FLAD1 encodes human FAD synthase (FMN adenylyltransferase; FMNAT; FADS; EC 2.7.7.2),
the enzyme that catalyzes the final, committed step of FAD-cofactor biosynthesis:
adenylylation of flavin mononucleotide (FMN) with ATP to yield flavin adenine
dinucleotide (FAD) plus diphosphate. FAD is the cofactor of hundreds of flavoenzymes
(respiratory chain Complex I/II, acyl-CoA dehydrogenases, MTHFR, dimethylglycine
dehydrogenase, monoamine oxidase, etc.). The reaction requires a divalent metal,
optimally Mg2+.
[file:human/FLAD1/FLAD1-uniprot.txt "FAD synthase catalyzes the adenylation of flavin mononucleotide (FMN) to form flavin adenine dinucleotide (FAD) coenzyme"]
[file:human/FLAD1/FLAD1-uniprot.txt "Reaction=FMN + ATP + H(+) = FAD + diphosphate"]
PMID:16643857
PMID:16643857
The human enzyme is bifunctional and multidomain, distinguishing it from the
single-domain yeast Fad1p:
- C-terminal PAPS-reductase domain (FADSy; ~residues 355-587) = the FAD-forming
catalytic (FMN adenylyltransferase) domain.
- N-terminal molybdopterin-binding-resembling (MPTb) domain (with a KH domain) =
a FAD hydrolase / FAD diphosphatase (FADHy) activity (EC 3.6.1.18), and, at lower
efficiency, an NADH diphosphatase / NADH pyrophosphatase activity (EC 3.6.1.22).
[file:human/FLAD1/FLAD1-uniprot.txt "This enzyme has two activities: FAD diphosphatase activity"]
[file:human/FLAD1/FLAD1-uniprot.txt "In the N-terminal section; belongs to the MoaB/Mog family"]
[file:human/FLAD1/FLAD1-uniprot.txt "In the C-terminal section; belongs to the PAPS reductase"]
PMID:26277395
PMID:38688286
FMN + ATP + H+ = FAD + diphosphate (EC 2.7.7.2, RHEA:17237) — the physiological/biosynthetic direction.FAD + H2O = FMN + AMP + 2 H+ (EC 3.6.1.18, RHEA:13889).NADH + H2O = reduced NMN + AMP + 2 H+ (EC 3.6.1.22, RHEA:48868).[file:human/FLAD1/FLAD1-uniprot.txt "Reaction=FAD + H2O = FMN + AMP + 2 H(+)"]
[file:human/FLAD1/FLAD1-uniprot.txt "PATHWAY: Cofactor biosynthesis; FAD biosynthesis; FAD from FMN: step"]
Note the direction convention: GO:0003919 FMN adenylyltransferase activity is defined
ATP + FMN = diphosphate + FAD (the synthase direction), while GO:0047884 FAD
diphosphatase activity is FAD + H2O = AMP + FMN (the hydrolase direction). Both are
distinct, well-supported activities of this bifunctional protein.
FAD synthase activity requires a divalent metal (Mg2+ optimal; Co2+ comparable; lower
with Mn/Ca/Zn). FAD diphosphatase activity instead requires Co2+ (Ni/Mn partial;
Mg/Ca/Cu cannot substitute) and is stimulated by K+.
[file:human/FLAD1/FLAD1-uniprot.txt "The FAD synthase activity requires a divalent metal"]
[file:human/FLAD1/FLAD1-uniprot.txt "Magnesium or cobalt supports the highest FAD synthase activity"]
[file:human/FLAD1/FLAD1-uniprot.txt "The FAD diphosphatase activity requires cobalt"]
KM(FMN) ~0.35-1.5 uM; KM(ATP) ~15 uM; kcat(FAD synthesis) ~0.69 s-1 (E. coli-purified).
kcat is low, consistent with FAD release being rate-limiting and with a "FAD chaperone"
delivery step.
Beyond catalysis, hFADS2 physically interacts with client apo-flavoproteins and
directly transfers newly made FAD to them (a "FAD chaperone" role). Demonstrated for
the nuclear demethylase KDM1A/LSD1 and mitochondrial dimethylglycine dehydrogenase
(DMGDH).
[PMID:25954742 "hFADS is able to operate as a FAD \"chaperone.\""]
PMID:25954742
[file:human/FLAD1/FLAD1-uniprot.txt "Interacts with KDM1A; which promotes"]
[file:human/FLAD1/FLAD1-uniprot.txt "Interacts with DMGDH; which promotes DMGDH holoenzyme formation"]
PMID:20060505
[file:human/FLAD1/FLAD1-uniprot.txt "Isoform 1 and 2 are located in mitochondria and cytosol, respectively"]
[file:human/FLAD1/FLAD1-uniprot.txt "SUBCELLULAR LOCATION: Nucleus"]
[file:human/FLAD1/FLAD1-uniprot.txt "SUBCELLULAR LOCATION: [Isoform 1]: Mitochondrion matrix"]
[file:human/FLAD1/FLAD1-uniprot.txt "SUBCELLULAR LOCATION: [Isoform 2]: Cytoplasm, cytosol"]
Crystal structure (PDB 8ROM/8RON) of full-length hFADS2 and its C-terminal PAPS domain
in complex with FAD. The enzyme forms a stable C2-symmetric dimer; packing of one
protomer's KH domain against the other's N-terminal domain creates the adenosine-specific
hydrolytic active site.
