CYB5A

Summary: Retain the source-supported PMP22 interaction as non-core.

Reason: Three source-specific IntAct records support the exact CYB5A–A0A6Q8PF08 pair by array, pooling and validated yeast two-hybrid assays, with sequence-matched ROGIDs. A0A6Q8PF08 is the 217-residue unreviewed PMP22 product. Unlike the verified kinesin partner, its membrane/myelin association does not establish a narrower binding target class or a myelination function for CYB5A. Retain the supported contact as non-core under the project policy.

Summary: Refine the interaction with KIF1B isoform 2 to kinesin binding.

Reason: Three source-specific IntAct records support the exact CYB5A–O60333-2 pair by yeast two-hybrid assays. The sequence-verified 1,770-residue KIF1B beta product is a kinesin, and every matching record specifies its sufficient binding region 1192–1770. GO:0019894 classifies the bound protein, so this tail-region interaction supports kinesin binding without implying motor-domain contact, motor activation or native transport of CYB5A. The other screen partners lack an equally established narrower target class at the inspected scope.

Proposed replacements: kinesin binding

Summary: Retain the source-supported wolframin interaction as non-core.

Reason: Three source-specific IntAct records support the exact CYB5A–O76024 pair by yeast two-hybrid assays, with sequence-matched ROGIDs. O76024 is the 890-residue human wolframin product. Its role in ER calcium homeostasis does not establish a receptor, channel or other narrower binding class for this interaction. Retain the positive contact as non-core without transferring a transport or disease process.

Summary: Retain the source-supported progranulin interaction as non-core.

Reason: Three source-specific IntAct records support the exact CYB5A–P28799 pair by yeast two-hybrid assays. The partner ROGID matches the 593-residue progranulin precursor; it does not identify an isolated processed granulin peptide. Unlike kinesin identity, precursor identity alone does not establish cytokine binding or an extracellular signaling function. Retain the supported contact as non-core.

Summary: Retain the source-supported RNF11 interaction as non-core.

Reason: Three source-specific IntAct records support the exact CYB5A–Q9Y3C5 pair by yeast two-hybrid assays, with sequence-matched ROGIDs. Q9Y3C5 is human RNF11. Its official functional record describes ubiquitin-editing complex assembly and recruitment of other enzymes; a RING domain alone does not establish the active ubiquitin-ligase class for this binding assay. Retain the positive contact as non-core without assigning CYB5A ubiquitination activity.

Summary: Refine the soluble-product localization from cytoplasm to cytosol.

Reason: The literal SL-0086 mapping is positive but broader than the established soluble-product localization. PMID:4030743 explicitly identifies sequenced human erythrocyte cytochrome b5 as soluble and present in the cytosol; PMID:1712589 and the official P00167-2 sequence distinguish the transcript lacking the membrane anchor. GO:0005829 therefore specifies the demonstrated compartment. This refinement preserves the gene-level source and its absent isoform field; it does not assign every splice product to the soluble pool.

Proposed replacements: cytosol

Summary: The modeled mitochondrial location requires source-specific resolution.

Reason: The actual Reactome 198824 target path contains P00167 residues 2–134 in mitochondrial-outer-membrane cytochrome-b5/heme entities (198781 within 198772/198808); the annotation is not an accession-free pathway inference. However, the accessible abstract of PMID:8513896 distinguishes separate genes for ER and mitochondrial cytochrome b5, whereas dual localization of the reductase is attributed to one gene. PMID:7668255 measures reductase deficiency and ascorbate regeneration, without resolving the CYB5A product at this location in the accessible text. Full originals were not recovered. The model/source identity and compartment tension remains unresolved; no wrong-gene removal or universal exclusion of CYB5A from mitochondria is inferred from abstracts. Reactome 198824 attributes the myristoyl anchor specifically to CYB5R3, so that mechanism cannot establish a CYB5A membrane anchor. In contrast, the 198845 summary explicitly places CYB5A at the mitochondrial outer membrane, as do the actual target entities. This disagreement between the model and accessible literature remains unresolved; the reductase mechanism alone does not prove a misassigned CYB5A annotation.

Summary: The modeled mitochondrial location requires source-specific resolution.

