CYB5R3

Summary: The conserved reductase activity is supported; NADH is the informative donor.

Reason: The reconstructed 820-node PAINT tree and MSA place sequence-identical human CYB5R3 below the exact GO:0004128 IBD at PTN000452227. Current experimental donor records ground the ancestral activity, including legitimate human self-evidence. Human WT kinetics and the independently characterized CYB5R3 reaction support the narrower NADH-specific term. This is a chemical-specificity refinement, not a challenge to donor count or node placement.

Proposed replacements: cytochrome-b5 reductase activity, acting on NADH

Summary: Human reductase kinetics support the NADH-specific activity.

Reason: The complete PMID1400360 abstract describes purified recombinant human WT and Arg57Gln/Val105Met enzymes and compares NADH catalytic efficiency. The full original was not recovered, so unreported assay details are not invented. The positive curated activity is consistent with the measured NADH kinetics and independent human biochemistry; the specific NADH term improves the broad NAD(P)H label.

Proposed replacements: cytochrome-b5 reductase activity, acting on NADH

Summary: CYB5R3 has NADH-dependent reductase activity that is inhibited by ufmylation.

Reason: The actual PMID36543799 Methods assay NADH-dependent ferricyanide reduction, while the paper identifies the physiological electron-transfer role toward cytochrome b5. Ufmylation opens the FAD/NADH-domain arrangement and reduces activity. This particular endpoint is not a direct cytochrome-b5 reduction experiment; the positive curated class is retained and refined using the independently established human NADH reaction. The 275-residue ΔN26 construct is not official isoform 2.

Proposed replacements: cytochrome-b5 reductase activity, acting on NADH

Summary: The erythroid transcript supplies soluble NADH-cytochrome-b5 reductase.

Reason: PMID9639531 explicitly reports human M/S-transcript expression and HeLa translation products, with S-transcript-derived soluble reductase at late erythroid maturation. Its complete abstract supports the curated function and compartment distinction; the full original was not recovered. The canonical donor is NADH, supporting the specific child without inventing exact internal-initiation experiments.

Proposed replacements: cytochrome-b5 reductase activity, acting on NADH

Summary: Human WT and C203Y enzyme studies support NADH-dependent reductase activity.

Reason: The complete PMID10807796 abstract reports purified E. coli-expressed human enzymes, NADH catalytic measurements and reduced thermal stability of C203Y. WT activity is positively established; mutant instability is not a loss of the gene’s catalytic class. Refine the broad NAD(P)H term to the established NADH reaction, retaining historical residue numbering as reported.

Proposed replacements: cytochrome-b5 reductase activity, acting on NADH

Summary: The D239G study distinguishes the WT NADH reaction from altered mutant specificity.

Reason: PMID15953014 directly characterizes human recombinant reductase and its D239G variant. The variant binds stoichiometric FAD but decreases NADH preference and increases relative NADPH specificity. The gene-level WT activity supports GO:0090524; this does not assert that every disease mutant is unable to use NADPH or erase the mutant result.

Proposed replacements: cytochrome-b5 reductase activity, acting on NADH

Summary: The Ser127Pro study directly measures the human NADH/cytochrome-b5 reaction.

Reason: The complete PMID1898726 abstract reports human recombinant Ser127Pro enzyme and NADH kinetics with cytochrome b5 as electron acceptor. The specific NADH term is supported by the actual reaction; reduced mutant affinity does not negate WT catalysis.

Proposed replacements: cytochrome-b5 reductase activity, acting on NADH

Summary: Human cysteine-mutant kinetics retain the NADH-dependent catalytic class.

Reason: PMID2019583 reports human site-directed variants, NADH/FAD kinetic measurements and the conclusion that the studied cysteines are not catalytically essential. Refine to the specific NADH reaction without converting the mutant perturbations into a different physiological donor class.

Proposed replacements: cytochrome-b5 reductase activity, acting on NADH

Summary: Human WT and F298-deletion studies support the NADH-specific reductase.

Reason: The complete PMID8119939 abstract describes recombinant human WT and deletion enzymes, altered NADH kinetics and stability. The historical deletion numbering is retained as reported. These positive enzymatic data justify the more informative NADH-specific activity rather than the broad NAD(P)H parent.

Proposed replacements: cytochrome-b5 reductase activity, acting on NADH

Summary: The cited human kinetic study supports the specific NADH activity.

Reason: This TAS assertion cites the same PMID1400360 human WT/variant biochemical study as the separate IDA row. Both literal source objects are retained. NADH kinetic evidence and independent human biochemical characterization justify the child term; no unpublished assay or full-text access is claimed.

Proposed replacements: cytochrome-b5 reductase activity, acting on NADH

Summary: CYB5R3 associates with hemoglobin alpha in vascular and purified-protein experiments.

Reason: Actual PMID23123858 Results and Fig.4 text report Hbα association with CYB5R3 using purified proteins, vascular-cell cocultures and arterial lysates. Hbβ is not detected in this endothelial context. GO:0031721 hemoglobin alpha binding fits the demonstrated partner; the broader hemoglobin-binding definition describes an oxygen-carrying tetramer and is not substituted for the actual alpha-chain evidence. The literal P69905 source partner remains unchanged.

