MTRNR2L12 is a nuclear human locus on chromosome 3 (3q11.2) that resembles the humanin small open reading frame of the mitochondrial large-subunit rRNA gene MT-RNR2, one of thirteen such MT-RNR2-like sequences in the human genome that are generally interpreted as nuclear insertions of mitochondrial DNA and are classed by HGNC as pseudogenes. Its predicted product, humanin-like 12 (HN12), is a 24-residue peptide (MAPRGFSCLLLSTSEIDLPVKRRA) that matches humanin at 23 of 24 aligned positions, differing only by Leu12->Ser, and is identical in sequence to the predicted product of MTRNR2L8; a coding polymorphism at this position would make it identical to humanin itself. Every residue humanin is known to require for activity or secretion is retained. The locus is the most frequently named of the thirteen in transcriptomic studies, appearing as a differentially expressed transcript or co-expression hub in blood mononuclear cells of adults with Down syndrome and in platelets during COVID-19, where it groups with the mitochondrially encoded transcripts MT-ND1 and MT-ND5. UniProt lists it at protein existence level 3 (inferred from homology), and whether a peptide is made from it, and what it would do, is unestablished: no HN12 peptide has been detected in human tissue or fluid, no molecular activity has been measured for it, and no binding partner or receptor interaction has been reported. The humanin peptide from which it is named is cytoprotective and antiapoptotic, binds BAX, BID, BIM and IGFBP3 intracellularly and signals through FPRL1/FPR2 and a CNTFR-alpha/WSX-1/gp130 complex when secreted, but those properties were measured on the mitochondrially encoded peptide and on synthetic analogues of it, not on this gene product.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005576 extracellular region | IEA GO_REF:0000044 | MARK AS OVER ANNOTATED | Summary: Electronic mapping of UniProt's Secreted subcellular-location line for P0DMP1. That line is itself annotated ECO:0000250 by similarity to humanin (UniProtKB:Q8IVG9), so the row carries no observation made on MTRNR2L12. Reason: MTRNR2L12 is known only by homology (UniProt protein existence level 3); no peptide from this locus has been detected in any compartment, by mass spectrometry or otherwise. UniProt's Secreted and Cytoplasm lines for P0DMP1 are themselves ECO:0000250 by-similarity statements transferred from humanin (file:human/MTRNR2L12/MTRNR2L12-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 MTRNR2L12 both of those segments are intact - this is the closest of the thirteen copies to humanin, differing at Leu12 only - so unlike the diverged loci there is no residue-level argument against secretion, and I do not remove the row. What is wrong with it is that it is asserted of this locus on the strength of humanin alone, for a gene product that has never been shown to exist. The transcriptomic literature naming this locus measures RNA, not peptide, and does not demonstrate locus specificity against the near-identical copies or the mitochondrial rRNA. Supporting Evidence: PMID:35372353 HN was found to be an extracellularly secreted factor whereby two amino acid structures, Leu9-Leu11, and Pro19-Va120, appear to be essential for the secretion of HN peptide PMID:19477263 Proline vs serine in position 19 may determine whether the peptide is secreted or not, while threonine in position 13 may be important for cell surface receptor binding. file:human/MTRNR2L12/MTRNR2L12-uniprot.txt SUBCELLULAR LOCATION: Secreted |
| GO:0005576 extracellular region | ISS GO_REF:0000024 | MARK AS OVER ANNOTATED | Summary: Curator transfer of humanin's secreted location to MTRNR2L12 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: MTRNR2L12 is known only by homology (UniProt protein existence level 3); no peptide from this locus has been detected in any compartment. The sequence argument that defeats this transfer in several sibling loci does not apply here: HN12 retains Leu9-Leu11 and Pro19-Val20, the two segments humanin needs for secretion [PMID:35372353], and differs from humanin at Leu12 only. So the claim may well be right if the peptide exists. But an ISS asserts that a donor observation holds for the target, and here the premise that there is a target gene product at all is untested, which makes the row an over-reach rather than an error. Note also that the two most-cited studies naming this locus report transcript abundance in blood mononuclear cells [PMID:25720405] and in platelets [PMID:35421970]; neither localises a peptide, and the second finds the transcript clustering with the mitochondrially encoded MT-ND1 and MT-ND5, which is what read mis-assignment from the parent mitochondrial locus would look like. Propagation Review Root cause: PROPAGATION BAD Failure modes: WRONG ORTHOLOG OR PARALOG Sources checked: UniProtKB:Q8IVG9 · MT-RNR2 SUPPORTS SOURCE BUT NOT TARGET Humanin, the mitochondrially encoded peptide. Its secreted location is directly observed and the source annotation is sound. What does not transfer is the premise: this nuclear MT-RNR2-like sequence is a paralogous genomic copy rather than an ortholog of a protein-coding gene, and no peptide product of it has been observed, let alone observed outside the cell. The residues required for secretion are, unusually for this family, all present. Supporting Evidence: PMID:35372353 HN was found to be an extracellularly secreted factor whereby two amino acid structures, Leu9-Leu11, and Pro19-Va120, appear to be essential for the secretion of HN peptide PMID:35372353 It is predicted that these various isoforms might contribute to differential neuroprotective effects and receptor binding, but their individual roles remain to be investigated. PMID:35421970 Regarding the magenta module, MT-ND1, MT-ND5, and MTRNR2L12 were selected as hub genes. |
