Genetics of Movement Disorders in the Next-Generation Sequencing Era

Genetics of Movement Disorders In the Next-Generation Sequencing Era

Simone Olgiati The work described in this thesis was performed at the Department of Clinical Genetics, Erasmus MC, Rotterdam, The Netherlands.

The studies presented in this thesis were financially supported by: Erasmus MC, The Netherlands; Stichting ParkinsonFonds, The Netherlands; BGI-Shenzen, China; The Netherlands Organization for Scientific Research (NWO, VIDI Grant, project n. 91786395).

The production costs of this thesis were entirely supported by:

ISBN: 978-94-6299-316-7 Cover: Simone Olgiati Layout: Simone Olgiati Printing: Ridderprint BV, Ridderkerk, The Netherlands

Copyright © S. Olgiati, 2016 All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means, without prior permission of the author. The copyright of the published papers remains with the publisher

2 Genetics of Movement Disorders in the Next-Generation Sequencing Era

Genetica van bewegingsstoornissen in het next-generation sequencing tijdperk

Thesis

to obtain the degree of Doctor from the Erasmus University Rotterdam by command of the rector magnificus

Prof.dr. H.A.P. Pols

and in accordance with the decision of the Doctorate Board.

The public defense shall be held on Tuesday 12 April 2016 at 13:30 hrs

Simone Olgiati born in Busto Arsizio, Italy Doctoral Committee

Promotors: Prof.dr. V. Bonifati Prof.dr. R.M.W. Hofstra

Other members: Prof.dr. R. Krüger Prof.dr. C.C.W. Klaver Prof.dr. J.C. van Swieten

4 Table of contents

List of abbreviations 6

Part I – Introduction and scope of the thesis Chapter 1.1 Introduction 11 Scope of the thesis 31

Part II – Parkinson’s disease Chapter 2.1 DNAJC6 mutations associated with early-onset Parkinson’s disease 41 Chapter 2.2 Early-onset parkinsonism caused by alpha-synuclein gene triplication: 67 clinical and genetic findings in a novel family Chapter 2.3 An exome study of Parkinson’s disease in Sardinia, a Mediterranean 79 genetic isolate

Part III – Atypical parkinsonism Chapter 3.1 Mutation in the SYNJ1 gene associated with autosomal recessive, 105 early-onset parkinsonism Chapter 3.2 PARK20 caused by SYNJ1 homozygous Arg258Gln mutation in 127 a new Italian family Chapter 3.3 A new Turkish family with homozygous FBXO7 truncating mutation 141 and juvenile atypical parkinsonism

Part IV – Other movement disorders Chapter 4.1 Paroxysmal exercise-induced dystonia within the phenotypic 155 spectrum of ECHS1 deficiency Chapter 4.2 A recurrent mutation in the C19orf12 gene causes MPAN in Turkey 177

Part V – General discussion Chapter 5.1 General discussion 191

Appendix Summary, Samenvatting, Riassunto 204 Curriculum vitae 210 List of publications 211 PhD portfolio 213 Acknowledgments 215

5 List of abbreviations ACMG American College of Medical Genetics and Genomics AD Autosomal dominant AR Autosomal recessive BAC Bacterial artificial chromosome BFMRS Burke-Fahn-Marsden dystonia rating scale bp Base pairs BPAN Beta-propeller protein-associated neurodegeneration CCV Clathrin-coated vesicle Chr Chromosome CME Clathrin-mediated endocytosis CNV Copy number variant Comp Het Compound heterozygous CoPAN CoA synthase protein-associated neurodegeneration DBS Deep brain stimulation DNA Deoxyribonucleic acid dNTP Deoxynucleotide triphosphate ERM Ezrin/radixin/moesin EVS NHLBI Exome Variant Server ExAC Exome Aggregation Consortium FAB Frontal assessment battery FISH Fluorescence in situ hybridization FLAIR Fluid-attenuated inversion recovery GC Guanine-cytosine GoNL Genome of The Netherlands GRCh Genome Reference Consortium human build GWAS Genome-wide association studies Het Heterozygous Hom Homozygous ICC Immunocytochemistry Indels Insertions and deletions KO Knock-out LD Linkage disequilibrium LOD Logarithm of odds MAF Minor allele frequency MLPA Multiplex ligation-dependent probe amplification MMSE Mini-mental state examination MPAN Mitochondrial-membrane protein associated neurodegeneration MP-RAGE Magnetization prepared rapid acquisition gradient-echo MRI Magnetic resonance imaging NBIA Neurodegeneration with brain iron accumulation