Proposed replacements: hemoglobin alpha binding

Summary: The source-specific CANX association supports chaperone binding.

Reason: The exact current IntAct record joins human CYB5R3 and calnexin P27824 under PMID30021884. It records crosslinking mass spectrometry in U2OS material, specifically the Triton-X100-soluble fraction of fractionated nuclei and Supplementary Table S2. Calnexin is a protein-folding chaperone, so its class supports GO:0051087. This is not a purified binary-binding assay, a folding-activity measurement by CYB5R3, or evidence that CYB5R3 is a nuclear core protein. The cached text lacks complete Methods/Results despite its full-text flag.

Proposed replacements: protein-folding chaperone binding

Summary: The CDIPT (O14735) interaction supports a specific partner-binding class.

Reason: CDIPT is phosphatidylinositol synthase. Enzyme binding is a broad but biologically defined catalytic-partner class, supported here by the positive pair and the independently established CDIPT identity; it is not selected merely to satisfy a validator. It does not assign CDIPT chemistry to CYB5R3 or show that CYB5R3 regulates the enzyme. Three records from PMID:32296183 match the human partner and CYB5R3 sequences/taxa through their ROGIDs and record array, pooling and validated yeast two-hybrid formats from one study. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: enzyme binding

Summary: The TNFRSF10C (O14798) interaction is retained as a contextual protein interaction.

Reason: TNFRSF10C is a TRAIL-binding decoy lacking the intracellular death domain. The positive HuRI association is retained as contextual protein binding; no apoptotic signaling role is inferred for CYB5R3. This conservative retention does not require CYB5R3-mediated receptor activation. Three literal records match the human participant sequences and taxa in the recorded yeast two-hybrid assay context; native-human pathway participation is not established.

Summary: The VAMP4 (O75379) interaction supports a specific partner-binding class.

Reason: VAMP4 is a vesicular SNARE; binding to this partner supports SNARE binding without establishing membrane-fusion activity by CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: SNARE binding

Summary: The LRCH4 (O75427) interaction is retained as a contextual protein interaction.

Reason: LRCH4 is a leucine-rich-repeat/calponin-homology protein. The pair is positive, but the assay does not establish a more specific binding function for CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The VAMP5 (O95183) interaction supports a specific partner-binding class.

Reason: VAMP5 is a SNARE-family vesicle-associated membrane protein; the supported partner class is SNARE binding. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: SNARE binding

Summary: The CNIH1 (O95406) interaction is retained as a contextual protein interaction.

Reason: CNIH1 is a cargo-trafficking protein. The interaction alone does not establish CYB5R3 as a cargo receptor or transporter. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The GAST (P01350) interaction is retained as a contextual protein interaction.

Reason: P01350 identifies the gastrin precursor. The experiment does not specify a processed gastrin peptide, so mature peptide-hormone binding is not inferred from the precursor name. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The TNF (P01375) interaction supports a specific partner-binding class.

Reason: The positive pair identifies the full 233-residue TNF precursor P01375, supporting the TNF-binding class rather than a demonstrated interaction with the processed soluble cytokine. The official sequence-based topology assigns residues 1–35 to the cytoplasm; the extracellular TNF-homology domain therefore does not make every possible contact with cytosol-facing CYB5R3 impossible. This is not evidence that CYB5R3 binds that cytoplasmic region. GO:0043120 does not require binding the extracellular cytokine surface or inducing TNF signaling. The three PMID:32296183 records are array, pooling and validated yeast two-hybrid record types from the same study, not independent physiological validations. Their ROGIDs match the annotated human protein sequences and taxa. The interaction interface, native colocalization and processing of the fusion proteins in yeast remain unresolved.

Proposed replacements: tumor necrosis factor binding

Summary: The ASGR1 (P07306) interaction supports signaling receptor binding.

Reason: ASGR1 is an endocytic glycoprotein receptor whose official functional record also describes ligand-dependent JAK2–STAT3 feedback regulating thrombopoietin expression after recognition of desialylated platelets. This independently established receptor class and the positive HuRI pair support GO:0005102. The term does not require CYB5R3 to be an agonist or recognize the physiological ligand; no CYB5R3 effect on JAK2–STAT3, receptor activation or native signaling is demonstrated. The recorded partner backbone is the 291-residue P07306 sequence, without retrofitting an isoform field into the source assertion. The three PMID:32296183 records are array, pooling and validated yeast two-hybrid record types from the same study, not independent physiological validations. Their ROGIDs match the annotated human protein sequences and taxa. The interaction interface, native colocalization and processing of the fusion proteins in yeast remain unresolved.

Proposed replacements: signaling receptor binding

Summary: The VAMP1 (P23763-3) interaction supports a specific partner-binding class.

Reason: The literal partner is VAMP1 product P23763-3 (officially named isoform 2) (P23763-3), whose verified sequence retains the SNARE identity. Preserve that product rather than substituting canonical VAMP1. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: SNARE binding

Summary: The STOM (P27105) interaction is retained as a contextual protein interaction.

Reason: STOM is stomatin, a membrane-organizing protein. Binding is retained as a contextual interaction without an unsupported catalytic or transporter-binding refinement. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The NSG1 (P42857) interaction is retained as a contextual protein interaction.

