1
Table of content
Table of content ... 1
List with figures ... 2
List with tables ... 2
List with abbreviations ... 3
Abstract ... 6
Introduction ... 8
A. Frontotemporal lobar degeneration and amyotrophic lateral sclerosis ... 8
1. Clinical and pathological features ... 8
2. Multiple genes play a key role in FTLD and ALS ... 10
3. ALS and FTLD represent a disease continuum ... 12
B. TDP-43: a key protein in the pathogenesis of neurodegenerative diseases ... 14
1. Structural features of TDP-43 ... 14
2. Physiological role of TDP-43 in the cell ... 15
3. Pathological features of TDP-43 in ALS and FTLD ... 17
4. A prion-like mechanism for TDP-43? ... 19
C. Description of the study ... 20
Aim ... 22
Aim 1: To decipher seeding properties of pathological TDP-43 extracted from brain samples obtained from patients with FTLD pathology. ... 22
Aim 2: Study toxicity of pathological TDP-43 in neuronal cells. ... 22
Results ... 24
A. SarkoSpin is a tool to isolate pathological TDP-43 from the brain ... 24
B. HEK293T and HeLa cells expressing hemagglutinin-tagged TDP-43 serves as model to investigate seeding properties of pathological TDP-43 seeds ... 25
C. Different FTLD-TDP subtypes show distinct density profile of pathological TDP-43 27 D. Primary neuron culture as a second cellular model to asses FTLD-TDP subtype toxicity ... 28
Discussion ... 31
Material and methods ... 35
References ... 42
Figure Legend: ... 52
2
List with figures
Figure 1: Overview of genetic causes in sporadic and familial ALS and FTD cases ... 52 Figure 2: Representation of the clinical, genetic and pathological continuum in ALS and FTLD ... 52 Figure 3: Organisation of TDP-43 and equilibrium in the nucleus and cytoplasm in
physiological and pathological conditions ... 52
Figure 4: Sarkospin isolates pathological TDP-43 from cortical brain samples. ... 52 Figure 5: Seeding of pathological TDP-43 on HEK293T cells triggers more endogenous
aggregation. ... 53
3
List with abbreviations
AD Alzheimer’s disease
ALS Amyotrophic lateral sclerosis
AraC Arabinoside
BBS α-bungarotoxin binding site
bvFTD behavioural variant frontotemporal dementia C9ORF72 chromosome 9 open reading frame 72 gene CHMP2B charged multivesicular body protein 2B CLSM Confocal laser scanning microscopy
CTD C-terminal domain
DDT Dihiotreitol
DMEM Dulbecco’s Modified Eagle’s Medium
DN Dystrophic neurites
Dox Doxycycline
ESCRTs Endosomal sorting complexes required for transport
fALS Familial ALS
FCS Fetal calf serum
fFTLD Familial FTLD
FTD Frontotemporal dementia
FTLD Frontotemporal lobar dementia
FUS/TLS Fused in sarcoma/translated in liposarcoma GFAP Glial fibrillary acidic protein
GRN Granuline
HBSS Hanks balanced salt solution
HIV-1 Human immunodeficiency virus type 1 hnRNP Heterogeneous ribonucleoprotein
HS Homogenization-Solubilzation
IBMPFD-ALS Inclusion body myopathy associated with Paget disease of bone and frontotemporal dementia
ILVs Intralumenal vesicles
4
LMN Lower motor neurons
mAb Monoclonal Antibody
MAPT Microtubule-associated protein tau
miRNA microRNA
MND Motor neuron disease
MVB Multivesicular body biogenesis
NB Neurobasal medium
NCI Neuronal cytoplasmic inclusions
NF Neurofilaments
NLS Nuclear localisation signal
NTD N-terminal domain
pAb Polyclonal Antibody
PBS Phosphate buffered saline PDL Poly-D-Lysine Hydrobromide
PGRN Progranulin
PNF Phosphoneurofilaments
Pri-miRNA Primary microRNA PSD-95 Post-synaptic density 95 pTDP-43 Phosphorylated TDP-43
PTM Post-translational modifications
RAN Repeat-associated-non-ATG
rcTDP-43 Recombinant TDP-43
RISC RNA-induced silencing complex
RRM RNA recognition motif
sALS Sporadic ALS
Sarkosyl N-lauroyl-sarcosine sFTLD TAR Sporadic FTLD Transactive response SG Stress granule
SOD1 Cu/Zn superoxide dismutase
5 TDP-43 Transactive response DNA binding protein
UBQLN2 Ubiquilin-2
UCL University College of London
UMN Upper motor neurons
UPS Ubiquitin proteasome system UZH University of Zurich
VCP Valosin containing protein ZMB
SKS
Center for Microscopy and Image Analysis Sarkospin
6
Abstract
Amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD) are two neurodegenerative diseases, with different clinical pictures, but with common cellular mechanisms. Among those, aggregation of the RNA-binding protein TDP-43 was identified as a pathological hallmark of both conditions. Recent work from the Polymenidou lab uncovered that different species of TDP-43 aggregates correlate with specific disease subtypes. During my master project, I contributed to the development of new cellular models to investigate the pathogenesis of these heterogeneous TDP-43 pathologies and explore the consequences of different TDP-43 aggregate species in the outcome of the disease. I used patient-derived pure pathological TDP-43 seeds to study the sequence of events leading to the pathology. To do so, I took take advantage of the lab expertise in HEK293T, HeLa cell lines and primary neuronal culture to explore the seeding properties and intracellular toxicity of pathological aggregates. Further, I characterized biochemical properties of the TDP-43 aggregates that correlate with distinct subtypes. I used tools such as SarkoSpin and confocal laser scanning microscopy to decipher the pathological mechanisms. I used ultracentrifugation for biochemical profiling of the pathological aggregates. With this study, I aimed to unravel cellular mechanisms related to the TDP-43 neurotoxicity building invaluable primary data that could help in identifying new drug targets. This is of great value since disease-modifying treatments for TDP-43 associated proteinopathies are currently lacking.
7
Acknowledgments
I would like to thank Prof. Dr. Magdalini Polymenidou for the opportunity if letting me grow as a scientist in her lab and trusting me with such an interesting and diverse project.
I would further like to thank Prof. Dr. Geert von Loo who agreed to support and supervise my journey.
I also want to thank Dr. Pierre de Rossi for his supervision, his input, all the help and scientific advice he has given me.
A huge thank you to Johanna, who was my partner in crime in the lab. Late nights and weekends were always a pleasure when such competent and fun company is around.
I want to thank Julien for all the organization he did for me, Marian for his amazing explanation and guidance, Manu for introducing me to cell culture, Elena for all the help and guidance with the neurons, Sonu for all her Kraken moments, Kathy for good discussions during lab meetings, Tomas and Mischa for being cool room bodies and Auri for her help.
I also want to thank my parents, family and friends for supporting me in every way they could.
And finally I want to thank Han and Xavi, who are both the loves of my life. Thank you for all the help, patience and “peptalks”.
8
Introduction
Neurodegenerative diseases are characterized by progressive and irreversible loss of neurons in the central and peripheral nervous system. The hallmark of these diseases is the accumulation of intra- or extracellular pathological proteins leading to neuronal damage, loss of function and neuronal death (Forman et al, 2007). Alzheimer’s disease (AD), the most common cause of dementia, is characterized by accumulation of extracellular plaques containing β-amyloid depositions and intracellular neurofibrillary tangles composed of hyperphosphorylated microtubule-associated protein tau (MAPT) (Long & Holtzman, 2019). Similarly, typical features for Parkinson’s disease, another neurodegenerative disease, are the cytoplasmic presence of Lewy bodies enriched in α-synuclein (Deng et al, 2018). Huntington disease is characterized by the expansion of the CAG trinucleotide repeat in the
huntingtin gene, leading to accumulation of a toxic variant of huntingtin protein (McColgan
& Tabrizi, 2018). Frontotemporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS) are further examples of neurodegenerative diseases. Tough characterized by cytosolic accumulation of proteins, the diversity of involved proteins leading to neurodegeneration is amongst the most challenging to understand in the neurodegenerative field. Therefore ALS and FTLD represent a thought-provoking subclass of neurodegeneration yet to be characterized and deciphered (Forman et al, 2007).