6 NGS Next-generation sequencing OR Odds ratio PBS Phosphate-buffered saline PCR Polymerase chain reaction PD Parkinson’s disease PED Paroxysmal exercise-induced dyskinesia PET Positron emission tomography PKD Paroxysmal kinesigenic dyskinesia PxDs Paroxysmal dyskinesias qPCR Real-time quantitative PCR RFU Relative fluorescence unit RNA Ribonucleic acid RR Relative risk SNP Single nucleotide polymorphism SNV Single nucleotide variant SPECT Single photon emission computed tomography STR Short tandem repeat UPDRS Unified Parkinson’s disease rating scale UTR Untranslated region VUS Variant of unknown significance WES Whole-exome sequencing WGS Whole-genome sequencing XD X-linked dominant XR X-linked recessive

PART I Introduction and Scope of the Thesis 10 1.1 Introduction and Scope of the Thesis

Adapted from: “Genetics of Movement Disorders in the Next-Generation Sequencing Era” S. Olgiati,1 M. Quadri,1 and V. Bonifati.1,# Published in Movement Disorders 2016, doi: 10.1002/mds.26521

# Corresponding author.

1) Department of Clinical Genetics, Erasmus MC, Rotterdam, The Netherlands.

11 Chapter 1.1

Movement Disorders Movement disorders are defined as neurologic syndromes in which there is either an excess of movements or a paucity of voluntary and automatic movements, unrelated to weakness or spasticity [1]. This group of disorders comprises neurological conditions that are clinically, pathologically, and genetically heterogeneous, but they all present with defective control, planning, or execution of movements. The frequency of these syndromes in the general population is very diverse but, overall, movement disorders are common neurological problems, with the most common conditions being restless legs, essential tremor, and Parkinson’s disease [2-4]. The majority of the movement disorders are associated with dysfunctions of the basal ganglia, a group of gray matter nuclei with a primary function in the control of movements, but having also cognitive, behavioral, and emotional functions. Exceptions to this general concept are disorders of movement due to pathologies of the cerebellum (ataxia, asynergia, and dysmetria). The role of basal ganglia and cerebellum is to orchestrate the adequate control of movements, based on internal and external stimuli. Tradi- tionally, only disorders of the basal ganglia and/or cerebellum with clear consciousness are listed as movement disorders. However, disorders of movement can also arise from pathologies of cerebral cortex, brainstem, and spinal cord (several forms of myoclonus and tremor) [5,6]. Movement disorders can be classified into syndromes with too little movements or syndromes with an excess of movements. The first group, the syndromes with too little movements, are commonly referred to as hypokinesias (decrease amplitude of movements), but the terms bradykinesia (slowness of movements) and akinesia (loss of movements) have also been used to describe this category. The most frequent hypokinetic syndromes are the parkinsonisms. The second group, the syndromes with an excess of movements, are referred to as hyperkinesias (excessive movements), but the term dyskinesia (abnormal movements) has been used interchangeably [6]. Many different neurological conditions are listed in this category, including tremor, dystonia, chorea, myoclonus, and tics.

Genetics of Movement Disorders in the Next-Generation Sequencing Era During the past few years, the advent of innovative and extremely powerful methods for sequencing nucleic acids (DNA and RNA) has produced a major boost in our understanding of the role of genes and genetic variability in biology and pathology, and is offering novel research and clinical opportunities for all fields of medicine, including

12 Introduction and Scope of the Thesis the movement disorders. These new methods, collectively called Next-Generation Sequencing (NGS) technologies, provide a way to rapidly generate an unprecedented amount of DNA or RNA sequences at very low costs. As a consequence, new research 1.1 and diagnostic approaches have been made available for the exploration of the genetic bases of diseases. As a result of these novel approaches, the pace of discovery of novel disease-causing or disease-predisposing genes is accelerating (Table 1). Furthermore, the phenotypic spectra associated with previously-known disease-genes are expanding. For instance, ATP1A3 mutations, initially recognized as the cause of rapid-onset dystonia parkinsonism (DYT12), have been more recently identified also as the cause of alternating hemiplegia of childhood, [7-9] and the phenotype of PRRT2 mutations, initially considered to consist of paroxysmal kinesigenic dyskinesias, has been later expanded to also include migraine and epileptic syndromes [10-12]. Last, the NGS technologies provide novel tools for rapid, cheap, and comprehensive genetic testing in the clinical setting. There are great expectations about the opportunities offered by these novel technologies. However, their use in both research and clinical settings comes with certain challenges and limitations that should also be taken into account. Here, we provide an update on the current DNA sequencing technologies, focusing on their possible applications for gene-finding in both research and diagnostic settings in the field of the movement disorders.

