Cryogenic transmission electron microscopy (cryo-EM) is a technique for high-resolution imaging of biological macromolecules under near-native conditions. In recent years this method has seen dramatic development and has now become the method of choice for structural biology of large macromolecular complexes used by hundreds of academic and industrial research laboratories worldwide. To image and obtain a three-dimensional structure, an aqueous solution containing the protein of interest must be rapidly cooled to form a uniform layer of amorphous ice, ideally not much thicker than the particle itself (~20-100 nm), in which the particles are partitioned in random orientations. The prevailing method for making thin specimens suitable in cryo-EM has not changed substantially since its introduction several decades ago: a filter paper is pressed against one or both sides of a thin, fenestrated support film spanning a metallic mesh grid across which a droplet of the biological sample is suspended, and then rapidly plunged into a liquid cryogen. A recognised shortcoming of this method is that a reproducibly uniform ice layer can often not be achieved. As a result, it may require weeks of optimisation to find conditions suitable for high-resolution imaging, and these conditions are typically different for each sample. Even if such conditions are found, additional problems encountered in practice include preferential orientation of particles, preferential interaction of particles with the support film and denaturation of particles due to exposure to the large air-water interface. This severely limits throughput and impedes any attempts at automation of sample preparation. It is therefore widely appreciated that sample preparation is the major remaining bottleneck precluding cryo-EM from realizing its full potential.
We have developed a novel approach for cryo-EM sample prepraration using a Micro-Electromechanical-Systems (MEMS)-based nanofluidic chip with a free-standing electron transparent observation window that can be filled with protein solution. In our design, specimen thickness reproducibly constrained by the dimensions of the nanofluidic channel. In addition, the destructive effect of the air-water interface on macromolecules is entirely avoided. We expect this approach to be capable of mitigating many of the challenges that are limiting present cryo-EM sample preparation methods: It requires only picoliter sample volumes, it can robustly provide uniform thickness gradients across the sampling area without the need for any optimisation or expensive instrumentation, which together will allow complete automation of the sample preparation process.
