The role of the RNA-binding protein FUS in axonal organisation and disease

In this thesis, I discuss the role of the RNA-binding protein Fused in sarcoma (FUS) in axonal organisation and disease. FUS aggregates in forms of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD), and ALS-associated mutations change its phase transitions and subcellular localisation. This is known to compromise axonal local protein synthesis (LPS), but it is not well-understood which critical LPS-dependent processes are affected. This is important, as these processes represent potential therapeutic targets.

In my work, I built on insights from the field of developmental neurobiology to formulate hypotheses on the physiological consequences of altered FUS phase behaviour. Specifically, I focussed on axonal survival signalling and cytoskeletal remodelling. I tested these hypotheses using the Xenopus laevis retinal ganglion cell (RGC) model system, which is an established model system for combining in vitro and in vivo studies into LPS-dependent axonal properties.

First, I studied the effect of FUS mutation on LPS of axonal survival-associated proteins, as changes in the local activity of these could be linked directly to axonal degeneration. I focussed on the essential axonal survival factor nicotinamide mononucleotide adenylyltransferase 2 (NMNAT2), as its mRNA is a potential FUS binding target. I hypothesised that axonal NMNAT2 levels are sustained by LPS, as this protein is known to have a very short half-life. However, by combining in situ hybridisation and quantitative PCR approaches, I found that nmnat2 mRNA does not localise to RGC axons, falsifying this hypothesis. Therefore, I shifted the focus of my research to axonal cytoskeletal reorganisation, as this is known to be regulated by LPS, and ALS is associated with mutations in cytoskeleton-linked genes.

I next hypothesised that reorganisation of the axonal cytoskeleton is compromised by mutant FUS, as this process is strongly LPS-dependent, and so studied the dynamic axon terminal. I used two previously reported ALS/FTD X. laevis models: the juvenile ALS-associated nuclear localisation signal mutant FUS(P525L) and an artificial FTD-mimic of hypomethylated FUS, FUS(16R). Mutant FUS compromised the actin cytoskeleton of the growth cone in vitro, resulting in considerable changes in growth cone mechanoproperties. This occurred without detectable reductions in axonal microtubule density or RNA transport function, indicating this defect may be specific to the dynamic actin cytoskeleton. However, axonal cytoskeletal organisation is dependent on local cues in vivo, and ALS is considered a non-cell-autonomous disease. Therefore, I subsequently studied whether this actin change correlates with in vivo phenotypes, hypothesising that mutant FUS would compromise actin-dependent axon branching in vivo. Mutant FUS indeed reduced axon complexity in vivo, and FUS(P525L) additionally induced axon looping, pointing to the occurrence of signalling defects. This loss of branching complexity has implications for synaptic connectivity, changes in which are a key feature of neurological disorders. Therefore, altered actin remodelling could be a central phenotype on which multiple pathomechanisms associated with ALS/FTD converge.

These results have wider conceptual implications for our understanding of the nature of ALS/FTD. In particular, the present work demonstrates that mutant FUS-associated changes to axonal organisation can occur early in development already, which supports the emerging concept that neurodevelopmental and neurodegenerative diseases may have overlapping molecular players and may therefore share aspects of pathomechanisms. I conclude this thesis by discussing how mutant FUS-induced developmental phenotypes may be connected to onset of symptoms in adolescence or adulthood, and which future research directions emerge from this perspective.

Francesca Wilhelmina van Tartwijk December 2022