## HSPB3
- **UniProt:** Q12988 · **batch:** proteostasis-batch-2026-06-07b · **review status:** COMPLETE
- **PN placement:** two rows — `Cytonuclear proteostasis|Chaperone|small HSP system|small HSP` (type) and `Mitochondrial proteostasis|Chaperone|small HSP` (group) ; **PN-node mapping:** both sHSP nodes → mapped/ok_for_propagation_to_go GO:0044183 protein folding chaperone (new_to_goa); class/branch nodes no_mapping.
- **Consistency:** Notes ↔ review fully consistent — HSPB3 is the most divergent human sHSP (HSP20/alpha-crystallin family), an ATP-independent holdase partnering HSPB2 in muscle, cytoplasm+nucleus, R7S causes HMND4 (PMID:9858786, 8972725, 20142617, 19464326). The PN mitochondrial row is questionable: neither review, notes, UniProt nor GOA places HSPB3 in mitochondria (documented compartments are cytoplasm/nucleus), so the "Mitochondrial proteostasis" placement is a PN-internal contradiction with the evidence.
- **PN story / NEW pressure:** GOA has NO molecular-function annotation for HSPB3 (only GO:0006986 response to unfolded protein, TAS). PN projects GO:0044183 protein folding chaperone as new_to_goa. Verified real, but its OLS definition is a foldase ("binds an unfolded protein **to fold it**... do not confuse with unfolded protein holdase activity"). HSPB3 is a holdase, not a foldase. The review's core MF is GO:0051082 unfolded protein binding (holdase); the precise term is GO:0140309 unfolded protein holdase activity (verified real). So an MF gap exists, but GO:0044183 over-reaches the wrong way.
- **Mapping strategy:** Keep status=mapped but the target is wrong for sHSPs. GO:0044183 (foldase) is broader/different from the holdase activity sHSPs actually have. Prefer GO:0140309 unfolded protein holdase activity (or GO:0051082). Drop or re-scope the mitochondrial-row projection (no mito evidence for HSPB3).
- **Evidence alignment:** PN rows carry no reference titles for HSPB3; review PMIDs (9858786, 8972725, 20142617, 19464326) are sHSP-identity/localization/disease papers — no divergence, just no overlap to compare.
- **Verdict:** sHSP placement sound but PN MF target mischaracterizes holdases as foldases; mito row unsupported. **Recommended edits:** [MAP] retarget GO:0044183 → GO:0140309 unfolded protein holdase activity (or GO:0051082) for sHSP nodes on Q12988; [MAP] do not project a mitochondrial chaperone term onto HSPB3 (no mitochondrial localization evidence).