Reason: NSG1 is neuronal vesicle trafficking-associated protein 1. This assay supports association, not a trafficking activity performed by CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The PNLIPRP1 (P54315) interaction is retained as a contextual protein interaction.

Reason: PNLIPRP1 is the inactive pancreatic lipase-related protein 1. Its lipase-family name does not establish a catalytically active enzyme partner, so generic positive binding is retained. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The ADAM33 (Q08AM2) interaction is retained as a contextual protein interaction.

Reason: The exact 692-residue partner is unreviewed ADAM33 entry Q08AM2, whose official record cautions that conserved features cannot be propagated. Binding remains positive, but active metalloprotease identity is not assumed from the gene name. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The TMEM115 (Q12893) interaction is retained as a contextual protein interaction.

Reason: TMEM115 is a membrane trafficking-associated protein; a specific binding class is not demonstrated by this pair. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The BNIP2 (Q12982) interaction is retained as a contextual protein interaction.

Reason: BNIP2 is the BCH-domain protein, not a BCL2-family protein merely because of its name. The pair supports contextual binding without a BCL2-binding refinement. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The AQP6 (Q13520) interaction supports a specific partner-binding class.

Reason: AQP6 is a transmembrane channel. Its established transporter identity supports transmembrane transporter binding without assigning transport or channel regulation to CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: transmembrane transporter binding

Summary: The LAPTM5 (Q13571) interaction is retained as a contextual protein interaction.

Reason: LAPTM5 is a lysosome-associated transmembrane protein. The assay does not establish a substrate-translocating function for this partner, so its membrane location alone is not used to infer transporter binding. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The GRM2 (Q14416) interaction supports a specific partner-binding class.

Reason: Q14416 is the type 2 metabotropic glutamate receptor GRM2. Its official sequence-based topology includes cytoplasmic loops 591–604, 656–679 and 748–760 and a cytoplasmic tail 820–872, so an extracellular ligand-binding domain does not exclude every possible cytoplasmic contact. The positive pair supports the receptor-binding class without identifying any of these regions as the CYB5R3 interface or establishing agonism or receptor activation. The three PMID:32296183 records are array, pooling and validated yeast two-hybrid record types from the same study, not independent physiological validations. Their ROGIDs match the annotated human protein sequences and taxa. The interaction interface, native colocalization and processing of the fusion proteins in yeast remain unresolved.

Proposed replacements: type 2 metabotropic glutamate receptor binding

Summary: The WFDC2 (Q14508) interaction is retained as a contextual protein interaction.

Reason: WFDC2 is the WAP-domain secreted protein HE4. The screen does not establish which processing state is bound or a specific inhibitory-partner role. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The FXYD3 (Q14802-3) interaction is retained as a contextual protein interaction.

Reason: The literal FXYD3 isoform 3 partner is recorded as residues 58–144. This fragment-level positive interaction is retained without extrapolating to binding an intact ion-transport complex. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The SLC10A1 (Q14973) interaction supports a specific partner-binding class.

Reason: SLC10A1 is the sodium/bile-acid cotransporter. Its established substrate-translocation class supports transporter binding; no bile-acid transport by CYB5R3 is implied. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: transmembrane transporter binding

Summary: The SLC10A6 (Q3KNW5) interaction supports a specific partner-binding class.

Reason: SLC10A6 is a sodium-dependent organic-anion transporter, supporting transmembrane transporter binding at the exact partner scope. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: transmembrane transporter binding

Summary: The COL8A2 (Q4VAQ0) interaction is retained as a contextual protein interaction.

Reason: The tested Q4VAQ0 is a 427-residue unreviewed COL8A2 product with a C1q domain, not an experimentally identified mature collagen VIII assembly. Retain the positive construct interaction without extending it to binding native extracellular collagen. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The ORMDL2 (Q53FV1) interaction is retained as a contextual protein interaction.

Reason: ORMDL2 is a regulatory component associated with serine palmitoyltransferase. It is not the catalytic enzyme itself; generic binding is retained rather than assigning enzyme binding from complex association. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The TMEM201 (Q5SNT2-2) interaction is retained as a contextual protein interaction.

Reason: The literal TMEM201 isoform 2 is a nuclear-envelope-associated membrane protein. No narrower functional binding class is established by this pair. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The AGTRAP (Q6RW13-2) interaction is retained as a contextual protein interaction.

Reason: The partner is AGTRAP isoform 2, an angiotensin-receptor-associated protein rather than the angiotensin receptor itself. The positive interaction does not support receptor binding by name association. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The MFSD6 (Q6ZSS7) interaction is retained as a contextual protein interaction.

Reason: MFSD6 is recorded as residues 329–791. Its major-facilitator-family identity and this partial construct do not establish a defined transport activity of the assayed partner, so the supported interaction remains non-core. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The HSD17B13 (Q7Z5P4) interaction supports a specific partner-binding class.

Reason: HSD17B13 is a biochemically characterized dehydrogenase. Enzyme binding is supported at the exact partner scope; isoform-dependent retinol activity is not generalized from this interaction, and no substrate-specific catalytic role is assigned to CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: enzyme binding

Summary: The NRM (Q8IXM6) interaction is retained as a contextual protein interaction.

