Loss‐of‐Function Variants in HOPS Complex Genes VPS16 and VPS41 Cause Early Onset Dystonia Associated with Lysosomal Abnormalities

The majority of people with suspected genetic dystonia remain undiagnosed after maximal investigation, implying that a number of causative genes have not yet been recognized. We aimed to investigate this paucity of diagnoses.

Objectives: The majority of people with suspected genetic dystonia remain undiagnosed after maximal investigation, implying that a number of causative genes have not yet been recognized. We aimed to investigate this paucity of diagnoses. Methods: We undertook weighted burden analysis of whole-exome sequencing (WES) data from 138 individuals with unresolved generalized dystonia of suspected genetic etiology, followed by additional case-finding from international databases, first for the gene implicated by the burden analysis (VPS16), and then for other functionally related genes. Electron microscopy was performed on patient-derived cells. Results: Analysis revealed a significant burden for VPS16 (Fisher's exact test p value, 6.9 × 10 9 ). VPS16 encodes a subunit of the homotypic fusion and vacuole protein sorting (HOPS) complex, which plays a key role in autophagosomelysosome fusion. A total of 18 individuals harboring heterozygous loss-of-function VPS16 variants, and one with a microdeletion, were identified. These individuals experienced early onset progressive dystonia with predominant cervical, bulbar, orofacial, and upper limb involvement. Some patients had a more complex phenotype with additional neuropsychiatric and/or developmental comorbidities. We also identified biallelic loss-of-function variants in VPS41, another HOPS-complex encoding gene, in an individual with infantile-onset generalized dystonia. Electron microscopy of patient-derived lymphocytes and fibroblasts from both patients with VPS16 and VPS41 showed vacuolar abnormalities suggestive of impaired lysosomal function. Interpretation: Our study strongly supports a role for HOPS complex dysfunction in the pathogenesis of dystonia, although variants in different subunits display different phenotypic and inheritance characteristics. ANN NEUROL 2020;88:867-877 D ystonia is a common movement disorder associated with significant disability and increased risk of mortality. 1,2 It is characterized by sustained or episodic muscle contractions, which cause abnormal, often repetitive movements and twisting postures affecting the limbs, trunk, neck, and face. 3 Despite significant advances in nextgeneration sequencing technologies, over 85% of people with suspected genetic dystonia remain undiagnosed after whole-genome sequencing, 4 implying that the majority of genetic dystonias currently remain unrecognized. The reasons for this are multifactorial, attributed to locus heterogeneity, incomplete disease penetrance, and the current limitations of next-generation sequencing technologies.
Here, we report a cohort of individuals with loss-offunction (LOF) mutations in 2 components of the homotypic fusion and vacuole protein sorting (HOPS) complex, a highly conserved complex required for endosome-lysosome and autophagosome-lysosome fusion. 5 We describe a series of patients with generalized dystonia associated with heterozygous LOF variants in VPS16 and also report biallelic LOF variants in a second HOPS complex gene, VPS41, in a child with a severe infant-onset dystonic disorder.

Generalized Dystonia Cohort for Burden Analysis
A consecutive series of 138 unrelated individuals with generalized dystonia (57 men and 81 women, all self-identifying as European) was recruited into the study. Diagnoses were established in accordance with the dystonia consensus criteria 3 at movement disorder specialty centers in Austria, Czechia, and Germany. The clinical characteristics of the cohort are summarized in Supplementary Table S1. We excluded patients from the cohort who had (i) a known genetic diagnosis or (ii) an acquired form of the disease.

