CYLD is a causative gene for frontotemporal dementia – amyotrophic lateral sclerosis

Dobson-Stone et al. identify a missense mutation in CYLD at a recently discovered disease locus for FTD-ALS. The CYLD mutation gives rise to increased deubiquitinase activity, inhibition of NF-κB, impaired autophagy, increased cytoplasmic localization of TDP-43 and altered axonal morphology.

Abstract

Frontotemporal dementia and amyotrophic lateral sclerosis are clinically and pathologically overlapping disorders with shared genetic causes. We previously identified a disease locus on chromosome 16p12.1-q12.2 with genome-wide significant linkage in a large European Australian family with autosomal dominant inheritance of frontotemporal dementia and amyotrophic lateral sclerosis and no mutation in known amyotrophic lateral sclerosis or dementia genes. Here we demonstrate the segregation of a novel missense variant in CYLD (c.2155A>G, p.M719V) within the linkage region as the genetic cause of disease in this family. Immunohistochemical analysis of brain tissue from two CYLD p.M719V mutation carriers showed widespread glial CYLD immunoreactivity. Primary mouse neurons transfected with CYLD_M719V exhibited increased cytoplasmic localization of TDP-43 and shortened axons. CYLD encodes a lysine 63 deubiquitinase and CYLD cutaneous syndrome, a skin tumour disorder, is caused by mutations that lead to reduced deubiquitinase activity. In contrast with CYLD cutaneous syndrome-causative mutations, CYLD_M719V exhibited significantly increased lysine 63 deubiquitinase activity relative to the wild-type enzyme (paired Wilcoxon signed-rank test P = 0.005). Overexpression of CYLD_M719V in HEK293 cells led to more potent inhibition of the cell signalling molecule NF-κB and impairment of autophagosome fusion to lysosomes, a key process in autophagy. Although CYLD mutations appear to be rare, CYLD’s interaction with at least three other proteins encoded by frontotemporal dementia and/or amyotrophic lateral sclerosis genes (TBK1, OPTN and SQSTM1) suggests that it may play a central role in the pathogenesis of these disorders. Mutations in several frontotemporal dementia and amyotrophic lateral sclerosis genes, including TBK1, OPTN and SQSTM1, result in a loss of autophagy function. We show here that increased CYLD activity also reduces autophagy function, highlighting the importance of autophagy regulation in the pathogenesis of frontotemporal dementia and amyotrophic lateral sclerosis.

Introduction

Frontotemporal dementia (FTD) is the second most common cause of younger-onset (<65 years old) dementia after Alzheimer’s disease (Ratnavalli et al., 2002). FTD is a heterogeneous disorder characterized by personality and behavioural changes, and/or progressive deficits in language expression and comprehension (Seelaar et al., 2011). Around 15% of individuals diagnosed with FTD also develop amyotrophic lateral sclerosis (ALS), a progressive degeneration of upper and lower motor neurons that leads to paralysis and ultimately death (Lillo and Hodges, 2009).

Approximately 40% of FTD cases have an affected relative, suggesting a genetic cause (Goldman et al., 2005). Mutations in several genes have been identified as causes of familial FTD. MAPT and GRN mutations together account for 10–40% of familial FTD cases (Hutton et al., 1998; Baker et al., 2006; Pottier et al., 2016), and a hexanucleotide repeat expansion in C9orf72 is the most common genetic cause of familial FTD and ALS (∼12% and 24% of familial FTD and familial ALS cases, respectively) (DeJesus-Hernandez et al., 2011; Renton et al., 2011). Several other genes have been identified as rarer causes of both FTD and ALS, including VCP, OPTN, SQSTM1, TBK1, CCNF and TIA1 (Watts et al., 2004; Maruyama et al., 2010; Fecto et al., 2011; Freischmidt et al., 2015; Williams et al., 2016; Mackenzie et al., 2017). Several of these genes have been functionally implicated in autophagy, a process by which the cell removes aggregated or misfolded proteins and damaged organelles (reviewed in Deng et al., 2017; Evans and Holzbaur, 2019). Both optineurin (OPTN) and SQSTM1 (p62) are autophagy receptors that are recruited to ubiquitinated substrates and bind the autophagosome protein LC3 (microtubule associated protein 1 light chain 3, encoded by MAP1LC3A) to induce engulfment (Lamark et al., 2017; Markovinovic et al., 2017). In addition, TANK binding kinase 1 (TBK1) is known to phosphorylate and activate both p62 and OPTN (Wild et al., 2011; Pilli et al., 2012).

The aim of this study was to determine the causative mutation in a large European Australian family with FTD-ALS, with no mutations in known dementia or ALS genes. Our previous work provided genome-wide significant evidence of a disease locus on chromosome 16p12.1-16q12.2 [maximum multipoint logarithm of odds (LOD) score of 3.0], with a maximal critical region of ∼38 Mb (GRCh37/hg19 chr16:17 473 796–55 359 315) and a smaller suggestive disease haplotype, defined by recombination in an elderly unaffected individual, of ∼24 Mb (chr16:31 273 792–55 359 315) (Dobson-Stone et al., 2013). Using a comprehensive strategy combining genetic linkage analysis, DNA sequencing, brain immunohistochemical analysis, enzymatic assay and cell culture analyses of axonal morphology, TAR DNA binding protein 43 kDa (TDP-43) localization, nuclear factor kappa B (NF-κB) activity and autophagic flux, we provide evidence that CYLD lysine 63 deubiquitinase (CYLD; NCBI Gene 1540) at chromosome 16q12.1 is a causal gene for FTD-ALS.

Materials and methods

Subject information

One FTD-ALS family (Family Aus-12, Fig. 1) and four FTD and/or ALS patient cohorts were included in this study. Clinical notes for Family Aus-12 pedigree members have been published previously (Dobson-Stone et al., 2013). Cohort sizes were determined by availability of DNA samples and/or whole-exome sequencing data.

Figure 1.

Family Aus-12 has a shared disease haplotype at chromosome 16p12.3–16q12.2. Black symbols show individuals with presenile dementia: clinically diagnosed as Alzheimer’s disease (Individual II:2), FTD (Individuals IV:5, IV:7 and IV:23), or unspecified dementia (Individuals I:2, III:1, III:7, III:9, III:12, III:17, IV:4 and IV:6); grey symbols show individuals with ALS. A diagonal line indicates deceased subjects. Individuals with DNA available are indicated with an asterisk. Inferred haplotypes are in parentheses. Allele data for 12 microsatellite markers and single nucleotide variants in the region are shown, including g.50825515A>G (= CYLD p.M719V).