Reason: NRM is nurim, a nuclear-envelope membrane protein. Positive binding is retained without manufacturing a more specific molecular-function class. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The CSGALNACT2 (Q8N6G5) interaction supports a specific partner-binding class.

Reason: CSGALNACT2 is a chondroitin N-acetylgalactosaminyltransferase. Its catalytic identity supports enzyme binding without glycosyltransferase activity by CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: enzyme binding

Summary: The SLC35B2 (Q8TB61) interaction supports a specific partner-binding class.

Reason: SLC35B2 is a transmembrane PAPS transporter, supporting transporter binding rather than sulfonation chemistry by CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: transmembrane transporter binding

Summary: The GPR152 (Q8TDT2) interaction supports a specific partner-binding class.

Reason: Q8TDT2 is officially named Probable G protein-coupled receptor 152. Its GPCR-family classification is inferred by PROSITE-ProRule PRU00521, and its official GO record also carries GO-Central GPCR-activity IBA; neither is a new direct ligand/coupling assay. The positive pair supports this partner-binding class. The official sequence-based topology includes cytoplasmic regions 55–65, 127–148, 216–257 and 303–470, without identifying any as the CYB5R3 interface. There is no demonstrated ligand, receptor activation or native signaling effect of CYB5R3. The three PMID:32296183 records are array, pooling and validated yeast two-hybrid record types from the same study, not independent physiological validations. Their ROGIDs match the annotated human protein sequences and taxa. The interaction interface, native colocalization and processing of the fusion proteins in yeast remain unresolved.

Proposed replacements: G protein-coupled receptor binding

Summary: The SLC13A3 (Q8WWT9) interaction supports a specific partner-binding class.

Reason: SLC13A3 is a sodium-dependent dicarboxylate transporter; its established transport function supports transmembrane transporter binding. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: transmembrane transporter binding

Summary: The FAM3C (Q92520) interaction is retained as a contextual protein interaction.

Reason: FAM3C is a secreted signaling-associated protein. The tested sequence does not by itself establish a mature cytokine-binding function or a signaling consequence for CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The NINJ1 (Q92982) interaction is retained as a contextual protein interaction.

Reason: NINJ1 is a membrane-associated adhesion and membrane-rupture protein. The interaction does not establish CYB5R3 participation in membrane rupture or a specific adhesion mechanism. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The NXPE3 (Q969Y0) interaction is retained as a contextual protein interaction.

Reason: NXPE3 is a neurexophilin/PC-esterase-domain protein. A domain name alone does not demonstrate catalytic esterase activity of the tested partner. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The SLC35A4 (Q96G79) interaction supports a specific partner-binding class.

Reason: SLC35A4 has directly supported CDP-ribitol transport despite its broader probable UDP-sugar-transporter name. The verified transmembrane-transporter class supports the refinement without guessing another transported substrate. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: transmembrane transporter binding

Summary: The VMP1 (Q96GC9) interaction supports a specific partner-binding class.

Reason: VMP1 is an experimentally characterized phospholipid scramblase that moves lipids between membrane leaflets. This is a transporter class; binding does not make CYB5R3 an autophagy effector or scramblase. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: transmembrane transporter binding

Summary: The FAXDC2 (Q96IV6) interaction is retained as a contextual protein interaction.

Reason: FAXDC2 contains a fatty-acid-hydroxylase domain, but the retrieved official functional evidence describes differentiation rather than a demonstrated catalytic reaction. The domain name alone is insufficient for an enzyme-binding refinement. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The MIMS2 (Q96KR6) interaction is retained as a contextual protein interaction.

Reason: MIMS2 is mitochondrial import inner membrane translocase subunit 10B. Its participation in an import assembly is not equated with an independently catalytic or transmembrane-transporter activity. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The ZDHHC15 (Q96MV8) interaction supports a specific partner-binding class.

Reason: ZDHHC15 is a palmitoyltransferase. The exact catalytic partner class supports enzyme binding without demonstrating CYB5R3 palmitoylation or acyltransferase activity. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: enzyme binding

Summary: The TMEM79 (Q9BSE2) interaction is retained as a contextual protein interaction.

Reason: TMEM79 is a membrane protein involved in epidermal biology. The positive interaction does not establish a narrower binding class or a skin-barrier process role for CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The ALG8 (Q9BVK2) interaction supports a specific partner-binding class.

Reason: ALG8 is a dolichyl-linked oligosaccharide glucosyltransferase. Enzyme binding captures the exact partner class without assigning glycosylation chemistry to CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: enzyme binding

Summary: The FUNDC2 (Q9BWH2) interaction is retained as a contextual protein interaction.

Reason: FUNDC2 is a mitochondrial membrane protein. This positive yeast two-hybrid interaction is distinct from the independently sourced BioPlex association and does not establish a shared complex mechanism. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The ELOVL4 (Q9GZR5) interaction supports a specific partner-binding class.

Reason: ELOVL4 is a very-long-chain fatty-acid elongation enzyme. Its catalytic identity supports enzyme binding, without making the screen an assay of fatty-acid elongation by CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: enzyme binding

Summary: The TMEM60 (Q9H2L4) interaction is retained as a contextual protein interaction.

