ALS-associated missense and nonsense TBK1 mutations can both cause loss of kinase function

Mutations in TBK1 have been linked to amyotrophic lateral sclerosis (ALS). Some TBK1 variants are nonsense and are predicted to cause disease through haploinsufficiency, however many other mutations are missense with unknown functional effect. We exome sequenced 699 familial ALS patients and identified 16 TBK1 novel or extremely rare protein changing variants. We characterised a subset of these: p.G217R, p.R357X and p.C471Y. Here we show that the p.R357X and p.G217R both abolish the ability of TBK1 to phosphorylate two of its kinase targets, IRF3 and OPTN and to undergo phosphorylation. They both inhibit binding to OPTN and the p.G217R, within the TBK1 kinase domain, reduces homodimerisation, essential for TBK1 activation and function. Lastly, we show that the proportion TBK1 that is active (phosphorylated) is reduced in five lymphoblastoid cell lines derived from patients harbouring heterozygous missense or in-frame deletion TBK1 mutations. We conclude that missense mutations in functional domains of TBK1 impair the binding and phosphorylation of its normal targets, implicating a common loss of function mechanism, analogous to truncation mutations. Lymphoblastoid cell lines (LCLs) derived from FALS patients and healthy controls were obtained from the European Collection of Authenticated Cell Cultures (ECACC). LCLs were grown in RPMI media (Gibco, Life Technologies) complemented with 10% FBS (Fetal Bovine Serum, Life Technolgies), 5% PenStrep (penicillin 100 U/ml and streptomicin 100 U/ml, Life Technologies) and 5% L-Glutamine (Life Technologies). These cells grow in suspension and were, therefore, kept in upright T25 flasks (Nunc, Life Technologies) in a water-jacketed 5% CO 2 incubator. We identified 16 potentially deleterious protein-changing variants in TBK1 , which were novel, or had an ExAC Non-Finnish European (NFE) carrier frequency of <1:20,000 individuals. A similar filtering strategy applied to the NFE subset of ExAC identified 54 variants from a total of 33,075 individuals, revealing a significant overabundance of protein-changing TBK1 variants in our familiar cohort (p=1.02e -10 , Fisher’s two-tailed test). Thirteen of these variants were absent from the following databases: 1000 genomes, UK10K, Exome Variant Server (EVS), and ExAC databases (n > 72,000). Two variants (p.R357Q, p.C471Y) were found once and one variant (p.R358H) was found seven times. Out of the 16 variants identified: four were nonsense, three in-frame deletion and nine were missense variants (Supplementary table 1). The p.G217R variant is present in two Dutch cases predicted to be first degree relatives (King kinship coefficient=0.314, vcftools Ajk=0.451), and is their only shared novel variant in a gene or pathway previously linked to ALS. The variant p.R357X is also present in a single FALS case in the ALS data browser (ALSdb) however, this is unlikely to be closely related to the p.R357X carrier identified in this study as they lack any other shared rare variants. Amongst the previously excluded FALS samples, another novel missense variant (p.Y394D) was identified in a patient who harbours the known pathogenic TARDP mutation, p.M337V, which segregated in their affected sibling. However, the available exome data did not have sufficient coverage to determine if the sibling also shared the TBK1 variant and DNA was not available to determine segregation by Sanger sequencing. Furthermore, the TBK1 variant p.R358H was present in both 1 st -degree relatives of a kindred, but both were also carriers for the particularly aggressive FUS p.R521C


