ACP7

Summary: Acid phosphatase activity, by automatic annotation from the EC number, with UniProt recording the reaction as RHEA:15017. Provenance, stated once and applying to every annotation. ACP7 has no biochemical or functional characterisation: UniProt's FUNCTION, EC number, catalytic activity and both cofactor statements are all ECO:0000250 by similarity to UniProtKB:P80366, the Fe(3+)-Zn(2+) purple acid phosphatase of the kidney bean Phaseolus vulgaris. That cross-kingdom template is not arbitrary, which is the key point. PMID:16793224 identified ACP7 as a novel PLANT-LIKE human purple acid phosphatase, built a structural model of the human enzyme on the red kidney bean structure specifically, and showed the catalytic centre is present in it. UniProt's AltName 'Purple acid phosphatase long form' is ECO:0000303 from that paper. So the kidney bean is the deliberate, published choice of template for a protein argued to belong to a plant-like subfamily - a considered inference rather than a distant accident. What remains absent is functional testing: no activity assay, no substrate, no phenotype. The protein itself is not hypothetical - UniProt records PE 1, evidence at protein level, with proteomics identification - so ACP7 is uncharacterised rather than unstudied. Accepted as a well-founded inference: the metallophosphoesterase fold is supported by several InterPro signatures, the iron- and zinc-coordinating residues are individually annotated, and a published structural model places the catalytic centre in the human protein. But it is a modelled prediction, not a measured activity.

Summary: Extracellular region, from the UniProt SubCell mapping of a Secreted assertion. That assertion is ECO:0000305, inferred by a curator from the ECO:0000255 predicted signal peptide at residues 1-26, so it is an inference resting on a prediction. Accepted - a cleaved signal peptide with no downstream transmembrane segment is a reliable indicator, the protein is glycosylated as a secreted protein would be, and it is detected in proteomics - but the localisation itself has not been demonstrated experimentally.

Summary: Hydrolase activity, from InterPro. Correct but two levels above the acid phosphatase term derived from the same prediction, and it conveys nothing the specific term does not. Retained as a non-core parent.

Summary: Metal ion binding, from InterPro, and the best-supported annotation on this gene: UniProt annotates individual BINDING features for the Fe cation (141, 170, 173, 335) and for Zn(2+) (170, 205, 286, 333), and the binuclear Fe-Zn centre defines the purple acid phosphatase family rather than being a generic fold property. PMID:16793224's structural model shows that centre is present in the human protein. Still ECO:0000250 in origin, but the residue-level conservation makes it the least speculative call here. Note the more specific GO:0008270 zinc ion binding is proposed under proposed_new_terms; no ferric-specific term is proposed because UniProt records the ligand as 'Fe cation' without an oxidation state.