Cohort 1 comprised 132 European ancestry Australian individuals with younger-onset (age <65 years) dementia (51 with family history of younger-onset dementia) recruited or referred for DNA studies of dementia since 1994, via specialist dementia clinics, neurologists, psychiatrists, geriatricians or genetic clinics. Subjects were negative for mutations in MAPT and GRN and repeat expansions in C9orf72.

Cohort 2 comprised 91 European Australian individuals (18 with ≤1 relative with neurologist/geriatrician-confirmed FTD or younger-onset dementia or ALS) who had undergone precise clinical phenotyping at FRONTIER, an FTD research clinic in Sydney (28 behavioural variant FTD, 23 semantic dementia, 14 progressive non-fluent aphasia, two progressive supranuclear palsy, seven corticobasal syndrome and 17 FTD-ALS). Patients were negative for mutations in MAPT and GRN and repeat expansions in C9orf72.

Cohorts 3 and 4 comprised 81 Australian individuals from 61 families and 219 unrelated British individuals, respectively, with probable or definite ALS diagnosed according to revised El Escorial criteria (Brooks et al., 2000), with at least one relative known to have ALS and/or FTD. Individuals known to harbour mutations in causative ALS genes were excluded from analysis.

Whole-exome data from two larger patient datasets (one FTD, one ALS) were used for gene burden analysis. The FTD cohort was accessed through the International FTD Genomics Consortium (https://ifgcsite.wordpress.com/) and was a subset (n = 1105, 53.3% male) of cases examined for a genome-wide association study of FTD (Ferrari et al., 2014). All cases were diagnosed with behavioural variant FTD only, i.e. no concomitant ALS. Subjects were of European ancestry and samples were collected from sites in Australia, North America and Europe via neurologist referral or (for ∼3% samples) after neuropathological diagnosis. The familial ALS cohort (n = 1138) was accessed through the ALS Variant Server (http://als.umassmed.edu/). Case and control characteristics are described in Nicolas et al. (2018). Cases were diagnosed with ALS and were of European ancestry, recruited at specialist clinics located in Australia, North America and Europe.

Clinical and case note review of neurological phenotypes was performed as part of baseline assessments for 16 consented patients with documented CYLD loss-of-function mutations (n = 15) or fitting clinical diagnostic criteria (n = 1) who were recruited for a clinical trial of targeted topical tumour treatment (TRAC) (Danilenko et al., 2018).

Brain tissue from two sporadic corticobasal degeneration (CBD) cases was sourced from the Sydney Brain Bank for comparison with Family Aus-12 Individuals IV:5 and IV:7. Clinical and neuropathological details are provided in the Supplementary material.

This study was approved by the ethics committees of the respective institutions. Biological samples were obtained after informed consent from subjects and/or their proxies.

Genetic analysis of Family Aus-12

Microsatellite genotyping of Family Aus-12 Individual IV: 14 was performed at the Australian Genome Research Facility (Melbourne, Australia). Parametric multipoint LOD scores were calculated using MERLIN v1.1.2 (Abecasis et al., 2002), with parameters detailed in the Supplementary material. Whole-genome sequencing was performed using 100-bp paired-end sequencing on the Illumina HiSeq2000 with 30× coverage, by Macrogen. The Integrative Genomics Viewer was used to visualize sequencing reads [https://software.broadinstitute.org/software/igv/ (Thorvaldsdottir et al., 2013)]. Analysis of intronic variants, large- and small-scale structural variants and repeat expansion was performed as described in the Supplementary material.

Immunohistochemical analysis

Formalin-fixed paraffin-embedded 10 μm frontal cortex and hippocampal tissue sections were analysed using immunohistochemistry to determine the distribution and localization of CYLD (MABF108, mouse, Merck, 10 μg/ml). Distribution of CYLD in these areas was visualized with DAB (3,3′-diaminobenzidine) peroxidase labelling, using Vector antigen unmasking solution prior to the Novolink Polymer Detection System (Leica, RE7140-K). Double-labelling immunofluorescence was also performed using mouse anti-CYLD and rabbit anti-TDP-43 (10782-2-AP, Protein Tech, 2 μg/ml) and mouse anti-CYLD and rabbit anti-Human Tau (A0024, Dako, 0.4 μg/ml). Detection was achieved using Alexa Fluor secondary antibodies followed by nuclear DAPI (4′,6-diamidino-2-phenylindole) staining. Autofluorescence eliminator reagent (Merck, #2160) was used with VECTASHIELD^® Antifade Mounting Medium (Vector Laboratories H-1000) prior to visualization on a Zeiss Axio Imager M2 Fluorescence microscope.

Mutagenesis

CYLD mutations were introduced into the pCMV6-CYLD construct (Origene) by site-directed mutagenesis using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent). Mutated CYLD cDNA sequence was subcloned into pCMV6-AC, pCMV6-Entry and pCMV6-AC-GFP vectors (Origene) for expression of untagged, C-terminal myc-FLAG-tagged or C-terminal turbo green fluorescent protein (GFP)-tagged CYLD protein, respectively. All clones were verified by restriction digestion and sequence analysis.

Preparation of mouse primary neurons

Animal experiments were approved by the Animal Ethics Committee at Macquarie University. Primary cortical neurons were harvested from C57BL/6 mouse embryos at embryonic Days 16–18, and hippocampal neurons at embryonic Day 16.5, dissociated and plated out on to glass coverslips according to methods detailed in the Supplementary material, followed by incubation at 37°C with 5% CO₂ in a humidified incubator.

TDP-43 localization analysis

After 5 days in vitro, mouse primary cortical neurons were transfected with constructs encoding GFP-tagged empty vector, wild-type or mutant CYLD using Lipofectamine 2000 (Invitrogen) following the manufacturer’s instructions. Neurons were then incubated for 48 h before fixation in 4% paraformaldehyde in phosphate-buffered saline (PBS). Cells were then permeabilized with 0.1% Triton™ X-100 in PBS and non-specific background staining was blocked using 3% bovine serum albumin (BSA) in PBS. Cells were then incubated overnight at 4°C with rabbit anti-TDP-43 antibody (1:100) and mouse anti-MAP2 (1:200, ThermoFisher Scientific 13-1500) antibodies. Detection was achieved using Alexa Fluor secondary antibodies (1:300) followed by nuclear staining with 0.5 μg/ml Hoechst 33342 reagent (Sigma). Coverslips were mounted onto slides in fluorescent mounting medium (Dako) and cells were photographed with the 40× (numerical aperture = 0.75) objective on an AxioImager Z2 fluorescent microscope equipped with a monochrome AxioCam HRm digital CCD camera (Zeiss). MAP2-positive neurons expressing GFP were counted and the proportion of GFP-expressing cells with observable cytoplasmic staining of TDP-43 was calculated. Cytoplasmic TDP-43 was scored by a researcher blind to the DNA construct used. Mean proportions were calculated from at least 50 neurons from three independent experiments.