M A N U S C R I P T
A C C E P T E D ACCEPTED MANUSCRIPT

Introduction
Amyotrophic Lateral Sclerosis (ALS) is an adult onset and progressive neurodegenerative disorder that targets the upper and lower motor neurons in the brain and spinal cord. Death usually occurs within three to five years from the symptom onset and treatment is largely palliative (Morgan and Orrell, 2016). ALS is often associated with cognitive changes linked to mild frontotemporal dementia (FTD) (Gijselinck et al., 2015) and up to 50% of the FTD cases develop signs of motor neuron disease (MND) (van der Zee et al., 2017).
Approximately 10% of ALS cases have a familial history of ALS or FTD (fALS, fALS/FTD) (Tiwari et al., 2005). To date, more than 40 genes have been identified to be associated with ALS through linkage studies, Genome Wide Association Studies (GWAS), Whole Exome Sequencing (WES) and Whole Genome Sequencing (WGS). Four genes account for over 50% of fALS cases: SOD1, C9ORF72, TARDBP and FUS/TLS in population of European ancestry and most other genes are rare, each accounting for ~1% of the cases (Taylor et al., 2016).
Tank Binding Kinase 1 (TBK1, NAK, T2K) codes for a protein kinase involved in many pathways including the immune response and autophagy (Weidberg and Elazar, 2011). TBK1 is composed by four domains: a kinase domain (KD), responsible for its kinetic activity, an ubiquitin-like domain (ULD), a scaffold dimerization domain (SDD) and a C-terminal domain (CTD), involved in TBK1 association with binding partners such as optineurin (OPTN), an important autophagy receptor (Tu et al., 2013) (Figure 1). TBK1 has been shown to homo-dimerise through a central axis formed by the two SDD domains interacting with each other (Figure 1c). This structure is stabilised by the ULD and the KD that interact with each other and with the SDD axis, forming a globular head that stabilises the whole structure (Tu et al., 2013). The interactions between the ULD, the SDD and their linker region are highly hydrophobic and prevent the homodimer from being dissociated when carrying out its functions. On the other hand, interactions of the KD within this structure are mainly polar (Tu et al., 2013). TBK1 activation has been demonstrated to be a multistep process that begins with the Lys-63-linked polyubiquitination, which is required for Ser172 phosphorylation within the activation loop. This causes a critical change in protein conformation promoting the active position of the C-helix in the SDD domain and facilitating the final step of homo-dimerisation, essential for mature kinase activity (Ma et al., 2012;Tu et al., 2013).
Mutations in TBK1 have been recently linked with ALS and FTD by two WES/WGS independent studies (Cirulli et al., 2015;Freischmidt et al., 2015). Many ALS-linked TBK1 mutations generate premature stop M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT codons, leading to nonsense-mediated mRNA decay and haploinsufficiency that is predicted to impair autophagy (Freischmidt et al., 2016). However, the pathogenicity and mechanism of missense mutations is unclear (Freischmidt et al., 2016). Here we describe 16 novel or extremely rare, potentially deleterious variants in TBK1, and demonstrate that missense mutations can lead to a loss of TBK1 kinase activity by either disrupting homodimer formation, phosphorylation of itself and its targets Optineurin (OPTN) and interferon regulatory factor 3 (IRF3), implicating a loss of function pathogenic mechanism.

Patients and DNA samples
All patients and controls individuals gave full patient consent for research purpose. DNA was extracted from 932 patient samples primarily of European ancestry of which 757 were index cases and 175 were affected relatives. All patients had a diagnosis of ALS following revised El Escorial criteria (Brooks et al., 2000) with at least one family member affected by ALS and/or FTD. Any sample positive for mutations in known ALS genes (eg. SOD1, C9orf72, TARDBP, FUS, PFN1, UBQLN2, OPTN, VCP, and ANG) were excluded from further analysis, resulting in a final cohort of 699 probands. Exome sequence data for 102 FALS cases in this cohort were obtained, with permission, from the dbGAP (database of Genotypes and Phenotypes) repository (National Institutes of Health (NIH) Exome Sequencing of FALS, National Institute of Neurological Disorders and Stroke (NINDS), phs000101. v4.p1, Traynor).

