRNA splicing (BP): keep but mark as
  non-core / over-annotated homology-only inferences — do NOT elevate to core function;
  document the PWI-absence + yeast-splicing-machinery caveat in the review reason. (Per
  guidelines, IBA over-propagations may be argued against on biological grounds.)
• ND molecular_function / cellular_component / biological_process root terms: these are
  "no data" placeholders; keep as-is (they honestly encode "unknown"). Mark KEEP_AS_NON_CORE
  / note that they reflect the dark-gene status.
• description: project-independent — disordered RS/low-complexity protein of unknown
  function, SR-repetitive-matrix family by composition, phosphoprotein, non-essential, WGD
  paralog YOR019W. No curation commentary.
• core_functions: only the defensible minimum (essentially none can be asserted with
  confidence; possibly a cautious statement that no core molecular function can be assigned).
• knowledge_gaps: the primary deliverable (MF, localization, BP, phenotype, splicing
  hypothesis).

Update after falcon deep research (2026-07-05)

The falcon deep-research report (JIP4-deep-research-falcon.md, Edison, 20 citations,
~23 min) is consistent with the assessment above and adds one useful hypothesis about the
gene name:

• The "Jumonji-interacting protein 4" name most plausibly reflects identification of JIP4 as
  a Gis1-associated factor. Gis1 is a JmjC (Jumonji-domain) zinc-finger transcription
  factor that acts downstream of the Rim15/TOR/Sch9 nutrient-signaling and
  calorie-restriction response pathway [file:yeast/JIP4/JIP4-deep-research-falcon.md
  "Gis1 is a Jumonji-domain zinc-finger transcription factor acting downstream of
  Rim15/TOR/Sch9 nutrient signaling"]. The report explicitly flags that "direct
  mechanistic characterization remains limited", i.e. this is a naming-origin inference, not
  an established function — treated here as a hypothesis in knowledge_gaps / suggested
  questions, not as core function.
• The report reconfirms: merged ORF (YDR474C+YDR475C), non-essential, paralog YOR019W,
  Ser/Arg repetitive matrix (IPR052225) domain class, uncharacterized status, and that the
  splicing/RNA-processing role is a low-to-moderate-confidence domain-based inference only.
• CAVEAT: the report states a "165 aa" size for a "merged gene model/extension" — this is
  inconsistent with the authoritative UniProt length of 876 aa (Q03361) and is not used;
  it appears to be a confused artifact from the ORF-merge re-annotation history.