Neuronal morphology analysis

After 3 days in vitro, mouse primary hippocampal neurons were transfected by incubation for 1 h with GFP-tagged constructs (1 μl of Lipofectamine^® 3000 reagent and 0.5 μg of plasmid) and fixed at 72 h post-transfection with 4% paraformaldehyde. The fixed cells were permeabilized and non-specific binding blocked as described above and stained with antibodies: rabbit anti-GFP (1:1000, ab290, Abcam), mouse anti-MAP2 (1:500, M4403, Merck) and chicken anti-β-tubulin III (1:250, ab9354, Merck). Detection was achieved using Alexa Fluor^® secondary antibodies (1:500; anti-rabbit-488, anti-mouse-647 and 1:250; anti-chicken-555). Cell nuclei were stained with DAPI (1:500, D1306, Invitrogen). Images for analysis were taken using a Zeiss Axio Imager Z1 Upright Fluorescence Microscope with a 40× air objective. Representative images were taken using a Zeiss 880 laser scanning microscope with a 40× oil objective. Images were processed using Zen 2.6 software. Axonal morphology was measured by a researcher blind to the construct used.

Deubiquitinase assay

HEK293 cells (ATCC CRL-1573) were maintained in Dulbecco’s modified Eagle medium with 10% foetal bovine serum (Sigma-Aldrich) at 37°C with 5% CO₂ in a humidified incubator, and were authenticated by short tandem repeat analysis. Cells were seeded into 75 cm² flasks at 2.5 × 10⁶ cells/flask and immediately transfected with LacZ-expressing, wild-type or mutant CYLD-mycFLAG-expressing, cDNA constructs using Lipofectamine™ 2000 (Thermo Fisher Scientific) according to manufacturer’s directions. Forty-eight hours after transfection, cells were harvested by EDTA treatment and washed in PBS. CYLD-mycFLAG protein was purified as described for the co-immunoprecipitation analyses (Supplementary material), with the addition of 1 μg/μl BSA to the elution buffer. Eluted CYLD-mycFLAG was used to digest K63-linked tetra-ubiquitin or K48-linked tetra-ubiquitin chains (Enzo Life Sciences) in a reaction containing the following: 100 ng ubiquitin chains diluted in 9 μl deubiquitinase buffer (50 mM Tris HCl pH 7.6, 5 mM dithiothreitol, 1 μg/μl BSA); and 9 μl CYLD-mycFLAG eluate. Reactions were incubated at 37°C for the specified time and stopped by addition of 5 μl 5× Laemmli sample buffer and heating to 85°C for 5 min. Proteins were resolved by SDS-PAGE (sodium dodecyl sulphate-polyacrylamide gel electrophoresis) on a 12% polyacrylamide gel and immunoblotting was performed as for the co-immunoprecipitation analyses, using mouse anti-ubiquitin (MAB1510, Merck, 1:5000) or mouse anti-CYLD (1:1000) primary antibodies. Immunoblot signal intensity was quantified for each sample using ImageJ software (Abràmoff et al., 2004). The level of ubiquitin was calculated at each time point as the ratio of ubiquitin tetramer to dimer signal intensity, in order to adjust for differences in gel loading between lanes. Deubiquitinase activity was quantified by plotting log₁₀-transformed tetramer:dimer ratios from the 15-, 30- and 45-min time points and then calculating the regression slope, which represented the deubiquitinase activity of each enzyme.

NF-κB luciferase assay

NF-κB activity was assayed by luciferase reporter assay using the pGL4.32 construct (Promega), which expresses firefly luciferase protein under the control of a NF-κB responsive promoter, and normalizing to Renilla luciferase expression from the pRL-TK construct (Promega). HEK293 cells were seeded into 96-well plates at 1.23 × 10⁴ cells/well and transfected after 16–24 h using Lipofectamine™ 2000 according to manufacturer’s directions. Cells were co-transfected with 6.5 ng pGL4.32, 32.5 ng pRL-TK, and 40 ng untagged CYLD wild-type or mutant cDNA construct per well. Twenty-four hours after transfection, NF-κB expression was induced by treatment with 10 pg/μl recombinant human tumour necrosis factor alpha (Promega) for 5 h. Induced cells were washed with PBS and luciferase expression was assayed using the Dual-Luciferase Reporter Assay System (Promega) and the CLARIOStar microplate reader (BMG Labtech), according to manufacturers’ directions. For each construct combination, the ratio of NF-κB-driven firefly luciferase activity to Renilla luciferase activity was calculated per well and mean firefly: Renilla luciferase ratio was calculated across six replicate wells.

Autophagosome maturation assay

HEK293 cells were seeded in 24-well plates containing 13-mm round coverslips. Cells were co-transfected with CYLD constructs and pBABE-puro mCherry-EGFP-LC3B (Addgene plasmid #22418) (N’Diaye et al., 2009) using Lipofectamine™ 2000 according to manufacturer’s instructions. Forty-two hours after transfection, cells were serum starved for 6 h to induce autophagy and then fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton™ X-100, and blocked with 3% BSA in PBS. Cells were then incubated with mouse anti-FLAG (1:250, Merck) at 4°C overnight. For visualization, secondary antibody Alexa Fluor^® 647 conjugated anti-mouse (1:1000, Thermo Fisher Scientific) was used. Coverslips were mounted using fluorescence mounting medium with DAPI (Agilent). Immunofluorescence microscopy was performed using a Zeiss LSM 880 laser scanning microscope with a 100×/1.4 Plan-Apochrome oil objective lens. In dual-channel imaging, photomultiplier sensitivities and offsets were set to a level at which bleed-through effects from one channel to another were negligible. All images were processed using Zen software (Carl Zeiss). Only single cells expressing both CYLD-mycFLAG and mCherry-GFP-LC3 were selected for analysis. For each cell, all autophagosomes (yellow-labelled vesicles with both mCherry and GFP tags fluorescing) and autolysosomes (red-labelled vesicles with only mCherry tag fluorescing) were counted on a merged image. Eighty-eight cells in total were scored for each construct combination across three biological replicates (at least 28 cells scored per replicate) and the mean percentage of autolysosomes was calculated as the number of red-labelled vesicles divided by the number of red- or yellow-labelled vesicles. The scorer was blind to the construct combination used.