Exome sequencing and variant analysis
Exomes were captured from the UK samples using the Roche-Nimblegen SeqCap EZ Exome probe library and sequenced on an Illumina HiSeq 2000 producing 100 bp paired-end (PE) reads. All other exomes were provided as FASTQ files, captured with a variety of probe sets and sequenced to produce either 50, 75, or 100 bp Illumina PE reads. Novocraft NovoAlign was used to align the FASTQ files to the hg19 human reference, variants were called with SAMtools v1.1mpileup then normalised with bcftools v1.1 norm.
Individual variant call files (VCF) were filtered by the following criteria: DP≥10, QUAL>20, GQ≥50, and MQ≥50 then merged to a single cohort VCF. Common ancestry between samples was taken from existing familial annotation where available and also deduced from inheritance by descent (IBD) analysis in vcftools (Yang et al., 2011) and King (Manichaikul et al., 2010), using only variant positions covered to a depth >10 in >85% of FALS cases, and recoding all missing data to a heterozygous reference genotype (0/0). Functional annotation, pathogenicity predictions, AdaBoost (ADA) & Random Forest (RF) splicing predictions (Jian et al., 2014) and matches to 1000 genomes were added with table_annovar.pl (Wang et al., M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT 2010), whereas all other annotations including variant frequencies in Exome Sequencing Project (ESP, http://evs.gs.washington.edu/EVS), Exome Aggregation Consortium (ExAC, http://exac.broadinstitute.org) and UK10K (www.uk10k.org), were added via custom perl scripts. Variants were removed if they had a carrier frequency of greater than 1 in 20,000 in the Non-Finnish European (NFE) subset of ExAC (MAF>0.0025%) or were predicted benign by at least 15 of the 20 pathogenicity prediction algorithms.
Synonymous and intronic variants were assessed by NetGene2 and GeneSplicer and excluded if no changes in scores were observed compared to the reference allele at locations matching to known Refseq acceptor or donor splice sites. 5' and 3' UTR variants were excluded from consideration in this analysis. To assess the relative abundance of TBK1 variants in our cohort compared to ExAC, a burden test (Fisher's Exact, two tailed) was performed between the number of FALS and ExAC NFE variants remaining after annotation and filtering by the above criteria.

Plasmid and Cloning
HA-tagged TBK1 wild type (WT) and FLAG-tagged OPTN pCMV3 expression vectors were purchased from Creative Biogene Biotechnology. Single amino acid changes (p.G217R, p.R357X, p.C471Y) were introduced in HA-tagged TBK1 WT by site direct mutagenesis using Q5® Site-Directed Mutagenesis Kit according to manufacturer's protocol (New England Biolabs). All constructs were verified by Sanger sequencing.

Antibodies
Mouse and rabbit HA-tag monoclonal antibodies were used at a dilution of 1/1000 for western blot and 1/500 for immunocytochemistry (ICC) (cat no. 2367 and 3724, Cell Signaling Technology). Rabbit anti-TBK1 monoclonal antibody was used at a dilution of 1/1000 (cat no ab40676, Abcam). Rabbit antiphospho-TBK1 (S172) monoclonal antibody was used at a dilution of 1/500 for western blot and 1/50 for ICC (cat no. 5483, Cell Signaling Technology). Rabbit anti-Interferon regulatory factor 3 (IRF3) polyclonal antibody was used at a dilution of 1/200 (cat no. A022993, Bioassay Technology Laboratory).

M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT
Lymphoblastoid cell lines (LCLs) derived from FALS patients and healthy controls were obtained from the European Collection of Authenticated Cell Cultures (ECACC). LCLs were grown in RPMI media (Gibco, Life Technologies) complemented with 10% FBS (Fetal Bovine Serum, Life Technolgies), 5% PenStrep (penicillin 100 U/ml and streptomicin 100 U/ml, Life Technologies) and 5% L-Glutamine (Life Technologies). These cells grow in suspension and were, therefore, kept in upright T25 flasks (Nunc, Life Technologies) in a water-jacketed 5% CO 2 incubator.

RNA extraction and RT-PCR
Total RNA was extracted using RNeasy Mini Kit (Qiagen) according to manufacturer's protocol. The extracted RNA was used as a template for the synthesis of complementary DNA (cDNA) through reverse transcription, using SuperScript® III Reverse Transcriptase (Life Technologies) following manufacturer's protocol. Oligo dT were used to synthesise cDNA. cDNA was amplified using the PCR primers ATGCAGAGCACTTCTAATCATCTGTGGC and CTAAAGACAGTCAACGTTGCGAAG and Sanger sequenced using the sequencing primers TTGAAGGGCCTCGTAGGAAT and TCAGCCATCGTATCCCCTTT.

Immunocytochemistry (ICC)
48 hours after transfection cells were fixed with 4% PFA at room temperature for fifteen minutes, permeabilised with 0.2% Triton-X-100 for 30 minutes and blocked with 5% Goat Serum (Sigma) for one hour at room temperature. Samples were incubated with primary antibody (anti-HA tag 1/500, anti-pTBK1 were incubated with primary antibody only or secondary antibody only (data not shown). DAPI (4', 6diamidino-2-phenylindole) was used to detect the nuclei. Coverslips were mounted on microscope slides (Thermo Scientific) and imaged using Leica confocal SP5 microscope (Leica).

