At the beginning of this thesis there was no high-resolution single-particle cryo-EM structure of a small membrane protein. In those days, the structure of the TRPC3 channel was solved to a resolution of only 15 Å with more than 150.000 particles (Mio et al, 2007). In the time period of this work the resolution limit of cryo-EM changed dramatically due to the development of direct electron detectors and better software for particle alignment (Kühlbrandt, 2014). Using a high-end electron microscope equipped with a direct electron detector, only the particle number and the intrinsic flexibility of a protein limit the modern single-particle cryo-EM structure determination.
In this work single-particle cryo-EM was used to determine the structure of three different types of membrane proteins. The first project dealt with the human TRP channel polycystin-2 (PC-2), which was expressed in GnTI- cells. The expression of PC-2 resulted in crystalloid ER, a formation that was previously only obtained in vitro by the overexpression of HMG-CoA reductase in UT-1 cells. Although it seems that proteins involved in lipid biogenesis as well as proteins that form intracellular interactions seem to support crystalloid formation we are still far away from understanding what triggers the formation of crystalloid versus karmellea and whorl ER.
PC-2 was produced in GnTI- cells and different strategies for protein purification were tested.
The use of a StrepII-tag and LMNG/CHS as detergent mixture resulted in a pure protein sample suitable for cryo-EM data acquisition. A dataset of PC-2 in amphipol A8-35 comprising ~860 micrographs and ~120.000 particles was taken and yielded a 4.6 Å structure. The structure of PC-2 displays the open conformation and gives first insights into its extracellular domain.
To gain higher-resolution the dataset of PC-2 has to be increased. With a resolution below 4 Å an atomic model of PC-2 could be build, which would help to understand how specific mutations could cause ADPKD. Further, structure determination of the closed state of PC-2 in the absence of Ca2+ would allow understanding how PC-2 changes its conformation. On the basis of the opened and closed state and the knowledge of the residues involved in causing ADPKD, a drug-based treatment can be developed. This would help thousands of people suffering from ADPKD especially because currently kidney transplantation is their only hope of healing.
The structural investigation of the secondary-active transporter BetP by single-particle cryo-EM was the second project of this work. BetP was incorporated into amphipol A8-35 before a cryo-EM dataset was acquired which yielded in a 6.8 Å structure. The cryo-EM structure of BetP showed a new state in which the C-terminal domains are located differently compared to all known X-ray structures in which one of the three C-terminal domains always forms an essential crystal contact. Since no potassium and glycine betaine were present, the cryo-EM map most likely shows an inactive state of BetP. As in the case of the human γ-secretase, the collection of more data would most likely result in a high-resolution structure of BetP (Bai et al, 2015b; Lu et al, 2014). However, the extensive knowledge of BetP together with the cryo-EM map already allowed the generation of a very detailed model of the inactive state.
Single-particle cryo-EM of BetP in the presence of K+ is a structural approach to investigate the regulation and activation cycle of the protein. This would reveal the activated structure of BetP in solution and the movement of the C-terminal domain upon activation. It has to be considered that in the presence of K+ the binding sites of these ions are important. The determination of these sites requires a resolution of about 3.5 Å and thus a very large dataset, comprising approximately 1.500.000 particles.
In the third project of this work, the structure of the N-type ATPase rotor ring was determined.
The N-type ATPase c-ring was a suitable membrane protein to investigate the impact of amphiphatic molecules on single-particle cryo-EM structure determination. The choice of the amphiphatic molecule was decisive for structure determination of this small membrane protein complex. The particles were only accurately aligned against projections of a reference if the rotor ring was embedded in a detergent of a low density and small micelle size. By using LDAO as detergent, a 6.1 Å structure of the N-type ATPase c17 ring was solved. Based on the c-ring stoichiometry, a possible function of this enzyme was discussed. Although it is not known whether the B. pseudomallei N-type ATPase is an ATP synthase or a proton pump, it can be concluded that this enzyme can work efficiently in ATP synthesis in the presence of a low electrochemical gradient.
A single-particle cryo-EM analysis on the entire N-type ATPase would answer many more questions about this type of ATPase such as the number of outer stalks and how ions are transported through the Fo domain.
The limit of single-particle cryo-EM was estimated by Richard Henderson on the basis of
determine an atomic structure of a 100 kDa protein from approximately 10.000 reprojections acquired under ideal conditions. Considering the impact of the detergent it would be interesting to see how many particles would be necessary to obtain the structure of a 100 kDa membrane protein. It has to be mentioned that the technical development of direct electron detectors as well as the software development for single-particle cryo-EM is still ongoing. I assume that the next generation of detectors, which have more pixels and a better quantum efficiency, will enable high-resolution structure determination of proteins with a molecular weight even below 100 kDa by single-particle cryo-EM.
There are already centres for cryo-EM sample freezing and data-collection allowing more scientists access to this method. Combining the new developments in cryo-EM with NMR spectroscopy, the structure of every protein could, theoretically, be investigated in solution without the need of protein crystals. Additionally, the sample amount necessary for cryo-EM is much lower compared to X-ray crystallography and NMR. However, to take advantage of this method the protein has to be expressed and purified in high quality, which is still a challenge for most membrane proteins.