COX15 converts heme O to heme A in the mitochondrial inner membrane. The current heme A synthase definition specifies oxidation of the C8 methyl group to a formyl group and lies under oxidoreductases acting on CH or CH2 groups. The older CH-CH subclass therefore needs correction to GO:0120547. Heme A synthesis is the integrated catalytic core; supplying a prosthetic group to complex IV does not by itself demonstrate a separate insertase or scaffold activity.
The original human study tested sequence-verified COX15.1 in patient fibroblasts. Oxidase activity increased from 28% to 61% of control. The heme A/B ratio increased from 0.030 to 0.090, compared with 0.064 in untransduced controls and 0.139 in controls overexpressing COX15. Thus the rescue is substantial but not uniform normalization. Heme O accumulated in patient heart but was undetectable in patient fibroblasts. The native-gel antibodies detected respiratory subunits, not COX15 itself. See PMID:12474143(https://pmc.ncbi.nlm.nih.gov/articles/PMC378614/), Methods “Cell Culture and Retroviral Expression” and “Electrophoretic Methods,” Results Tables 2–3 and Figure 3, and the final isoform discussion.
The official products contain 410 and 388 residues. VSP_011281 replaces canonical positions 368–410 with a different 21-residue tail, changing the C-terminal membrane architecture and a by-similarity heme-binding position. Sequence change alone establishes neither loss of activity nor equal activity. The human rescue tested only COX15.1; the suggestion that both products may be needed for optimal function remains a hypothesis.
PMID:26940873(https://pmc.ncbi.nlm.nih.gov/articles/PMC4858986/) directly tests tagged yeast Cox15, conserved histidines and linker variants. Its native mammalian oligomer analysis uses porcine kidney; the human patient-cell assays measure steady-state COX15 abundance. The original Results on heme depletion show that oligomerization can persist without available heme, and histidine mutations do not by themselves distinguish catalytic from ligating roles. No direct purified human heme-binding measurement follows from those experiments.
GO:0070069 means a protein complex containing a redox-active heme protein and is broader than mature respiratory complex IV. The 2003 human study does not establish the exact COX15 complex assertion, while the independently observed oligomers make a broader cytochrome-containing assembly plausible. The exact human IDA assertion remains unresolved. GO:0098803 specifically concerns a respiratory-chain complex; its PMID:9878253 basis remains unresolved because the complete original was not recovered. The complete abstract positively supports mitochondrial import and protease protection. This is an access limit, not evidence that COX15 can never associate with a respiratory assembly.
PMID:29128334(https://doi.org/10.1016/j.cels.2017.10.010) reports a mitochondrial interaction map using tagged human proteins and antibody-based immunoprecipitation. The original author-lab PDF, pages 565–567 and STAR Methods e1/e3/e7–e9, describes the constructs, cross-linking, purification and assay context. The exact COX15–PRDX6 records include two spoke-expanded cross-link associations and one anti-bait co-IP record in HEK293 cells. Both official sequences match the interaction-record ROGIDs; the COX15 sequence is the displayed 410-residue product.
Human PRDX6 is an independently established phospholipase A2 as well as a peroxidase. Its class is documented in the reviewed P30041 record, and the complete PMID:10893423 abstract describes separate catalytic activities and mutational controls. GO:0043274 specifies binding to a phospholipase. This refines the positive association by counterpart class without asserting regulation, a particular contacted active site, native binary affinity or activity of the shorter COX15 product. The original target-specific supplementary image and table were not independently inspected; their assay attribution rests on the literal curated records.
The current Human Protein Atlas ICC/IF summary identifies mitochondria as the main location with supported reliability, and nucleoplasm as additional. The returned summary, rather than image pixels, was assessed. The MitoCoP study PMID:34800366(https://pmc.ncbi.nlm.nih.gov/articles/PMC8664129/) combines subtractive/spatial proteomics, importomics and literature curation. Its selected original Methods and Results were read. During follow-up the original Table S1 workbook was recovered and authenticated against the original XML supplement MD5. Sheet (A), row 239 reports COX15/Q7KZN9 group 7459 with subtractive Class 1, spatial Cluster 1 and importomics Class 1 assignments; sheet (B), row 214 confirms MitoCoP inclusion. The Integral membrane category is an additional observation, but the group includes Q7KZN9-2 and does not resolve isoform function or directly specify the inner membrane. The earlier target-row access limitation is now closed. The publisher workbook and exact cell locators/hashes are recorded under mitocop_table_s1 in the evidence index.