Statistical analysis

Regression slopes of CYLD_wt and CYLD_M719V deubiquitinase activity were compared across 10 biological replicate assays by paired Wilcoxon signed-rank test using SPSS version 24. For NF-κB assays, mean luciferase ratios were compared in the presence of wild-type and mutant CYLD across five biological replicate assays by Student’s paired-sample t-test using Microsoft Excel. For the TDP-43 localization assay, the mean percentage of GFP-positive neurons with cytoplasmic TDP-43 from at least 50 neurons per group across three biological replicates was compared by one-way ANOVA and Tukey’s multiple comparisons test using GraphPad Prism version 7. For the neuronal morphology analyses, 20 neurons per group from two independent biological replicates were analysed and used for group comparisons. Outliers in the data were identified using the ROUT method (Q = 1), and data normality was tested using the Shapiro-Wilk test. Means were compared using either one-way ANOVA followed by Bonferroni’s multiple comparisons test for parametric data, or a Kruskal-Wallis test followed by Dunnett’s multiple comparisons test for non-parametric data, using GraphPad Prism version 8.0.2. For the autophagosome maturation assay, the mean percentage of autolysosomes was compared for CYLD wild-type and mutant constructs across 88 cells by one-way ANOVA and Tukey’s multiple comparisons test using GraphPad Prism version 7. Significance for all tests was set at P < 0.05.

Data availability

The authors are willing to provide non-identifying raw data related to this study upon request.

Results

Reclassification of Family Aus-12 Individual IV:14 and analysis of linkage

Our previous study identified a large European Australian kindred (Family Aus-12) with autosomal dominant inheritance of frontotemporal dementia and/or ALS. Presenting symptoms in family members with dementia often mimicked Alzheimer’s disease but neuropathological diagnosis of two individuals confirmed that the family’s dementia was due to frontotemporal lobar degeneration (FTLD) (Dobson-Stone et al., 2013). Available family history at the time indicated that Individual IV:14 was unaffected. However, subsequent clinical follow-up discovered that this person had been diagnosed with ALS and had provided a blood sample for DNA studies. Microsatellite genotyping of selected chromosome 16 markers and analysis of rare single nucleotide variants from whole-exome sequencing data revealed that this person shared the 37.9 Mb common disease haplotype on chromosome 16, flanked by markers D16S3103 and D16S489 (Fig. 1). Multipoint linkage analysis with MERLIN including this affected individual yielded a genome-wide significant maximum LOD score of 3.3 for markers 31447539G>A through to rs190770139 (Supplementary Table 1).

Genetic analysis of the Family Aus-12 critical region

As previously reported (Dobson-Stone et al., 2013) and detailed in Supplementary Table 2, whole-exome sequencing identified only one variant that segregated with disease and was absent from public databases (dbSNP, 1000 Genomes, Exome Variant Server, gnomAD): chr16: g.50825515A>G, within the Family Aus-12 critical region, which leads to the substitution of methionine to valine at amino acid position 719 of the cylindromatosis lysine 63 deubiquitinase protein, CYLD (p.M719V). As in silico analysis had predicted that this substitution was not likely to be pathogenic, and no CYLD mutations had been reported in an independent ALS family linked to the same region (Abalkhail et al., 2003), our initial conclusion was that CYLD M719V was not the disease-causing variant in Family Aus-12 (Dobson-Stone et al., 2013). We next performed whole-genome sequencing of DNA from affected Individuals IV:5 and IV:23, plus three unrelated controls. To ensure we had comprehensive data for all regions likely to harbour pathogenic point mutations, we performed Sanger sequencing of any low coverage exonic or proximal intronic regions within the Family Aus-12 maximal critical region. We defined ‘low coverage’ regions as those that were not represented by at least 10 reads in any affected individual in the whole-exome or whole-genome sequencing data. No additional exonic or canonical splice-site variants were identified that were absent in public databases, and of the six novel deep intronic variants that were identified, none were determined to affect gene splicing (Supplementary Table 3). We also screened affected individuals for structural variation using PennCNV (Wang et al., 2007), Breakdancer (Chen et al., 2009), STRetch (Dashnow et al., 2018) and exSTRa (Tankard et al., 2018) algorithms. No structural variants or significant (adjusted P-value < 0.05) repeat expansions were found in common between affected individuals within the critical region. Following these analyses, CYLD M719V thus remained the only novel protein-altering variant that segregated with disease.

CYLD mutation analysis in clinical and neuropathological cohorts

We screened CYLD coding exons and flanking intronic sequences by Sanger sequencing or examination of whole-exome or whole-genome sequencing data in two European ancestry Australian FTD or dementia cohorts (n = 223), a European Australian familial ALS cohort (n = 81 from 61 families) and a British familial ALS cohort (n = 219). We did not identify any additional variants predicted to affect CYLD coding sequence or splicing that were absent from or with minor allele frequency (MAF) < 0.0001 in the Genome Aggregation Database (gnomAD v2.1.1, http://gnomad.broadinstitute.org/) (Lek et al., 2016).

We also examined CYLD variants in whole-exome data from four larger patient datasets: 1105 individuals with behavioural FTD accessed through the International FTD Genomics Consortium (https://ifgcsite.wordpress.com/); 1138 individuals with familial ALS at the ALS Variant Server (http://als.umassmed.edu/); 2800 individuals in the ALS Data Browser (http://alsdb.org/); and 247 familial ALS cases from the NINDS Genome-Wide Association Study of ALS (dbGaP accession no. phs000101.v5.p1). We identified 26 rare missense variants (MAF < 0.0001 in gnomAD v2.1.1 non-Finnish Europeans without neurological diseases) in ALS or FTD patients (Supplementary Table 4).

We performed gene burden testing of CYLD in the 1105 FTD patients relative to gnomAD non-Finnish European non-neurological exome controls (n = 44 779) as implemented in the TRAPD software package (Guo et al., 2018), comparing the frequency of rare (MAF < 0.0001) missense variants. To mitigate problems arising from comparison of independently processed sample cohorts, we restricted our analyses to variants of high quality (GATK Genotype Quality score ≥ 95) in positions with at least 10× coverage in ≥90% of the cohort. No enrichment of qualifying variants was observed in FTD cases (6/1105, 0.54%) relative to controls (170/44 779, 0.38%) in the whole gene (Fisher’s exact test two-sided P = 0.251). However, analysis restricted to the deubiquitinase domain of CYLD (amino acids 593–948) revealed a significant enrichment (4/1105 cases, 0.36%; 47/44 779 controls, 0.10%; two-sided P = 0.034). A similar analysis comparing the familial ALS Variant Server cohort with 19 494 jointly called controls (Nicolas et al., 2018) did not detect an enrichment of rare missense variants in CYLD, either in the whole gene (3/1138 cases, 0.26%; 124/19 494 controls, 0.64%; Fisher’s exact test two-sided P = 0.974) or the deubiquitinase domain (1/1138 cases, 0.09%; 27/19 494 controls, 0.14%; P = 0.796).

