Building a bridge between biophysics and neurobiology: A synergic approach to develop Alzheimer’s disease translational models and research tools

Alzheimer's disease (AD) is the most prevalent neurodegenerative disorder, accounting for 60-70% of dementia cases diagnosed worldwide. The impact of AD is rapidly increasing, with deaths more than doubling from 2000 to 2021. As populations age, especially in industrialized nations, the socio-economic burden of AD is massively escalating, necessitating the urgent development of effective treatments to delay, halt or reverse this fatal disease. In 1906, Dr. Alois Alzheimer identified the two main cardinal lesions of AD brains: (i) extracellular amyloid deposits or plaques; and (ii) intracellular neurofibrillary tangles (NFTs). We now know that these lesions result from the misfolding and aggregation of the intrinsically disordered proteins amyloid-β (Aβ) and tau, respectively. For decades, these proteins have been the focal point of research efforts and drug discovery initiatives. Recently, the U.S. Food and Drug Administration approved the first three disease-modifying drugs, which target amyloid plaques: aducanumab (2020), lecanemab (2021), and donanemab (2024). However, the scientific community remains divided over their clinical impact, given the limited cognitive improvements and severe side effects reported. The reduced efficacy of current therapeutics can be attributed to at least two critical factors. Firstly, treatment intervention often occurs at a late disease stage, when the patient's therapeutic window has already significantly narrowed. Secondly, the drugs target Aβ aggregate species that may not be one of the primary AD cytotoxic agents. The Aβ aggregation cascade involves a series of complex processes, wherein misfolded monomers transition through a wide range of intermediate oligomeric states to form assemblies such as protofibrils and mature fibrils, characterized by cross-β core secondary structures. Oligomers, rather than mature fibrils, are increasingly recognized as the primary toxic species in AD. However, due to their heterogeneous and transient nature, capturing these species to characterise their mechanisms of formation and cytotoxicity is a frustrating task. Moreover, Aβ can undergo heterogeneous nucleation and aggregation with other peptides and brain biomolecules, which increases the complexity of existing aggregate species. For these reasons, Aβ oligomers are often pushed to the periphery of drug discovery pipelines. In the first sections of this thesis, I present the development of tools and models to make Aβ oligomers more accessible targets for drug discovery and preclinical setups. To contribute new tools towards unravelling the structural determinants of the cytotoxicity of oligomer species, in Chapter 3, I introduce a step-by-step method to stabilize physiologically-relevant, off-pathway Aβ42 oligomers using Zn(II), a brain-enriched cation. This approach enables a robust analysis of oligomer structure-toxicity relationships and modeling AD-relevant phenotypes – such as calcium influx, reactive oxidative species (ROS) production, and mitochondrial dysfunction – in human cell systems. Additionally, I detail a workflow for fine-tuning experimental parameters to generate kinetically-trapped oligomers from any aggregation-prone peptide of interest. Traditional in vitro kinetic studies of Aβ aggregation inform on the microscopic mechanisms and rates of formation of on-pathway Aβ oligomers, but fail to capture the complexity of the cellular environment and the biological effects of oligomer formation. To address this limitation, in Chapter 4, I present a protocol to control the in situ formation of on-pathway Aβ42 oligomers via secondary nucleation (the microscopic mechanism that has the major impact in oligomer production) using human neuronal cultures. This is accomplished by seeding the aggregation of the Aβ42 monomer using pre-formed amyloid fibrils, directly on cells. This model bridges the gap between in vitro kinetic studies and cellular biology, providing an effective platform for studying Aβ oligomerization and testing potential inhibitors. To move from methods relying on exogenously-driven Aβ oligomerisation to approaches that mimic the endogenous Aβ aggregation cascade of human brain cells, in Chapter 5 I work on the development of robust and scalable human cell systems that can recapitulate endogenous Aβ production and aggregation, along with associated dysfunctional phenotypes. Creating human induced pluripotent stem cells (hiPSC)-derived neurons that accurately recapitulate endogenous Aβ aggregation and associated pathologies in a scalable, high-throughput and robust format, is challenging. This is due to the intrinsic variability of iPSC systems and the difficulties to mimic endogenous Aβ aggregation in neuronal systems without relying on complex three-dimensional cultures or artificial overexpression of several familial AD variants. To this aim, I established