Confrontation of AlphaFold models with experimental structures enlightens conformational dynamics supporting CYP102A1 functions.

Conformational dynamics plays a critical role for the function of multidomain electron transfer complexes. While crystallographic or NMR approaches allow detailed insight into structures, lower resolution methods like cryo-electron microscopy can provide more information on dynamics. In silico structure modelling using AlphaFold was recently successfully extended to the prediction of protein complexes but its capability to address large conformational changes involved in catalysis remained obscure. We used bacterial CYP102A1 monooxygenase homodimer as a test case to design a competitive modelling approach (CMA) for assessing alternate conformations of multi-domain complexes. Predictions were confronted with published crystallographic and cryo-EM data, evidencing consistencies but also permitting some reinterpretation of experimental data. Structural determinants stabilising the new type of domain connectivity evidenced in this bacterial self-sufficient monooxygenase were analysed by CMA and used for in silico retro-engineering applied to its eukaryotic bi-component counterparts.