Non-uniform refinement: Adaptive regularization improves single particle cryo-EM reconstruction

Ali Punjani; Haowei Zhang; David J. Fleet

The Na v 1.7 channel [33] is a voltage-gated sodium channel found in the human nervous system. Na v channels are fundamental to the generation and conduction of action potentials in neurons, are mutated in various diseases, and are targeted by toxins and therapeutic drugs (e.g., for pain relief). We present results of uniform and non-uniform refinement on Na v 1.7 data originally collected and processed by the authors of [33] , available publicly as raw microscope movies (EMPIAR-10261). In this dataset, the Na v 1.7 channel is bound to two Fabs, resulting in a 245kDa C2 symmetric complex solublized in detergent. The data contains 25,084 movies from a Gatan K2 Summit direct electron detector camera in counting mode, with a pixel size of 0.849Ã…. We processed the entire dataset through motion correction, CTF estimation, particle picking, 2D classification, ab-initio reconstruction and heterogeneous refinement in cryoSPARC v2.11 [22] . A total of 738,436 particle images were extracted from the movies. These were curated using 2D classification, yielding 431,741 particle images. As in [33] , we detected two discrete conformational states corresponding to the active and inactive states of the channel. 130,982 particles were found in the inactive state, and the remaining 300,759 were found to be in the active state. We obtained state-of-the-art resolution reconstructions for both the active and inactive states, but here we focus solely on the active state.