Intrinsic RNA targets of Fused in sarcoma inhuman biology and disease & mechanism of splicing regulation by RNA-binding motif protein 39

Abstract

In humans, more than 1500 RNA-binding proteins (RBPs) and accessory factors assemble with RNA of numerous functional classes to form ribonucleoprotein particles (RNPs). While some RBPs are structural components of RNPs or endow them with enzymatic activities to enable highly diverse functions, others regulate the processing, localisation and stability of their targeted RNAs. One extraordinary mechanism which not only shapes, but also diversifies the transcriptional output of most eukaryotic genomes is alternative splicing, the differential excision of non-coding introns and alternative exons from precursor messenger RNA (pre-mRNA) and long non-coding RNA. Its underlying general splicing reaction is orchestrated andcatalysed by the spliceosome, a remarkably dynamic and complex RNP machine comprising 5 small nuclear RNAs (snRNAs) and over 200 unique proteins. To understand how RBPs regulate alternative splicing, their interactions with primary transcripts and components of the splicing machinery must be identified and rigorously characterised. Besides their physiological roles, dysfunction of RBPs has been linked to cancers of the myeloid lineage and neurodegenerative disorders such as amyotrophic lateral sclerosis (ALS), indicating that altered RNA metabolism may contribute to the disease phenotype.

The first part of this thesis focuses on Fused in sarcoma (FUS), a highly abundant and ubiquitously expressed RBP of the hnRNP family. The FUS molecule comprises two functional modules: An N-terminal low complexity (LC) region that interacts with other proteins and drives liquid-liquid phase separation of FUS and a C-terminal module that binds to RNA. Although FUS was discovered more than two decades ago, its cellular activities remain incompletely understood. Here, I aimed to identify central RNA targets of FUS and investigate how the LC region affects the recruitment of FUS to RNA on a transcriptome-wide scale. To this end, I performed crosslinking and immunoprecipitation coupled to high-throughput sequencing (CLIP-Seq) with full-length FUS as well as an artificial deletion construct encompassing only the RNA-binding module (FUS-RBD). While the full-length protein predominantly bound to introns, this regional preference was partially abrogated for FUS-RBD, indicating that the LC domain promotes the co-transcriptional recruitment of FUS to pre-mRNAs. Consistent with a modest intrinsic preference for stem loop motifs, both FUS constructs extensively crosslinked to the highly structured snRNAs, with the U1 snRNA being the top target in our enrichment analysis. In collaboration with the group of Frédéric Allain, we uncovered the molecular basis of the FUS -U1 snRNP interaction: The RRM of FUS recognises stem loop 3 of the U1 snRNA, which is
solvent exposed in the crystal structure of the U1 snRNP. This unexpected mode of interaction with the splicing machinery could explain how FUS couples transcription to splicing or regulates 5’-splice site selection.

Mutations in the FUS gene cause early-onset ALS, an inevitably fatal neurodegenerative disease affecting motor neurons in the motor cortex and spinal cord. Typically, these mutations destroy the nuclear localisation signal of FUS and thus mis-localise the protein to the cytoplasm, where it initiates a complex disease cascade via an unknown toxic gain-of-function. To identify RNA targets that could be linked to the pathomechanism of FUS-linked ALS, I performed CLIP with a cytoplasmic FUS-RBD construct. Intriguingly, the U1 snRNA was also considerably bound by cytoplasmic FUS, however, the mode of RNA binding was different: Besides the interaction between RRM and stem loop 3, the zinc finger domain recognised a GGU motif overlapping with the Sm-site. This bi-partite binding results in a high-affinity interaction that prevents the assembly of the heptameric Sm ring onto the Sm-site in vitro, a critical step in the biogenesis of snRNPs that occurs in the cytoplasm. Besides this aberrant interaction with FUS, cellular stress interferes with cytoplasmic snRNP homeostasis via the sequestration of biogenesis intermediates into stress granules. This dual insult could explain the cytoplasmic accumulation of U1 snRNA in spinal motor neurons of a FUS-ALS mouse model and provides a molecular link between the motor neuron diseases ALS and spinal muscular atrophy.

The second part of this thesis deals with RNA-binding motif protein 39 (RBM39), an SR-like family member that regulates alternative splicing events required for transformation and tumour growth in acute myeloid leukemia and has therefore emerged as a promising drug target. Yet, the mechanism of RBM39-dependent splicing and especially the role of its three RNA-recognition motifs (RRMs) in this process has not been addressed. Here, I assessed the contribution of each RRM in RBM39 knockdown and rescue experiments and uncovered a critical role for RRM1 in the splicing of endogenous RBM39 targets, whereas RRM2 and RRM3 only mildly contribute to RBM39’sfunctionin splicing. Intriguingly, preliminary evidence from the Allain lab indicates that RRM1 interacts with the U1 snRNA. Furthermore, I found that RBM39 promotes the inclusion of a poisonexon into its pre-mRNA, which produces a non-functional transcript that is rapidly degraded by the nonsense-mediated decay machinery.

A unifying theme of these two projects is that both FUS and RBM39 could regulate alternative splicing by simultaneously contacting the U1 snRNA component of the splicing machinery as well as the pre-mRNA substrate. Such RNA-bridging interactions are a novel concept in splicing regulation and extend the role of RNA-binding domains beyond providing substrate specificity.

Date of Award	1 Sept 2020
Original language	English
Awarding Institution	King's College London
Supervisor	Marc-David Ruepp (Supervisor)