Periodic Reporting for period 3 - CellStructure (Structural cell biology in situ using superresolution microscopy)
Okres sprawozdawczy: 2020-06-01 do 2021-11-30
Supra-molecular protein machineries control diverse cellular processes. Knowing their structural organization is crucial for understanding their function. As classical structural biology techniques are limited in studying such assemblies in their natural cellular environment, there is a critical methodological gap inhibiting a direct link between structure and function. Consequently, it is highly challenging to visualize how large multi-protein complexes are structurally organized and change this organization during their activity. Recent advances in fluorescence microscopy, in particular the development of groundbreaking superresolution microscopy (SRM) methods, can now help bridge this gap. In the context of this interdisciplinary ERC grant, my group is developing unique and innovative optical, biological and computational imaging technologies to determine the structural organization of multi-protein assemblies in their functional cellular context.
To reach this vision to address questions in structural biology with optical microscopy we are developing new technologies to overcome several technical limitations. Specifically, we need to reach highest spatial resolution in all three dimensions, we need to be able to accurately measure distances and count protein copy number in a quantitative way, we need to be able to relate our measurements to the functional state of the protein machine by integrating information on the cellular context and we need to develop new data analysis approaches to extract the maximum information on the biology from the data.
A stereotypic example for a complex and dynamic protein machine is the machinery that drives endocytosis. Endocytosis is the process by which cells take up molecules from the environment, an essential process for any eukaryotic cell. To understand how thousands of proteins self-organize in such a complex machinery that carries out a complex function, we set out to use our new superresolution technologies to determine the time-resolved 3D structural organization of the entire endocytic proteome in S. cerevisiae as a basis for a physical model of endocytosis.
To reach this vision to address questions in structural biology with optical microscopy we are developing new technologies to overcome several technical limitations. Specifically, we need to reach highest spatial resolution in all three dimensions, we need to be able to accurately measure distances and count protein copy number in a quantitative way, we need to be able to relate our measurements to the functional state of the protein machine by integrating information on the cellular context and we need to develop new data analysis approaches to extract the maximum information on the biology from the data.
A stereotypic example for a complex and dynamic protein machine is the machinery that drives endocytosis. Endocytosis is the process by which cells take up molecules from the environment, an essential process for any eukaryotic cell. To understand how thousands of proteins self-organize in such a complex machinery that carries out a complex function, we set out to use our new superresolution technologies to determine the time-resolved 3D structural organization of the entire endocytic proteome in S. cerevisiae as a basis for a physical model of endocytosis.
Since the start of the ERC grant we made excellent progress towards all objectives. We realized that to develop technologies that go beyond the state of art, we need new sensitive assays to quantitatively assess quality in superresolution microscopy, something critically missing in the field. Thus, we introduced new reference standards for superresolution microscopy based on nuclear pore complexes and developed assays to use them to measure resolution and performance of microscopes and to optimize all aspects of a superresolution workflow including imaging conditions and labeling efficiencies of the dyes (Thevathasan, Kahnwald et al, Nature Methods 2019). These new cell lines are also well-suited to count the number of proteins in a complex, because their precise stoichiometry and characteristic shape and size makes them excellent counting reference standards. We demonstrated an unprecedented robustness in accuracy and precision for intracellular counting applications.
To push the 3D resolution of superresolution microscopy to scales relevant for structural biology we pursued two approaches. On one hand we could improve the precision of standard 3D superresolution microscopy by developing a new data analysis framework that uses experimentally derived calibrations to extract precise coordinates for single molecules (Li et al, Nature Methods 2018). To overcome systematic errors introduced by the sample when imaging deeper inside cells, we developed a postprocessing workflow to correct for these artifacts to allow for accurate and absolute spatial measurements (Li et al, Biomed Opt Express 2019). On the other hand, we made big steps towards reaching the full potential of Supercritical Angle Localization Microscopy, a conceptionally new 3D superresolution technology that we developed. By combining a re-designed and largely improved microscope with new data analysis algorithms we could achieve an unprecedented optical 3D resolution with single nanometer localization precision in all dimensions (manuscript in preparation). As this technology relies on freely rotating fluorophores, we assembled a microscope that can measure the precise orientation of single emitters.
To integrate information about the cellular context and functional state of a protein machine, we developed a new approach for multi-color superresolution imaging. We could recently push the number of colors that can be measured simultaneously to four by using a new detection scheme and by developing new analysis software (Zhang et al, Nature Methods, in press and unpublished results). On the other hand, we made progress towards correlative superresolution microscopy and electron microscopy by developing a workflow to combine dynamic superresolution imaging of in vitro structures with electron tomography on the same samples (manuscript in preparation).