A. Frontotemporal lobar degeneration and amyotrophic lateral sclerosis
1. Clinical and pathological features
Frontotemporal lobar degeneration (FTLD) is the most common neurodegenerative disease under the age of 65. The disease occurs with an overall estimated prevalence of 10-20/100,000 for people between age of 45 – 65 and incidence of 8.9/100,000 for people be-tween age of 60 – 69 (Ratnavalli et al, 2002; Harvey et al, 2003; Rosso et al, 2003; Knopman
et al, 2004; Borroni et al, 2010; Onyike & Diehl-Schmid, 2013). However, these numbers
were recently contested (Coyle-Gilchrist et al, 2016; Nelson et al, 2019) and would require more statistical studies. The affected regions in FTLD are the frontal and anterior temporal lobes of the brain, which play an important role in emotion control, social behaviour, memory and language. Therefore, FTLD is clinically characterised by behavioural and
per-9 sonality changes along with loss of language skills. (Neumann et al, 2006; Van Langenhove et
al, 2012; Ling et al, 2013).
Approximately 30% of the FTLD cases show clear family history indicating inheritance also called familial FTLD (fFTLD) while 70% of the cases appear sporadically (sFTLD) (Laferrière et
al, 2015). The key pathological feature of FTLD is the presence of intraneuronal inclusions in
degenerating neurons and glial cells. Sub-classification of FTLD can be done based on the main protein accumulated in the pathological inclusions which are FTLD-TDP (45%), FTLD-tau (45%), FTLD-FUS (9%) and FTLD-UPS (ubiquitin proteasome system) (1%) (Ling et al, 2013). In contrast to FTLD, key features in ALS are amyotrophy (muscle degeneration) and lateral sclerosis (hardening of the anterior and lateral corticospinal tracts) due to degeneration of both upper (UMN) and lower (LMN) motor neurons causing progressive paralysis and respir-atory failure, typically leading to death within 1 to 5 years after onset.1 (Van Langenhove et
al, 2012; Ling et al, 2013; Laferrière et al, 2015). The disease is also known as Charcot’s (in
Europe) or Lou Gehrig’s (in North America) disease and is the most common motor neuron disease (MND). ALS occurs with an incidence of approximately 2 – 3/100,000 and the preva-lence for male is slightly higher than for female (Ratio M/F ~ 1.5/1). Around 10% of ALS are familial ALS (fALS) while 90% occurs sporadically (sALS) (Wijesekera & Leigh, 2009; Laferrière & Polymenidou, 2015).
In ALS, amyotrophy is caused by denervation as their corresponding anterior horn cells degenerate while the lateral sclerosis is due to degeneration of motor neurons and replacement by astrocytic gliosis. Pathology in UMN is characterized by depopulation of Betz cells and astrocytic gliosis in the motor cortex (grey matter and underlying subcortical white matter) as well as corticospinal tract and axonal loss within the pyramidal motor pathway. In LMN, the ventral horn of motor neurons in the spinal cord and brain stem are affected. At autopsy control, the number of neurons is reduced by 50% and the remaining neurons are anthropic containing intraneuronal inclusions such as Bunina bodies, ubiquitinated inclusions and neurofilament inclusions (Rowland & Shneider, 2001). Additionally, supporting cells such as surrounding glia cells, can significantly contribute to the progression of the disease in a non-cell autonomous manner (Ilieva et al, 2009).
1 While the UMN are situated in the cerebral cortex, the LMN in the anterior horn of the spinal cord. They UMN
10 ALS shows a genetic and phenotypic heterogeneity with indolent to aggressive variants determined by three main features: site of onset, rate of progression and relative number of UMN and LMN deficits. Classical sporadic ALS, familial ALS (autosomal dominant in their inheritance) and pacific type ALS (associated with dementia) are three generally accepted clinical subtypes of the disease.
2. Multiple genes play a key role in FTLD and ALS
Despite the fact that 25% of fALS, 90% of sALS, approximately 25% of fFTLD and 88% of sFTLD have unknown etiology, major progress has occurred in the past two decades to understand the potential genetic basis for ALS and FTLD (Figure 1) (Neumann et al, 2006; Mackenzie et al, 2011).
The first mutation identified in ALS was in the gene of the CU/Zn superoxide dismutase (SOD1), a homodimeric metalloenzyme (Cu/Zn) that is involved in ROS detoxification mechanisms. It accounts for approximately 20% of all fALS and less than 1% of sALS (Rosen
et al, 1993; Bruijn et al, 2004).
Over a decade later several other mutations were identified in two functionally similar nucleotide binding proteins called TAR DNA binding protein of 43kDA (TDP-43) and fused in sarcoma (FUS).TDP-43 is a nuclear located protein of 43kDa that is identified as a member in the complex of heterogeneous ribonucleoprotein family (hnRNP), sharing similarity with components of Drosha microprocessor complexes (Ling et al, 2010). It is involved in multiple biological processes by its ability of the protein to bind DNA, RNA and/ or proteins (Lillo & Hodges, 2009; Buratti & Baralle, 2012). Most dominantly TDP-43 is associated with RNA homeostasis. RNA targets includes genes associated with neuronal development, synaptic function and RNA metabolism as well as proteins that play an important role in neurodegeneration such as TDP-43 itself but also FUS, PGRN, Tau and Ataxan-1 & -2. Mutations in TDP-43 are found in 5% of familial ALS (Kabashi et al, 2008; Sreedharan et al, 2008; Van Deerlin et al, 2008) and are rare in patients with FTLD (Borroni et al, 2010). However, the pathological protein is found in more than 90% of ALS cases and 45% of the FTLD cases (Figure 1).
FUS is another protein involved in the pathology of ALS and FTLD. The protein consists of 526 amino acids with the ability to bind DNA/RNA. The protein is structurally and functionally
11 related to TDP-43. Normally, FUS is predominantly located in the nucleus while in pathology, mutant form of FUS accumulate in the cytoplasm of neurons forming inclusions, due to high aggregation propensity in an N-terminal prion-like domain (Lagier-Tourenne et al, 2010; Ling
et al, 2013; Yang et al, 2015; Laferrière & Polymenidou, 2015). Mutations in the FUS gene
account for 5% of fALS and is rare in FTLD patients, however FUS pathology shows significant severity (Kwiatkowski et al, 2009). Mutations are mainly located in the C-terminal nuclear localisation signal of the protein leading to accumulation of the protein in the cytoplasm. Other mutations located in the prion-like domain of FUS may directly increase the aggrega-tion propensity forming intracellular aggregates (Dormann et al, 2010; Van Langenhove et al, 2012). However, TDP-43 and FUS are similar RNA-binding proteins, their aggregation and pathogenic mechanism due to mutations might differ (Sun et al, 2011).
The most common genetic cause for both fFTLD and fALS are hexanucleotide (GGGGCC) repeats in the first non-coding introgenic region of the C9ORF72 (DeJesus-Hernandez et al, 2011; Renton et al, 2011; Hodges, 2012). These mutations are highly penetrant and decreased expression of C9ORF72 suggest haploinsufficiency as one of the underlying disease mechanisms (Gijselinck et al, 2012). Additionally, a gain of function disease cause has been proposed as well by either accumulation of toxic RNA or protein products. RNA transcripts of the expanded GGGGCC repeat and antisense CCCCGG repeat form nuclear loci in c9FTLD/ALS patients. Repeat-associated-non-ATG (RAN) translation results in the production of aggregation-prone proteins. Translation of GGGGCC repeats results in the synthesis of poly(GP), poly(GA) and poly(GR) proteins while translation of CCCCGG repeats results in poly(PR), poly(GP) and poly(PA) proteins which are all associated with toxicity (Gendron et al, 2013; Sareen et al, 2013).