CYLD loss-of-function mutations are not associated with dementia

CYLD cutaneous syndrome (also known as Brooke-Spiegler syndrome) is associated with heterozygous germline mutations in CYLD (Bignell et al., 2000; Rajan and Ashworth, 2015). These patients typically develop multiple benign hair follicle tumours called cylindromas, and are not recognized to have a neurological phenotype (Rajan et al., 2009). The majority of mutations in these pedigrees result in predicted truncation (Nagy et al., 2015), and CYLD loss of heterozygosity is frequently demonstrated in cylindroma skin tumours. We reviewed the clinical files of 16 subjects with CYLD loss-of-function mutations, including six who were older than the maximum known age of onset for Family Aus-12 (62 years) (Supplementary Table 5). No patients met diagnostic criteria for or had been prescribed medicine to treat dementia, and none were noted to have signs of cognitive impairment or neurological disease.

CYLD neuropathology analysis

We sought to determine whether CYLD showed an abnormal staining pattern in brain tissue from affected Family Aus-12 members. We have previously described the neuropathological analysis of two cases from this family: Individual IV:5, with clinical diagnosis of ‘FTD with aphasia’; and Individual IV:7, with clinical diagnosis of FTD-ALS, Paget’s disease and parkinsonism (Dobson-Stone et al., 2013). These individuals met neuropathological criteria for both CBD FTLD with tau pathology (CBD FTLD-tau) and FTLD with type B TDP-43 pathology (type B FTLD-TDP). In the present study, we examined fixed brain tissue from Family Aus-12 Individuals IV:5 and IV:7, plus two disease control cases (one sporadic CBD FTLD-tau, and one sporadic CBD FTLD-tau with concomitant TDP-43 pathology). CYLD immunohistochemistry revealed staining in the blood vessels (Fig. 2A) and the cytoplasm of balloon neurons (Fig. 2B) in all cases. Both Family Aus-12 cases demonstrated widespread glial CYLD-immunoreactivity in the white matter, which was particularly marked in Individual IV:7 (Fig. 2D) and was absent in the sporadic cases (Fig. 2C). Patchy areas of diffuse neuronal cytoplasmic CYLD staining were seen in the hippocampus and the frontal cortex of Individuals IV:5 and IV:7 only (Fig. 2E). In addition, CYLD-immunoreactivity was seen in the nuclei of pyknotic neurons in the hippocampus and deep frontal cortex layers for the Family Aus-12 cases only, most noticeably in Individual IV:7 (Fig. 2F). Double-labelling immunofluorescence did not detect co-localization of CYLD with either tau or TDP-43 inclusions (Fig. 2G and H).

Figure 2.

CYLD_M719V affects CYLD immunoreactivity in brain tissue. Photomicrographs demonstrating CYLD immunoreactivity in the frontal cortex and hippocampus. (A) Parahippocampal white matter showing CYLD peroxidase labelling of blood vessels, which was seen in all cases. (B) CYLD staining was seen in balloon neurons in all cases. Image shows representative image taken from Sydney Brain Bank Case 2. (C and D) CYLD-positive glia were common in Family Aus-12 (D) but not sporadic cases (C). Images were taken in the frontal white matter of Sydney Brain Bank Case 1 and Individual IV:7, respectively. (E) Diffuse, neuronal cytoplasmic CYLD staining of a neuron in the hippocampus of Individual IV:5. (F) CYLD-positive nuclear staining of a pyknotic neuron in the deep layers of the frontal cortex of Individual IV:7. (G and H) Double labelling immunofluorescence in the dentate gyrus of Individuals IV:5 (G) and IV:7 (H). TDP-43 (green, arrowheads in G) and Tau (green, H) did not co-localize with CYLD (red, G and H) in any case. Scale bar = 100 μm in A, 20 μm in B and E–H, 40 μm in C and D.

CYLD_M719V increases cytoplasmic localization of TDP-43

A cardinal neuropathological feature of half of FTD patients and almost all ALS patients, including Family Aus-12 affected members, is movement of TDP-43 from the nucleus to the cytoplasm (Neumann et al., 2006; Dobson-Stone et al., 2013). We examined the effect of the CYLD_M719V variant on TDP-43 localization by transfection of primary mouse cortical neurons with GFP-tagged CYLD wild-type (CYLD_wt) or mutant constructs and examination by immunofluorescence microscopy (Fig. 3). A significantly higher proportion of neurons transfected with CYLD_wt-GFP showed cytoplasmic TDP-43 staining (34.4 ± 0.5% of transfected cells) relative to empty vector (21.2 ± 0.8%, one-way ANOVA with Tukey’s multiple comparisons test P = 8.8 × 10⁻⁴) (Fig. 3B). Transfection with CYLD_M719V-GFP led to an even higher proportion of neurons with cytoplasmic TDP-43 staining (54.9 ± 2.7%, one-way ANOVA with Tukey’s multiple comparisons test P = 3.7 × 10⁻⁵ relative to CYLD_wt-GFP). In contrast, cells transfected with catalytically inactive CYLD cutaneous syndrome mutant CYLD_D681G (Almeida et al., 2008) showed a similar proportion of cytoplasmic TDP-43-positive cells relative to the empty vector condition (19.4 ± 0.6%).

Figure 3.

Expression of CYLD_M719V alters neuronal TDP-43 localization. (A) Representative fluorescence images of mouse cortical neurons at 7 days in vitro, overexpressing GFP or GFP-tagged CYLD_wt, CYLD_D681G or CYLD_M719V (green). TDP-43 was detected by immunofluorescent staining (red) and nuclei were visualized with DAPI (blue). Arrowheads indicate cytoplasmic TDP-43 staining in cells transfected with CYLD_wt and CYLD_M719V. (B) Quantification of the percentage of transfected cells with observable cytoplasmic TDP-43 staining, showing a significantly higher proportion of cytoplasmic TDP-43-positive cells in the CYLD_M719V condition relative to CYLD_wt. Horizontal bars indicate mean values from three independent experiments. Scale bars = 10 μm. **P < 0.005, ****P < 0.00005.