Reason: The TMEM60 partner is explicitly recorded as residues 3–133. Binding to that fragment remains positive but does not establish the physiology of an intact membrane assembly. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The REEP4 (Q9H6H4) interaction is retained as a contextual protein interaction.

Reason: REEP4 is a membrane-shaping/endoplasmic-reticulum protein. The screen supports binding without a specific enzymatic or cargo-receptor function for CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The CERS4 (Q9HA82) interaction supports a specific partner-binding class.

Reason: CERS4 is a ceramide synthase. The verified catalytic class supports enzyme binding without attributing ceramide synthesis or its regulation to CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: enzyme binding

Summary: The TMEM14B (Q9NUH8) interaction is retained as a contextual protein interaction.

Reason: TMEM14B is a mitochondrial membrane protein. The pair does not establish a substrate-translocating function, so a transporter-binding refinement is not inferred from membrane residence. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The SLC38A7 (Q9NVC3) interaction supports a specific partner-binding class.

Reason: SLC38A7 is a lysosomal amino-acid transporter, supporting transmembrane transporter binding at the exact partner scope. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: transmembrane transporter binding

Summary: The AIG1 (Q9NVV5-2) interaction supports a specific partner-binding class.

Reason: The exact AIG1 partner Q9NVV5-2 matches the current displayed 238-residue enzyme sequence, which has fatty-acid-ester hydrolase activity. Retain the literal isoform suffix; enzyme binding does not imply CYB5R3 catalyzes that reaction. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: enzyme binding

Summary: The TMEM51 (Q9NW97) interaction is retained as a contextual protein interaction.

Reason: TMEM51 is a membrane protein without a demonstrated narrower catalytic or transporter class in the retrieved official record. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The TMEM14C (Q9P0S9) interaction is retained as a contextual protein interaction.

Reason: TMEM14C participates in erythroid heme metabolism, but this interaction does not establish that CYB5R3 performs a porphyrin-transport step or binds a proven transporter. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Summary: The UBIAD1 (Q9Y5Z9) interaction supports a specific partner-binding class.

Reason: UBIAD1 is a prenyltransferase. Its catalytic partner identity supports enzyme binding without assigning prenyltransferase chemistry to CYB5R3. Three literal PMID32296183 records independently match the human partner and CYB5R3 sequences/taxa through their ROGIDs. They record array, pooling and validated yeast two-hybrid assays; yeast is the host. The source supports a biophysical interaction, not a native-human pathway or localization claim.

Proposed replacements: enzyme binding

Summary: The BioPlex CYB5R3–stomatin association is retained.

Reason: One exact current PMID33961781 record joins CYB5R3 with STOM/P27105 in anti-tag co-immunoprecipitation/mass-spectrometry material from HEK293T cells, with the recorded C-terminal tandem tag. This supports contextual association, distinct from the separate HuRI pair. It does not establish direct purified binding or a specific redox/membrane-organizing function for the association.

Summary: The BioPlex CYB5R3–FUNDC2 association is retained.

Reason: One exact current PMID33961781 record joins CYB5R3 with FUNDC2/Q9BWH2 in anti-tag co-immunoprecipitation/mass-spectrometry material from HEK293T cells. The positive association is retained independently of the separate HuRI source. Neither copurification nor shared mitochondrial context establishes a specific catalytic or mitophagy role for this pair.

Summary: The original OpenCell dataset directly records CYB5R3 in the CANX pulldown.

Reason: Although the current literal IntAct pair is absent, the genuine author dataset contains the CANX-bait/CYB5R3-prey interaction at CSV row 2364. The original Methods describe endogenously tagged HEK293T targets, digitonin-solubilized lysates, anti-mNeonGreen immunoprecipitation and triplicate mass spectrometry. The prey protein group contains P00387, P00387-2, P00387-3 and B1AHF3, so the association is not isoform resolved. Calnexin’s verified folding-chaperone class supports the refinement without claiming a purified binary interaction or chaperone activity of CYB5R3.

Proposed replacements: protein-folding chaperone binding

Summary: The exact extracellular region claim remains unresolved despite positive granule-associated evidence.

Reason: Reactome R-HSA-6798751 explicitly includes human CYB5R3(2–301) in the lumen input and extracellular output sets. Independent original PMID25244098 Supplementary Table 1 identifies CYB5R3/IPI00328415 in the MPO-enriched neutrophil alpha fraction (control/PLS ratio 138.6; 0.27% of alpha-fraction protein intensity). The actual Methods use whole-organelle Percoll fractionation and gel proteomics, which do not resolve target-specific release into the extracellular region. Product-specific peptides were not independently reconstructed. Expected cytoplasmic-facing topology does not prove impossibility, and the NGAL-focused abstract of the event citation does not invalidate the model. Preserve the positive assertions and target detection while leaving this precise compartment unresolved.

Summary: The soluble erythroid CYB5R3 product supports cytoplasmic localization.

Reason: The complete PMID9639531 abstract directly identifies S-transcript-derived soluble protein in transfected HeLa cells and late erythroid expression. The official 278-residue product lacks canonical residues 1–23; this differs from the 275-residue recombinant truncations used elsewhere. Positive cytoplasmic localization is retained without treating all three products as equivalent.