Further mutations were found in genes of protein involved in the protein clearance pathway, either in the autophagy or the ubiquitin proteasome system. One example is Ubiquilin-2 (UBQLN2), gene encoding ubiquilin-2, a protein part of the ubiquilin family that regulates degradation of ubiquitinated proteins. Impaired function of UBQLN2 leads to abnormal protein accumulation, such as TDP-43, associated with aggregation and neurodegeneration (Deng et al, 2011). Another example is VCP, which plays a key role in maturation of ubiquitin-containing autophagosomes and thereby involved in protein clearance pathways or maintenance of proper protein homeostasis. Mutant VCP mediates partly through TDP-43
12 which is redistributed from the nucleus to the cytoplasm. For both UBQLN2 and VCP mutations have been identified in fFTLD and fALS cases. (Watts et al, 2007; Johnson et al, 2010; Ling et al, 2013)
Approximately 20% of all fFTLD and 3-4% of sFTLD show mutations in the microtubule- associated protein tau (MAPT) gene, which encodes the protein tau, involved in microtubules assembly and stabilization, creating a firm cytoskeleton in neuronal cells (neurons). The cytoskeleton contributes to the rigidity of the cell, ensures cell division and transport of molecules through the cell. Missense mutations in MAPT are characterized by cytoplasmic neurofibrillary inclusions composed of hyperphosphorylated tau (Hutton et al, 1998; Baker et al, 2006).
For 25% of fFTLD and 1-2% of sFTD, mutation in the progranulin gene (PGRN) have been reported. PGRN encodes a secreted growth factor involved in the regulation of multiple physiological processes including development, wound repair and inflammation. Most likely, mutations induce PGRN haploinsufficiency leading to neurodegeneration because of reduced PGRN-mediated neuronal survival. (Baker et al, 2006; Cruts et al, 2006).
Charged multivascular body protein 2B (CHMP2B) is a subunit from the endosomal sorting complexes required for transport (ESCRTs). This complex mediates cytokinesis and multivesicular body (MVB) biogenesis. Ubiquitinated entities are recognized, deubiquitinated and packed into intralumenal vesicles (ILVs), budding into late endosomes and evolving in MVBs. Point mutations in CHMP2B are mostly associated with FTLD, but some cases have been reported within the ALS spectrum disorders (Skibinski et al, 2005; Parkinson et al, 2006; Henne et al, 2012).
3. ALS and FTLD represent a disease continuum
Despite apparent different clinical picture, ALS and FTLD are now considered as a disease continuum as they share clinical, genetic and pathological features (Figure 2) (Van Langenhove et al, 2012; Ling et al, 2013; Laferrière et al, 2015). Multiple studies showed cognitive and behavioural abnormalities within ALS (Lomen-Hoerth et al, 2003; Appel, 2005; Ringholz et al, 2005) and motor system dysfunction in patients within FTLD (Lomen-Hoerth
et al, 2002; Burrell et al, 2011) linking both disease on the clinical level. The finding of
13 in both diseases, suggesting similar pathobiological pathways (Neumann et al., 2006, Aria et al., 2006). The final finding convincing the field of a disease spectrum was the finding of the most common genetic cause for both fALS and fFTD in the C9ORF72 gene (DeJesus-Hernandez et al, 2011; Renton et al, 2011; Hodges, 2012).
All this evidence suggests that ALS and FTLD are two extremes of a disease spectrum rather than two distinct diseases, making the understanding of these neurodegenerative diseases a challenge.
14
B. TDP-43: a key protein in the pathogenesis of neurodegenerative diseases
TDP-43 is a main component found in the ubiquitinated inclusions of ALS (>97%) and various forms of FTLD (45%) (Neumann et al., 2006; Aria et al., 2006). However, other neurodegenerative diseases have shown TDP-43 pathology including AD, hippocampal sclerosis, Huntington disease and pick’s disease (Amador-Ortiz et al, 2007; Schwab et al, 2008; Freeman et al, 2008). Given the abundance of this protein found accumulated in neurodegenererative diseases, TDP-43 has become one of the focus points to understand pathobiology and potential terapeutic approches in the neurodegenerative field.
1. Structural features of TDP-43
TDP-43 is encoded by the TARDBP gene, which is highly conserved in human, mouse,
Drosophila melanogaster and Caenorhabditis elegans, suggesting that small genetic
mutations might destabilize its structural properties (Wang et al, 2004). TDP-43 is composed of 414 amino acids and is ubiquitously expressed. Knockdown of TDP-43 leads to fatalities, highlighting the importance of this protein. TDP-43 shows the characteristic multidomain architecture of the hnRNP protein family, composed of 2 RNA recognition motifs (RRMs) flanked by an N-terminal domain (NTD) and glycine-rich C-terminal domain (CTD) (Figure 3a) (Lukavsky et al, 2013; Wang et al, 2013; Afroz et al, 2017).
The NTD is responsible for oligomerization of physiological TDP-43 into head-to-tail homo-oligomers which represents the functional form in vivo. This NTD-driven organisation into a dynamic solenoid-like structure are more resistant to cellular stress than its monomeric form, suggesting an antagonistic role by separating the aggregation prone CTD (Figure 3b) (Afroz et al, 2017). Structurally, the two RRM domains of TDP-43 have distinct RNA/DNA binding features. RRM1 is necessary and sufficient for recognizing and binding its target carrying a minimum of six UG (or TG) single-stranded nucleotide repeats meaning that its target is bound in a sequence dependent manner. The binding affinity increases with the number of UG (TG) repeats. The role of RRM2 is believed to enhance the binding affinity of RRM1 (Lukaky et al, 2013). Lastly, the CTD also known as a low complexity domain (LCD) is glycine-rich, physiologically non-folded and is important for phase separation and
protein-15 protein interaction. The domain is aggregation-prone and believed to have the capability to sequester native protein into a misfolded/aggregated state (Ayala et al, 2008b).
TDP-43 performs its physiological functions in both nuclear and cytoplasmic compartments and therefore localisation is equilibrated between both cellular locations (Figure 3b). This is possible due to the cytoplasmic-nuclear shuttling property of TDP-43. While the nuclear import is mediated through a bipartite nuclear localisation signal (NLS) found in the linker between the NTD and RRM1 (Figure 3a) (Ayala et al, 2008b), the export has been shown to occur through passive diffusion (Ederle et al, 2018). This ability of a dynamic relocalization of TDP-43 enabled the protein to perform functions in both the nucleus and cytoplasm. How the shuttling is regulated and how dysregulation and redistribution of the protein are linked to pathology remain elusive. This is a dynamic equilibrium and disturbances in the mechanism contribute to the pathology (Svahn et al, 2018).
