Using cryo-electron microscopy maps for X-ray structure determination

Abstract

The combination of X-ray crystallography and cryo-electron microscopy (cryo-EM) reflects an important trend in structural biology. It can be achieved either by docking crystal structures into the cryo-EM map to generate pseudo-atomic resolution models or by using cryo-EM maps as search models to solve the phase problem of X-ray crystallography. In this study, I present a hybrid method for X- ray structure determination with the initial phases generated by the corresponding components of cryo-EM maps. The whole strategy involves three parts: (i) cryo-EM map replacement: the envelope provided by the map is used as an input for FSEARCH. The correct rotation and translation solution are found and the initial phases are calculated. (ii) Phase extension: the initial phases are extended to a higher resolution by Phenix.resolve. (iii) Model building: the phase-extended density map is utilized for IPCAS to perform model extension iteration. The general applicability of the hybrid method has been evaluated by ten test cases. For four test cases in Chapter 3 (the lowest cryo-EM map resolution being 6.9 Å), the corresponding model of the cryo-EM map and the target X-ray structure were the same. The hybrid method proved to be successful in dealing with such cases and generated almost complete models with reasonable Rwork/Rfree. However, if the target structure of X-ray crystallography is not identical but homologous to the corresponding molecular model of the cryo-EM map, the decrease in accuracy of starting phases makes the whole process more difficult. The initial phases can be more uncertain and the phase-extended map can be unreliable and uninterpretable. In Chapter 4, six test cases have been studied with the sequence identity/similarity ranging from 34%/52% to 86%/91% between the corresponding molecules in cryo-EM and X-ray structures. For most case studies, a standard procedure as described above was used. However, for case study 2, when the resolution gap between the cryo-EM and crystallographic data was large and the sequence identity was low, an intermediate stage of model building was performed. This hybrid method consistently produced models with reasonable Rwork/Rfree which agree well with original X-ray structures for all test cases, thus indicating the general applicability of the method to use cryo-EM maps for X-ray structure determination of homologues. We prove that the cryo-EM map could provide a general route for the structure determination of X-ray crystallography whether the cryo-map corresponds to exactly the target protein or a homologue. The hybrid method should streamline the use of a cryo-EM map as the starting point for high-resolution X-ray structure determination.