Enabling direct nanoscale observations of biological reactions with dynamic TEM

Abstract

Biological processes occur on a wide range of spatial and temporal scales: from femtoseconds to hours and from angstroms to meters. Many new biological insights can be expected from a better understanding of the processes that occur on these very fast and very small scales. In this regard, new instruments that use fast X-ray or electron pulses are expected to reveal novel mechanistic details for macromolecular protein dynamics. To ensure that any observed conformational change is physiologically relevant and not constrained by 3D crystal packing, it would be preferable for experiments to utilize small protein samples such as single particles or 2D crystals that mimic the target protein's native environment. These samples are not typically amenable to X-ray analysis, but transmission electron microscopy has imaged such sample geometries for over 40 years using both direct imaging and diffraction modes. While conventional transmission electron microscopes (TEM) have visualized biological samples with atomic resolution in an arrested or frozen state, the recent development of the dynamic TEM (DTEM) extends electron microscopy into a dynamic regime using pump-probe imaging. A new second-generation DTEM, which is currently being constructed, has the potential to observe live biological processes with unprecedented spatiotemporal resolution by using pulsed electron packets to probe the sample on micro- and nanosecond timescales. This article reviews the experimental parameters necessary for coupling DTEM with in situ liquid microscopy to enable direct imaging of protein conformational dynamics in a fully hydrated environment and visualize reactions propagating in real time.

Fig. 1.

Sample geometry similarities. (a) Schematic of in situ liquid S(TEM) chamber using two thin membranes to enclose the sample. (b) Cryo-EM layout showing the vitrified ice layer with trapped sample. (c) Simplified comparison of edge contrast from in situ and cryo-EM experiments from a generic 10 nm diameter protein.

Fig. 2.

Simulations of single particles. (a and b) Individual and averaged (n = 100) bright field cryo-EM simulations of apo-ferritin within an ice layer of a 100 nm thickness and using a defocus of −500 nm with C_s = 2.0 mm. (c and d) Individual and averaged (n = 100) bright field in situ liquid DTEM simulation of apo-ferritin between two 5 nm thick silicon nitride membranes and within a 50 nm fluid path length and with C_s = 0.005 mm. Both images assumed a total dose of 10 electrons per Ų and a signal-to-noise ratio of 4%. Note the lack of the strong black halo artifact in c and d due to the closer to focus imaging conditions made possible with aberration correction. Scale bar is equivalent for all panels and represents 10 nm.

Fig. 3.

In situ liquid STEM of soluble proteins. (a and b) Bright field and corresponding dark field STEM images of purified ferritin complexes with iron oxide core. Note the two particles at the top of each image that suffered from Brownian motion or beam-induced charging during image rastering (white arrows). Particles in the lower half of the image appear stable but may be tumbling end-over-end in solution. Scale bars represent 25 nm.

Fig. 4.

Illustration of the second-generation Dynamic TEM being installed at PNNL that is optimized for biological imaging.

Fig. 5.

DTEM pump-probe simulation of protein dynamics. Simulated electron diffraction patterns from 2D crystals of bacteriorhodopsin in the ground (a) and cytoplasmically open (b) M-intermediate conformational states. (c) Difference pattern between A and B with overlaid rings indicating 30 (inner) and 10 Å (outer) resolution.