2. Physiological role of TDP-43 in the cell
TDP-43 is an essential protein for the cell. It is involved in a variety of DNA regulatory path-ways, in stress response mechanisms and, most importantly, in RNA homeostasis where it regulates pre-mRNA splicing, mRNA transport, mRNA stability and translation (Ayala et al, 2008a; Dewey et al, 2011; Polymenidou et al, 2011; Tollervey et al, 2011; Buratti & Baralle, 2012; Ederle & Dormann, 2017). TDP-43 was first discovered as a transcriptional repressor of the transactivation response (TAR) element of the human immunodeficiency virus type 1 (HIV-1), giving the protein its name (OU et al, 1995). Even though this study has been rein-vestigated questioning the direct repression of gene transcription, other examples of the role of TDP-43 in transcriptional regulation has been proposed over the years (REF). Fur-thermore, TDP-43 serves as a splicing regulator as shown by genome wide studies identifying many sequence-specific TDP-43 RNA splicing targets in human and mouse brains. A signifi-cant proportion of alternative mRNA isoform plays a key role during neuronal development or have been associated with neurological disorders. Therefore, splicing has become an im-portant topic in the field (Polymenidou et al, 2011; Tollervey et al, 2011). TDP-43 can control its own expression through a negative feedback loop called ‘autoregulation’. The RNA-binding properties of TDP-43 are essential since the protein can bind to the 3’UTR in its own
16 mRNA thereby changing pre-mRNA splicing and promoting mRNA instability. These two mechanisms triggering nonsense-mediated RNA degradation. Thus, cellular levels of TDP-43 are tightly regulated. Disease-associated TDP-43 disrupt autoregulation contributing to the pathogenesis (D’Ambrogio et al, 2009; Ayala et al, 2011; Polymenidou et al, 2011; Eréndira Avendaño-Vázquez et al, 2012)
More, TDP-43 plays a key role in the biogenesis of microRNA (miRNA), small non-coding RNAs of ~ 20 – 22 nucleotides. miRNA synthesis starts from a long primary miRNA (pri-miRNA) transcript containing a small strand dsRNA and a loop. The long transcript is cleaved by the nuclear Drosha complex into pre-miRNA and subsequently transported to the cyto-plasm. A dicer complex is responsible for further cleavage and generating mature miRNA. Lastly, these are incorporated into the RNA-induced silencing complex (RISC) which assist the miRNA towards its target whose expression will be modified. TDP-43 plays a role in both nucleic and cytoplasmic compartment during miRNA biogenesis. Interaction of TDP-43 with the nuclear Drosha complex and binding towards the relevant pri-miRNA facilitates the pro-duction of pre-miRNA. Then, cytoplasmic TDP-43 promotes maturation by associating with the Dicer complex. The involvement of TDP-43 in miRNA biogenesis is shown and suggested to be indispensable in neurons (Kawahara & Mieda-Sato, 2012).
TDP-43 functions related to nucleotide binding is tought to be mediated by the binding of sequence-specfic nucleotides by the RRMs and the LCD which serves as a platform to re-cruite additional hnRNPs which assist in splicing activity, transciptional regulation and miRNA maturation (Buratti & Baralle, 2001; Buratti et al, 2001; D’Ambrogio et al, 2009; Ayala et al, 2011; Bhardwaj et al, 2013).
Additionally, it is shown that TDP-43 is involved in regulation of axon growth and axonal transport, neuronal activity and regulates factors promoting motor neuron survival (Wang et
al, 2008; Fallini et al, 2012; Pelletier et al, 2012; Baskaran et al, 2018; Kuliyev et al, 2018;
Burk & Pasterkamp, 2019). Microtubules are a key component in the cytoskeleton essential for stability, shape, transport and function of the neuron. As an intrinsic property of these structures is the ability to switch between rapid polymerisation and shrinkage. TDP-43 might play a role in regulating this mechanism due to suggested interaction with microtubules and microtubule associated proteins (Oberstadt et al, 2018; Melamed et al, 2019)
17 Lastly, if a cell is exposed to stress such as oxidative stress, cytoplasmic foci are transiently formed called ‘stress granules’ (SGs). These store house-keeping protein encoding mRNA and allow selective translation of stress-response proteins such as heat shock proteins and chaperones to enhance stress survival. SGs contain RNA-binding proteins such as TDP-43 seem in certain cases beneficial for the cells, making TDP-43 an important protein in the stress respond pathway (Colombrita et al, 2009; Bentmann et al, 2013).
3. Pathological features of TDP-43 in ALS and FTLD
Analysis of post mortem brains obtained from patients with ALS and FTLD revealed a loss of nuclear TDP-43 and appearance of misfolded TDP-43 ‘inclusions’ or aggregations, both nuclear (rare) and cytoplasmatic, affecting neurons and glia (Arai et al, 2006; Neumann et al, 2006; Lillo & Hodges, 2009; Ederle & Dormann, 2017). Aggregated TDP-43 has signature post-translational modifications (PTM) such as hyperphosphorylation and polyubiquitination. Additionally, TDP-43 can undergo proteolytic cleavage whereby aggregation prone C-terminal fragment are released. (Laferrière et al, 2019; Wang et al, 2013; Igaz et al, 2008; Neumann et al, 2009). Macroscopically, the TDP-43 containing inclusions show both granular amorphous non-filamentous morphology or filamentous structure representing structural heterogeneity within the ALS-FTLD spectrum (Neumann, 2009; Robinson et al, 2013).
Next to a classification based on the main protein accumulated in the pathological inclu-sions, FTD cases can be classified in subtypes characterized by different clinical and genetic features as well as differences in TDP-43 biochemical density, morphology and neuroana-tomical distribution. The first subtype, FTLD-TDP-A, is characterized by neuronal cytoplasmic inclusions (NCI), dystrophic neurites (DN) and neuronal intranuclear inclusions in the upper cortical layers. This results typically in a behavioural variant frontotemporal dementia (bvFTD) without MND. These are associated with mutations in granulin (GRN) and
chromo-some 9 open reading frame 72 gene (C9ORF72). FTLD-TDP-B as a second subtype, is
de-scribed by primarily granular NCI across all cortical layers. Clinical features are typically a combination of bvFTD and MND. Additionally, association with mutations in C9ORF72 have been described. FTLD-TDP-C shows long and thick DN and a few NCI only present in superfi-cial cortical layers. Typical features for this subtype are semantic dementia without MND. FTLD-TDP-D has been associated with mutations in valosin-containing protein (VCP) which
18 can induce familial inclusion bodies, Paget’s disease as well. Overall, this type has been called either inclusion body myopathy associated with Paget disease of bone and fronto-temporal dementia (IBMPFD-ALS) or multisystem proteinopathy. However, other subtypes are described as well, indicating that the full spectrum of heterogeneity is not been fully understood yet (Neumann et al, 2007; Lee et al, 2017; Mackenzie & Neumann, 2017; Laferrière et al, 2019)
In contrast to FTLD no different features in pathological TDP-43 are shown for the classification of subtypes in ALS (Strong & Rosenfeld, 2003; Picher-Martel et al, 2016; Laferrière et al, 2019).
Several mechanisms have been proposed in the formation and progression of the TDP-43 pathology. A first hypothesis suggested that the highly aggregation-prone cleaved C-terminal fragments can capture full length TDP-43 and form detergent-insoluble aggregates (Igaz et
al, 2008; Neumann et al, 2009). More detailed research showed distinct banding patterns for
C-terminal fragments upon protease treatment suggesting that differences in the TDP-43 molecules can contribute to the different pathologies in and clinical phenotypes in TDP-43 proteinopathies (Tsuji et al, 2012; Li et al, 2015; Kametani et al, 2016). TDP-43 serves as a splicing regulator and binds its own mRNA which can cause alternative splicing of its own mRNA. The second hypothesis suggested that an isoform of TDP-43 mRNA lacking an exon and consequently showing an alternative start codon ATGMet85 generates a 35 kDA
N-terminal truncated molecule. This molecule shows reduced solubility, a preferential cytoplasmic distribution and aggregation together with full length TDP-43 (Xiao et al, 2015). Both hypotheses suggested the retention of C-terminal TDP-43 particles (truncated or fragmented TDP-43) in the cytoplasm after loss of the NLS sequence as a molecular mechanism for aggregation. Additionally, abolishment of oligomerization could introduce aberrant equilibrium between nucleus and cytoplasm (Figure 3) (Afroz et al, 2017).
Other mechanisms based on post translational modifications has been proposed to influence aggregation as well. Hyperphosphorylation and ubiquitination of TDP-43 pathological inclusions influence the aggregation propensity and solubility depending on their position. How the PTMs are involved in disease initiation and progression is not clear yet and need further investigation (Hasegawa et al, 2008; Li et al, 2011).
