C0HM83 (SHMOOSE) review notes

Identity

• UniProt C0HM83 (SHMOS_HUMAN), "Protein SHMOOSE" / "Small human mitochondrial ORF over
  serine tRNA", 58 aa, PE1, encoded on mtDNA (OG Mitochondrion). UniProt gives no gene symbol,
  so the folder and gene_symbol use the accession (alt-ORF convention, projects/MICROPROTEINS.md).
• Essentially one primary paper: Miller et al. 2023 Mol Psychiatry PMID:36127429, plus an
  erratum [PMID:36658336, abstract only, content of correction not available]. Later PubMed hits
  (search "SHMOOSE", 2026-10-03: 8 records) are reviews or plasma-level association studies,
  e.g. PMID:41883858 (Mediterranean diet vs plasma SHMOOSE/humanin, same USC group). No
  independent replication of the mechanistic findings found.

ORF location and genetic code (bioinformatics, C0HM83-bioinformatics/RESULTS.md)

• The UniProt sequence maps once on rCRS, + (heavy-strand-encoded) sense, m.12234-12410 (incl. TAA
  stop); overlaps MT-TS2 (tRNA-Ser(AGY)), MT-TL2 (tRNA-Leu(CUN)) and the 5' end of MT-ND5 in a
  different frame. Same coordinates are given in PMID:41883858.
• No TGA/ATA/AGA/AGG codons, so standard and vertebrate mitochondrial codes give the identical
  58-aa product. Paper agrees: PMID:36127429. So the
  code cannot tell whether it is made by mitoribosomes or cytosolic ribosomes (unlike MOTS-c),
  and the authors leave this open PMID:36127429.
• My own inference (not tested): the start codon lies inside tRNA-Ser(AGY). Standard tRNA
  punctuation processing of the H-strand polycistron would separate the start from the rest of the
  ORF, so translation would need an unprocessed or alternatively processed RNA. Flag as a question.
• m.12372G>A (rs2853499) changes codon 47 GAC->AAC (D47N); synonymous for ND5 per the paper
  PMID:36127429.

Existence

• Endogenous peptide: WB ~6 kDa in mito and nuclear fractions of SH-SY5Y; absent in rho0 cells; two
  peptides by MS after IP from mitochondria [PMID:36127429 "SHMOOSE was detected in neuronal
  mitochondria and nuclei fractions at the predicted ~6kDa molecular weight"; "in rho zero cells
  (i.e., cells without mtDNA), SHMOOSE was not detected"]. Detected in CSF by ELISA.

Variant association (not function)

• D47N associated with AD in ADNI with modest effect PMID:36127429. rs2853499 is a haplogroup U marker; authors concede
  PMID:36127429. The same base also lies in the overlapping ND5 sequence (synonymous)
  and in the H-strand transcript, so association alone cannot attribute risk to the peptide. Genetic
  association is not a GO function.

Functional experiments (all with exogenous synthetic peptide)

• Mitofilin (IMMT, Q16891) binding: spiked synthetic peptide into lysate -> IP-MS (98 proteins, IMMT a
  top hit), reciprocal co-IP after treating cells with 1 uM peptide, and dot blot with recombinant
  proteins PMID:36127429. siRNA IMMT abolished peptide effect on superoxide. D47N still binds.
• Exogenous peptide raised basal OCR ~20% and MTT signal; protected SH-SY5Y against oligomeric Abeta42;
  ICV in rats changed hypothalamic transcriptome. These are pharmacological responses to added peptide,
  not loss-of-function of the endogenous product; no knockout/knockdown of endogenous SHMOOSE exists
  (mtDNA editing would be needed).

GO decisions

• GO:0005515 protein binding IPI (IMMT) -> REMOVE per project policy (no informative MF; there is no
  "MICOS complex binding" GO term in QuickGO search). Interaction is reported in description.
• mitochondrion IDA + IEA -> ACCEPT.
• nucleus IDA + IEA -> KEEP_AS_NON_CORE (fractionation-only, one study, no function there).
• No NEW: OCR/neuroprotection data are exogenous-peptide pharmacology; participation test not met.

2026-10-24 ORGANELLES batch 2 (mitochondrion rows)

• Rows: GO:0005739 IDA (PMID:36127429) ACCEPT; GO:0005739 IEA (GO_REF:0000044, SL-0173) ACCEPT. No sub-compartment rows. Actions unchanged; reasons extended.
• IEA mapping check: the UniProt location is curated from the same paper [file:human/C0HM83/C0HM83-uniprot.txt "SUBCELLULAR LOCATION: Mitochondrion {ECO:0000269|PubMed:36127429}"], so the feature the mapping assumes is present. Added the UniProt file to references.
• Rule 4: no presequence, no predicted TM (only MobiDB-lite disorder/proline-rich features); the cached full text has no mitoplast, protease-protection or sub-mitochondrial fractionation (grep for inner membrane/matrix/proteinase/mitoplast finds only the IMMT description and transcriptome GO enrichment). Binding of exogenous peptide to inner-membrane IMMT is one indirect line PMID:36127429. Stays at the root.