Reply: CYLD variants in frontotemporal dementia associated with severe memory impairment in a Portuguese cohort

We thank Tábuas-Pereira and colleagues for their interest in our study demonstrating a causative role for CYLD mutations in frontotemporal dementia (FTD) and amyotrophic lateral sclerosis (ALS) (Dobson-Stone et al., 2020). Tábuas-Pereira et al. (2020) report the identification of rare CYLD missense variants in 2 of 65 Portuguese clinically diagnosed FTD patients. Neither variant was observed in the patient cohorts we examined previously. However, it is of interest that the p.P229S variant affects the amino acid immediately adjacent to the p.P228L variant identified in the ALSdb database (http://alsdb.org/) (Dobson-Stone et al., 2020). CYLD missense mutations have also been reported in CYLD cutaneous syndrome (Bignell et al., 2000; Rajan and Ashworth, 2015). CYLD cutaneous syndrome-causative mutations have the opposite effect on CYLD deubiquitinase activity to the FTD-ALS-causative p.M719V mutation and are not associated with dementia (Dobson-Stone et al., 2020). We reviewed the scientific literature and the Leiden Open Variation Database (https://www.lovd.nl/) and confirmed that neither p.P229S nor p.S615F have been reported in CYLD cutaneous syndrome patients.

Both patients described by Tábuas-Pereira et al. exhibited severe episodic memory impairments, and early memory impairment was also a feature for several patients in the FTD-ALS family with the CYLD p.M719V mutation in our study. As the authors note, this is an Alzheimer’s disease-like phenotype and it is possible that other patients with an Alzheimer’s disease clinical diagnosis harbour pathogenic CYLD variants. Subsequent to the acceptance of our manuscript, we examined the Alzheimer’s Disease Sequencing Project dataset [accessed through the Database of Genotypes and Phenotypes (dbGaP), https://www.ncbi.nlm.nih.gov/gap/], comprising whole exome sequencing from 10 913 individuals. We used the peddy ancestry prediction program (Pedersen and Quinlan, 2017) to filter for individuals with European ancestry, leaving 5544 cases and 4531 controls. We performed gene burden testing of CYLD using the TRAPD software package (Guo et al., 2018) using previously described parameters (Dobson-Stone et al., 2020) and removing variants within the ReFiNE full 0.01 blacklist (Maffucci et al., 2019). No enrichment of qualifying variants (missense variants with maximum population minor allele frequency < 0.0001 in gnomAD non-neurological cohort) was observed in Alzheimer’s disease cases (16/5544, 0.29%) relative to controls (21/4531, 0.46%; Fisher’s exact test two-sided P = 0.185). Interestingly, one of the variants observed in an Alzheimer’s disease case was p.P229S, the variant observed in Patient 1 of Tábuas-Pereira et al. (2020). TRAPD analysis restricted to the deubiquitinase domain of CYLD (amino acids 593–948) in the Alzheimer’s Disease Sequencing Project dataset revealed a significant enrichment of rare missense variants, but in controls rather than cases (cases 1/5544, 0.02%; controls 8/4531, 0.18%; P = 0.014). The reason for this is unclear, but it does emphasize the necessity of functional assays to aid in the interpretation of pathogenicity of any CYLD variant, given that they may have opposing effects on gene function and disease phenotypes.

Further, in silico analysis may have limited predictive value for FTD-ALS-causative CYLD mutations: the CYLD p.M719V mutation was not predicted to be deleterious by common pathogenicity prediction programs PolyPhen-2 and SIFT (Dobson-Stone et al., 2013), in contrast to most CYLD cutaneous syndrome-causative mutations. Also, no family members were available for either patient reported by Tábuas-Pereira et al. to determine segregation of the variants with disease. Since our manuscript was accepted, we have examined the effect of CYLD p.M719V on cell death in L929 cells by detection of dead cells with SYTOX^TM Green (Invitrogen) after transfection with wild-type or mutant CYLD cDNA constructs (Fig. 1). Transfection of CYLD_M719V led to a significantly higher percentage of dead cells [mean ± standard error of the mean (SEM) 55.5 ± 2.7%] relative to wild-type CYLD (44.0 ± 2.6%, t = 3.653, P = 0.035) and empty vector (36.2 ± 1.5%, t = 7.636, P = 0.0047). We examined L929 cells as they have been extensively studied with respect to necroptosis (Degterev et al., 2014), a cell death pathway that CYLD has been implicated in Hitomi et al. (2008) and O’Donnell et al. (2011); however, we note that they are subcutaneous fibroblastoid cells and thus are not closely related to the cell types affected in FTD and ALS. Nevertheless, we share this assay in the hope that it could be used to help prioritize CYLD variants to be screened in other functional assays to determine pathogenicity.

Figure 1.

Expression of CYLD_M719V increases cell death. L929 cells were seeded into 96-well plates at 10⁴ cells/well and transfected with 100 ng CYLD wild-type (wt) or mutant cDNA construct per well 16–24 h after seeding using X-tremeGENE^TM HP (Merck). Dead cells were detected 48 h after transfection by incubation with 125 nM SYTOX^TM Green nucleic acid reagent (Invitrogen) and fluorescence detection on the CLARIOStar microplate reader (BMG Labtech). Dead cells were calculated as a percentage of the total cell count, with the latter determined by SYTOX^TM fluorescence after 1 h treatment with 1% Triton^TM X-100. Horizontal bars indicate mean values from four independent experiments, with transfections performed in quadruplicate per experiment. *P < 0.05; **P < 0.01.

Data availability

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