19 Lastly, oxidative stress induces formation of SGs containing TDP-43 is thought to play a role in the pathogenesis of proteinopathies through different mechanisms or a combination of them: 1) SG may serve as a seed for pathological inclusions. 2) SG mediates translational repression 3) SG traps essential RNA-binding proteins (Colombrita et al, 2009; Bentmann et
al, 2013). Dysregulated or mutant TDP-43 may influence the dynamics of SG formation and
therefore also contribute to the progression of TDP-43 proteinopathies (Dewey et al, 2011; McDonald et al, 2011)
All these events may result in misfolding, loss of normal function and thereby perturbation of RNA and protein homeostasis or gain of function in the toxic aggregates (Lagier-Tourenne
et al, 2010; Ling et al, 2013). However, it remains unclear whether loss of normal function
and or gain of toxicity leads or the combination of both leads to neuronal defects and eventually neurodegeneration.
4. A prion-like mechanism for TDP-43?
Protein aggregation and neuronal dysfunction during the early stages of neurodegeneration stay in a confined area of the nervous system. Yet, in later stages, the pathological alterations becomes more generalized and diffused which may suggest that misfolding protein propagation underlie the progression of neurodegenerative disease (Polymenidou & Cleveland, 2012, 2011). Some proteins are prone to aggregation due to the presence of aggregation-prone stretches of amino acids called prion-like domains, which are usually present in RNA-binding domains (Walker & Jucker, 2015) In contrast, the prion-like domain of TDP-43 is localized in the low complexity domain.
A prion-like mechanism has already been described for amyloid-β peptide in Alzheimer’s disease (Kane et al, 2000; Meyer-luehmann et al, 2006; Aguzzi, 2009; Aguzzi & Rajendran, 2009; Long & Holtzman, 2019) and α-synuclein in Parkinson’s disease (Aguzzi, 2009; Aguzzi & Rajendran, 2009; Li et al, 2008). Currently, there is no in vivo evidence that TDP-43 follow a true prion-like amplification. However, TDP-43 aggregates have shown prion-like behaviour in cells (Nonaka et al, 2013).
20
C. Description of the study
A consequence of the lengthening life-expectancy in developed countries is a rising preva-lence of neurodegenerative diseases. These disorders are a growing source of suffering for the patients as well as their families and friends. Currently, there is no disease modifying treatment available for TDP-43 associated pathologies. Therefore, it is an urgent challenge to understand the molecular mechanisms involved in neurodegenerative diseases, in order to find useful targets for drug development. In vitro studies have identified that pathological TDP-43 from FTLD-TDP and ALS brains can act as seeds to promote formation and propaga-tion of TDP-43 aggregates in culture (Nonaka et al, 2013; Feiler et al, 2015). But currently, only scarce in vivo evidence suggested that pathological TDP-43 follow a similar prion-like amplification (Porta et al, 2018). In Alzheimer’s disease, exploring the cellular mechanism involved in the Tau pathology generated a lot of interest, to understand how the pathology spreads through the brain, the cell-to-cell propagation and trans-synaptic transport of hy-perphosphorylated Tau (Calafate et al, 2016; Wu et al, 2016).
In this study, a prion-like mechanism for TDP-43 was explored using HEK293T and HeLa cell lines and advanced microscopy. Pathological TDP-43 was extracted from post-mortem brain tissue from patients with ALS or FTLD using the SarkoSpin method. This is a new method recently developed in the Polymenidou lab for extremely pure biochemical isolation of pathological TDP-43. They used the SarkoSpin to isolate pathological TDP-43 of different FTLD-TDP patients with the same subtype and pooled these together to characterize the seeding properties. In contrast to this, I used pathological TDP-43 aggregates from one single FTLD-TDP patient for seeding on HA-TDP-43 expressing cells. This approach allows to differ-entiate between individual patients. Seeding properties of the pathological TDP-43 seeds was evaluated over time. Two approaches were used to evaluate the capability of pathogen-ic TDP-43 to trigger more aggregation: immunoblotting after SarkoSpin on the cells and CLSM. Aggregation of endogenous HA-TDP-43 was triggered by seeding and CLSM showed aggregates containing both HA-TDP-43 and pathological hyperphosphorylated TDP-43.
Further, it has been proposed that the SarkoSpin extracted TDP-43 assemblies exhibit dis-tinct cytotoxicity that may be linked to duration of pathology typically for the disdis-tinct sub-types of FTLD-TDP (Laferrière et al, 2019). I assessed the biochemical properties like intrinsic density of the toxic assemblies by density flotation. The FTLD-TDP subtypes show a distinct
21 density profile of TDP-43. Additionally, I evaluated the cytotoxicity of pathogenic TDP-43 on neuronal primary culture by evaluating cell viability and morphology. This showed an in-crease of neuronal cell death overtime for FTLD-TDP subjects and aberrant morphology of neuron and glial cells.
22
Aim
The research in the Polymenidou lab focuses on understanding molecular mechanisms leading to neurodegeneration in FTLD and ALS. My project aimed to explore the cellular mechanisms involved in these TDP-43 associated proteinopathies. In particular, I studied the seeding properties of pathological TDP-43 and I explored the cytotoxic mechanisms induced by pathological TDP-43 on primary neurons and glial cells.
Aim 1: To decipher seeding properties of pathological TDP-43 extracted from brain samples obtained from patients with FTLD pathology.
ALS and FTLD are characterized by TDP-43 containing aggregates. In this first aim, I investigated the hypothesis that pathogenic TDP-43 triggered aggregation of endogenous TDP-43 over time. Therefore, both stable HEK293T and HeLa cell lines expressing HA-tagged TDP-43, described in Laferrière et al (2019) and Tighe et al (2004,2008), respectively were used to test the seeding properties of the pathological TDP-43. Pathological TDP-43, isolated from post-mortem brains tissue from patient with FTLD-A and FTLD-C, were seeded in both HEK and HeLa cells, to test its ability to trigger aggregation of HA-TDP-43. The aggregation of TDP-43 tagged with HA was evaluated after isolation followed by biochemical characterization (immunoblotting as described previously in Laferrière et al (2019)) and microscopy analysis (confocal laser scanning microscopy (CLSM)).
Using two different stable cell lines, differences in seeding properties of the aggregates were shown and allowed us to define universal prion-like mechanism. Additionally, different seeding properties between subtypes of FTLD-TDP (A and C) were evaluated.
Aim 2: Study toxicity of pathological TDP-43 in neuronal cells.
For the second aim, the hypothesis was that pathogenic TDP-43 displays differential biochemical properties linked to distinct subtypes and that these show differential toxicity on neuronal primary cell culture. Therefore, hippocampal and cortical primary neuronal cultures were used to define toxicity mechanisms associated with pathological TDP-43. Therefore, pathological TDP-43 isolated from post-mortem brains tissue from patient with FTLD-A and FTLD-C were seeded on hippocampal and cortical primary neuron culture. I evaluated viability using the LIVE/DEAD Viability/cytotoxicity assay to assess the toxicity of each TDP-43 species on neuron survival. I performed morphological characterization of
23 neurites using CLSM in live imaging to investigate the precise mechanisms involved in the neurodegeneration associated with TDP-43 pathologies. Here, we evaluated differences in toxicity between FTLD-A and FTLD-C subjects’ derived aggregates.
This project will expose primary data of the early cellular and molecular mechanisms involved in TDP-43 associated pathologies. Other studies have already shown that iPSCderived neurons can integrate TDP-43 aggregates and propagate them to other cells.
24
Results
A. SarkoSpin is a tool to isolate pathological TDP-43 from the brain
The Polymenidou lab developed a method, SarkoSpin (SKS), for biochemical isolation of pathological TDP-43 and to understand the molecular basis of the heterogeneity of FTLD and ALS (Laferrière et al, 2019). SarkoSpin is a method that combines a harsh sarkosyl-based solubilization together with a nuclease treatment. Pathological TDP-43 are resistant to detergents such as sarkosyl, and after centrifugation of the lysate, end up in the pellet. These isolated pathological TDP-43 are called ‘seeds’. In this project, SarkoSpin (Figure 4a) was performed on post mortem cortical brain samples obtained from patients diagnosed with FTLD. Cortical brain samples from age-matched individuals without diagnosed neurodegenerative disease were used as control patients.