Whole-Exome Sequencing
The generalized dystonia cohort underwent whole-exome sequencing (WES) at the Helmholtz Center Munich (Munich, Germany) according to previously described methods. 6 In brief, the exonic portions of genomic DNA were enriched in solution and indexed with Agilent (Agilent Technologies) SureSelect Human All Exon kits, version 5 and 6. Sequencing was carried out as 100-bp paired-end runs with HiSeq2500/4000 equipment (Illumina). Read processing and variant annotation used an inhouse pipeline based on BWA, SAMtools, PINDEL, GATK, ExomeDepth, and custom scripts (Helmholtz Center Munich and Technical University of Munich). Variant filtering was done per standard pipeline analyses, integrating data from online repositories (1000 Genomes Project, gnomAD, dbSNP, ClinVar, and HGMD) and in-house control-exome collections. For the 138 exomes, we obtained on average 13.6 Gb of sequence, resulting in a mean read depth of 143.6-fold with 98.6% of the target nucleotides covered at least 20-fold. Sequences were visualized with IGV. Across the cohort, the exome data were used to exclude causative variants in known disease genes, as described. 6 Case-Control Rare-Variant Collapsing Analysis (Burden Test) Gene-based collapsing analysis of rare variants in patients with dystonia versus controls was performed using TRAPD (Test Rare vAriants with Public Data), 7 a robust method for detecting gene-disease associations. 8 We searched for genes with excess mutational burden by comparing genotype counts from 138 generalized dystonia cases with those from gnomAD control subjects (non-Finish European [NFE] cohort, N = 64,603). We coded case and control subjects according to the presence or absence of at least one qualifying variant in any of the 20,000 consensus coding sequence (CCDS) genes and focused on the following genetic models: (1) dominant LOF, in which qualifying variants were defined as stop-gain, frameshift, and splice-site-altering (± 2 nucleotides of exon boundary) alleles; and (2) dominant nonsynonymous, in which qualifying variants were defined as LOF and missense alleles. The minor allele frequency (MAF) threshold of qualifying variants was set at < 0.0005, with the frequency of minor alleles determined from gnomAD (NFE cohort) and 4,000 non-neurological in-house control exomes for variants present in dystonia case subjects, and in gnomAD (NFE cohort) for variants present in control subjects. To detect differences in the carrier rate of qualifying variants between case and control subjects, we used a one-sided Fisher's exact test. Exome-wide significance was defined at a p value of < 1.25 × 10 6 , correcting for 20,000 CCDS genes studied in two individual case-control comparisons. To avoid spurious results, we undertook extensive quality control and harmonization analyses, as described earlier 7,8 : (i) variants overlapping low-complexity regions were filtered out; (ii) sites with a read depth of < 10-fold in either of the 2 cohorts were ignored; and (iii) rare synonymous variant burden testing was conducted. On the basis of the latter, only the top 85% of sites in terms of quality-by-depth (QD) scores for the case sequencing cohort and the top 95% of sites in terms of QD scores for gnomAD were included in the analysis.

Identification of Additional Cases
Using GeneMatcher 9 and direct communications, international collaborators were requested to screen their genomic databases for additional cases. Details of the WES/whole genome sequencing (WGS) methods used differed slightly among each center and can be provided on request. Cases were identified from the UCL Great Ormond Street Institute of Child Health neurogenetic movement disorders cohort (London, UK); the Kolling Institute of Medical Research (Sydney, Australia); the Carlo Besta Neurological Institute (Milan, Italy); the Koios Database of the Queen Square Genomics Group at University College London (London, UK); the Genomics England 100 K Genomes Project dataset (UK), and Radboud University Medical Centre (Nijmegen, The Netherlands). Databases from Cardiff (Wales) and Dublin (Ireland) were also checked but no additional cases were found there. Variants identified through WGS or WES and familial segregation were verified by Sanger sequencing. Details of protocols, reagents, and primer sequences are available on request. All variants are given with reference to the GRCh38 build.
We subsequently undertook a targeted search of the databases above for any additional individuals with mutations affecting other HOPS complex genes not previously associated with disease, namely VPS18, VPS39, and VPS41.

Electron Microscopy
Whole blood samples were obtained in EDTA and centrifuged to produce a buffy coat. Patient fibroblasts were obtained from skin biopsies and cultured in Ham's F10 medium with 12% fetal calf serum. Penicillin and streptomycin were added to the medium for transfer of fibroblasts. Following culture, cells were disaggregated using 0.2% trypsin for microscopy, then centrifuged to form solid clusters. Clusters were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer followed by secondary fixation in osmium tetroxide. Tissues were dehydrated in graded ethanol, transferred to an intermediate reagent, propylene oxide, and then infiltrated and embedded in Agar 100 epoxy resin. Polymerization was undertaken at 60C for 48 hours. Ninety (90) nm ultrathin sections were cut using a Diatome diamond knife on a Leica Ultracut UCT microtome. Sections were transferred to copper grids and stained with alcoholic uranyl acetate and Reynolds lead citrate. The fibroblasts were examined using a JEOL 1400 transmission electron microscope.