Summary: The official subcellular-location mapping correctly includes cytoplasm.

Reason: The literal SL-0086 source corresponds to cytoplasm and is supported by the soluble erythroid product and PMID9639531. It is a product-dependent location, not a claim that the myristoylated membrane product or the 334-residue alternative product behaves identically.

Summary: Cytoplasmic localization is consistent with the human alternative-transcript system.

Reason: The complete PMID9207238 abstract describes alternative first exons in human reticulocytes, liver, brain and HL-60 cells; that earlier study did not identify an erythroid-specific transcript. The later PMID9639531 work directly establishes the soluble S-transcript-derived product. Retain the positive broad location holistically while preserving the original TAS citation and distinguishing the two papers’ transcript conclusions.

Summary: Mitochondrial localization is supported, with the original proteomic-table limit explicit.

Reason: PMID20833797 is a human skeletal-muscle mitochondrial phosphoproteomics study. Its full target table was not recovered, so the exact historical peptide/isoform assignment remains unverified. The curated HDA location is biologically corroborated by the independently inspected PMID34800366 target protein group and the membrane-associated CYB5R3 product. This does not assign CYB5R3 to the inner membrane or matrix from the paper title.

Summary: CYB5R3 is explicitly present in the high-confidence MitoCoP dataset.

Reason: The genuine PMID34800366 workbook includes CYB5R3/P00387 in the MitoCoP gene list (sheet B, row 252) and its quantified protein group (sheet A, row 279), with classification and membrane-association measurements. The protein group contains multiple CYB5R3 products, so it is not isoform resolved. The selected Methods describe human cell-line mitochondrial proteomics and filtering; the data support mitochondrion localization without narrowing it to matrix or inner membrane.

Summary: The actual PAINT ancestry supports conserved mitochondrial localization.

Reason: The exact GO:0005739 IBD at metazoan PTN001064780 is on the reconstructed 20-node target path, with sequence-identical target MSA and no intervening loss record. RGD:2502 resolves through the official cross-reference to rat P20070; its current exact-term experimental query is empty, a recorded contemporary-source gap. Human experimental mitochondrial proteomics remains present and independently substantiated by MitoCoP. Legitimate human self-evidence and the phylogenetic placement are not circularity or donor-count defects.

Summary: The membrane-associated CYB5R3 product has an established mitochondrial outer-membrane location.

Reason: The literal SL-0172 mapping is consistent with the official myristoylated membrane product, the two independently traced Reactome catalyst assemblies and mitochondrial proteomics. The cytoplasmic-facing topology is partly supported by ortholog evidence in the official record; this row does not demonstrate an identical location for soluble isoform 2 or uncharacterized isoform 3.

Summary: Reactome explicitly places CYB5R3:FAD at the mitochondrial outer membrane.

Reason: The actual R-HSA-198824 catalyst path is 198863 to complex 198850 to human CYB5R3 entity 198771, carrying the P00387 reference and residues 2–301 with myristoylation. This is an explicit target component and compartment, not a location inferred merely from a pathway title. It supports the membrane product and does not identify every alternative product.

Summary: The mARC-associated Reactome assembly explicitly contains outer-membrane CYB5R3.

Reason: The actual R-HSA-8936442 catalyst path is 8936434 to complex 8936430 to the same human membrane CYB5R3 entity 198771. CYB5R3 supplies reducing equivalents through cytochrome b5; mARC performs reduction of the terminal N-hydroxylated substrate. The literal target membership and compartment support this location without transferring mARC catalysis to CYB5R3.

Summary: Human Protein Atlas immunofluorescence supports endoplasmic-reticulum localization.

Reason: The genuine HPA record reports supported ER localization in A-431, U-251 MG and U2OS cells with antibody 1566. This source is not isoform resolved and no image pixels were independently assessed. The ER location is corroborated by the distinct endogenous fractionation and tagged-protein experiments in PMID36543799.

Summary: Endogenous and expressed CYB5R3 localize to the ER membrane in the source study.

Reason: Actual PMID36543799 Results show endogenous CYB5R3 principally in HEK293T microsomes and colocalization of tagged protein with PDI in HeLa cells. The ΔN26 construct is cytoplasmic, supporting the N-terminal membrane-association requirement. The result supports the membrane product; it is not evidence for identical distribution of all official splice products.

Summary: The membrane-associated human transcript supports ER-membrane localization.

Reason: PMID9639531 explicitly distinguishes the membrane-associated and soluble erythroid products, although its full original localization Methods were not recovered. The positive curated ER-membrane assertion is independently corroborated by human PMID36543799 and official product architecture. This is curator deference with positive corroboration, not an inferred absence of unseen evidence.

Summary: The electronic ER-membrane mapping matches actual target conditions and experimental localization.

Reason: Both current ARBA00028685 condition sets 5 and 21 match the target: the former uses IPR001433/IPR001834 plus Euarchontoglires, and the latter uses the two exact FunFam signatures plus that taxon. SL-0097 agrees. All predicates were evaluated against actual target entries, and human PMID36543799 independently supports the membrane location; no unknown predicate or branch is silently assumed.