I first assessed the amount of pathological TDP-43 obtained by SarkoSpin purification. I used purified full length recombinant TDP-43 protein (with MBP-tag bound, 88kDa) (figure 4b), to quantify the amount of pathological TDP-43 present in a single SarkoSpin pellet. A SDS-PAGE comparing known amount of full length TDP-43 with SarkoSpin pellet fraction was performed and immunoblotted against TDP-43 (Figure 4c) and pTDP-43 (Figure 4d). Semi-quantification, based on the standard curve of recombinant TDP-43 (Figure 4e) showed that control, FTLD-TDP-A and FTLD-TDP-C patients contained respectively 1.7 µg, 2.5 µg, and 2.7 µg of TDP-43 and 0.01 µg, 0.18 µg, 0.24 µg of phosphorylated TDP43 (pTDP-43) (Figure 4f). The phosphorylated TDP-43 (pTDP-43) represents 1%, 7%, 9% of the SarkoSpin seed in control, FTLD-TDP-A and FTLD-TDP-C respectively (Figure 4g). This technique is a first approach to normalize the amount of pathological TDP-43 use for seeding. During the purification process, slightly different regions of the brain and different patients can express various amounts of pathology. Thus, this type of quantification is advised to be performed after every single SarkoSpin purification.
25
B. HEK293T and HeLa cells expressing hemagglutinin-tagged TDP-43 serves as model to investigate seeding properties of pathological TDP-43 seeds
To determine and compare seeding properties of FTLD-A and FTLD-C, specific TDP-43 overexpressing HEK293T and HeLa stable cell lines were used. These cell lines were generated using the Flp-In T-rex technology (Invitrogen) as described in Laferrière et al (Laferrière et al, 2019), and characterized by doxycycline (dox) inducible expression of C-terminally tagged TDP-43 protein. This resulted in a homogenous population of cells expressing traceable TDP-43. The nine-residue hemagglutinin (HA)-tag (YPYDVPDYA) was
placed at the C-terminal domain to minimize any interference with protein structure, which might hinder protein seeding. The advantage of these cell lines is the induced expression of HA tagged TDP-43 by low dox concentration with no apparent aggregation in the cell. This is advantageous when compared to other overexpressing model, as aggregation of TDP-43 is concentration dependent.
Pathological TDP-43 aggregates isolated from FTLD-TDP-A and FTLD-TDP-C samples, from a unique patient for each condition, were seeded2 on mitotically arrested cells. After 3, 4, 5
and 6 days, cell samples were harvested for biochemical analysis and fixed for microscopic analysis to evaluate the seeding properties of the pathological TDP-43 seeds (Figure 5a). After cell lysis, BCA protein quantification was performed in order to normalize the initial protein concentration used for SarkoSpin analysis. The SarkoSpin protocol was performed on the HEK293T cells of the 4 timepoints. The different obtained fractions (total, supernatant & pellet for D3, D4, D5 & D6) were analyzed by SDS-PAGE and subsequently, immunoblotted against HA-tag (Figure 5b). The total and supernatant fraction showed HA-TDP-43 which was expressed by the cell (Figure 5b, total, supernatant). We observed an increase of TDP-43-HA present in the SarkoSpin fraction after incubation with FTLD-TDP-A and C aggregates. Furthermore, the amount of TDP-43-HA appeared to increase over time (Figure 5b, pellet). These results suggested that the original pathological seeds successfully triggered aggregation in HEK293T cells.
The cells were mitotically arrested to avoid that the proliferation and high metabolism of HEK293T cells would mask the effect of exogenous aggregates. HEK293T cells were arrested
26 with cytosine arabinoside (AraC) that is incorporated into the DNA during DNA synthesis and arrest the cell cycle in the S phase. However, AraC is an anti-metabolic reagent that shows a certain toxicity, leading to cellular stress and cellular death 6 days after mitotic arrest, limiting the timeframe for the experiment. Another agent to arrest mitotic cell cycle is palbociclib, a selective inhibitor of the cyclin-dependent kinases CDK4 and CDK6 which arrest the cells in the G1 phase. However, after testing it on HEK293T cells, palbociclib showed no effect on cell proliferation with high toxicity at high dose. Palbociclib was previously used on HeLa cells. Therefore, inducible HA-TDP-43 expressing HeLa cells were used to explore seeding of TDP-43 after mitotic arrest induced by palbociclib. This treatment is less toxic than with AraC which may be advantageous to push the 6-days limit. The goal was to explore the universality of TDP-43 seeding properties, using multiple cell lines and models.
Therefore, biochemical analysis was performed using HA-TDP-43 expressing HeLa cell line. The different fractions ((total, supernatant & pellet at D3, D4, D5 & D6) were analyzed as previously described for HEK293T cells (Figure 5b). Unfortunately, no clear HA-TDP-43 expression was observed in the total and supernatant fractions (Figure 7, total, supernatant). Further, the SarkoSpin fraction did not show enrichment of HA-TDP-43, as previously observed for HEK293T model (Figure 5b). These results did not show expected outcome. We hypothesized that 1) the cell quality of the HA-TDP-43 expressing HeLa cells was insufficient for a successful seeding; 2) the original SarkoSpin isolation might have produced a low amount of pathological seeds and/or bad quality of seeds to induce neo-aggregation; 3) The variability of pathology among different patients can lead to different seeding properties; 4) HeLa stable cell lines have a lower expression of TDP-43-HA than HEK293T and are less stable overtime, leading to the loss of TDP-43-HA after few passages. More tests will be required to explore the seeding properties of TDP-43 seeds on this cell line.
To go further, I coupled the biochemical analysis with microscopic analysis of the seeded cells using HEK293T and HeLa cells. The cells were fixed 3, 4, 5 and 6 days after seeding, to visualize the seeding properties of the pathological TDP-43 by CLSM (Figure 6, a-c). Pathological TDP-43 were identified using pTDP-43 antibody, which can also detect the original seeds incubated with the cells. Neo-aggregates formed by the cells were identify using HA staining. Colocalization of both HA and pTDP-43 was consider as a pathological neo-aggregate. HEK293T cells seeded with FTLD-TDP-A seeds showed a higher increase in
pTDP-27 43 over time when compared to FTLD-TDP-C (Figure 6d). Both FTLD subtypes showed higher level of pathological TDP-43 compared to control. An increase in the level of pTDP-43 was also observed in the control condition, confirming the toxicity effect of the AraC treatment. Nevertheless, the phosphorylation shown in the FTLD-TDP patient conditions suggested that patient derived pathological TDP-43 may trigger more aggregation. More biological replicates using different patients will be necessary to draw statistically supported conclusions. These results are in line with the biochemical results previously described (Figure 5).
Surprisingly, seeding on HeLa cells showed a high amount of pathological pTDP-43 at day 3 (Figure 7a-d). Despite the unexpected results for day 3, HeLa cells seeded with FTLD-TDP-C pathological aggregates showed a trend of a higher increase in pTDP-43 over time when compared to FTLD-TDP-A (Figure 7d). Both FTLD subtypes showed higher level of pathological TDP-43 compared to control. These results show proof of principle that seeding of pathological TDP-43 can be universal. Repeating both biochemical and microscopic analysis of seeding of patient derived pathological TDP-43 aggregates to generate more conclusive results in future.
C. Different FTLD-TDP subtypes show distinct density profile of pathological TDP-43
Intrinsic density of TDP-43 aggregates was characterized for both A and FTLD-TDP-C and revealed differences in the biochemical density of the pathological aggregates between both subtypes (Laferrière et al, 2019). Brain cortex homogenates were processed for SarkoSpin solubilization followed by a 17h ultracentrifugation to reach isopycnic equilibrium (Figure 10a). Next, gradients were fractionated into 16 equal fractions and immunoblotted against TDP-43 to determine the density profile (Figure 8b).
