MTRNR2L3

Summary: Electronic mapping of UniProt's Secreted subcellular-location line for P0CJ70. That line is itself annotated ECO:0000250 by similarity to humanin (UniProtKB:Q8IVG9), so the row carries no observation made on MTRNR2L3.

Reason: MTRNR2L3 is known only as a transcript (UniProt protein existence level 2); no peptide from this locus has been detected in any compartment, by mass spectrometry or otherwise. UniProt's Secreted and Cytoplasm lines for P0CJ70 are themselves ECO:0000250 by-similarity statements transferred from humanin (file:human/MTRNR2L3/MTRNR2L3-uniprot.txt), so both the ISS row (donor UniProtKB:Q8IVG9) and the subcellular-location IEA row state the same humanin claim, once directly and once via the controlled-vocabulary mapping. Secretion of humanin is not a generic property of the sequence: it depends on Leu9-Leu11 and on Pro19-Val20 [PMID:35372353], and Bodzioch et al. proposed on the alignment that position 19 is what decides whether a nuclear isoform is secreted at all [PMID:19477263]. In MTRNR2L3, Leu9-Leu11 is intact but Pro19 is replaced by serine, so one of the two segments humanin needs for secretion is altered. Because the asserted location depends on secretion determinants that are altered in this peptide, the row is not merely unverified but positively doubtful, and I remove it.

Summary: Curator transfer of humanin's secreted location to MTRNR2L3 on sequence similarity. Humanin's own extracellular annotations are IDA/EXP from several studies, so the donor side is sound; the question is only whether the transfer is safe.

Reason: MTRNR2L3 is known only as a transcript (UniProt protein existence level 2); no peptide from this locus has been detected in any compartment, by mass spectrometry or otherwise. UniProt's Secreted and Cytoplasm lines for P0CJ70 are themselves ECO:0000250 by-similarity statements transferred from humanin (file:human/MTRNR2L3/MTRNR2L3-uniprot.txt), so both the ISS row (donor UniProtKB:Q8IVG9) and the subcellular-location IEA row state the same humanin claim, once directly and once via the controlled-vocabulary mapping. Secretion of humanin is not a generic property of the sequence: it depends on Leu9-Leu11 and on Pro19-Val20 [PMID:35372353], and Bodzioch et al. proposed on the alignment that position 19 is what decides whether a nuclear isoform is secreted at all [PMID:19477263]. In MTRNR2L3, Leu9-Leu11 is intact but Pro19 is replaced by serine, so one of the two segments humanin needs for secretion is altered. The transfer therefore fails on the target side, and I remove it.

Summary: Electronic mapping of UniProt's Cytoplasm line, which is itself a by-similarity statement transferred from humanin.

Reason: MTRNR2L3 is known only as a transcript (UniProt protein existence level 2); no peptide from this locus has been detected in any compartment, by mass spectrometry or otherwise. UniProt's Secreted and Cytoplasm lines for P0CJ70 are themselves ECO:0000250 by-similarity statements transferred from humanin (file:human/MTRNR2L3/MTRNR2L3-uniprot.txt), so both the ISS row (donor UniProtKB:Q8IVG9) and the subcellular-location IEA row state the same humanin claim, once directly and once via the controlled-vocabulary mapping. A translated 24-28-residue peptide would plausibly be cytosolic, which is why this row is the least objectionable of the four; but it is asserted from humanin, not observed here, and it adds nothing to the record of this locus.

Summary: Curator transfer of humanin's cytoplasmic location on sequence similarity; humanin's own cytoplasm annotations are IDA/EXP.

Reason: MTRNR2L3 is known only as a transcript (UniProt protein existence level 2); no peptide from this locus has been detected in any compartment, by mass spectrometry or otherwise. UniProt's Secreted and Cytoplasm lines for P0CJ70 are themselves ECO:0000250 by-similarity statements transferred from humanin (file:human/MTRNR2L3/MTRNR2L3-uniprot.txt), so both the ISS row (donor UniProtKB:Q8IVG9) and the subcellular-location IEA row state the same humanin claim, once directly and once via the controlled-vocabulary mapping. A translated 24-28-residue peptide would plausibly be cytosolic, which is why this row is the least objectionable of the four; but it is asserted from humanin, not observed here, and it adds nothing to the record of this locus.

Summary: Phylogenetic propagation of humanin's receptor antagonist activity to MTRNR2L3 from the humanin-family node PANTHER:PTN002141596. The grounding experiment is humanin competing with amyloid-beta at FPRL1 [PMID:15153530]; nothing comparable has been done with HN3.

Reason: The IBD behind this row places the activity at the humanin-family node, and the placement is reasonable for the mitochondrial peptide: humanin is a genuine FPRL1 ligand and the antagonism of amyloid-beta at that receptor is directly observed [PMID:15153530]. Arguing with the row means asking whether MTRNR2L3 sits inside the clade that inherited the function, and the difficulty is that the question is not really phylogenetic. The nuclear MT-RNR2-like loci are copies of a mitochondrial rRNA segment, so descent in this tree tracks genomic sequence rather than a protein-coding lineage, and membership of the node does not establish that the ORF is translated, that the peptide is stable, or that it reaches a receptor. On residues: HN3 loses Pro3, at which humanin Pro3->Ala abolishes neuroprotective activity, and Leu12 (file:human/MT-RNR2__Q8IVG9/MT-RNR2__Q8IVG9-uniprot.txt); Thr13 and Ser14, the two positions most clearly required in humanin, are retained. Bodzioch et al., who defined these loci, singled out position 13 as the one that may matter for cell-surface receptor binding [PMID:19477263]. A second, term-level point applies to every member of this node: receptor antagonist activity describes humanin's competition with amyloid-beta at FPRL1, which is a property measured in a specific assay, whereas humanin also carries receptor ligand activity by IDA - so even for the donor the term captures one facet of the interaction. The node placement is defensible and the row may yet prove right, so this is an over-annotation rather than an error: it states an activity for a gene product that has never been shown to exist, let alone to engage a receptor.

Summary: Phylogenetic propagation of negative regulation of the execution phase of apoptosis to MTRNR2L3 from the humanin-family node. The co-listed donor is UniProtKB:P0CJ72 (MTRNR2L5), whose own evidence for this term is an IMP from an assay on synthetic HN5 peptide [PMID:19477263].

Reason: The antiapoptotic function is the best-supported property of this family, and the IBD placement is sensible: humanin blocks BAX, BID and BIM and is antiapoptotic in many systems, and synthetic HN5 peptide - the product of one of these very nuclear loci - is cytoprotective against staurosporine-induced apoptosis [PMID:19477263]. That last point is important and cuts against a blanket dismissal: a nuclear humanin-like sequence, carrying the same Leu12->Ser substitution that most of these loci carry, does retain cytoprotective activity when synthesised. What the row still lacks for MTRNR2L3 is anything specific to this locus. Its transcript is detectable and staurosporine-responsive [PMID:19477263], but a transcript is not a peptide: no HN3 peptide has been observed, no apoptosis assay has used it, and the systematic human-genetics lookup across all thirteen loci found no association with disease and no expression QTL that would argue for a functional product [PMID:31753007]. On residues: HN3 loses Pro3, at which humanin Pro3->Ala abolishes neuroprotective activity, and Leu12 (file:human/MT-RNR2__Q8IVG9/MT-RNR2__Q8IVG9-uniprot.txt); Thr13 and Ser14, the two positions most clearly required in humanin, are retained. I mark the row as an over-annotation: plausible, untested for this locus, and currently an assertion about the humanin sequence rather than about this gene product.