Abstract
High resolution information is important for accurate structure modelling. However, this level of detail is typically difficult to attain in macromolecular crystallography because the diffracted intensities rapidly fade with increasing resolution. The problem cannot be circumvented by increasing the fluence as this leads to detrimental radiation damage. Previously, we demonstrated that high quality MicroED data can be obtained at low flux conditions using electron counting with direct electron detectors. The improved sensitivity and accuracy of these detectors essentially eliminate the read-out noise, such that the measurement of faint high-resolution reflections is limited by other sources of noise. Inelastic scattering is a major contributor of such noise, increasing background counts and broadening diffraction spots. Here, we demonstrate that a substantial improvement in signal-to-noise ratio can be achieved using an energy filter to largely remove the inelastically scattered electrons. This strategy resulted in sub-atomic resolution MicroED data from proteinase K crystals, enabling accurate structure modelling and the visualization of detailed features. Interestingly, filtering out the noise revealed diffuse scattering phenomena that can hold additional structural information. Our findings suggest that combining energy filtering and electron counting can provide more accurate measurements at higher resolution, providing better insights into protein function and facilitating more precise model refinement.