Humanin (Q8IVG9, HUNIN_HUMAN) — curation notes

Folder follows the <HOST>__<ACC> convention for alternative-ORF peptides
(CLAUDE.md, "Alternative-ORF peptides"): UniProt files humanin under the host
symbol MT-RNR2, the mitochondrial 16S rRNA gene, and the peptide has its own
accession and its own GOA rows (76).

1. What the gene is

Humanin (HN) is a 24-residue peptide (MAPRGFSCLL LLTSEIDLPV KRRA) whose ORF
lies inside the mitochondrial 16S rRNA gene MT-RNR2. It was found by functional
screening of an Alzheimer brain cDNA library for a factor protecting neurons
from familial AD genes PMID:11371646.

2. The coding question is genuinely unresolved

UniProt itself carries a CAUTION on this entry. Three distinct uncertainties:

  1. Genetic code / peptide length. The ORF read with the cytoplasmic code
    gives 24 aa; read with the mitochondrial code the last three residues are
    lost (21 aa). UniProt: "If translation of the mRNA occurs in the
    mitochondrion rather than in the cytoplasm, then the usage of the
    mitochondrial genetic code would lead to the production of a shorter peptide
    lacking the last three C-terminal residues."
  2. Where translation happens. No mechanism has been shown for export of an
    mtDNA-encoded mRNA to the cytoplasm, nor for secretion of a peptide made in
    the matrix. UniProt: "The mechanisms allowing the production and the
    secretion of humanin remain unclear."
  3. Which locus actually makes the active peptide. 13 nuclear MT-RNR2-like
    loci (NUMT-derived MTRNR2L1-13) are predicted to encode 15 full-length
    HN-like peptides, and at least ten are expressed PMID:19477263. Guo et al. made the same point from the other
    direction PMID:12732850. In-vivo peptide detection used an anti-HN antibody
    that cannot distinguish the mitochondrial product from the nuclear paralogs
    PMID:12009529.

Consequence for curation: every immunodetection-based localization row on this
entry (nucleus, mitochondrion, sperm compartments) is in principle a statement
about "HN immunoreactivity", not about this accession specifically. I have not
used this to remove well-replicated locations, but it is the reason I treat the
single-study nuclear and perinuclear rows as over-annotations.

A second, pervasive caveat: most functional work uses synthetic peptide added
exogenously, frequently at micromolar concentrations, and frequently the
S14G ("HNG") analogue, which is 2-3 orders of magnitude more potent than HN
PMID:12787071.

3. Direct molecular activities (what HN itself does)

Receptor ligand. Two receptor systems, both with direct binding data.
- GPCRs FPR2/FPRL1 and FPR3/FPRL2: PMID:15465011, with direct membrane binding ("We demonstrated by binding
experiments using [(125)I]-W peptide that HN and fHN directly interacted with
hFPRL1 on the membrane") and EC50 3.5 nM for HN at FPRL1. Functional output:
chemotaxis of mononuclear phagocytes PMID:15153530.
HN also competes with Abeta42 for the same receptor, which is the basis of the
receptor antagonist activity annotation.
- A gp130-family cytokine receptor complex: PMID:19386761. Downstream signalling
is JAK/STAT3 PMID:16005025, confirmed in RPE cells
PMID:26990160.
Note PMID:16005025 explicitly dissociates the two systems: FPR2 knockdown did
not abolish HN rescue in F11 cells.

BH3-domain / BCL-2-family binding. The intracellular arm.
- BAX: PMID:12732850.
- BID/tBID: PMID:15661737.
- BimEL only: PMID:15661735.
- Mechanism: co-fibrillation and sequestration PMID:31690630; PMID:33106313. Both are in-vitro recombinant studies.

Amyloid-beta binding / anti-fibrillization.
PMID:28282805;
PMID:27349871.

IGFBP3 binding. High-affinity, functionally consequential, but GO has no
term for it (no "insulin-like growth factor binding protein binding").
PMID:14561895; Kd 5.05 uM and competition with importin-beta1
PMID:26216267. Mapped to the IGFBP3 heparin-binding
domain, and competitive with hyaluronan for the same segment
PMID:30184438.

Homodimerization. Required for neuroprotection, Ser7-dependent
PMID:12787071.

Low-value interactions. TRIM11 (Y2H; destabilizes HN, so TRIM11's substrate
rather than HN's function) PMID:12670303; alpha-actinin-4 (bacterial two-hybrid
+ GFP-fusion colocalization) PMID:15619032; MPP8 (Y2H/co-IP, no functional
consequence shown) PMID:23532874; VSTM2L,
a secreted antagonist PMID:21393573.

4. Localization

Secretion is well supported PMID:11371646 and residues required for it were mapped PMID:12860203. HN circulates and declines with
age PMID:19623253. Cytoplasmic pool is needed for the BAX arm. Mitochondrial signal is
reported by EM-immunohistochemistry PMID:15567815 and by uptake of labelled
peptide PMID:26990160 — i.e. exogenous peptide entering mitochondria, not evidence of
matrix synthesis. Sperm midpiece/flagellum localization is replicated by two
groups [PMID:20542501; PMID:30920769 "Humanin was expressed in the midpiece of
the spermatozoa"].

5. Downstream phenotypes (the over-annotation zone)

A single HUVEC study generates five GOA rows (IL-1 production, IL-18 production,
inflammatory response, NLRP3 inflammasome assembly, response to oxidative
stress) PMID:32923762. Others: astrocyte neuroinflammation PMID:23277413, PGC1A-driven
mitochondrial biogenesis in MIN6 beta-cells PMID:29432738, ER-stress
protection via mitochondrial GSH PMID:27783653, insulin sensitivity PMID:19623253. These are consequences of receptor signalling in whichever cell
type was assayed; they are not separate activities of the peptide, and the
IL-1/IL-18-production and oxidative-stress rows in particular are two steps
downstream of anything HN touches.

Two rows are flatly unsupported:
- GO:0006879 intracellular iron ion homeostasis (NAS) rests only on humanin
immunoreactivity being seen near iron deposits in villonodular synovitis
PMID:15567815. Proximity to iron
is not iron homeostasis. REMOVE.
- GO:0048471 perinuclear region of cytoplasm rests on GFP-HN overexpression
colocalizing with endogenous actinin in HEK293 PMID:15619032.

6. Decisions taken

2026-11-03 ORGANELLES batch 4 (mitochondrion rows)