Native gel electrophoresis
Cells were harvested in PBS complemented with phosphatase inhibitors and proteinase inhibitors. Samples were then processed using NativePAGE™ Novex® Bis-Tris Gel System according to manufacturer's protocol. Protein were transferred on a polyvinylidene difluoride (PVDF) membrane, previously activated in methanol, using wet transfer system (300mA one hour cell were harvested in immunoprecipitation (IP) buffer (50 mM tris (pH 7.4), 150 mM NaCl, 1% Triton X-100, and 100 mMCaCl 2 with protease and phosphatase inhibitors). Lysates were partly harvested and diluted in loading buffer complemented with 250 mM 1,4-Dithiothreitol (DTT, Thermo Scientific). The lysates were pre cleaned through incubation with Dynabeads ® Protein G for Immunoprecipitation (Life Technologies) at 4°C for two hours. The beads were then discarded and the lysate incubated with Dynabeads ® Protein G and anti-HA tag antibody (1/100) at room temperature for two hours. As an additional negative control two samples transfected with OPTN WT only or TBK1 WT only were incubated with beads and no antibody, to reveal any unspecific binding. The beads were separated from the flowthrough (FT) through magnetic separation and washed with IP buffer six times before elution in loading buffer complemented with DTT. Lysates and IP fractions were analysed by western blot.

Phosphatase assay
Cells were transfected with TBK1 WT and mutant plasmids together with OPTN WT plasmids. Additional controls untransfected or transfected with either TBK1 WT or OPTN WT only were used. After 48 hour transfected cells were harvested in RIPA buffer (50 mM Tris pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulphate with protease inhibitor) and sonicated for ten seconds.
Six µg of protein per sample was added to 3µl of CIP buffer (100 mM NaCl, 50 mM Tris-HCl, 10 mM MgCl 2 , 1 mM dithiothreitol, pH to 7.9 at 25°C) and 3µl of alkaline phosphatase (Roche) were added to the phosphatase positive samples, according to Abcam protein dephosphorylation protocol (http://www.abcam.com/protocols/protein-dephosphorylation-protocol). All the samples were incubated for 30 minutes at 37 °C and ran on NuPAGE Novex 3-8% Tris-Acetate Midi Protein Gels (Life Technologies).
Membrane imaging was conducted with fluorescent secondary antibodies and a LI-COR Odyssey.

Statistical Analysis
Statistical analysis of western blot data was performed using GraphPadPrism software. One way ANOVA analysis followed by Dunnett's post-test was applied to datasets. A t-test was used to compare the mean of two groups of data. T tests were unpaired, two tailed with 95% confidence intervals.

Exome Sequencing in familial ALS detects 16 protein-changing TBK1 variants
We exome sequenced 699 index cases from a cohort of fALS from eleven countries, negative for mutations in all known ALS genes (including SOD1, TDP43, C9ORF72, FUS, PFN1, UBQLN2, OPTN, VCP, and ANG), and the intronic C9ORF72 repeat expansion.    Figure 1C). The p.G217R mutation is located in the kinase domain and was predicted to be damaging by 17/20 of applied algorithms. The p.R357X mutation is located in the ULD and found to remove the entire SDD. The p.C471Y is located in the SDD and may therefore impair TBK1 homodimerisation (Figure 1 A,B).

ALS-linked TBK1 variants decrease the phosphorylation of the TBK1 target IRF3
Some ALS and FTD associated TBK1 variants have previously been shown to diminish or abolish phosphorylation of the TBK1 target IRF3 (Freischmidt et al., 2015;Kim et al., 2016;Pozzi et al., 2017;Tsai et al., 2016). In order to test the efficiency of p.G217R, p.R357X and p.C471Y on IRF3 phosphorylation, we transiently transfected HEK293T cells with wild-type (WT) or mutant TBK1 and quantified IRF3 phosphorylation by Western blot. The expression levels of total IRF3 were comparable for WT and mutant constructs (Figure 2A,B), however, levels of phospho-IRF3 (pIRF3) were significantly reduced in p.G217R and p.R357X variant compared to the WT by western blot (Figure 2A,B) and immunocytochemistry ( Figure 2D). Interestingly, the p.C471Y variant, predicted to be pathogenic by our bioinformatic tools, showed no difference from WT. Thus, both missense p.G217R and nonsense p.R357X mutations, but not the p.C471Y variant, abolished TBK1 kinase activity on its target IRF3.