Ethics
Ethical approval for genetic research was obtained by each center separately as follows: Great Ormond Street (Family 7): approved by the London Bloomsbury Research Ethics Committee (ref: 13 /LO/0168); Generalized dystonia cohort including Families 1-6: all subjects provided written informed consent, and the study protocol was approved by the institutional ethics review boards at the Technical University of Munich, Medical

Weighted Burden Analysis and Case Identification
The weighted burden analysis of 138 individuals with etiologically unresolved generalized dystonia (Supplementary  Table S1) identified a single study-wide significant signal, VPS16, with a Fisher's exact test p value of 6.9 × 10 -9 (Fig 1, Supplementary Table S2). In addition to 5 heterozygous LOF alleles uncovered in the burden test (carrier rate of 3.6% in case subjects), we found one individual with a VPS16-encompassing microdeletion in the cohort. Through international collaboration, an additional 13 cases with heterozygous LOF variants in VPS16 were identified. All 19 patients (from 14 families) had VPS16 variants predicted to result in haploinsufficiency (Table 1,  Supplementary Table S3). One proband (Patient 14) had a second, non-truncating VPS16 variant but phasing of the variants could not be established as parental samples were unavailable. For the other 18 probands, detailed genomic analysis did not identify a second potentially pathogenic VPS16 variant. Moreover, 9 individuals from 5 multigenerational families (Families 3, 7, 8, 9, and 13) confirmed a clearly dominant pattern of disease inheritance (Fig 2). Segregation analysis was possible in 9 families: of these, de novo occurrence was confirmed in 1 family; inheritance from a symptomatic parent was found in 4 families; and inheritance from an apparently nonmanifesting parent in 4 families, indicating incomplete penetrance.

Clinical Features of Patients With VPS16
Affected individuals presented with a progressive early onset dystonia (median age 12 years, range 3-50 years), with prominent oromandibular, bulbar, cervical, and upper limb involvement (Fig 3A, Tables 2 and 3). Progressive generalization ensued in most cases, although most remained ambulant, and only a minority (16%) lost the ability to walk in adulthood (Supplementary Videos S1-S3). Additional clinical features of mild to moderate intellectual disability and neuropsychiatric symptoms were present in approximately one-third of patients, and 50% of families had a positive family history of dystonia (Supplementary Table S4). A degree of interfamilial and intrafamilial phenotypic variability was evident, both with regard to age of symptom onset and dystonia severity. Routine diagnostic testing was unremarkable. In 4 individuals, magnetic resonance imaging (MRI) showed bilateral and symmetrical hypointensity of the globi pallidi and sometimes also the midbrain and dentate nuclei on MRI sequences known to demonstrate susceptibility (T2-weighted, T2*-weighted, and susceptibility-weighted datasets), suggestive of iron deposition. 10 Mild generalized cerebral atrophy was also apparent in 4 individuals. Although not grossly abnormal, caudate nuclei and putamina appeared relatively small and bright on T2 (Fig 3B,   . A Fisher's exact test was used to determine differences in the carrier rate of qualifying variants between cases and controls. Qualifying variants were defined as stop-gain, frameshift, and essential splice-site variants with a minor allele frequency of < 0.0005, whereas exome-wide significance was set to a p value of < 1.25 × 10 −6 (Bonferroni-corrected threshold, see Methods). We observed a significant mutational burden (minimal genomic inflation) for VPS16, in which 5 individuals with dystonia had a qualifying variant.   November 2020 871 residues (p.Ile129_Lys150del; Fig 4). This patient presented in infancy with global developmental delay and generalized dystonia. He attained a few words of speech and voluntary limb movements but never sat unsupported. He had pale optic discs and an axonal neuropathy. From 6 years of age, his condition began to deteriorate, with reduced motor abilities and alertness. An MRI of the brain showed atrophy of the superior cerebellar vermis and slimming of the posterior limb of the corpus callosum (Fig 4).

Electron Microscopy
Electron microscopy (EM) was performed on patientderived fibroblasts and peripheral lymphocytes from patients with both VPS16 and VPS41 mutations in order to determine the impact on lysosomal and vacuolar morphology. When compared to age-matched controls, VPS16 patient cells (fibroblasts n = 6; and lymphocytes n = 4) contained increased clusters of vacuoles, with some containing inclusions in the form of particulate or laminated material (Fig 5, see Supplementary Fig S1). EM analysis of patientderived fibroblasts and lymphocytes from the patient with VPS41 showed numerous membrane-bound vacuoles containing granular material and, in some cases, fine electrondense laminated strands. A large number of small pinocytic vesicles arising from the plasma membrane were also seen. In both VPS16-related and VPS41-related disease, the EM changes seen in patient-derived tissue were consistent with lysosomal dysfunction.