The first 6-7 lanes in the TDP-43 profiles (Figure 8b-d) are empty showing that fractions with highest buoyancy do not contain TDP-43. Further, the TDP-43 profile for non-neurodegenerative patients (Figure 8b) show presence of endogenous TDP-43 in physiological conditions in fractions 8 - 13. In FTLD-A, fractions 9 to 14 contained TDP-43 whereas in FTLD-TDP-C, fractions 9 to 13. As previously published, pathological TDP-TDP-43 is
28 characterized by C-terminal cleavage products of 25 and 35kDa and ubiquitination which are present in the immunoblots of both FTLD-TDP subtypes and absent in the immunoblot of the control sample (Figure 8b-d) (Laferrière et al, 2019). The next phase is to immunoblot against pTDP-43 to investigate the amount of pathological TDP-43 in each fraction. Since hyperphosphorylation is a hallmark of TDP-43 inclusions, this approach would reveal different features between distinct subtypes of FTLD-TDP (Laferrière et al, 2019) (Due to technical issues, these immunoblots were not generated.) We expect the immunoblots to show absence of pTDP-43 in control samples while in FTLD-TDP samples pTDP-43 would be present. Even more, the density profile of FTLD-FTD-A and C would be unique for each. Using this technique, it will be possible to explore whether biochemical properties of subtypes of FTLD can be reproduced in the cells after seeding. This will be a first step in validating the prion-like property of TDP-43 as well as the use of cell line to amplify pathological material from post-mortem tissue. This technique will reduce the needs for autopsy material and would allow for normalization of the aggregates for further experiments.
D. Primary neuron culture as a second cellular model to asses FTLD-TDP subtype toxicity
Hippocampal and cortical primary neuron were used as a more disease-relevant system to study the toxicity of pathological TDP-43.
First, toxicity of SarkoSpin isolated pathological aggregates was investigated on hippocampal primary culture. Pathological seeds from FTLD-TDP-A and FTLD-TDP-C patients were seeded on maturated 15DIV (days in vitro) primary neurons. The effect of exogenous SarkoSpin extracts on neuron viability was then monitored using live imaging with fluorescence microscopy. Ethidium homodimer-1 and Hoechst were imaged at 0h, 33h and 57h (Figure 9a). Before incubation with the seeds, a low ratio of dead cells was recorded (Figure 9b). I then analyzed the evolution of the cellular population by quantifying the amount of dead cells per field (331.5µmx331.5µm). In the first hours after seeding, all conditions showed an increase of cellular death, suggesting a toxic effect of the neuroporter alone. However, at timepoint 33h, the amount of cell death was significantly increased in the FTLD-TDP-A condition (2 way ANOVA, F (6, 137) = 4.092, p = 0.0008, Post hoc uncorrected Fisher’s LSD test: control vs FTLD-TDP-A, p = 0.0016; neuroporter vs FTLD-TDP-A, p = 0.02)
29 After 57h, the amount of dead cells was significantly higher in FTLD-TDP seeded cells compared to others (2 way ANOVA, F (6, 137) = 4.092, p = 0.0008, Post hoc uncorrected Fisher’s LSD test: control vs FTLD-TDP-A, p = 0.0314; neuroporter vs FTLD-TDP-A, p < 0.0001, neuroporter vs FTLD-TDP-C, p < 0.0001). Furthermore, higher toxicity in the control condition compared to the neuroporter condition was recorded (neuroporter vs control, p = 0.018), suggesting a toxicity of the SarkoSpin material, independently from the load of TDP-43.
These results suggested an intrinsic cytotoxic effect of the TDP-43 seeds. Seeds isolated from FTLD-TDP-A showed faster and stronger cytotoxic potential compared to seeds isolated from FTLD-TDP-C. Analysis of earlier timepoints would be interesting to explore how fast the cell viability is influenced directly after adding the toxic seeds. Later timepoints could potentially show how the cells further reacted on the pathological seeds. However, limitations of primary neuron cell culture must be taken into account. Primary cell cultures are fragile and are not a proliferating model the numbers of cells are limited and cell quality is extremely variable. However, the use of primary cell culture is desirable since they are not tumor-derived and more likely to recapitulate the properties of neuronal cells in vivo. The use of mature neurons is also important in the context of neurodegenerative disease, as these diseases do not usually affect the development and maturation of the cells. Further analysis and optimization of the assay and increased expertise will lead to more conclusive results.
To explore more specifically the effect of the seeds on neuronal survival, live neurons expressing EGFP were live imaged using Spinning Disk microscopy. Neurons were imaged at T=0 (before incubation with seeds) and for multiple timepoints after incubation up to 42h (Figure 10). EGFP was imaged from T=0. In addition, starting at T=18, Syr-Tubulin was added to allow imaging of all the cells present in the field. Syr-Tubulin was not added at earlier timepoint to avoid artificial stabilization of the cytoskeleton. Both EGFP and Syr-Tubulin showed significant loss of expression by T=42 compare to T=0, suggesting that the technique itself could induce cellular toxicity. However, our results suggested a major neuronal toxicity for FTLD-TDP-A (figure 10b) and C (figure 10c) seeds. We observed neuronal fragmentation and neurites retraction after only 18h. This effect was observed in control condition too, but with a delay of 6-8h. I also notice that FTLD-TDP-A seeds were showing major cellular death (neurons and glia), whereas FTLD-TDP-C seems to target specifically neurons with less
30 noticeable glial toxicity in the time frame imaged. To go deeper, I performed a quantification of the area occupied by the neurites using Syr-Tubulin staining and the area of the neurons (using EGFP mask) overtime (figure 10d). My results showed that, indeed FTLD-TDP-C showed significant neuronal toxicity (2 way ANOVA, post-hoc uncorrected Fisher’s LSD, control vs FTLD-TDP-C, p<0.05 for t24, 26, 28, 30, 32 and 34). Surprisingy, FTLD-TDP-A did not show significant toxicity compare to the control group. Despite obvious toxicity of the imaging technique, these results are in line with my previous observation using live-dead assay and with published results from the polymenidou group (Laferriere et al, 2019). Altogether, these results showed different behaviour of FTLD-TDP seeds depending on their origin. More experiments will be needed to decipher these mechanisms, but my preliminary results are encouraging and show potential in the techniques used.
31
Discussion
In my project, I isolated pathological TDP-43 from patients who were diagnosed with FTLD-TDP-A or C using the SarkoSpin method. The Sarkospin fraction containing the pathological TDP-43 seeds was composed by 7% and 9% of TDP-43 for FTLD-TDP-A and C, respectively. The pathological features as described before were all present in the pellet. These pathological fraction containing seeds were subsequently used to test my hypotheses.
The first hypothesis in this project was that pathological TDP-43 could act as a universal seed to promote formation and propagation of TDP-43 aggregates cell lines. Every seed was isolated from one single patient in contrast to what previously used in Laferrière et al (Laferrière et al, 2019). It was shown that original pathological TDP-43 seeds were potent to trigger aggregation of endogenous HA-TDP-43 in both HEK293T and HeLa cells (Figure 6-9). The advantages of the HEK293T and HeLa cell lines are that they are well characterized and mimic endogenous TDP-43 expressing which is traceable by presence of the small HA-tag. Further, these cell lines are easily handable and the seeding experiment can be performed in a short time frame (2 weeks). It is the first time that endogenous aggregates are directly visualized as a result of seeding. These results are in line with previous studies showing that the sarkosyl-insoluble TDP-43 act as seeds triggering aggregation in a self-templating manner (Nonaka et al, 2013, feiler et al, 2015). Our results therefore support the hypothesis that TDP-43 associated proteinopathies could have similar mechanisms to prion disease. However, the exact spreading mechanism between the cells remains to be discovered. Both microscopic and biochemical analysis of seeding in HEK293T show a similar pattern which may suggest FTLD-TDP-A’s potency to trigger aggregation faster than FTLD-TDP-C, however more biological replicates using different patients should be performed to provide statistical evidence. In HeLa cells, biochemical analysis did not show enrichment of endogenous aggregation over time however microscopic analysis showed presence of pathological aggregates over time. To provide further evidence, seeding optimization could be obtained by increasing the quality of the HeLa cells and increase quality of the SarkoSpin pellet. Additionally, by exploring more single patients, variation between patients of the same FTLD-TDP subgroups may be exposed.
