Atomic force microscopy methods to study protein phase transitions

Abstract

Proteins occupy many regions of the phase space, with the ability to form dense liquids and gels, such as viscoelastic condensates, and solids, such as disordered oligomers and ordered amyloid fibrils. This rich phase behaviour of polypeptides is necessary for biological function; however it also renders them a formidable experimental challenge, due to their dynamic, heterogeneous nature, and dimensions ranging from nanometre to micrometre scale. It remains a crucial problem to monitor the molecular structures formed, as aberrant phase transitions have been implicated in neurodegenerative disease. Therefore, this thesis aims to uncover the molecular and structural changes associated with protein phase changes via atomic force microscopy, infrared spectroscopy, and other biophysical techniques. AFM is a powerful surface-based technique which can provide nanometre-resolved morphological and mechanical information on protein aggregates. As such, it is capable of visualising diverse species which are below the resolution limit of light, making it particularly well-suited to capture small assembly states with nanometre dimensions. Therefore, we first applied it to the investigation of small, soluble oligomeric species extracted from human brains with neurodegenerative disease. We identified species which had interesting morphologies; notably, we observed protofibrillar structures in a loop-like conformation. This reminded us of ‘amyloid pores’ which were first reported over two decades ago. Much attention was focused on identifying the precise structural features and biological relevance of these ‘pores’ however no clear consensus was ever reached. Therefore, we aimed to perform the first structural characterisation of amyloid rings extracted from human AD brains. However, these are precious samples, and these loop-like structures exist in relatively low abundance within a complex mixture, meaning they are not amenable to structural study using bulk techniques. Therefore, we sought to access structural information by characterising their mechanical properties based on the bending fluctuations of protofibrils, borrowing concepts from polymer physics. We first demonstrated that these loop structures possess high levels of conformational flexibility. Then, by exploiting the established correlation between the β-sheet content of amyloid fibrils and their mechanical properties, we were able to determine that these loop forming structures lack the extended β-sheet content characteristic of mature amyloid fibrils. This allows them to bend to adopt many conformations, including for the two ends to meet to form loop structures. The ends of the chains can also dissociate to adopt extended conformations such that they can exist in both open and closed states. We show that this process is very well-modelled by theory of semi-flexible polymers, which quantitatively predicts the range in which loop structures exist. These findings suggest that the formation of amyloid loops is a generic process, governed by the fact that intermediate protofibrillar species behave as semiflexible polymers, thus adding to our overall understanding of the formation of disease-relevant protein assembly states. Thus, AFM imaging allowed us to understand the structure and formation of amyloid loop structures, clarifying our understanding of these ‘pores’. In this case, the deposition of the heterogeneous sample onto a 2-dimensional surface enabled us to capture and characterise an assembly state which is not readily detectable using bulk methods. However, the deposition of samples onto a surface generally represents an inherent limitation of surface-based techniques. Indeed, it is accepted that the sample preparation process affects the conformational state of biomolecules, especially protein aggregates which are incredibly sensitive to their environment. Therefore, we sought to improve the sample deposition process to allow us to access the ‘solution-like state’ of proteins on surfaces. To this end, we developed a microfluidic spray deposition method which exploits controlled, ultra-fast sample deposition to access biological processes which occur on timescales 1000x faster than traditional deposition timescales. We demonstrate that this method is able to maintain the heterogeneity of complex biological samples, while also preserving the conformation as in solution. We then exploit this to identify the monomeric and oligomeric species present at the very earliest timescales of self-assembly, which are believed to be involved in disease onset. These early assembly states are typically inaccessible by surface-based characterisation, due to their transient nature. By applying ultra-fast deposition, we preserve the full heterogeneity of these protein species, as well as their secondary and quaternary structure. Thus, we were able to acquire quantitative information on the initial stages of protein aggregation. The ability to quantitatively characterise heterogeneous protein phases using AFM and other surface-based techniques represents an exciting advance. Therefore, this spray depo- sition technology served as the platform for the majority of work presented in this thesis, allowing us to access structural information on protein phases which are notoriously difficult to study. We first applied this method to uncover the secondary and quaternary structural changes that arise from fibril self-association during time-dependent maturation processes, via the use of AFM and infrared spectroscopy. This provided a molecular basis for under- standing how amyloids evolve over time. We then sought to extend this knowledge to the study of amyloid formation within protein condensates. Condensates are dense biomolecular assemblies which have complex, liquid-like material behaviour. The material properties of these condensates play important roles in their cellular functions, with aberrant liquid-to-solid phase transitions having been implicated in neurode- generative diseases. Using the fused in sarcoma (FUS) protein as an example, it has been suggested that amyloid fibrils may form within the dense phase. However, as of yet, it has been difficult to provide an in-depth structural characterisation of these amyloid formation processes in physiologically-relevant conditions. Unfortunately, there currently exist limited tools which are able to go ‘inside’ the condensate to understand the liquid-to-solid transition. Therefore, we sought to apply AFM and IR spectroscopy to understand the morphologi- cal, mechanical and conformational changes associated with the liquid-to-solid transition. Through our studies of fibril maturation, we demonstrated that these techniques are capable of detecting the molecular changes associated amyloid formation. However, we quickly found that condensates were not amenable to systematic studies on surfaces, as the presence of a solid support interferes with the conformational state. Therefore, we hypothesized that the spray deposition technology would allow us to maintain relevant structural features of condensates on surfaces, thus enabling us to characterise how changes in the morphological and material properties correspond with molecular changes at the protein level. Indeed this hypothesis was correct, and we first demonstrated that we were able to maintain relevant structural features of condensates in physiologically-relevant conditions on surfaces. Then, we characterised the spatio-temporal changes in condensate structure and mechanical properties to reveal local liquid-to-solid phase transitions in individual condensates. We were also able to clarify the nature of the solid assemblies by demonstrating that they lack the properties of traditional amyloid fibrils. Rather, these solid structures are composed of heterogeneous, non-amyloid β-sheets, which are stabilised by distinct interactions compared to the fluid state. Overall, this allowed us to identify the molecular conformations associated with different physical states of condensates, clarify the nature of the solid state, and establish a technology platform to understand the role of phase behaviour in condensate function and dysfunction. I hope the key takeaway of this thesis is the need for creative methods to generate robust information about these complex protein systems. In this context, rather than invent new instrumentation, we simply adapted well-established methods to precisely suit the biological needs of our sample. The primary example of this is the development of the spray deposition approach; by being more considerate of the limitations of the sample deposition, we were able to improve the capabilities of surface-based techniques and unlock their application to the robust structural study of condensates on surfaces for the first time. Overall, this approach has allowed us to clarify the structural properties of various protein phases, providing a better understanding of the role of phase transitions in neurodegeneration.