Discussion
We report a cohort of 20 individuals with mutations in 2 related genes, VPS16 and VPS41, which encode vacuolar protein sorting-associated proteins 16 and 41, respectively, both key components of the HOPS complex. The HOPS complex mediates autophagosome-lysosome and endosomelysosome fusion through several different interactions with SNARE proteins, including catalyzing the formation of the SNARE complex 11 and protection of the trans-SNARE complex from disassembly once formed (Fig 6). 12 Our observation of incomplete penetrance in VPS16-related disease (a common hindrance to gene discovery) suggests that, like most other genetic dystonias, 13 additional genetic, epigenetic, and/or environmental factors are likely to play an important role in disease manifestation. Indeed, weighted burden analysis suggests a wider role for VPS16 in conferring genetic susceptibility in a broader group of patients with dystonia. Although adolescent-onset dystonia has been reported in a single family harboring a homozygous missense mutation in VPS16, 14 our data suggest that VPS16 haploinsufficiency (dominant inheritance with incomplete penetrance) is a much more common genetic mechanism for VPS16-related disease.
Autosomal dominant VPS16-related disease appears to be an early-onset, progressively generalizing dystonia, which may occur in isolation or in combination with neuropsychiatric and neurodevelopmental features. In this study, it clinically resembles other "classic" genetic dystonias, such as those related to KMT2B or TOR1A: indeed, at least 1 family (Family 7) in our study had initially been referred for KMT2B testing. Radiologically, too, there is a degree of overlap with KMT2B disease, with basal ganglia hypointensity seen on T2 (and other related MRI sequences) in a proportion of patients 15 : whether this reflects a common pathophysiological mechanistic end point remains to be determined. Our series does not identify any therapeutic option as reliably beneficial for all patients with VPS16-related disease but it is notable that some patients did derive significant benefit from deep brain stimulation (DBS), a treatment that has also proved very useful for both TOR1A and KMT2B-affected patients. Three patients also reported some degree of levodopa responsivity, which, although far from conclusive, may be worth pursuing for mutation-positive patients.
There are clear differences between VPS16-related and VPS41-related disease, although both involve subunits of the HOPS complex and manifest with dystonia as a prominent symptom. Whereas the cases of VPS16-related dystonia we report involve monoallelic variants, predicted to cause haploinsufficiency, the child with VPS41-related disease has biallelic LOF mutations. He also has a correspondingly more profound phenotype, with very early onset of symptoms (presentation in infancy compared to the VPS16 patient cohort, median age of presentation 12 years), severe neurodevelopmental impairment, and evidence of clinical deterioration from during childhood. The differing MRI findings (cerebellar vermis atrophy in VPS41 vs subtle basal ganglia changes in VPS16) also suggest some divergence of pathophysiological pathways. Corroborating our findings, we note that a paper not yet published but recently deposited with bioRxiv describes   an additional family where 2 siblings with homozygous missense variants in VPS41 were affected by dystonia and ataxia, with similar MRI findings to our proband, and lysosomal abnormalities in patient-derived fibroblasts. 16 Thus, our study further supports the emerging role of biallelic LOF VPS41 mutations in early-onset movement disorders.
The microscopic vacuolar changes we observed in both VPS16 and VPS41-patient-derived cells are consistent with lysosomal dysfunction. Vacuolar changes have also been observed in fibroblasts from patients with mucopolysaccharidosis-plus syndrome due to biallelic variants in VPS33A, another subunit of the HOPS complex. 17 These observations are in keeping with in vitro studies on human cell lines, where depletion of both VPS16 18 and VPS41 19 have been separately shown to impair endosomallysosomal fusion. Furthermore, the accumulation of vacuoles has been observed in Drosophila pigment cells in a dVps16A knockdown model 20 and yeast cells expressing mutant vps41 protein are reported to contain many small membrane-bound compartments. 21 It has been suggested that VPS41, through its contribution to autophagocytosis, plays a role in suppression of neurodegenerative processes, especially those mediated by toxic accumulation of aberrant proteins: overexpression of human VPS41 has been shown to be protective in Caenorhabditis elegans models of both Parkinson's 22 and Alzheimer's diseases. 23 Although other components of the HOPS complex have been reported in human disease (specifically VPS33A in mucopolysaccharidosis-plus syndrome 17 and VPS11 in hypomyelinating leukodystrophy type 12) 24 unlike VPS16 and VPS41, none have been associated with dystonia phenotypes, thereby identifying a new pathway in dystonia pathogenesis. We postulate that impairment of endosomal-lysosomal fusion may hinder key cellular processes within core neural networks governing motor control (see Fig 6). Overall, our study provides compelling evidence for the role of VPS16 and VPS41 in the physiological control of movement, mediated through its role in the HOPS complex and lysosomal function.  European Fund for Regional Development from the European Union (01492947), and the province of Friesland, Dystonia Medical Research Foundation, from Stichting Wetenschapsfonds Dystonie Vereniging, from Fonds Psychische Gezondheid, from Phelps Stichting, and an unrestricted grants from Actelion and AOP Orphan Pharmaceuticals AG.