Comprehensive molecular and clinical analysis of human electron transfer flavoprotein variants in multiple Acyl-CoA dehydrogenase deficiency

Abstract Human electron transfer flavoprotein (hETF) is a critical mitochondrial enzyme essential for fatty acid β-oxidation and amino acid catabolism. Genetic variations in hETF subunits cause Multiple Acyl-CoA Dehydrogenase Deficiency (MADD), a severe metabolic disorder. This study integrates computational analyses with clinical variant classification to propose a mechanism of pathogenicity and explore its diagnostic applicability. We investigated three naturally occurring single nucleotide polymorphisms (SNPs) at the critical α/β-subunit interface—hETF-βE165K, hETF-αN259S, and hETF-αN259D—using comprehensive in-silico analyses including homology modeling, interprotein interaction profiling, stability predictions (ΔΔG), and theoretical small-angle X-ray scattering (SAXS). Concurrently, we applied the ACMG/AMP framework to classify three clinical variants: rs1441796075, rs897415970, and rs1371983074. Our findings reveal a mechanistic model: all three interface SNPs disrupt native inter-subunit interactions (hydrogen bonds, salt bridges, and van der Waals contacts), driving the protein toward an "open" conformational state. This structural shift significantly increases solvent accessibility of the FAD cofactor pocket (> 15% ΔSASA), rendering it susceptible to pathogenic oxidation to 8-formyl-FAD (8f.-FAD). Stability predictions using multiple methods (DynaMut, mCSM, DUET) revealed variant-specific profiles: αN259S showed consensus destabilization across 3/3 methods, while βE165K showed consensus destabilization across 2/3 methods, while αN259D predictions were mixed, indicating uncertain energetic consequences. SAXS analysis revealed subtle conformational differences consistent with increased flexibility. Clinical classification identified rs1441796075 as Likely Pathogenic (PM2 + PP3 + PP5), while rs897415970 and rs1371983074 remain Variants of Uncertain Significance (VUS) despite extreme rarity and deleterious predictions, highlighting the critical gap that molecular evidence can fill. This integrated approach provides strong computational evidence for a structure–function-disease model: interface destabilization → open conformation → FAD vulnerability → MADD pathology. Our methodology demonstrates potential as a high-throughput in silico screening approach for prioritizing VUS for further investigation and suggests a testable therapeutic strategy—small-molecule interface stabilizers—that warrants preclinical evaluation for MADD patients carrying interface-destabilizing variants.