Summary: The precise lipid-droplet assertion remains unresolved at the original target-table scope.

Reason: The PMID14741744 abstract describes lipid-droplet-enriched fractions from human HuH7 cells but does not identify CYB5R3 among its named results. Bounded original/table recovery did not retrieve the target evidence. ER-membrane localization is not itself a refutation of lipid-droplet association, and the curated IDA assertion is preserved pending the actual protein identification and fraction-purity evidence.

Summary: Hb-alpha association and redox maintenance do not establish membership in the oxygen-carrying globin complex.

Reason: GO:0005833 describes the iron-containing oxygen carrier assembled from globin chains. CYB5R3 supplies redox maintenance and can associate with Hb alpha; it is not thereby assigned the globin-subunit role of this defined complex. The complete PMID:1400360 abstract measures reductase variant kinetics and stability, while its full original remains unrecovered. Independently read PMID:23123858 Results and Figure 4 establish Hb-alpha association using lysates and purified proteins. These positive reductase/binding findings do not support the more specific oxygen-carrying-complex assertion. Mark the term as over-annotated rather than claim that hemoglobin association is false or that the inaccessible original contains no relevant assay. The original TAS source remains unchanged.

Summary: CYB5R3 performs electron delivery in the reconstituted nitrite-to-NO pathway.

Reason: Actual PMID24500710 Methods use His-tagged human 278-residue CYB5R3 isoform 2, cytochrome b5 and recombinant mARC. CYB5R3 supplies electrons from NADH through b5; mARC molybdenum catalyzes nitrite reduction to NO. This is a performed reductase contribution to the process, not merely necessity or a substrate role. The reconstituted and hypoxic assay context supports a non-core NO-biosynthesis annotation without asserting canonical arginine-dependent nitric-oxide-synthase chemistry.

Summary: CYB5R3 supports a contextual vascular redox role.

Reason: The original PMID1400360 citation concerns methemoglobinemia-associated reductase instability, with full experimental details unavailable. Independent PMID23123858 directly links CYB5R3/Hbα redox control to NO signaling and vascular reactivity in cocultures and arteries. This positive mechanistic corroboration supports contextual circulation involvement while preserving the original citation and avoiding a general circulatory-development function.

Summary: The broad membrane annotation is consistent with the established membrane product.

Reason: PMID19946888 is an experimental YTS natural-killer-cell membrane proteomics source; its exact CYB5R3 target table was not recovered. The positive broad membrane class is independently established by CYB5R3 architecture, ER fractionation and mitochondrial proteomics. Retain the curated source with that access limit rather than inferring an NK-specific organelle or isoform assignment.

Summary: Human alternative-first-exon expression includes the membrane-associated reductase.

Reason: The complete PMID9207238 abstract examines alternative human first exons and tissue expression. The broad membrane assertion agrees with the known M product and later directly observed ER localization. It is not a claim that the earlier study established the later erythroid-specific transcript conclusion.

Summary: The broad InterPro oxidoreductase mapping can be refined to the human NADH reaction.

Reason: All four literal InterPro sources are genuine target matches and legitimately support oxidoreductase activity. Their families span several electron-transfer systems; domain membership alone would not specify the complete reaction. Human biochemical evidence independently establishes NADH-cytochrome-b5 reductase activity, justifying the informative child term without disputing the broad electronic mapping.

Proposed replacements: cytochrome-b5 reductase activity, acting on NADH

Summary: The curated mitochondrial-membrane location is positively corroborated.

Reason: The full PMID9639531 original was not recovered, so its exact mitochondrial-localization assay is not claimed. The historical IDA is consistent with the independently verified outer-membrane Reactome target entities, official membrane-product localization and human MitoCoP data. Retain the broader positive source term without silently converting it to an isoform-specific experiment.

Summary: The exact azurophil granule lumen claim remains unresolved despite positive granule-associated evidence.

Reason: Reactome R-HSA-6798751 explicitly includes human CYB5R3(2–301) in the lumen input and extracellular output sets. Independent original PMID25244098 Supplementary Table 1 identifies CYB5R3/IPI00328415 in the MPO-enriched neutrophil alpha fraction (control/PLS ratio 138.6; 0.27% of alpha-fraction protein intensity). The actual Methods use whole-organelle Percoll fractionation and gel proteomics, which do not resolve lumen rather than granule-membrane placement. Product-specific peptides were not independently reconstructed. Expected cytoplasmic-facing topology does not prove impossibility, and the NGAL-focused abstract of the event citation does not invalidate the model. Preserve the positive assertions and target detection while leaving this precise compartment unresolved.

Summary: The b5/mARC electron-transfer system does not establish the cytochrome-c-defined nitrite-reductase class.

Reason: The original contributes_to assertion is assessed as a genuine complex contribution, not as autonomous terminal catalysis by CYB5R3. PMID:24500710 reconstitutes NADH–CYB5R3–CYB5B–mARC nitrite reduction, with mARC active-site controls and no NO production when mARC is omitted. Current GO:0050421 instead specifies a cytochrome-c redox reaction. Even under the weaker complex-contribution interpretation, the actual b5 system does not establish that particular catalytic class, so this term is over-annotated. The positive reductase contribution and contextual NO-biosynthesis judgment are retained. GO:0098809 is not forced as a replacement because its literal reaction wording remains problematic; the ontology question preserves that separate uncertainty.