Analysis of superresolution microscopy data relies heavily on computational analysis. Over the last years we spent a lot of effort developing a modular open-source software platform that spans all aspects of data analysis from fitting of single molecules to rendering of superresolution images to extracting quantitative biological information from such data (github.com/jries/SMAP manuscript in preparation). To extract the maximum information from superresolution data we are developing a new approach to fit a parametrized model to such data using maximum likelihood estimation (manuscript in preparation).
We applied these methods to get some comprehensive understanding of the structure and function of the yeast endocytic machinery (Mund et al, Cell 2019). To obtain statistically relevant information, we automated our entire data acquisition and analysis pipeline to run our microscopes in an unsupervised fashion around the clock. This allowed to acquire >100’000 superresolution images 23 different proteins in yeast endocytosis to reconstruct their structural organization. We developed timing markers to obtain dynamic information from fixed structures. Finally, we used these data as the basis for a computational model of yeast endocytosis. We discovered that actin nucleation is pre-patterned by a nano-template of WASP, whose assembly in a ring allows actin to pull in a membrane vesicle even against the high turgor pressure present in yeast. This addressed the long-standing question of how the force is generated and transmitted to the membrane to form a vesicle in yeast endocytosis. This work was the first application of high-throughput superresolution microscopy to biology, and it provided many fundamental insights into assembly and function of the yeast endocytic machinery.
To push the 3D resolution of superresolution microscopy to scales relevant for structural biology we pursued two approaches. On one hand we could improve the precision of standard 3D superresolution microscopy by developing a new data analysis framework that uses experimentally derived calibrations to extract precise coordinates for single molecules (Li et al, Nature Methods 2018). To overcome systematic errors introduced by the sample when imaging deeper inside cells, we developed a postprocessing workflow to correct for these artifacts to allow for accurate and absolute spatial measurements (Li et al, Biomed Opt Express 2019). On the other hand, we made big steps towards reaching the full potential of Supercritical Angle Localization Microscopy, a conceptionally new 3D superresolution technology that we developed. By combining a re-designed and largely improved microscope with new data analysis algorithms we could achieve an unprecedented optical 3D resolution with single nanometer localization precision in all dimensions (manuscript in preparation). As this technology relies on freely rotating fluorophores, we assembled a microscope that can measure the precise orientation of single emitters.
To integrate information about the cellular context and functional state of a protein machine, we developed a new approach for multi-color superresolution imaging. We could recently push the number of colors that can be measured simultaneously to four by using a new detection scheme and by developing new analysis software (Zhang et al, Nature Methods, in press and unpublished results). On the other hand, we made progress towards correlative superresolution microscopy and electron microscopy by developing a workflow to combine dynamic superresolution imaging of in vitro structures with electron tomography on the same samples (manuscript in preparation).
Analysis of superresolution microscopy data relies heavily on computational analysis. Over the last years we spent a lot of effort developing a modular open-source software platform that spans all aspects of data analysis from fitting of single molecules to rendering of superresolution images to extracting quantitative biological information from such data (github.com/jries/SMAP manuscript in preparation). To extract the maximum information from superresolution data we are developing a new approach to fit a parametrized model to such data using maximum likelihood estimation (manuscript in preparation).
We applied these methods to get some comprehensive understanding of the structure and function of the yeast endocytic machinery (Mund et al, Cell 2019). To obtain statistically relevant information, we automated our entire data acquisition and analysis pipeline to run our microscopes in an unsupervised fashion around the clock. This allowed to acquire >100’000 superresolution images 23 different proteins in yeast endocytosis to reconstruct their structural organization. We developed timing markers to obtain dynamic information from fixed structures. Finally, we used these data as the basis for a computational model of yeast endocytosis. We discovered that actin nucleation is pre-patterned by a nano-template of WASP, whose assembly in a ring allows actin to pull in a membrane vesicle even against the high turgor pressure present in yeast. This addressed the long-standing question of how the force is generated and transmitted to the membrane to form a vesicle in yeast endocytosis. This work was the first application of high-throughput superresolution microscopy to biology, and it provided many fundamental insights into assembly and function of the yeast endocytic machinery.
All technologies that we developed are beyond the state of the art and outperform other published approaches. Until the end of the project we expect to further push the 3D resolution and multi-color abilities of our microscopes and software to obtain ever more detailed information on the yeast endocytic machinery. Specifically, we are working towards a complete dynamic reconstruction of this important protein machine and to investigate specific aspects of its function.