Reason: The actual Reactome 198845 target path contains P00167 residues 2–134 in mitochondrial-outer-membrane cytochrome-b5/heme entities (198781 within 198772/198808); the annotation is not an accession-free pathway inference. However, the accessible abstract of PMID:8513896 distinguishes separate genes for ER and mitochondrial cytochrome b5, whereas dual localization of the reductase is attributed to one gene. PMID:7668255 measures reductase deficiency and ascorbate regeneration, without resolving the CYB5A product at this location in the accessible text. Full originals were not recovered. The model/source identity and compartment tension remains unresolved; no wrong-gene removal or universal exclusion of CYB5A from mitochondria is inferred from abstracts. Reactome 198824 attributes the myristoyl anchor specifically to CYB5R3, so that mechanism cannot establish a CYB5A membrane anchor. In contrast, the 198845 summary explicitly places CYB5A at the mitochondrial outer membrane, as do the actual target entities. This disagreement between the model and accessible literature remains unresolved; the reductase mechanism alone does not prove a misassigned CYB5A annotation.

Summary: Refine supported ER localization to the ER membrane.

Reason: The literal mouse P56395/ENSMUSP00000124480 transfer supports ER localization, with current same-term IDA PMID:31006538; its target-specific experiment was not recovered beyond the abstract. Independent human membrane topology and the retained cytb5-ops28 ER-insertion assay support the more precise GO:0005789 compartment for the anchored product. This is a refinement of the positive target localization, not a claim that the donor was misannotated or that its original paper demonstrated an uninspected finer term. The soluble erythrocyte product remains a separate class.

Proposed replacements: endoplasmic reticulum membrane

Summary: Retain the actual inherited ER-membrane placement.

Reason: The genuine 1,037-node PTHR19359 tree and MSA place the exact 134-residue target at PTN002500004 below PTN000451786. The current GO:0005789 IBD is literally at that ancestral node and cites RGD:620558, resolved in the tree and official identity to rat P00173/Cyb5a. Current experimental rat records include same-term IDA PMID:12761189. This is a valid inherited localization judgment, not a pairwise similarity inference weakened by having one donor. The human C-terminal anchor and ER insertion evidence are compatible; the soluble splice product remains a distinct localization context.

Summary: Retain the ER-membrane subcellular-location mapping.

Reason: The original SL-0097 mapping agrees with the official membrane-bound product, including its C-terminal transmembrane helix and cytoplasmic heme domain. The actual cytb5-ops28 insertion assay detects ER integration independently of Sec61 depletion. The historical PMID:2752049 topology model is not used to assert that every C-terminal residue remains cytoplasmic; engineered glycosylation and product-specific topology must be distinguished.

Summary: Retain the directly curated cytosolic localization.

Reason: Current HPA assay metadata for ENSG00000166347 reports enhanced cytosol and vesicle localization using two antibodies in HepG2, RT4 and U2OS; five of six assay entries explicitly list both locations and one entry has no listed location. This is corroborating assay metadata, not an independent image-pixel interpretation or isoform-specific antibody assignment. The soluble human product is compatible with cytosolic localization. Cytosol is a functional location of mature soluble CYB5A, independently supported by human erythrocyte sequencing; the HPA assay is not assigned to a specific isoform. This differs from the transient P00167-1 biogenesis context modeled by the SGTA source.

Summary: Retain the cytosolic tail-anchored-protein biogenesis context.

Reason: Reactome 9609921 uses input set 9609909 and output SGTA complex 9610465; both target-bearing paths reach entity 9609919, explicitly P00167-1 residues 2–134 in cytosol. This is the full-length membrane-bound product before ER insertion, not proof that the mature anchor-containing protein is constitutively soluble or that the model tested erythrocyte product 2. The transient biogenesis location is positive but secondary to the mature protein's redox function.

Summary: The exact CYB5A identification in the NK-cell membrane proteome remains unverified.

Reason: The complete abstract describes membrane fractions from human NK-like YTS cells and 1,843 identified proteins, but does not name CYB5A. The original target table could not be recovered: both saved supplementary-table routes returned HTTP 403. Membrane localization is independently plausible and supported, but it does not authenticate this specific high-throughput identification or its peptides. Preserve the HDA source assertion with uncertainty, rather than remove it or claim its target table was inspected.

Summary: Refine the demonstrated cytb5 membrane insertion to ER membrane.