ALS-linked TBK1 variants decrease binding to OPTN and its phosphorylation
TBK1 is known to phosphorylate and regulate the activity of OPTN, a key receptor for poly-ubiquitinated proteins and mitochondria in autophagy pathways (Richter et al., 2016). TBK1 binds to the N-terminal region of OPTN (26-119) , via its C-terminal domain (residues 677-729) (Li et al., 2016) and phosphorylates it on Ser177 (Wild et al., 2011) and Ser473 (Heo et al., 2015;Richter et al., 2016). Since mutations in OPTN are also linked to ALS, we tested whether our ALS-associated TBK1 variants affected its ability to bind to OPTN and phosphorylate it. Co-immunoprecipitation (Co-IP) of HA tagged WT and p.C471Y TBK1 consistently pulled down flag tagged OPTN ( Figure 2C), however, p.G217R and p.R357X dramatically reduced TBK1 binding to OPTN. This finding was validated by the observation that the same two mutants also failed to phosphorylate OPTN. While TBK1 WT and p.C471Y generated a higher band on western blot that disappeared in the presence of alkaline phosphatase, the higher band is absent following co-transfection with OPTN WT and TBK1 p.G217R or p.R357X ( Figure 2E). We conclude that both missense p.G217R and nonsense p.R357X mutations, but not p.C471Y, impair TBK1 binding to and phosphorylation of OPTN.

M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT
TBK1 is activated by the phosphorylation of Ser172, which causes a critical change in protein conformation promoting the active position of the C helix (Tu et al., 2013). We therefore tested whether the ALSassociated TBK1 variants affect the phosphorylation and autophosphorylation of TBK1 itself. Western blots of HEK293T cells, transfected with each variant, were probed with an antibody specific for phospho-S172.
Total TBK1 expression was similar for WT and all of the ALS-associated variants ( Figure 3A,C). Robust levels of phospho-S172 TBK1 were evident in WT and p.C471Y transfected cells but were absent in cells expressing p.G217R and p.R357X ( Figure 3A,B). Similarly, transfected HEK293T cells stained for phopsho-S172 TBK1 by immunocytochemistry confirmed that TBK1 phosphorylation was absent in cells expressing p.G217R and p.R357X mutants ( Figure 3E). This indicates that the missense p.G217R and nonsense p.R357X but not the p.C471Y variant abolished the capacity of TBK1 for autophosphorylation.

ALS-linked variant p.G217R disrupts TBK1 homodimerisation
TBK1 has to homodimerise in order to be functional and does so via a central axis formed by aligning the two SDD domains in a parallel orientation. The ULD and KD domains interact with each other at one end of the dimer creating a globular structure and stabilising the homodimer (Tu et al., 2013) (Figure1C). We, therefore, investigated whether any of our variants affected the homodimerisation of TBK1 by transfecting HEK293T cells with TBK1 WT, p.G217R, p.R357X and p.C471Y and ran the lysates on non-denaturing gels ( Figure 3B). Native blots revealed two strong high and low molecular weight bands for WT and p.C471Y TBK1, indicating that a similar proportion exists as a homodimer and monomer. Little dimerization, however, was evident for the p.G217R mutant and no band was visible for the p.R357X truncation mutant ( Figure 3B,D). Quantification of the dimer/monomer ratio confirmed that the missense p.G217R kinase domain mutation showed a significantly lower TBK1 homodimerisation compared to WT (p<0.05) ( Figure 3D).
The skipping of exon 8 was confirmed by Sanger sequencing.
Quantification of western blots confirmed that total TBK1 was expressed at a similar level in all of the LCLs. Probing the same blots for phospho-S172-TBK1 revealed that the ratio between pTBK1 and total TBK1 is significantly different between patient and control derived LCLs (p=0.0229, Figure 4). Therefore, missense TBK1 mutations lead to reduced phosphorylation by self-interaction or with other kinases.

