The success of cryo-electron microscopy (cryo-EM) is based on many inconspicuous, but groundbreaking research achievements over the last three decades, starting more than 30 years ago with the discovery of a way to freeze a large number of protein particles under close-to-physiological conditions by Jacques Dubochet [2]. The real boom of the method happened just a few years ago with the introduction of direct electron detectors, which allowed fast recording of image sequences with excellent sensitivity, so that the fine structure of the proteins could be captured before being destroyed by the electrons. Since then, the field has advanced rapidly, due to progress in image processing software that is accounting for ever more complex physical phenomena that give rise to recorded images, leading to an increasing number of interpretable details in the 3D reconstructions [3].

Notwithstanding the technical progress, there remain two major bottlenecks that cryo-EM researchers have to fight with: preparation of protein specimens suitable for high-resolution imaging remains challenging, and the processing of the recorded images continues to be a tedious and resource-consuming task. In order to arrive at a 3D protein density with interpretable amino-acid side chains, a thin layer of non-crystalline (vitreous) ice with millions of randomly rotated but identical protein particles has to be prepared and imaged in the electron microscope. To create such a layer, typically 3 µl of a specimen at a concentration of 1 mg/ml is applied to an EM grid, most of it is harshly blotted away with a standard filter paper, and the remaining thin liquid layer is rapidly plunge-frozen to liquid nitrogen temperature. 

Protein samples for cryo-EM have to be highly pure and sufficiently concentrated. Many proteins are still challenging purification targets. Structures of membrane proteins have been traditionally addressed by X-ray crystallography, but production of sufficient quantities was tedious and often took years. Despite novel methods such as high-throughput molecular cloning and robotic tools to assist in protein expression, purification, and crystallization, it remains very challenging to produce well-ordered protein crystals suitable for the high-resolution structure determination by X-ray diffraction.

In contrast, cryo-EM studies require a much smaller amount of protein (10-20 times less). In addition, for membrane proteins, the natural membranes can be replaced with amphiphilic moieties (detergents) in the course of specimen preparation, so those membrane proteins remain correctly folded for imaging in transmission electron microscopes without the need for growing 3D crystals.

The uncontrolled blotting step during specimen freezing takes only less than one second, but this is enough to allow the molecules in the thin water layer to come into contact with the water surface, where proteins often denature at the hydrophobic air/water interface. As a result, molecules can become involved in unfavorable redistributions of particles or orientations. Theoretically, only a few thousand perfect protein particle images are required for a high-resolution reconstruction, but the real-life imperfections of ice, insufficient distribution of particle orientations, and protein flexibility raise the required particle number to hundreds of thousands. A typical cryo-EM imaging session consists of recording 5’000 to 10’000 high-quality movies, which correspond to several terabytes of data that need to be rapidly processed. This imposes large requirements for processing resources and time despite the latest IT technology employed.

These problems call for new solutions. At C-CINA, the team of Thomas Braun has developed a microfluidic device, the Cryowriter, which allows blotting-free preparation of cryo-EM grids via microfluidic technology. This device consumes only nanoliters of specimen volumes, reducing thus the need for purification of large amounts of proteins. Most importantly, the method can be coupled to microfluidic protein purification from less than 1 µl of cell lysate, which opens completely new ways for isolation of sensitive, endogenous proteins and their visualization. Currently, we are looking for ways to apply this method to detergent-solubilized cell lysates and membrane proteins.