32 In the second part of the project I aimed to show distinct biochemical properties of pathological TDP-43 and their differential toxic effect by density floatation and seeding on neuronal primary cultures. Evidence for the second hypothesis was obtained by combining the SarkoSpin with biochemical characterization and imaging methodologies for neuronal primary culture.
Pathological TDP-43 was separated based on its intrinsic density. TDP-43 was shown in a defined range of fractions in which FTLD-TDP subtypes (A and C) showed the typical expected pattern for the pathological features of the aggregates (Figure 10). Previous data collection in the lab showed distinct pTDP-43 profiles for FTLD-TDP-A and C (Laferrière et al, 2019). Together, these results suggest that there is a link between the biophysical properties of pathological TDP-43 and the clinicopathological subtypes of FTLD-TDP (Laferrière et al, 2019). In future project, the density profiles for cell-derived TDP-43 from HEK293T and HeLa cells can be determined and compared to TDP-43 of the original patient. By this quality control, we can deduce whether the newly cellular generated aggregates show biophysical and clinic-pathological similarities to the patient derived aggregates. If this is the case, the amount of patient tissue used for experiments decreases and variability between experiments can be reduced by normalization of the aggregates.
To observe the potential seeding properties of cell-derived TDP-43, a re-seeding experiment can be performed. For this, an additional cell line should be created with TDP-43 coupled to a distinct tag like α-bungarotoxin binding site (BBS) in order to distinguish between the origins of TDP-43 in the obtained aggregates.
To examine the differential toxicities of TDP-43 from FTLD-TDP subtypes, hippocampal and cortical primary cultures were seeded with patient material. Primary neuronal culture is a more representative model to study mechanisms in neurodegeneration since they provide the proper cellular environment when compared to HEK293T and HeLa cell lines. Cell lines often differ genetically and phenotypically from their origin tissue whereas primary cells maintain many of the important markers and functions seen in vivo. The viability of hippocampal primary culture was monitored by live imaging (epifluorescence) and showed an increase of cell death due to toxicity of patient derived seeds over time (Figure 9). This suggest an intrinsic cytotoxic effect. Additionally, the morphology of seeded cortical derived neuronal cells was observed by CLSM. Neuronal fragmentation and retraction was detected
33 in the condition FTLD-TDP-A and C at 18h. The same effect was observed in control condition as well, however with a delay of 6 – 8h. Additionally, different cell types were specifically targeted. In FTLD-TDP-A conditions, both neurons and glial cells showed major cellular death whereas FTLD-TDP-C seemed to specifically target neurons with less glial toxicity. To understand the exact effect leading to cellular toxicity, a series of experiments should be performed looking at neurodegenerative markers like neurofilaments (NF), synaptic activity marker like post-synaptic density protein (PSD-95), neuronal and glial specific morphological markers like phosphoneurofilaments (PNF) and glial fibrillary acidic protein (GFAP), respectively by fixed and live imaging CLSM. Additionally, the universal property of the pathological seeds can be tested in a neuronal primary model as well.
However the use of primary neuronal culture shows some disadvantages. Culture are only relevant after maturation which takes 14 days, need more nutrients to mature and after one month you reach maximal culture time. Obtaining the cells from the P0 pups takes some experience since these cells as not as robust as HEK293T and HeLa cell line. An additional challenge was the use of P0 pups because the obtained neuronal cultures are more variable. Therefore, the use of late embryonic stage (e.g. E17) mice is recommended since those cells have better cell viability. Primary neurons from rat could be a better approach, as they tend to be more resilience and more relevant morphology (higher number of synapses). Another approach would be to use human neurons as a model to test seeding properties and toxicity of SarkoSpin isolates. Human neurons would be a perfect model to test the long term effects since FTLD and ALS are human disease and these neurons can survive in vitro for considerable time longer than primary culture. However, Human neuronal model takes more time to differentiate and becoming mature, which introduce more time constrain on the study.
Additional technical aspects need to be considered when performing live imaging. When using epifluorescence, the whole surface is scanned leading to exposure of all cells to the light source. In contrast to this, CLSM was used to select a specific location exposing less cells but with higher intensity of the laser. The later may introduce more damage to the targeted cells, however a higher resolution can be obtained. This advantage was used for the morphology experiment. A major drawback was that we were dependent on the availability of the microscopes in the ZMB facility of UZH. Between timepoints, the plate must be
34 removed causing changes in conditions (temperature, CO2 level, vibrations during transport). In the epifluorescence microscopy pictures, it was not possible to image the same cells between the different timepoints due to shift of the plate, but in the CLSM the software provided a tool to remember specific positions.
Despite these limitations, these results are in line with previous results by the Polymenidou lab (Laferrière et al, 2019). Additional literature shows that FTLD-TDP subtypes are linked to differential life duration (Lee et al, 2017, Laferrière et al, 2019). FTLD-TDP-A shows a more aggressive clinical picture leading to shorter disease duration when compared to the more chronic FTLD-TDP-C (Lee et al, 2017). All the obtained results together are in line with disease pattern of the different subtypes. It suggests that FTLD-TDP-A is linked with higher seeding potential shown by the increase of aggregation in cells over time, an earlier onset of toxicity and more morphological changes over time, compared to both FTLD-TDP-C and control conditions.
This study contributes to the current knowledge of pathological mechanisms of TDP-43 associated pathologies. The cellular model show clear incorporation of endogenous HA-TDP-43 never shown before in a non-overexpressing system enabling more physiological studies. More troubleshooting are necessary for the neuronal model, but my results showed encouraging outcome to continue exploring subtype differences.
In the future, these models will enable to study pathobiology of TDP-43 potentially enabling us to understand neurodegeneration and potentially develop new therapeutics to cure these fatal diseases.
35
Material and methods
Post-mortem brain samples: Brains were obtained from the University College of London (UCL). FTLD-TDP-A and C patients were formalized identified after autopsy. Frozen blocks of temporal lobes were used in this study. All transfer of material were covered by MTAs signed between the University of Zurich and UCL, and patient identity was anonymized by UCL before shipment.
Primary Antibodies:
Table 1: List of primary antibodies used for western blot and immunofluorescence.
Antibody (clone) Type Source Western
Blot
Immuno-fluorescence 6H6 (TDP-43) Mouse mAb Proteintech
60019-2-Ig
1:1000 1:1000
HA-tag Rabbit mAb Cell Signaling C29F4-3724S
1:1000 1:1000
HA-tag Rabbit pAb Proteintech 51064-2-AP
1:1000 1:1000
HA-tag biotinylated
Rabbit mAb Cell Signaling C29F4-5017S
1:1000 1:1000
pTDP-43 - - 1:1000 1:1000
β-tubulin Rabbit pAb Proteintech 10068-1AP
1:1000 1:1000
Secondary Antibodies:
Table 2: List of secundary antibodies used for western blot and immunofluorescence.
Antibody (clone) Type Source Western
Blot
Immuno-fluorescence Alexa Fluor 568
anti-mouse IgG
Donkey pAb Thermo Fisher Scientific A31573
1:1000 1:1000
Alexa Fluor 568 anti-rabbit IgG
Donkey pAb Thermo Fisher Scientific A10037
1:1000 1:1000
Alexa Fluor 647 anti-mouse IgG
Donkey pAb Thermo Fisher Scientific A10042
1:1000 1:1000
Alexa Fluor 647 anti-rabbit IgG
Donkey pAb Thermo Fisher Scientific A31571
1:1000 1:1000
Alexa Fluor 647 anti-human IgG
Goat pAb Thermo Fisher Scientific A21445
1:1000 1:1000
Alexa Fluor 488 Streptavidin Thermo Fisher Scientific S32354
