Cryogenic electron microscopy (cryo-EM) has become a powerful technique to understand the structure and function of proteins. Thanks to substantial technical advancements in the 2010s, the frequent determination of atomic structures is now possible. This thesis encompasses chapters that contribute to both the remaining technical challenges of cryo-EM, and to the advancement of biological insight through the application of cryo-EM to a motility machinery called a gas vesicle. As an overarching theme, the thesis is titled ’Imaging Life at the Molecular Scale using Electrons’. Chapter 1 provides an introduction into methods to study structure and function in biology at the molecular scale, and into the development of cryo-EM since its inception. A comprehensive review of previous structural work on gas vesicles follows.
Technological improvements in electron detectors have significantly enhanced the quality of protein images obtainable. However, predicting the success of cryo-EM structure determination still poses a challenge, as it is heavily dependent on the unique properties of the sample and its behavior during the sample preparation process. Proteins are typically forced into an thin liquid layer, which is frozen by quick immersion into a cryogen. The large air-water interface can cause proteins to aggregate, denature or adopt preferred orientation, hindering structure determination. The thickness of the layer is also hard to control - too thick layers lead to poor image contrast, too thin layers damage the protein.....
@en

Gas vesicles, micrometer-scale protein structures that function as bacterial buoyancy providers, encapsulate gas in a highly optimized manner. While their atomic structure has been elucidated through single-particle analysis of cryo-EM images, certain structural and functional details remain uncertain. Its biogenesis - the formation and growth mechanisms - consequently remains elusive. Here we apply automated segmentation methods originating from cell imaging to cryo-EM images of gas vesicles to analyze the positions of gas vesicle features in a context-preserving matter. Subsequent whole gas vesicle processing is able to transform accurate gas vesicle segmentations into high-confidence structural feature location picks and statistics. This enables the formation of a sizeable data set containing 86k whole gas vesicles, improved resolution 2D class averages, and the potential for improved structural modeling. Combining high sample number contextual information enables inference on the dynamical properties of gas vesicle growth. Our findings validate recent atomic structure propositions and lend support to a stochastic monomer insertion growth model.

Cyo-EM is a powerful technique in biophysics to model. By recording tens of thousands of these images and using computational methods to form a 3D density map. The interpretability of these cryo-EM maps decreases in high-resolution by using map sharpening to increase these areas of low contrast locally. It is possible to locally sharpen the cryo-EM map. While global sharpening methods would result in oversharpening of some areas should be and undersharpening in others, using a local sharpening method can adjust for a variation in contrast within a map. This resulted in the development of LocScale. Which is a local sharpening method using a (pseudo) atomic model. The main drawback of using a local sharpening method which would require an atomic model is the fact that it is impractical in cases where there is no model available as well as requiring atomic model refinement for sharpening, which takes a lot of resources. While other machine learning methods are able to be used to sharpen EM-maps. When decomposing the sharpened maps in Fourier space, it often becomes clear that not only the amplitudes but also the phases, instead of just the amplitudes of the EM-map have been changed. Resulting in hallucinations appearing, and ML sharpening methods are often implemented in such a way that the entire method is housed inside a black box, making it challenging to understand why certain parts of the macromolecule are sharpened. Using a machine learning-based approach to predicting local radial profiles it is possible to limit the use of machine learning to only one step, the step for determining the radial profiles, compared to the model-based method, LocScale. And use all other steps which would normally be associated with LocScale, vastly reducing the use of ML. Since the radial profiles describe the amplitudes of the EM-map it is possible to change these while leaving the phases unchanged. Reducing the chance of hallucinations appearing to zero.In this thesis, it is shown that it is possible to produce a machine learning method (named iLocScale) to effectively sharpen EM maps by predicting radial profiles. This resulted in a method which is almost as sharp as their model-based (target) counterpart (LocScale) in the modelled region and vastly outperforms LocScale in the unmodelled regions for sharpness. While also offering a large improvement in correlation coefficient for local resolution and B-factor between the iLocScale sharpened maps (0.3416) and unsharpened maps (0.0465). Moreover, iLocScale vastly outperforms LocScale in unmodeled regions. It also takes approximately 75% less time.

