The Ewald’s Sphere Correction And Low-voltage Electron Counting Of Electron Cryomicroscopy

Cryo-EM single particle analysis is one of the mainstream structural biology analysis methods.Single-particle analysis is mainly suitable for protein samples with homogeneity and good dispersion,but not for non-homogeneous samples and thick sections.The conventional single particle analysis cannot achieve the desired level of resolution for large proteins and small proteins.Their limiting factors are different.For large proteins,it is generally believed that the limiting factor is the flexibility of proteins and the Ewald sphere effect.Flexibility disrupts the requirement of homogeneity for single particle analysis,and the Ewald sphere effect disrupts the requirement of reconstructions.Traditionally,the Ewald sphere effect correction performed in Fourier space is limited by protein flexibility,thus being ineffective on real data.We give the resolution limit of the Ewald sphere effect on cryo-EM based on the real-space equivalent of the Ewald sphere effect-the depth-of-field effect,using a resolution criterion consistent with cryo-EM by simulations.Based on correction of depth-of-field effect,we developed the Block-based Reconstruction algorithm.When applying the Block-based Reconstruction algorithm on real datasets,we not only break the resolution limit caused by the Ewald sphere effect on single particle analysis for the first time,but also can effectively compensate the resolution loss caused by the flexibility of proteins on every dataset.As a result,we basically solve the limitation problems of cryo-EM for large proteins.For small proteins,the main limiting factor is that their mass is so small that the ability to scatter electrons is significantly lower,resulting in relatively low signal-tonoise ratio(SNR)images.Software of single particle analysis has difficulty finding the correct orientation of particles on low SNR images,which results in the inability to restore the correct signal no matter how much data is superimposed,thus limiting resolving small proteins.One possible way to enhance the SNR of small proteins is to reduce the acceleration voltage of cryo-EM.The lower the voltage,the greater proteins to scatter electrons,which can lead to more signal.However,lowering the voltage also introduces stronger radiation damage,which reduces the total number of available electrons.It was believed that for a certain thickness of the sample,after considering two factors of the elastic scattering and radiation damage,there exists a voltage that makes the general SNR maximum,which is called the optimal voltage.Current study showed that low voltage cryo-EM can greatly enhance the SNR of small proteins.However,the study was based on 2D-crystal diffraction,which is not suitable for the vitrified samples used for cryo-EM.We measure the radiation damage rate from vitrified samples at various voltages,correct the optimal voltage and SNR enhancement,and confirm that lowering the voltage of the cryo-EM enhance SNR,but to a more limited extent.Reducing the voltage of cryo-EM also reduces the detection efficiency of electrons of the direct detect device(DDD).Based on the fact that low-energy electrons produce more backscattering electrons,we infer that there DDD contains more low-SNR information at low voltage.We develop a cluster classification weighting filter,which can improve the SNR by about 80%near the Nyquist frequency comparing to the conventional electron counting algorithm at 120 kV.When applied to real data with Hybrid method of our filter,the resolution is improved effectively for several datasets at 120 kV or 200 kV.In the end,we show that without the envelop from chromatic aberration,the lowvoltage cryo-EM is able to achieve the same or better results than conventional 300 kV cryo-EM.Our study points the way to the development of low-voltage cryo-EM.