Reason: The indexed original Results explicitly test cytb5-ops28 integration into the ER, using a C-terminal glycosylation reporter, and report no loss of insertion after Sec61 depletion (Figure 2D and Table 1). The Methods describe reticulocyte-lysate translation, semipermeabilized cells, glycosylation and protease-protection assays. This supports ER membrane rather than an unspecified membrane. The exact species/sequence of the supplied cytb5-ops28 construct is not established in the read text; the curated human assignment is retained with that limitation, not inferred from a human host. The canonical cache contains Abstract/Introduction despite its full-text flag; the substantive Results/Methods were read separately from the indexed original.

Proposed replacements: endoplasmic reticulum membrane

Summary: Retain the positive CYP1A2 interaction without inferring activation.

Reason: The preserved IPI assertion names the verified human enzyme CYP1A2. PMID:15680923 explicitly reports this positive interaction in immobilized purified-protein affinity assays; full Methods and the exact b5 species/construct remain unavailable. The existing enzyme-binding class is supported, although no current IntAct/MINT literal source pair was recovered. Binding alone does not establish enhancement or transfer the P450 catalytic activity to CYB5A. The abstract specifically reports no enhancement of the tested CYP1A2 ethoxyresorufin turnover.

Summary: Retain the positive CYP2E1 interaction without inferring activation.

Reason: The preserved IPI assertion names the verified human enzyme CYP2E1. PMID:15680923 explicitly reports this positive interaction in immobilized purified-protein affinity assays; full Methods and the exact b5 species/construct remain unavailable. The existing enzyme-binding class is supported, although no current IntAct/MINT literal source pair was recovered. Binding alone does not establish enhancement or transfer the P450 catalytic activity to CYB5A.

Summary: Retain the positive CYP3A4 interaction without inferring activation.

Reason: The preserved IPI assertion names the verified human enzyme CYP3A4. PMID:15680923 explicitly reports this positive interaction in immobilized purified-protein affinity assays; full Methods and the exact b5 species/construct remain unavailable. The existing enzyme-binding class is supported, although no current IntAct/MINT literal source pair was recovered. Binding alone does not establish enhancement or transfer the P450 catalytic activity to CYB5A.

Summary: Retain the positive CYP2A6 interaction without inferring activation.

Reason: The preserved IPI assertion names the verified human enzyme CYP2A6. PMID:15680923 explicitly reports this positive interaction in immobilized purified-protein affinity assays; full Methods and the exact b5 species/construct remain unavailable. The existing enzyme-binding class is supported, although no current IntAct/MINT literal source pair was recovered. Binding alone does not establish enhancement or transfer the P450 catalytic activity to CYB5A.

Summary: Retain heme binding from the actual ancestral-node inference.

Reason: The exact PTHR19359 target sequence descends from root PTN000451724, where the current GO:0020037 IBD explicitly cites human CYB5B/O43169 and zebrafish cyb5a/A8WGS0 and cyb5b/Q6NY41 through the two literal ZFIN aliases. Current same-term experimental records include PMID:20861021 and PMID:31257865. The family-wide prosthetic-group chemistry is compatible with human CYB5A His44/His68 and its independently verified domain. This does not transfer CYB5B mitochondrial localization, equal splice-product activity, or every donor enzyme pathway.

Summary: Retain the InterPro heme-binding-site inference.

Reason: The actual IPR018506 entry is a heme-binding-site annotation and maps to GO:0020037. Its P00167 match spans residues 40–47, including the first axial ligand His44; official human features separately identify His44 and His68. The entry describes the conserved first heme-ligating histidine rather than merely a generic cytochrome fold. H44L functional impairment is corroborating human evidence, although heme loss in that study was predicted structurally rather than directly measured.

Summary: Propose the conserved cytochrome-b5 electron-transfer activity with explicit homolog evidence.

Reason: PMID:29888793 directly measures oxidation of full-anchor rabbit cytochrome b5 coupled to reduction of the human CYP17A1 oxy-complex in nanodiscs, with rat reductase as the other redox partner. Official rabbit P00169 is CYB5A; its 134-residue sequence shares 121 positions with human P00167, including 75 of 77 heme-domain positions and both axial histidines. This supports a manual ISS inference of the conserved activity, corroborated by human WT/H44L CYB5A functional assays in PMID:22170710 and redox-active versus Mn-b5 experiments in PMID:27297105. The human assay measures lyase output, not electron-transfer kinetics or H44L heme loss. Neither the exact experimental rabbit plasmid sequence nor equal activity of all human splice products is inferred. The function describes electron movement and is distinct from the existing heme/enzyme/protein-binding terms; substrate-dependent allostery can coexist.