PMID:38688286
[file:human/FLAD1/FLAD1-uniprot.txt "SUBUNIT: Dimer (PubMed:38688286)"]
Biallelic FLAD1 variants cause Lipid Storage Myopathy due to FAD Synthase Deficiency
(LSMFLAD; MIM:255100), presenting as combined respiratory-chain deficiency and multiple
acyl-CoA dehydrogenase deficiency (MADD-like). Some (frameshift, hypomorphic) cases are
riboflavin-responsive. (Primary reference PMID:27259049 not in local cache; disease
statement quoted from UniProt.)
[file:human/FLAD1/FLAD1-uniprot.txt "Lipid storage myopathy due to flavin adenine dinucleotide"]
[file:human/FLAD1/FLAD1-uniprot.txt "Some patients show significant improvement with riboflavin treatment"]
This entry supersedes conflicting interpretations in the preserved earlier journal. It does not rewrite that journal or add publication quotations. The detailed source access, assay boundaries, partner joins, natural-product checks and PAINT evidence are in FLAD1-source-evidence.yaml.
The current official snapshot contains 43 annotation contexts. Source-aware seeding retains the prior 41 records, adds two separately identified partners for PMID:25416956, and adds the current nucleus/cytosol GO_REF:0000120 assertions. The old GO_REF:0000044 nucleus and cytosol assertions are explicitly retired, keeping their original reference identifiers and complete historical reviews. Thus the document has 43 current plus two historical rows, rather than two newly discovered GO functions. Seventeen legacy source lists were populated by the supported seeder. All 38 inherited quotation/reference occurrences are conserved, including duplicated cofactor quotations moved into the synthase core. The title and quotation checks apply to inherited quotations, not newly manufactured evidence.
All 29 pre-existing shared normal source files were hydrated from the pinned main commit before the genuine gene fetch, and remain byte-identical. Five previously qualified absent publications were genuinely fetched: PMID:17049878, PMID:19369195, PMID:23946482, PMID:27259049 and PMID:9110174. Only the last two have full text in their fetched normal cache. The UniProt and GOA files were regenerated through the repository fetcher. No shared source was edited to improve an assay claim. Research used actual primary sources and independent consultations; known campaign provider authentication/quota failures were not retried, and no manual text is presented as provider-generated research.
The two core activities are FMN adenylyltransferase and FAD diphosphatase. ATP and divalent-metal requirements belong within the synthase core rather than being independent third/fourth core functions. The description no longer lists respiratory complex I as an FAD-dependent example. The broad InterPro catalytic-activity assertion is modified to the demonstrated FMN adenylyltransferase activity, not called biologically excessive merely for being general. Cytoplasm is a valid broad location and is refined by MODIFY to the experimentally supported cytosol and mitochondrial-matrix terms, retaining their distinct product scopes. Cytosol and mitochondrial matrix in the synthase description refer to the supported product-specific pools; they do not assign these locations to all five products. The matrix annotations are accepted as a compartment of the core synthase function with the longer-product assay scope. ACCEPT does not require a location to apply to every natural isoform; the broader mitochondrial-proteome annotation remains non-core.
Actual PMID:23443125 Methods/Results assign FAD synthesis to the human 329–587 C-terminal construct. Its reverse reaction uses pyrophosphate; it is not the water-dependent FAD diphosphatase reaction. PMID:26277395 and PMID:31351152 establish recombinant FAD hydrolysis under particular cobalt-dependent conditions. A separate consultation of the original PMID:38688286 author-repository PDF refines the active substructure to MPTb plus KH with an interprotomer site; MPTb alone binds nucleotides but is not sufficient for measured hydrolysis. Its 110–587 crystal construct is twelve residues shorter than natural isoform 2. Physiological hydrolysis remains uncertain, as the authors themselves discuss scarce free cobalt. The separately retrieved primary text is documented distinctly from the abstract-only normal cache.
NADH pyrophosphatase remains a reported non-core substrate activity. The current inspection does not establish a quantitative lower rate or a constitutive role in cellular NADH turnover. NADH inhibition of FAD hydrolysis, NADH binding/docking, and NADH substrate turnover are different claims. The inherited UniProt excerpt on the NADH row describes FAD hydrolysis and cannot by itself support NADH turnover. It is preserved for provenance, with this limitation made explicit.