The Landscape of NGS Technologies The term NGS refers to a number of different technologies that are rapidly overcoming the traditional way of performing DNA and RNA sequencing (the so-called Sanger sequencing method, or first-generation sequencing) [13]. These novel technologies, also called massively parallel sequencing or second-generation sequencing, have a broad range of applications in disparate areas of biology and medicine. These include: I) sequencing of individual DNA species (single fragments, one gene, gene panels, up to entire genomes) for variants discovery by comparison to reference (normal) known sequences; this process is also termed resequencing, and represents the main application of NGS in human and medical genetics; II) de novo assembly of long DNA sequences (up to entire genomes) without the comparison to other, previously determined sequences; III) analysis of single-cell genomes, and of somatic DNA variation in normal tissues and cancer; IV) evolutionary genetics and species classification [14]; V) investigation of the entire set of mRNA transcripts (transcriptome) of an organism, a tissue, or a single cell (RNA sequencing) [15]; VI) characterization of the epigenetic modifications of DNA,

13 Chapter 1.1

Table 1. Non exhaustive list of genes associated with movement disorders identified using NGS. Independent Gene Disease Inheritance Reference replication Vilariño-Güell et al [48]; VPS35 Late onset PD (PARK17) AD Yes Zimprich et al [49] DNAJC6 Juvenile atypical parkinsonism (PARK19) AR Yes Edvardson et al [42] Juvenile atypical parkinsonism (PARK20) and Quadri et al [43]; Krebs et SYNJ1 AR Yes severe epilepsy with developmental delay al [44] DNAJC13 Late onset PD (PARK21) AD Pending Vilariño-Güell et al [104] CHCHD2 Late onset PD AD Pending Funayama et al [105] Multiple-System Atrophy COQ2 Multiple system atrophy (MSA1) AR Pending Research Collaboration [38] ATP6AP2 Intellectual disability, epilepsy, and parkinsonism XR Pending Korvatska et al [106] RAB39B Intellectual disability with early-onset PD XD Yes Wilson et al [40] FUS Essential tremor (ETM4) AD Pending Merner et al [107] HTRA2 Essential tremor AD Pending Unal Gulsuner et al [108] TENM4 Essential tremor AD Pending Hor et al [109] SCN4A Essential tremor AD Pending Bergareche et al [110] Neurodegeneration with brain iron WDR45 XD Yes Haack et al [57] accumulation (NBIA5) - BPAN Neurodegeneration with brain iron COASY AR Pending Dusi et al [111] accumulation (NBIA6) - CoPAN PDGFRB Idiopathic basal ganglia calcification (IBGC4) AD Yes Nicolas et al [50] PDGFB Idiopathic basal ganglia calcification (IBGC5) AD Yes Keller et al [51] ISG15 Idiopathic basal ganglia calcification AR Pending Zhang et al [112] XPR1 Idiopathic basal ganglia calcification (IBGC6) AD Pending Legati et al [113] Chen et al [30]; Wang et al PRRT2 Paroxysmal kinesigenic dyskinesia AD Yes [31]; Li et al [32]; Lee Li et al [33] HPCA Generalized dystonia (DYT2) AR Pending Charlesworth et al [45] Lohmann et al [114]; TUBB4A Cranial-cervical dystonia (DYT4) AD Yes Hersheson et al [115] CIZ1 Cranial-cervical dystonia (DYT23) AD Pending Xiao et al [116] ANO3 Cranial-cervical dystonia (DYT24) AD Yes Charlesworth et al [117] GNAL Cranial-cervical dystonia (DYT25) AD Yes Fuchs et al [118] KCTD17 Myoclonus-dystonia (DYT26) AD Pending Mencacci et al [39] COL6A3 Cranial-cervical dystonia (DYT27) AR Pending Zech et al [119] PD, Parkinson’s disease; AD, autosomal dominant; AR, autosomal recessive; XD, X-linked dominant; XR, X-linked recessive; BPAN, beta-propeller protein-associated neurodegeneration; CoPAN, CoA synthase protein- associated neurodegeneration.

