Conformational diversity and interaction signatures of NADH across protein families

Nicotinamide adenine dinucleotide (NADH) is a ubiquitous redox cofactor that participates in a wide range of enzymatic and regulatory processes. These include metabolism, signalling, and diseases such as cancer and neurodegeneration. Despite the abundance of NADH-protein complex structures, the general principles governing how proteins shape NADH conformation and interaction modes remain unclear, limiting our ability to rationally interpret cofactor specificity, catalytic efficiency, and off-target effects of inhibitors. Here, we present a comprehensive structural analysis of NADH recognition across protein families using 345 NADH-bound crystal structures from the Protein Data Bank. We adopted a descriptor-driven strategy that quantitatively captures the internal geometry of NADH using angles, dihedrals, and interatomic distances, enabling direct comparison of cofactor shapes independent of protein fold. These studies reveal that 65% of structures preferred conformers with a conserved adenine-nicotinamide separation while allowing limited flexibility in the pyrophosphate. The interaction profiles demonstrate that NADH recognition is dominated by hydrogen bonding and electrostatic interactions involving nearly all heteroatoms, while most carbon positions remain non-interacting. Residue- and moiety-level analyses further show that the nicotinamide region serves as the primary interaction hotspot across enzyme classes, while only a handful of structures exhibit adenine-centric recognition. Together, this study establishes a unified biophysical framework that links NADH shape, interaction signatures, and protein context, providing rational insights for cofactor engineering and the design of NADH-targeted inhibitors.