CYLD_M719V alters axonal morphology

We examined the effect of CYLD_M719V on neuronal morphology by transfection of primary mouse hippocampal neurons with GFP-tagged CYLD wild-type or mutant constructs and examination by immunofluorescence microscopy (Fig. 4). Overexpression of CYLD_wt led to a small non-significant decrease in primary and total axon length relative to GFP-only vector condition [mean lengths (μm) ± SEM: 781.1 ± 72.5 primary, 2120 ± 228.7 total) (Fig. 4E, H). Overexpression of CYLD_M719V led to a further decrease in primary (627.3 ± 63.5) and total axon length (1460 ± 174.4) relative to GFP-only vector, which was significant [one-way ANOVA with Bonferroni’s multiple comparisons test: P = 0.026 (primary axon length), P = 0.003 (total axon length)].

Figure 4.

Expression of CYLD_M719V alters axonal morphology. (A₁–D₁) Representative fluorescence images of mouse hippocampal neurons at 6 days in vitro, overexpressing GFP (A₁), CYLD_wt-GFP (B₁), CYLD_D681G-GFP (C₁) or CYLD_M719V-GFP (D₁). (A₂–D₂) Inverted black and white images of GFP fluorescence of the transfected neurons shown in A₁–D₁, with pseudo-coloured axon labelling for clearer visualization of the axonal compartment. (E–H) Quantification of axonal morphology. Box plots represent median, interquartile range and minimum-maximum (n = 20 each group). *P < 0.05, **P < 0.005. Scale bars = 200 μm.

CYLD_M719V does not affect interaction with TBK1, OPTN or p62

CYLD has been reported to physically interact with proteins whose genes are mutated in ALS and FTD, namely TBK1 (Friedman et al., 2008), OPTN (Nagabhushana et al., 2011) and p62 (Jin et al., 2008). We sought to confirm these interactions and determine whether the M719V variant abrogated them by co-transfection of HEK293 cells with wild-type or mutant CYLD-GFP constructs and TBK1-mycFLAG, OPTN-mycFLAG, or p62-mycFLAG constructs, followed by immunoprecipitation with anti-FLAG antibody (Supplementary Fig. 1). A faint band corresponding to CYLD-GFP was detected in the immunoprecipitation eluate only when TBK1, OPTN or p62 was co-transfected, indicating that a small proportion of CYLD is bound to these proteins. An interaction was also observed with all three proteins for CYLD cutaneous syndrome mutant CYLD_D681G and the Family Aus-12 variant CYLD_M719V.

CYLD_M719V shows increased deubiquitinase activity

CYLD is a deubiquitinase with lysine 63 (K63) activity (Komander et al., 2008). The M719V variant is located within the deubiquitinase region of CYLD (amino acids 593–948). We compared the deubiquitinase activity of purified CYLD_wt-mycFLAG, CYLD_D681G-mycFLAG and CYLD_M719V-mycFLAG by incubation with synthetic K63-linked ubiquitin tetramers and western blot (Fig. 5A–C and Supplementary Fig. 3A). As expected, the catalytically inactive cylindromatosis mutant CYLD_D681G-mycFLAG showed no observable K63-deubiquitinase activity (Supplementary Fig. 3A). Interestingly, the Family Aus-12 variant CYLD_M719V-mycFLAG showed increased K63-deubiquitinase activity relative to CYLD_wt-mycFLAG (Fig. 5A), with a significant 3.7-fold increase in rate of conversion of K63-linked ubiquitin tetramers to dimers (paired Wilcoxon signed-rank test P = 0.005) (Fig. 5B and C). Wild-type and variant CYLD did not show observable K48-deubiquitinase activity (Supplementary Fig. 3B).

Figure 5.

CYLD_M719V shows increased K-63 deubiquitinase activity and inhibition of NF-κB. K63-linked ubiquitin tetramers (Ub4) were incubated with wild-type (wt) or M719V mutant CYLD-mycFLAG and digestion was observed. (A) Representative immunoblot showing appearance of bands corresponding to ubiquitin trimers (Ub3) and dimers (Ub2) after 15, 30 and 45 min of CYLD treatment. An uncropped image of this blot is available in Supplementary Fig. 4. (B) Quantification of ubiquitin cleavage expressed as log₁₀-transformed Ub4/Ub2 after treatment with CYLD_wt (open circles) or CYLD_M719V (filled squares) (mean ± SEM from 10 experiments). (C) Comparison of rate of K63-ubiquitin cleavage shows significantly increased rate of cleavage for CYLD_M719V relative to CYLD_wt. Box plots represent median, interquartile range and minimum-maximum from n = 10 independent experiments. (D) Quantification of NF-κB activity. HEK293 cells were transfected with NF-κB-responsive luciferase construct pGL4.32 plus empty vector or human (h) CYLD_wt, CYLD_D681G or CYLD_M719V constructs or mouse (m) equivalents Cyld_wt, Cyld_D680G or Cyld_M718V, and NF-κB expression was induced by tumour necrosis factor α treatment. Horizontal bars indicate mean values from five independent experiments. Values are normalized to human CYLD_wt or mouse Cyld_wt, and show significant increases for cylindromatosis mutant CYLD_D681G and mouse equivalent Cyld_D680G and decreases for Family Aus-12 mutant CYLD_M719V and mouse equivalent Cyld_M718V. **P < 0.005, ***P < 0.0005.

CYLD_M719V increases inhibition of NF-κB

CYLD is a negative regulator of NF-κB activity (Trompouki et al., 2003). Thus we sought to determine whether the M719V variant affected CYLD’s inhibition of NF-κB by co-transfection of wild-type or mutant untagged CYLD constructs and the NF-κB-responsive luciferase construct pGL4.32 into HEK293 cells, followed by assessment of luciferase activity (Fig. 5D). As expected, transfection of CYLD_wt led to a significant decrease in NF-κB activity relative to empty vector control (0.26 ± 0.05-fold, t = 4.97, P = 0.008), and transfection of cylindromatosis mutant CYLD_D681G led to a significant increase in NF-κB activity relative to wild type (2.39 ± 0.27-fold, t = 11.43, P = 3.3 × 10⁻⁴). Transfection of CYLD_M719V led to a significant decrease in NF-κB activity relative to wild-type (0.50 ± 0.05-fold, t = 7.63, P = 0.002). This increase in CYLD inhibitory activity was not due to any increase in CYLD protein levels of the M719V mutant relative to wild-type (Supplementary Fig. 5). This increased inhibitory effect on NF-κB was recapitulated using constructs with the equivalent mutation in mouse sequence Cyld (0.51 ± 0.05-fold activity of NF-κB with Cyld_M718V relative to Cyld_wt, t = 8.11, P = 0.001) (Fig. 5D and Supplementary Fig. 5), but not in 10 other missense variants from ALS or FTD patients (Supplementary Figs 3C and 5).