Vaccination is the most effective strategy in humanity’s fight against viruses. The concept of vaccination was first proposed by dr. Edward Jenner in the 18th Century, and its efficacy has been proven over time, providing unparalleled protection against viral infections. Large-scale vaccination campaigns have been successful in eradicating diseases such as smallpox in 1980. In recent times, the coronavirus pandemic has brought vaccines to the forefront of public, academic, and industry interest. Apart from conventional live vaccines, which were originally designed and employed by Jenner, there has been a significant expansion in vaccine types, including mRNA, viral vector, attenuated, and inactivated vaccines, each with their own advantages and limitations. Furthermore, virus-like particles (VLPs), a novel class of vaccines, have emerged as a promising alternative to traditional vaccine design strategies, potentially offering improved efficacy and safety. VLPs are multimeric nanoparticles derived from one or more viral structures. They are typically empty or devoid of genetic material, rendering them non-replicative and non-infectious. Despite the proposed absence of genetic material, VLPs possess immune-inducing surface patterns resembling those of the native virus, making them recognizable to the immune system. This unique feature can be exploited for vaccine purposes. VLP-based vaccines offer a safer alternative to traditional vaccines and present an opportunity to create vaccines against viruses that recognize specific viral surface structures instead of a single protein. Large-scale production of VLPs can potentially mitigate many of the drawback of current vaccines, such as extreme storage conditions (mRNA), vector immunity (vector), reversion (attenuated), and high production costs (inactivated). Therefore, VLP-based vaccines hold great promise in the fight against viral infection, by providing a safer and superior product. The safety and efficacy of VLP-based vaccines have been established through extensive research. At present, the market has numerous VLP vaccines available, with HPV VLP vaccines being the most prominent. Its success has set the standard and paved the way for the development of other VLP-based vaccines. In this work, we demonstrated the feasibility of producing and purifying virus-like particles that closely resemble enterovirus A71 (EV71) and coxsackievirus A6 (CVA6). Both viruses have a positive sense RNA genome of ~7.4 kilobase pairs (kbp), which encodes a 260 kDa polyprotein that is stepwise cleaved into eleven viral proteins. Enteroviruses, and most prominently EV71 and CVA6, are the main causative agents of hand, foot, and mouth disease (HFMD). HFMD is named after the characteristic lesions that develop on the hands, feet, mouth, and buttocks of infected individuals. In severe cases, especially among children, the disease can spread to the central nervous system (CNS), resulting in complications such as aseptic meningitis and encephalitis. By employing VLPs, a multivalent vaccine can be developed to target multiple viral strains simultaneously, providing an opportunity for the prevention and control of HFMD. We utilized the baculovirus expression vector system (BEVS) to produce the enterovirus-like particles. For the insect cell lines employed in the BEVS, cell counting is crucial for the maintenance and manipulation of cell cultures. It is a vital aspect of assessing cell viability and determining proliferation rates, which are critical to maintaining the health and functionality of the culture. In Chapter 2, we introduce a machine learning (ML) model based on YOLOv4, capable of performing cell counts with high accuracy (>95%) for Trypan blue-stained insect cells. The model was trained, validated, and tested using images of two distinctly different insect cell lines, Trichoplusia ni (High FiveTM; Hi5 cells) and Spodoptera frugiperda (Sf9). The model achieved F1 scores of 0.97 and 0.96 for alive and dead cells respectively, demonstrating substantially improved performance over other cell counters. Furthermore, the ML model is versatile, as an F1 score of 0.96 was also obtained on images of Trypan blue-stained human embryonic kidney (HEK) cells that the model had not been trained on. Our implementation of the ML model comes with a straightforward user interface and can image in batches, which makes it highly suitable for the evaluation of multiple parallel cultures (e.g., in Design of Experiments). Overall, this approach