Discussion
In this study, we systematically analysed samples from 699 index fALS patients and identified 16 TBK1 variants including 4 nonsense mutations, predicted to cause haploinsufficiency by nonsense mediated RNA decay or to be translated as truncated proteins. Three were in frame deletions and nine missense, which appear to cluster in the functional kinase and ubiquitin like domains known to play a role in TBK1 homodimerisation (Li et al., 2012;Tu et al., 2013). Five of these variants have never been published before: p.M623fs, p.Q629fs, p.T31A, p.R358H and c.992+1G>A (predicted to splice out the whole of exon 8 resulting in an in frame deletion within the ULD). The c.992+1G>A variant was reported once in ALSdb.
Of the other variants only p.R357Q, p.E643del and p.T79del have been functionally investigated (Freischmidt et al., 2015;Gijselinck et al., 2015;van der Zee et al., 2017). Interestingly, disease onset in the patient harbouring both the TBK1 p.Y394D and TARDP p.M337V mutations was 40 years (Table 1), which is 20 years earlier than the average age of onset described in literature (Pozzi et al., 2017). This double hit phenomenon was also described by Freischmidt and colleagues in the patients (all three affected relatives) harbouring TBK1 p.Y185X and FUS p.R524G (Freischmidt et al., 2015).
We characterised the functional impact of three ALS-linked variants selected on the basis of their predicted disruption to key functional domains within TBK1 ( Figure 1C). We chose p.G217R as it lies within the KD and was found in a putative affected sibling, p.R357X as it lies within the ULD and was found in an unrelated case in ALSdb and p.C471Y, which lies within the SDD. Only p.G217R and p.R357X TBK1 variants, but not p.C471Y, abolished the phosphorylation of IRF3. This has previously been described for other ALS-associated mutations (Freischmidt et al., 2015;Kim et al., 2016;Tsai et al., 2016). The same TBK1 variants, p.G217R and p.R357X, also abolished TBK1 binding to OPTN and prevented its phosphorylation. The inhibition of TBK1 binding to OPTN has been previously observed in ALS linked TBK1 variants, mainly located in the C-terminal region of the protein (Freischmidt et al., 2015;Kim et al., 2016;Pozzi et al., 2017;Tsai et al., 2016). However, the inhibition OPTN phosphorylation due to TBK1 mutations has never been shown before. A reduction in active OPTN phosphorylation would impair its M A N U S C R I P T

A C C E P T E D ACCEPTED MANUSCRIPT
function as a receptor for polyubiquitinated proteins and result in the accumulation of TDP-43 which has been observed in patients harbouring TBK1 mutations (Gijselinck et al., 2015;Pottier et al., 2015;van der Zee et al., 2017).
To become activated TBK1 must form a homodimer and be phosphorylated either by itself or by other kinases (Tu et al., 2013). Here we show that the p.G217R and p.R357X variants impair TBK1 autophosphorylation and its ability to be phosphorylated. This observation is consistent with a recent study that showed diminished TBK1 phosphorylation in ALS-associated TBK1 in-frame deletions (p.T79del, p.D167del, p.E643del) (van der Zee et al., 2017). We have also shown that p.G217R, although located in the KD, affects TBK1 ability to homodimerise. In contrast, the p.C471Y variant within the SDD is able to phosphorylate and homodimerise at equivalent levels to the wildtype TBK1 protein and shows no evidence of pathogenicity.
Lastly, we demonstrated that there is a significant difference between phospho-TBK1 and total TBK1 ratio in patient compared to control derived LCLs. This supports the hypothesis that disease-linked TBK1 variants might impair TBK1 autophosphorylation, disrupting its ability to bind and phosphorylate multiple partners including OPTN.

Conclusions
We have identified four novel and 12 previously described TBK1 variants in ALS patients. Our functional studies demonstrated that the missense mutation p.G217R in the KD has an almost identical profile as the truncation p.R357X in the ULD, and dramatically impairs the ability of TBK1 to form homodimers, autophosphorylate and function as a kinase. Furthermore, the proportion of TBK1 that is activated is significantly reduces in five lymphoblast ALS patient lines carrying missense or in-frame deletion mutations. Thus, missense mutations in critical functional domains may cause disease through a loss of TBK1 function supporting functional haploinsufficiency as a common TBK1 disease mechanism (Cirulli et al., 2015;Freischmidt et al., 2016Freischmidt et al., , 2015. Further investigation of how ALS-linked TBK1 variants alter TBK1 structure, phosphorylation and dimerisation will help unravel the disease pathogenesis and identify novel therapeutic targets.

Acknowledgement
We would like to thank people with motor neurone disease (MND) and their families for their participation in this project.

M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT
The authors have no actual or potential conflicts of interest.