SMIM45 (formerly the long non-coding RNA LINC00634, chromosome 22) is a bicistronic locus. Its upstream open reading frame encodes the 68-amino-acid protein represented by this entry, a predicted single-pass membrane microprotein with one hydrophobic helix near the N terminus. This 68-residue ORF is ultra-conserved across vertebrates, is supported by ribosome-profiling evidence of translation, and has no reported molecular function, partner or subcellular localization from experiment. A second, non-overlapping downstream ORF on the same mRNA encodes a 107-amino-acid protein that arose de novo in hominoids; that separate product is expressed in fetal brain, is found in both nucleus and cytoplasm, delays neuronal maturation when over-expressed in cortical organoids and mice, and in hepatocellular carcinoma cells binds and stabilises metadherin (MTDH). The functions reported for the 107-residue product have not been shown for the 68-residue protein.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005634 nucleus | IDA PMID:36593289 De novo genes with an lncRNA origin encode unique human brai... | REMOVE | Summary: Nuclear localization reported in PMID:36593289 for the protein encoded by ENSG00000205704. The protein studied was the 107-amino-acid downstream-ORF product, not the 68-amino-acid protein that A0A590UK83 now represents. Reason: The annotation was made (GOA date 2023-02-14) while A0A590UK83 carried sequence version 1, the 107-residue protein that the paper explicitly describes. In UniProt release 2023_03 the entry's sequence was replaced by the unrelated, ultra-conserved 68-residue upstream ORF, and the experimental rows were carried over unchanged. The two ORFs share no sequence (SMIM45-bioinformatics/RESULTS.md). The curator's reading of the paper was correct for the sequence at the time; the row is now attached to the wrong gene product. The localization belongs to the 107-residue product, which currently has no UniProtKB accession. Supporting Evidence: PMID:36593289 which encodes a putative protein of 107 amino acids located in both cytoplasm and nucleus PMID:41219895 A previous report showed that the transcript ENSG00000205704 of LINC00634 encoded a 107 amino acid peptide, which was localized in both the cytoplasm and nucleus file:human/SMIM45/SMIM45-bioinformatics/RESULTS.md The two share no sequence. |
| GO:0005634 nucleus | IEA GO_REF:0000044 | REMOVE | Summary: Electronic mapping of the UniProt subcellular-location statement "Nucleus", which is itself sourced from PMID:36593289 and therefore describes the 107-residue downstream-ORF product. Reason: Same root cause as the IDA row. The UniProt SUBCELLULAR LOCATION comment was retained when the entry's sequence was replaced by the 68-residue ORF, so the SubCell-to-GO mapping now asserts a nuclear location for a protein that was never localized. Supporting Evidence: PMID:36593289 which encodes a putative protein of 107 amino acids located in both cytoplasm and nucleus |
| GO:0005737 cytoplasm | IDA PMID:36593289 De novo genes with an lncRNA origin encode unique human brai... | REMOVE | Summary: Cytoplasmic localization reported in PMID:36593289 for the 107-amino-acid downstream-ORF product of ENSG00000205704, not for the 68-amino-acid protein of the current entry. Reason: As for the nucleus IDA row: the evidence is for the 107-residue protein, which was the sequence of A0A590UK83 when the annotation was made, and the row was not moved when UniProt replaced the sequence with the 68-residue ORF. A later study of the 107-residue peptide (PMID:41219895) also reports it as cytoplasmic, so the observation is sound for that product. Supporting Evidence: PMID:36593289 which encodes a putative protein of 107 amino acids located in both cytoplasm and nucleus PMID:41219895 Immunofluorescent staining further demonstrated SMIM45-107aa and MTDH was co-located in the cytoplasm of HCC cells |
| GO:0005737 cytoplasm | IEA GO_REF:0000044 | REMOVE | Summary: Electronic mapping of the UniProt "Cytoplasm" location, sourced from PMID:36593289 and describing the 107-residue downstream-ORF product. Reason: Same root cause as the IDA cytoplasm row; the location statement was carried over to the unrelated 68-residue sequence. Supporting Evidence: PMID:36593289 which encodes a putative protein of 107 amino acids located in both cytoplasm and nucleus |
| GO:0016020 membrane | IEA GO_REF:0000044 | ACCEPT | Summary: UniProt predicts a single helical transmembrane segment (residues 7-27) in the current 68-residue sequence, mapped to membrane. Reason: This is the only annotation derived from the sequence the entry now represents. The hydrophobic segment VYLVISVLILVGFGACIYYF is a clear transmembrane helix candidate, and the generic term makes no claim about which membrane. It is a prediction and has not been tested experimentally. |
| GO:0042551 neuron maturation | IMP PMID:36593289 De novo genes with an lncRNA origin encode unique human brai... | REMOVE | Summary: Knockout and over-expression of the hominoid-specific 107-amino-acid ORF of ENSG00000205704 changed the pace of neuronal maturation in cortical organoids (knockout accelerated it, over-expression and a mouse knock-in delayed it). Reason: The perturbed ORF is the 107-residue downstream ORF, not the 68-residue protein this accession now represents. The knockout genotyping primer begins at the start codon of the 107-residue ORF and the over-expression construct carried that ORF, while neither touches the 68-residue ORF (SMIM45-bioinformatics/RESULTS.md). The row was made on sequence version 1 (107 aa) and survived the 2023_03 sequence replacement. For the 107-residue product itself the observed effect is negative regulation (loss speeds maturation), so a regulation term would be the more accurate choice once that product has its own accession. Supporting Evidence: PMID:36593289 KO-GT-F: ATGTCTATGGCTGCCTGTCCTG PMID:36593289 The ORF sequence of ENSG00000205704 and the human codon-optimized HA PMID:36593289 exhibit a delayed neuronal maturation and a subsequent cortical expansion file:human/SMIM45/SMIM45-bioinformatics/RESULTS.md The KO genotyping primer KO-GT-F (ATGTCTATGGCTGCCTGTCCTG) begins at nt 817, the start codon of the 107-aa ORF |
Loading supporting content…
Download this section (compressed HTML)Q: Should UniProt give the 107-residue downstream-ORF product of SMIM45 its own accession (as for other alternative-ORF peptides) and move the PMID:36593289 FUNCTION, SUBCELLULAR LOCATION and GO rows onto it, leaving A0A590UK83 for the conserved 68-residue protein?
Q: What does the ultra-conserved 68-residue SMIM45 microprotein do, and in which membrane does it reside? Conservation back to cartilaginous fish implies strong constraint on a protein for which there is no functional data.
Experiment: Express epitope-tagged 68-residue SMIM45 at endogenous level (knock-in tag) in neural and non-neural human cells; determine localization by co-staining with organelle markers, carbonate extraction and protease protection; identify partners by proximity labelling. Test loss of function with a knockout that disrupts the 68-residue ORF while leaving the downstream 107-residue ORF intact.
Hypothesis: The 68-residue SMIM45 protein is a single-pass integral membrane protein of a specific organelle.
Type: localization and loss-of-function
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)