a robust hiPSC differentiation protocol with stringent quality control checkpoints to generate well-characterized cortical neuronal progenitor cells (NPCs) and derived cortical glutamatergic neurons. The protocol has been scaled up to produce batches of millions of high-quality cryopreserved progenitor cells, suitable for high-throughput applications. Using the pro-inflammatory stressor TNF-⍺, I developed a neuronal model with these standardized hiPSC cultures that demonstrates endogenous Aβ production and aggregation, along with dysfunctional phenotypes such as altered synapsis and hypermetabolism. The research presented in Chapters 3, 4, and 5, revealed that Aβ aggregates in the developed cell models mostly co-localize with cell membranes and progress from spherical to irregular, fibrillar structures over time. This morphological progression is typical of proteins undergoing phase separation, liquid-to-solid transition (LTST), and subsequent amyloid aggregation. Given that cell membranes are primarily composed of lipids, which have the ability to phase separate and bind to Aβ peptides, in Chapter 6, we investigated the role of lipids in regulating Aβ aggregation and phase separation. We show that Aβ forms biocondensates with lipids, a mechanism that favors primary nucleation and accelerates the conversion of monomers into fibrils. This process may dominate early amyloid-β aggregation and can form the basis for novel therapeutic intervention strategies. In the final sections of this thesis, I expand our focus from the pathological event of Aβ aggregation to a broader perspective on the nature of AD. While preclinical settings may benefit from translational tools to model Aβ aggregation and oligomer formation, it is crucial to recognize that AD extends beyond Aβ and tau pathologies. AD represents a highly complex and heterogeneous spectrum of disorders with multiple clinical and neuropathological manifestations that vary among patient subtypes, particularly in sporadic forms of the disease (sAD). We emphasize the importance of integrating this heterogeneity in preclinical settings to accelerate drug discovery programs towards more effective, personalized medicine. sAD accounts for approximately 90% of all AD cases and stems from a complex interplay of genetic and environmental factors that: (i) vary across patient groups, and (ii) propel cells towards a variety of early disease states. These disease states can be defined by transcriptomic fingerprints that functionally drive cells into the disease phenotype. Therefore, by integrating such fingerprints in human cells, one can generate panels of AD models. In Chapter 7, I worked towards disentangling the nature of such fingerprints and present a cell-based pipeline to identify early disease drivers from patient-derived transcriptomic signatures. This work resulted in the creation of a comprehensive model sporadic AD model in iPSC-derived neurons, based on the early downregulation of FBXO2, a gene encoding a subunit of the ubiquitin protein ligase complex SCF. This model captures key pathological features such as synaptic dysfunction, Aβ aggregation, and tau hyperphosphorylation. To accelerate the generation of AD preclinical models, in Chapter 8 I introduce the designs of a novel human cellular biological tool, the Recipient Line, that holds the essential orthogonal genetic elements to capture the complex and heterogenous molecular nature of AD in a standardised, time and cost-effective manner. The Recipient Line holds a well-characterised genetic background, the plasticity to be converted into the cell type from which patient-derived molecular fingerprints are predicted, and the genetic tools necessary to: (i) allow for a rapid and systematic site-specific insertion of fingerprint modulators without the need of applying cumbersome CRISPR gene editing approaches, and (ii) enable a temporal control of the manifestation of each molecular fingerprint, and therefore, the entry into patient-specific disease states. This approach aims to bridge the gap between computational predictions and biological validation, enhancing the translational value and predictive accuracy of artificial intelligence (AI) pipelines, ultimately accelerating the development of personalized therapeutics for AD.

Name dc:type.qualificationname

Doctor of Philosophy (PhD)

Level dc:type.qualificationlevel

Doctoral

Grantor dc:publisher.institution

University of Cambridge

Year dc:date.issued

2024

Author dc:creator

• Gonzalez Diaz, Alicia

Advisor dc:contributor.advisor

• Vendruscolo, Michele

• Amyloid-β
• iPSC
• Oligomers
• Phase separation
• Preclinical models
• Sporadic Alzheimer's disease
• TNF⍺

Licence

• https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/9c915f54-8053-4da1-9895-6fdbbd987284/download
• https://www.rioxx.net/licenses/all-rights-reserved/

Language dc:language

eng

DOI dc:identifier.doi

https://doi.org/10.17863/CAM.113748

OAI identifier oai:identifier

oai:www.repository.cam.ac.uk:1810/376550