| GO:0005737 cytoplasm | IEA GO_REF:0000044 | MARK AS OVER ANNOTATED | Summary: Electronic mapping of UniProt's Cytoplasm line, which is itself a by-similarity statement transferred from humanin. Reason: MTRNR2L12 is known only by homology (UniProt protein existence level 3); no peptide from this locus has been detected in any compartment. UniProt's Cytoplasm line for P0DMP1 is itself an ECO:0000250 by-similarity statement transferred from humanin (file:human/MTRNR2L12/MTRNR2L12-uniprot.txt), so this IEA row and the corresponding ISS row state the same humanin claim twice over. A translated 24-residue peptide with no signal sequence would plausibly be cytosolic, and no residue of HN12 argues against it; but the location is asserted from humanin, not observed here, and it adds nothing to the record of this locus. Supporting Evidence: PMID:35372353 It is predicted that these various isoforms might contribute to differential neuroprotective effects and receptor binding, but their individual roles remain to be investigated. file:human/MTRNR2L12/MTRNR2L12-uniprot.txt Cytoplasm {ECO:0000250|UniProtKB:Q8IVG9} |
| GO:0005737 cytoplasm | ISS GO_REF:0000024 | MARK AS OVER ANNOTATED | Summary: Curator transfer of humanin's cytoplasmic location on sequence similarity; humanin's own cytoplasm annotations are IDA/EXP. Reason: MTRNR2L12 is known only by homology (UniProt protein existence level 3); no peptide from this locus has been detected in any compartment. A translated 24-residue peptide would plausibly be cytosolic, and nothing in the HN12 sequence argues against it, so this is the least objectionable of the four rows; but the premise an ISS needs - that the target gene product exists - is untested here, so the row over-reaches rather than being demonstrably wrong. Propagation Review Root cause: PROPAGATION BAD Failure modes: WRONG ORTHOLOG OR PARALOG Sources checked: UniProtKB:Q8IVG9 · MT-RNR2 SUPPORTS SOURCE BUT NOT TARGET Humanin has direct cytoplasmic localisation evidence, so the source is sound; the transfer assumes a peptide product of this nuclear locus that has never been detected. Supporting Evidence: PMID:35372353 It is predicted that these various isoforms might contribute to differential neuroprotective effects and receptor binding, but their individual roles remain to be investigated. file:human/MTRNR2L12/MTRNR2L12-uniprot.txt Cytoplasm {ECO:0000250|UniProtKB:Q8IVG9} |
Loading supporting content…
Download this section (compressed HTML)Q: Is a peptide produced from the MTRNR2L12 locus? This is the locus where the question matters most, because its predicted peptide differs from humanin at a single position and a common polymorphism would erase even that difference [PMID:25720405]. One paper asserts that the peptide has been confirmed in brain tissue, but gives no source for the claim and I could find no primary report of it. Targeted mass spectrometry cannot distinguish HN12 from humanin on sequence alone, so the experiment has to be genetic: look for the Ser12-containing tryptic peptide in a donor homozygous for the serine allele.
Suggested experts: Alan Saghatelian, Pinchas Cohen
Q: Why does MTRNR2L12 carry no IBA rows when MTRNR2L8, whose predicted peptide is identical in sequence, carries both the receptor antagonist activity and the apoptosis rows from PANTHER:PTN002141596, as do the far more diverged MTRNR2L11 and MTRNR2L13? The difference tracks PANTHER subfamily assignment rather than sequence, which makes the family's IBA coverage inconsistent in a way visible from the alignment alone. The resolution should be to narrow the node to the sequences with functional data, not to extend the rows to the remaining copies.
Q: How should GO and UniProt treat a nuclear pseudogene whose transcript is repeatedly reported as differentially expressed, when the transcript cannot be distinguished by short reads from twelve near-identical copies and from an extremely abundant mitochondrial rRNA? MTRNR2L12 is the most frequently named locus of the thirteen in transcriptomic studies, and in at least one it clusters with mitochondrially encoded transcripts [PMID:35421970]. A locus-resolved quantification standard for this family would settle how all of that literature should be read.
Q: If HN12 is translated, is it functionally distinguishable from humanin at all? With a single conservative difference at Leu12, and with synthetic HN5 carrying the same Leu12->Ser still cytoprotective [PMID:19477263], the interesting possibility is not that this locus has a distinct function but that some of the activity attributed to mitochondrial humanin is contributed by nuclear copies.
Suggested experts: Pinchas Cohen
Experiment: Allele-resolved mass spectrometry for the HN12 peptide. Because HN12 differs from humanin only at position 12, use donors or cell lines genotyped at rs6484338 and search for the serine-containing peptide; the leucine allele produces a species indistinguishable from mitochondrial humanin, so genotype is the only available discriminator.
Experiment: Locus-resolved transcript quantification for MTRNR2L12 by long-read sequencing or by amplicon sequencing designed across the positions that differ between the nuclear copies, applied to the sample types where the gene has been reported as differentially expressed (Down syndrome blood mononuclear cells, COVID-19 platelets). This tests whether the reported signal is this locus rather than a sibling copy or mitochondrial rRNA.
Experiment: Ribosome profiling of the MTRNR2L12 transcript in a cell type where it is reported to be expressed, to test whether the predicted open reading frame is translated rather than merely transcribed.
Experiment: CRISPR deletion of the MTRNR2L12 locus in a human cell line, with apoptosis challenge and with measurement of total circulating humanin-like immunoreactivity, to test whether this copy contributes to the pool of humanin-like peptide. The thirteen near-identical loci make guide design the hard part, and off-target cutting at MTRNR2L8, which has the identical coding sequence, is the control that matters.
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)