PMID:25954742 directly demonstrates transfer from recombinant human hFADS2 to recombinant rat apo-DMGDH. Its LSD1 association assays use rat nuclear material, not a purified-human-LSD1 transfer experiment. Human HepG2 staining supports nuclear/cytoplasmic FADS localization without resolving every splice product. Rat donor D4A4P4 has its own nuclear experimental evidence in PMID:23946482 and PMID:25954742. Nuclear FAD-degrading activity in the former study is not automatically assigned to FLAD1. The human isoform-1 import and processing observations in PMID:20060505 are distinguished from its predicted exact transit-peptide cleavage boundary.
The five seeded natural products remain in their original order with all identifiers and names preserved. Their reconstructed sequences match official FASTA records at 587, 490, 446, 294 and 338 residues. Only the two legacy dangling-comma VSP notes were repaired through the repository helper, appending VSP_027952 and VSP_027953. PMID:27259049 paper isoforms 5/6 cannot be equated by number to current UniProt products. Its Met268/Met355 constructs were expressed in control fibroblasts; the Met268 activity observation is not a patient-cell rescue experiment. The endogenous small-band product identity remains partly speculative. Neither the natural products 3–5 nor a newly invented Q8NFF5-6 receives those effects. FAD-dependent stability recovery of a recombinant variant is distinguished from catalytic recovery and variable clinical riboflavin response.
The actual PTHR23293 tree contains 233 nodes. Human Q8NFF5 is leaf PTN002510965 with an exact 587-residue ungapped MSA match. Both incoming IBA terms join positive original IBDs at PTN000591373 on that target path. The current API release and original PAINT source are separately pinned; no whole-release identity is assumed. Human evidence among the descendants is expected and is not circular.
Eight of nine incoming interaction contexts have exact official IntAct pair/publication joins. The remaining PMID:40205054/Q5T6F2 context is retained on independent same-pair PMID:33961781 coimmunoprecipitation evidence, with the missing newer-source assay expressly disclosed. Generic associations do not establish adaptor, kinase-binding or other specific molecular activities. PMID:25416956 is the 2014 interactome study, not the later HuRI resource. Self-association is consistent with structural dimerization but is not a separate catalytic core. The exact PMID:34800366 FLAD1 target table row was not recovered; its inherited title quotation is not treated as a target-specific hit. Independent PMID:20060505 localization supports the non-core mitochondrial context.
The user campaign instructions retain supported generic binding as KEEP_AS_NON_CORE and prohibit new publication quotations. They take precedence over conflicting local skill/CLAUDE defaults; this is an explicit current instruction, not a legacy exemption. No validator behavior was changed. Canonical checks determine the computed status and advisories. No NEW annotation is proposed.
Independent scientific review also narrowed the PMID:25135855 finding to measured biochemical thiol dependence and a proposed physiological redox switch, while preserving its exact inherited quotation. Final ontology-enabled validation passed with only the six disclosed generic-binding policy advisories. The unchanged canonical helper confirms DRAFT with zero status mismatches. History schema validation and private HTML rendering also passed.
The mitochondrial location is now ACCEPT, consistent with the mitochondrial-matrix synthase pool and the repository's root-organelle exception. The precise PMID:34800366 target row remains unverified; independent PMID:20060505 import/localization evidence supports the product-scoped decision. The description now states the symmetric dimer observed for the human 110–587 construct and the demonstrated transfer from recombinant human FADS2 to rat apoDMGDH. It does not extend that transfer assay to human LSD1 or every FAD-dependent enzyme. PMID:17049878 is explicitly cited in the synthase core. The physiological contribution of cobalt-dependent hydrolysis remains an open question.
The two inherited NADH-row snippets have been moved unchanged into findings for their actual sources. The UniProt phrase records a protein/activity name; the PMID:31351152 sentence records NADH inhibition of FAD hydrolysis. Neither now appears as evidence for NADH substrate turnover. The NADH rows retain source-only references and defer to the original EXP curation, with the unrecovered turnover-assay limitation stated. NADH inhibition does not itself disprove turnover, and membership in the ADP-containing dinucleotide class does not identify an assayed substrate. The PMID:34800366 title excerpt is retained as explicitly labeled bibliographic provenance rather than a FLAD1 hit.
For the ClinGen Mendelian campaign, the explicit current user instructions require supported generic binding to remain KEEP_AS_NON_CORE and require preservation of every inherited quotation/reference occurrence without new publication quotations. These instructions outrank conflicting repository/skill defaults; they are not an exemption for untouched legacy rows. They are also recorded in FLAD1-source-evidence.yaml. All 38 occurrences remain, including the duplicate PMID:16643857 synthase excerpt: it records two inherited evidence occurrences, not two independent experiments. No validator was changed, and all six policy advisories remain visible. The prior journal and earlier history record are preserved; this section supersedes their earlier disposition where it differs.