CYLD_M719V impairs autophagosome maturation

Several FTD-ALS proteins have been implicated in autophagy, including TBK1, OPTN and p62 (Lamark et al., 2017; Markovinovic et al., 2017; Oakes et al., 2017), which interact with CYLD (Supplementary Fig. 1). We examined whether CYLD_M719V has an impact on this process. Autophagosome maturation was monitored at the single cell level by expressing a dual-labelled mCherry-EGFP-LC3 construct under serum starvation conditions (Fig. 6). The fusion of the autophagosome to the lysosome causes displacement of its contents within the acidic environment of the lysosome for degradation, resulting in quenching of the EGFP but not the mCherry signal for the mCherry-EGFP-LC3 reporter. Thus, co-localization of both mCherry and GFP fluorescence is indicative of improper fusion of the autophagosome to the lysosome, whereas mCherry fluorescence alone corresponds to proper fusion and mature autolysosome (Klionsky et al., 2016). A significantly increased proportion of autophagosomes and concomitant decreased proportion of autolysosomes were observed in cells co-transfected with mCherry-EGFP-LC3 and CYLD_M719V-mycFLAG compared to cells that were co-transfected with CYLD_wt-mycFLAG (one-way ANOVA with Tukey’s multiple comparisons test P < 10⁻⁴) (Fig. 6B). This suggests that autophagosome fusion is impaired in CYLD_M719V-mycFLAG-transfected cells.

Figure 6.

CYLD_M719V impairs fusion of autophagosomes with lysosomes. (A) Images of HEK293 cells cotransfected with tandem mCherry-GFP-tagged autophagy marker LC3 plus empty vector or mycFLAG-tagged CYLD_wt, CYLD_D681G or CYLD_M719V constructs. Cells were serum starved for 6 h to induce autophagy. Fusion of the autophagosome with the lysosome to produce autolysosomes quenches GFP but not mCherry fluorescence. In the merged image, autophagosomes appear as yellow vesicles, while autolysosomes appear as red vesicles. (B) Quantification of the percentage of autophagosomes (mCherry- and GFP-positive vesicles, left) and autolysosomes (mCherry-positive only vesicles, right) relative to total vesicle number, showing a significant decrease in proportion of autolysosomes in CYLD_M719V cells, but not CYLD_D681G cells relative to CYLD_wt. Box plots represent median, interquartile range and minimum-maximum (n = 88 each group). Scale bar = 10 μm. ***P < 0.0001.

Discussion

We demonstrate here that the M719V missense variant in CYLD segregates with disease in a large multigenerational FTD-ALS family. This variant is associated with altered CYLD immunoreactivity in brain tissue and significantly impacts two key roles of CYLD: K63 deubiquitination (Komander et al., 2008) and negative regulation of NF-κB (Trompouki et al., 2003). Furthermore, in vitro expression of CYLD_M719V recapitulates two cellular phenotypes of FTD and ALS: increased cytoplasmic TDP-43 (Neumann et al., 2006) and impaired autophagy (Deng et al., 2017; Evans and Holzbaur, 2019). As we did not have DNA available from relatives of other CYLD variant carriers, we were only able to demonstrate disease segregation of a CYLD mutation in one family, which is a limitation of this study, albeit not unusual for modern FTD and ALS gene identification studies (Williams et al., 2016). However, our observations of genome-wide significant linkage in this family, exclusion of other genomic variants in the linked region, and our neuropathological and in vitro analyses of CYLD_M719V, together indicate that mutation of CYLD is a cause of FTD-ALS.

No mutations were identified in the coding or flanking intronic regions of CYLD in a previously reported ALS family linked to an overlapping region on chromosome 16 (Abalkhail et al., 2003; Dobson-Stone et al., 2013). We note that the maximal LOD score for this family of 2.06 was suggestive but not statistically genome-wide significant, and a third family initially reported to be linked to this locus with maximal LOD score of 2.1 (Family F2 in Ruddy et al., 2003) was subsequently shown to exhibit genome-wide significant linkage to chromosome 9p, with affected family members all harbouring the pathogenic C9orf72 repeat expansion (Al-Sarraj et al., 2011).

We observed differences in CYLD immunostaining in CYLD mutation carriers relative to sporadic CBD cases, most notably widespread staining of glia in the white matter. CYLD was not observed in cytoplasmic inclusions, which contrasts with findings for mutation carriers of some other FTD-ALS genes [e.g. OPTN (Maruyama et al., 2010), SQSTM1 (Teyssou et al., 2013)] but not all [e.g. TBK1 (Pottier et al., 2015), TIA1 (Hirsch-Reinshagen et al., 2017)]. We previously reported that Family Aus-12 Individuals IV:5 and IV:7 met neuropathological criteria for both FTLD-TDP type B and CBD FTLD-tau (Dobson-Stone et al., 2013). TDP-43 deposition is characteristic of most other reported FTD-ALS genes (Maruyama et al., 2010; DeJesus-Hernandez et al., 2011; Teyssou et al., 2013; Pottier et al., 2015). However, it is of note that concomitant TDP-43 and tau pathology has also been reported in 3/11 TBK1 mutation carriers with neuropathological data reported thus far: one exhibited tau-immunoreactive tufted astrocytes and neuronal cytoplasmic inclusions (Van Mossevelde et al., 2016), and two exhibited argyrophilic grain disease (Koriath et al., 2017; van der Zee et al., 2017). Three cases showed no tau pathology (Van Mossevelde et al., 2016; van der Zee et al., 2017), and tau neuropathological analysis was not reported for the remaining five cases (Freischmidt et al., 2015; Pottier et al., 2015). Neuropathological examination of more CYLD and TBK1 mutation carriers will be required before concluding whether this is a chance finding, or indicative of some common predisposition for these mutations to cause abnormal deposition of tau in addition to TDP-43.

