Sequence constraints predispose Class D GPCRs to follow an atypical activation mechanism.

The biophysical principles underlying distinct conformational changes in proteins with similar topologies remain poorly understood. Class D G Protein-Coupled Receptors (GPCRs), fungal pheromone-sensing receptors essential for mating and survival, exhibit an atypical activation mechanism compared to other GPCR classes. Unlike Class A GPCRs, which activate through outward movement of TM6, Class D receptors undergo activation via outward displacement of TM7 coupled with inward movement of TM6. To investigate the origin of this atypical process, we employed state-specific generative AI sequence models to design protein sequences corresponding to unique active states, revealing that sequence constraints predispose Class D GPCRs toward this mechanism. We further explored the dynamic basis of activation through millisecond-scale atomistic simulations of STE2, a representative Class D GPCR and therapeutic target for fungal diseases. Using Maximum Entropy VAMPNets, an active learning based adaptive sampling strategy, we efficiently mapped the conformational free energy landscape of STE2. These simulations uncovered multiple intermediate states that have not yet been resolved experimentally and demonstrated that activation in these dimeric proteins involves fully decoupled monomers. Comparative simulations across Classes A, B, D, and F, totaling 4 milliseconds of all-atom molecular dynamics, revealed distinct patterns of electrostatic interactions. In Class D GPCRs, activation disrupts a dense TM7 interaction network while inward TM6 movement enhances electrostatic contacts. In contrast, Class A GPCRs such as CB1R display the opposite trend. Together, these findings show that sequence differences in TM6 and TM7 underlie the unique activation mechanism of STE2, offering new insights into GPCR conformational diversity.