Summary: The FAD-binding IBA has valid on-path placement and experimental grounding.

Reason: The exact GO:0071949 IBD at PTN000452227 lies on the human target path. The MSA matches P00387 exactly; the literal descendant set has current experimental FAD-binding evidence, including human structures/biochemistry and mapped rat P20070. Human self-evidence is legitimate descendant grounding, not circularity. No target-specific loss or unsupported node placement was found.

Summary: FAD binding is integral to the experimentally characterized human reductase.

Reason: The positive PMID1400360 IDA assertion is consistent with purified human WT/variant reductase biochemistry. Full original access remains unavailable, but independent human structural and stoichiometric FAD evidence directly corroborates the function; the annotation is not rejected from abstract silence.

Summary: The human crystal structure directly resolves bound FAD.

Reason: The recovered PMID15502298 author original and PDB 1UMK establish the FAD-containing human catalytic structure. The bacterial, thrombin-cleaved construct is 275 residues (paper positions 26–300), not the official 278-residue soluble product. FAD binding is supported without making those constructs identical.

Summary: The D239G biochemical study directly supports FAD binding.

Reason: The complete PMID15953014 abstract reports stoichiometric FAD in the recombinant human mutant while documenting altered nicotinamide-cofactor specificity. FAD association is positive despite diminished NADH catalytic efficiency; this is not an inference from a domain label alone.

Summary: Human WT and C203Y enzyme studies support NADH-dependent reductase activity.

Reason: The complete PMID10807796 abstract reports purified E. coli-expressed human enzymes, NADH catalytic measurements and reduced thermal stability of C203Y. WT activity is positively established; mutant instability is not a loss of the gene’s catalytic class. The existing specific NADH term matches that reaction; historical residue numbering remains as reported.

Summary: Human reductase kinetics support the NADH-specific activity.

Reason: The complete PMID1400360 abstract describes purified recombinant human WT and Arg57Gln/Val105Met enzymes and compares NADH catalytic efficiency. The full original was not recovered, so unreported assay details are not invented. The positive curated activity is consistent with the measured NADH kinetics and independent human biochemistry; the existing specific NADH term is supported.

Summary: The D239G study distinguishes the WT NADH reaction from altered mutant specificity.

Reason: PMID15953014 directly characterizes human recombinant reductase and its D239G variant. The variant binds stoichiometric FAD but decreases NADH preference and increases relative NADPH specificity. The gene-level WT activity supports GO:0090524; this does not assert that every disease mutant is unable to use NADPH or erase the mutant result.

Summary: The Ser127Pro study directly measures the human NADH/cytochrome-b5 reaction.

Reason: The complete PMID1898726 abstract reports human recombinant Ser127Pro enzyme and NADH kinetics with cytochrome b5 as electron acceptor. The existing specific NADH term is supported by the actual reaction; reduced mutant affinity does not negate WT catalysis.

Summary: Human cysteine-mutant kinetics retain the NADH-dependent catalytic class.

Reason: PMID2019583 reports human site-directed variants, NADH/FAD kinetic measurements and the conclusion that the studied cysteines are not catalytically essential. Retain the specific NADH reaction without converting the mutant perturbations into a different physiological donor class.

Summary: Human WT and F298-deletion studies support the NADH-specific reductase.

Reason: The complete PMID8119939 abstract describes recombinant human WT and deletion enzymes, altered NADH kinetics and stability. The historical deletion numbering is retained as reported. These positive enzymatic data support the existing NADH-specific activity.

Summary: The erythroid transcript supplies soluble NADH-cytochrome-b5 reductase.

Reason: PMID9639531 explicitly reports human M/S-transcript expression and HeLa translation products, with S-transcript-derived soluble reductase at late erythroid maturation. Its complete abstract supports the curated function and compartment distinction; the full original was not recovered. The canonical donor is NADH, supporting the existing specific term without inventing exact internal-initiation experiments.

Summary: The NADH-specific electronic reaction mapping is supported.

Reason: The literal RHEA:46680 and EC:1.6.2.2 mappings match the human NADH–cytochrome-b5 reductase reaction and independent experiments. Current ARBA00098155 retrieval returned no rule, so its historical predicate set is explicitly unresolved rather than reconstructed from its identifier. The convergent reaction mappings and human biochemical evidence support the existing activity despite that one unavailable rule source.

Summary: The demonstrated nitrite-reducing system is not the defined nitric-oxide-synthase complex.

Reason: Actual PMID24500710 reconstitutes NADH–CYB5R3–cytochrome-b5–mARC reduction of nitrite to NO. Current GO:1903958 is defined through nitric-oxide synthase GO:0004517, whose reaction uses L-arginine, NADPH and oxygen to produce citrulline and NO. The source’s use of nitric-oxide-synthase wording does not establish that distinct chemistry or complex class. Preserve the experimentally supported electron-transfer system and contributes_to role elsewhere; this is a source/definition mismatch, not denial of NO production.