Microfluidic protein isolation and sample preparation for transmission electron microscopy

The knowledge of atomic structures is essential to understand the mechanics and chemistry of proteins in fundamental research and is often the base for drug development. During the last decades, X-ray crystallography has been the primary method for determining atomic models providing an impressive number of molecular structures. Nevertheless, the technique is limited by the fact that the complexes of interest have to be crystallized. Nuclear magnetic resonance (NMR), which is used as an alternative to solve biomolecules in solution, has the drawback of consuming large amounts of protein, being labour intensive and challenging for large molecules.
In recent years, cryogenic electron microscopy (cryo-EM) has evolved as an important tool for protein structure determination. Technical advances in the instrumentation and increased computational power combined with better processing algorithms caused a massive improvement in the resolution of obtained structures. For these achievements Jacques Dubochet, Joachim Frank and Richard Henderson were awarded with a Nobel Prize in 2017. However, sample preparation methods lack behind and did not change a lot. A significant complication is the production of target proteins in sufficient amounts and quality. Although only some thousands to a few million protein particles must be imaged to solve a protein structure, much larger quantities are required to prepare specimens for cryo-EM. Conventional sample preparation methods are very wasteful with proteins and more than 99% of protein is lost during a paper blotting step. Thus, considerable amounts of purified proteins have to be produced using complex and costly procedures usually including several chromatography steps.
In this thesis, a novel sample preparation and purification system consuming only minute amounts of biological material is presented. The system allows the purification of proteins and the subsequent preparation of isolated targets for negative stain and cryo-EM. We constructed corresponding hardware and software described in Chapters 1 & 2. The application of the system on biological samples is demonstrated in Chapters 3 & 4. As an example, we purified endogenous human 20S proteasome starting with <1 μL HeLa cytosol and determined it’s 3D structure at a resolution of 3.5Å. In Chapter 5, we show the purification of recombinantly expressed proteins by the use of a novel crosslinker that was developed during the course of this thesis.