Current PTHR23289 topology places the exact human sequence below PTN000590786, with experimental support on both yeast lineages. The human target in location/process WITH/FROM is legitimate descendant evidence. The current v19 GO-slim chemistry fields differ from the current GOA assertion; they do not replay the historical IBD decision. The current ISS synthase donor is bacterial Bacillus subtilis CtaA/P12946. The inherited yeast-donor wording remains unverified because its historical WITH/FROM was omitted; the current donor does not reconstruct every historical source version.
IPR003780 and IPR023754 explicitly support membrane/heme A pathway mappings. The current ARBA membrane rule contains many FunFam-dependent branches for which target assignments were not obtained, so no particular current ARBA branch is claimed as confirmed. Independent InterPro and target evidence remain sufficient for the biological membrane judgment. The broad membrane record is refined to the inner membrane.
The historical InterPro-only membrane source is retained as retired, alongside the distinct active combined-source record. Nine historical objects omitted WITH/FROM values; their original fields remain, while the complete current objects and explicit identity mapping are available in the evidence index. Retirement and these omissions describe source history, not a loss of biological function.
The current heme pathway reaches the COX15 heme A reaction through an explicit catalyst/entity/reference chain. Three assembly events reach the same inner-membrane COX15 entity through the modeled metallochaperone complex. Those paths support location and pathway context; they do not establish autonomous COX15 copper insertion. Other-participant differences between event prose and component lists are retained as limits rather than copied into new mechanistic claims.
Priorities for experiments are product-specific heme synthesis and import, endogenous human oligomer/assembly composition, the physiological meaning of the PRDX6 association, and the transfer of synthesized heme A to nascent oxidase. Each can be tested without assuming that synthesis, binding and insertion are the same activity.
PR #836 explicitly changed GO:0016627, GO:0070069 and GO:0098803 to REMOVE. Those decisions were reconsidered rather than silently overlooked. The CH–CH chemical objection remains correct, but MODIFY to the independently supported heme A synthase term preserves the actual redox function. Five other sources already assert GO:0120547, so the correction consolidates existing coverage. For GO:0070069, the current definition covers complexes containing a redox-active heme protein. Excluding COX15 from mature oxidase is insufficient to exclude that broader assertion. The original 2003 native-gel experiments do not resolve it, and yeast/porcine oligomer observations cannot substitute for a human assay. UNDECIDED expresses this uncertainty even though the original 2003 paper was read. For GO:0098803, the complete 1998 original remains unavailable; its abstract establishes import, not the precise complex location or categorical exclusion from a transient respiratory assembly.
The 2003 cache's full_text_available: true flag does not describe its completeness: it contains Introduction, one Electrophoretic Methods paragraph and Discussion, but no Results body or Tables 2–3. Those results were read in the external PMC original. The cache remains unchanged. Short new primary-source anchors identify the PRDX6 phospholipase experiment, the yeast scope of the oligomer study and the import/protection assay. Reference identity, inherited findings and inherited quotations are preserved; annotation quotation count alone is not a measure of biological support.
The retired and current broad membrane assertions now both refine to mitochondrial inner membrane using independent target evidence. Their source identities remain distinct. Nine omitted historical WITH/FROM fields cannot be reconstructed by copying current values. Likewise, the current bacterial ISS donor does not establish which donor supported every historical source version, so the historical yeast-donor finding remains UNVERIFIED.
The positive PRDX6 association follows the standing binding instruction. The GO:0043274 definition requires binding to a phospholipase; it does not require dependence on that protein's phospholipase catalysis. PRDX6's peroxidase activity therefore does not negate the independently verified counterpart class. No phospholipase regulation, catalytic-interface specificity or additional binding term is inferred.
The authored source evidence index uses YAML and explicitly identifies manual compilation. The paint section supplies the original tree/MSA URLs and hashes, exact human PTN002510931 record and full path through PTN000590786. The arba section supplies the raw rule URL, response hash and exhaustive partitions: 1,269 branches, zero confirmed matches, 999 unresolved FunFam branches and 270 nonmatches. Local tmp paths are archival identifiers, not public download links. The index additionally preserves assay, product, ontology, donor, Reactome and source-version evidence beyond WITH/FROM. The research questions retain named experts and the experiments specify testable hypotheses and methods; heme handover, direct ligation and disease mechanisms remain proposals rather than asserted functions.