for accurate classification of cells provides a fast, bias-free alternative to manual counting. Previous studies have shown that the expression of the viral P1 structural proteins and the 3CD protease is sufficient to produce enterovirus-like particles in various organisms. However, there has been a lack of optimization based on the interplay between the three most commonly altered infection parameters, namely multiplicity of infection (MOI), viable cell density at the time of infection (VCD), and the infection period (tinf). In Chapter 3 we addressed this point by using Design of Experiments (DoE) to optimize the production of both EV71 and CVA6 VLPs. Our results indicated distinctively different preferences for infection parameters between the two types of VLPs, with EV71 VLP production preferring low MOI, low VCD, and long infection period, while CVA6 VLP production preferring for high MOI, high VCD and long infection period. Additionally, we developed a purification process for both VLPs, resulting in yields of 158 mg/l and 38 ml/l of culture volume for purified EV71 and CVA6 VLPs, respectively. These concentrations translate into thousands to tens of thousands of vaccines, highlighting the economic potential of enterovirus-like particles for vaccine purposes. Virus-like particles have been identified as a promising approach for the development of a multivalent vaccine. However, their stability is a major issue due to the significantly lower particle integrity lifetimes compared to inactivated vaccines. In Chapter 4, the VLPs produced using the optimized protocols described in Chapter 3 were subjected to biophysical characterization. We employed multiple biophysical techniques such as transmission electron microscopy and atomic force microscopy, to elucidate the origins of the reduced VLP stability (on average 1.5-2-fold lower) in comparison to native virions. Contrary to previous work on enterovirus VLPs, this study demonstrates that a substantial portion (31%) of the produced VLPs were able to encapsidate viral RNA (vRNA). Additionally, this work shows that the presence of vRNA in the capsids may not be the primary factor in enterovirus capsid stability. Furthermore, vRNA may not be the sole factor responsible for triggering the stabilizing viral maturation, and other underlying mechanisms may be at play. To achieve stability comparable to that of virions, artificial methods of inducing viral maturation or alternative means of stabilizing the capsids are of the utmost important to ensure success of VLPs as vaccine candidates. In Chapter 5, we present a protocol for the simultaneous investigation of RNA synthesis dynamics of hundreds of single polymerases with magnetic tweezers (MT). The protocol encompasses the entire process, starting from RNA construct preparation to quantitative and statistical analysis of the MT measurements of RNA synthesis kinetics. The protocol enables the measurement of hundreds of RNA tethers simultaneously, resulting in the characterization of single-molecule dynamics, which is presented in the subsequent chapter.
Chapter 6 of this dissertation showcases the potential of magnetic tweezers (MT) for the detailed mechanistic characterization of the viral RNA-dependent RNA polymerase (RdRp). By examining the pause dynamics and probabilities of each viral polymerase, we were able to decipher their individual mechanistic properties. In particular, we investigated the effects of the T-1106 triphosphate, a pyrazine-carboxamide ribonucleotide with antiviral properties, on the enterovirus A71 RdRp. Our result indicated that T-1106 incorporation into nascent RNA led to increased pauses and backtracking by the RdRp. Additionally, we identified the backtracked state as an intermediate used by the RdRp for copy-back RNA synthesis and homologous recombination, suggesting that pyrazine-carboxamide ribonucleotides function by promoting template switching and formation of defective genomes. Finally, we demonstrated that MT can scan promising antiviral candidates and indicate the most propitious ones for further development. The detailed mechanistic characterization of viral RdRp dynamics afforded by MT is a promising avenue for identifying and optimizing antiviral therapeutics. Chapter 7 of this dissertation provides concluding remarks and aims to illuminate potential avenues for subsequent studies. This work can serve as a basis for future investigations, not only from a biophysical perspective but also from a biochemical standpoint.
@en