The CYLD_M719V mutation leads to a significant increase in K63 deubiquitinase activity and inhibition of NF-κB. This is in contrast to CYLD missense mutations reported in cylindromatosis and similar disorders, which severely reduce or abrogate deubiquitinase activity and NF-κB inhibition (Blake and Toro, 2009), implying a different pathogenetic mechanism for these disorders. Consistent with this hypothesis, we found no evidence of neurological disease in patients with CYLD loss-of-function mutations, including six who were older than the maximum known age of onset for Family Aus-12. Conversely, review of available clinical records did not reveal any signs of a skin disorder in affected individuals from Family Aus-12, although we cannot exclude more subtle manifestations of CYLD cutaneous syndrome, such as milia (Bajwa et al., 2018). Our data are thus consistent with opposing effects of gain-of-function versus loss-of-function mutations for the same gene, as has been observed previously for other genes (e.g. SMCHD1; Gurzau et al., 2018).

We observed that overexpression of CYLD_M719V led to mislocalization of TDP-43 to the cytoplasm in primary neurons. No TDP-43-positive inclusions were observed, in contrast with similar studies performed with some other FTD and ALS genes, e.g. CHCHD10 (Woo et al., 2017) and PFN1 (Tanaka et al., 2016); but not all, e.g. VCP (Ritson et al., 2010). Diffuse cytoplasmic staining of TDP-43 is seen as a precursor to aggregate formation (Brandmeir et al., 2008; Mori et al., 2008; Kim et al., 2019): it is possible that longer-term transfection of mutant CYLD would lead to observable aggregation of TDP-43. We also observed that neuronal overexpression of CYLD_M719V led to a reduction of axonal length, which has been observed before with mutant ALS proteins, e.g. SOD1 (Takeuchi et al., 2002), PFN1 (Wu et al., 2012) and TDP-43 (Duan et al., 2011; Tripathi et al., 2014). The latter observation indicates that axonal morphology effects of CYLD_M719V might be mediated through altering TDP-43 localization. However, further studies would be needed to demonstrate whether there is a causal relationship between TDP-43 localization and altered axonal morphology as a result of CYLD_M719V expression. Both increased K63 deubiquitinase activity and inhibition of NF-κB may contribute to the reduced axonal growth in CYLD_M719V-expressing neurons. For example, K63 ubiquitination of muscleblind-like protein 1 (MBNL1) has recently been shown to be important for its localization to the cytoplasm, and promotion of axonal growth (Wang et al., 2018). K63 deubiquitination negatively regulates MBNL1 function and could contribute to the reduced axonal growth. Furthermore, NF-κB is a key regulator in mediating signalling cues that promote axonal growth (reviewed in Gutierrez and Davies, 2011) and increased inhibition of NF-κB may be a major factor, leading to the observed axonal phenotype in our study.

We confirmed that CYLD directly interacts with TBK1, OPTN and p62 (Friedman et al., 2008; Jin et al., 2008; Zhang et al., 2008; Nagabhushana et al., 2011), three proteins that are encoded by genes with mutations in FTD or ALS. Nagabhushana et al. (2011) reported that interaction between OPTN and CYLD is required for CYLD-mediated inhibition of NF-kB. The interaction of p62 with CYLD is required for the negative regulation of osteoclastogenesis (Jin et al., 2008). Lastly, CYLD and TBK1 are substrates of each other. CYLD negatively regulates TBK1-mediated antiviral innate immunity by removal of K63 ubiquitin chains from TBK1 (Friedman et al., 2008; Zhang et al., 2008). Conversely, CYLD phosphorylation by TBK1 has been observed upon T-cell receptor stimulation (Lork et al., 2018).

It is pertinent that TBK1, OPTN and p62 are all implicated in autophagy (Wild et al., 2011; Pilli et al., 2012; Lamark et al., 2017; Markovinovic et al., 2017) and that ALS and/or FTD-causative mutations in these and other genes impair autophagic function (Soo et al., 2015; Moore and Holzbaur, 2016; Webster et al., 2016; Lee et al., 2018). In the current study, expression of CYLD_M719V led to a significant decrease in relative autolysosome numbers compared to CYLD_wt, and concomitant increase in autophagosomes. This suggests that CYLD_M719V leads to an impairment in autophagosome maturation to autolysosomes. We hypothesize that overactive CYLD is another mechanism by which autophagy can become impaired in FTD and ALS.

CYLD has not hitherto been extensively studied in a neuronal context, but is highly enriched in the postsynaptic density (Dosemeci et al., 2013; Jin et al., 2019) and mediates long-term depression in cultured neurons, likely through deubiquitination of the synaptic scaffold protein PSD-95 (Ma et al., 2017). Recently, CYLD was identified as a mediator of programmed necroptosis in neurons in vitro, and post-traumatic brain injury neuronal death in vivo (Ganjam et al., 2018). It is therefore possible that the overactive CYLD_M719V mutant may at least in part lead to neurodegeneration via inappropriate necroptosis of neurons.

CYLD is a highly constrained gene (more intolerant to missense substitutions than 93% of genes; Lek et al., 2016) and mutations that increase its activity are likely extremely rare. We observed a statistically significant enrichment of CYLD deubiquitinase domain missense variants in individuals with FTD but not ALS. We note that gene burden tests in large case-control exome studies reliably detect enrichment of rare variants in major familial ALS genes such as SOD1 and TARDBP (Cirulli et al., 2015). However, rarer ALS genes, even those with previously demonstrated strong linkage and/or segregation in large pedigrees, often fail burden testing, e.g. PFN1, VAPB, MATR3 (Cirulli et al., 2015). Conversely, none of the CYLD variants we assayed apart from M719V showed a significant increase of NF-κB inhibition. This implies that either the enrichment in FTD patients is a chance observation, or that some FTD- or ALS-associated CYLD variants are pathogenic but do not act by affecting NF-κB inhibition. As we only examined cohorts with European ancestry, we cannot comment on the possible contribution of CYLD mutations to non-European FTD or ALS cohorts.

In conclusion, we have demonstrated that mutation of CYLD is a cause of FTD and ALS. Although CYLD mutations appear to be rare, its interaction with at least three other proteins encoded by established FTD and ALS genes suggests that CYLD also plays a central role in autophagy pathways contributing to the pathogenesis of these disorders. Further work on the central role of autophagy in FTD and ALS should be the focus of future studies.

Glossary

ALS =

amyotrophic lateral sclerosis

CBD =

corticobasal degeneration

GFP =

green fluorescent protein

FTD =

frontotemporal dementia

FTLD =

frontotemporal lobar degeneration

LC3 =

microtubule associated protein 1 light chain 3

NF-κB =

nuclear factor kappa B

TDP-43 =

TAR DNA binding protein 43 kDa