14 Introduction and Scope of the Thesis promoter occupancy by transcription factors, and chromatin structure (methyl–seq, DNase–seq, ChIP–seq, and others) [16,17]; and VII) analysis of microbial genomes, for example, characterization of the entire intestinal microbiota (metagenomics) [18]. 1.1 This introduction deals with the NGS applications in the area of DNA resequencing for investigating the genetic bases of disease, with a focus on the monogenic (Mendelian) forms of the movement disorders. Concerning the target regions to be sequenced, one can distinguish whole-genome sequencing (WGS), whole-exome sequencing (WES), and more focused, targeted sequencing [19,20]. WGS is the most comprehensive approach and it is expected to become the method of choice in the near future. Nevertheless, WGS still has limitations. Although prices are continuously dropping, the costs for WGS remain substantial for both sequencing and data storage. Moreover, interpreting the enormous amount of data generated by WGS poses major challenges. As a result, several protocols for target enrichment have been developed, which allow to sequence only specific genomic regions. WES is currently a cheaper and popular method to analyze the exome (the coding part, equal to ~2%, or ~50 millions of bases of the human genome). Targeted resequencing is a good strategy to screen panels of multiple candidate genes, and especially suitable for diagnostic purposes. NGS includes different technologies and each of them is characterized by specific chemistries, sample preparation protocols, and data analysis [21]. Readers interested in the technological aspects are referred to dedicated reviews [21,22]. Moreover, for the purpose of this introduction, we will focus on the more broadly-used, and commercially available platforms, omitting the so-called third-generation sequencing (or single-molecule sequencing) technologies, that have been reviewed elsewhere [23]. Despite their differences in chemistries and other technological aspects, all NGS technologies share a basic conceptual workflow in which an initial high-molecular-weight DNA sample (e.g., human genomic DNA) is shattered into a fragment library, and single-strand molecules are amplified and sequenced in parallel. First, DNA fragments (~150 to hundreds nucleotides) are generated by mechanical or enzymatic methods. In several applications, such as exome sequencing or resequencing of customized gene panels, an additional crucial step involves the enrichment of the target DNA fragments (capturing). Libraries are then obtained and platform-specific adaptors are added to both ends of each fragment. This step allows the fragments to be more easily PCR amplified using just one pair of primers, or to be hybridized to a surface using complementary adaptors. Then, the sequences of the fragments are read cyclically and in parallel using different chemistries that results, in most cases, in fluorescent or electrical signals. Last, this signal is detected by an imaging or a different sensing system coupled with a

15 Chapter 1.1 computer, which allows to simultaneously ascertain the sequences of millions of reads in parallel. The large amount of data generated is then processed using bioinformatic tools. For DNA resequencing, the first step of the bioinformatic analysis is the alignment of the raw reads against the reference sequence (alignment). Second, the aligned reads are compared against the reference in order to obtain a list of genomic variations (variant calling). Finally, all the identified variants are compiled with various descriptive information, such as name of the gene in which the variant is located, the predicted consequence of the variant at protein level, the predicted effect on the protein function, the frequency of the variant in population databases, and others (variant annotation). This process allows to confidently identify and report single nucleotide variants (SNVs) and short insertions and deletions (Indels). Currently, larger genomic rearrangements and copy number variants (CNVs) can also be identified using NGS data by using dedicated software [24]. However, these events are more commonly genotyped using single nucleotide polymorphisms (SNPs) array-based technologies, which are still the method of choice, particularly in the diagnostic laboratories. The currently available NGS platforms differ concerning throughput, read length, read accuracy, read depth, per-base costs, and other aspects [25,26]. Their choice is also determined on the basis of the experimental aims (e.g. de novo assembly or resequencing projects, size of the target genomes, etc.). Of note, every platform has some biases inherent to the specific chemistry or other technical aspects, such as a reduced sequence quality in GC-rich target regions and long stretches of the same nucleotide (homopolymers), variable base quality along the reads, and difficulties in detecting short or long insertions and deletions [25-27]. Such types of complex genomic regions or variations are generally difficult to capture and to sequence, often generating false positive and false negative results. These types of problems might have profound consequences for entire projects, such as failures in the identification of disease-causing mutations using NGS methods. For example, some of the PRRT2 mutations causing paroxysmal kinesigenic dyskinesia were missed by NGS-based approaches [28,29]; however, the first identification of this gene successfully relied on NGS [30-33]. Similarly, one of the SLC30A10 mutations causing the manganese transport disorder with dystonia/parkinsonism was originally missed by NGS, due to the combination of relatively low depth of coverage (the number of independent reads per DNA base), and the mutation being in a very high GC-rich region in the first exon of the gene [34]. Because of strong publication bias, it is difficult to appreciate the real extent of these phenomena. One way to at least reduce the amount of false negative and false positive results is to increase the depth of coverage of the NGS projects which of course leads to higher sequencing costs. Depths of >100-150× were

16 Introduction and Scope of the Thesis often mentioned in recent reports of successful exome research projects.

NGS Applications in Medical Genetic Research 1.1 Genetic variants of interest for medical genetics are conveniently classified according to their population frequency and effect-size [35]. Highly-penetrant variants that cause disease in a Mendelian fashion are usually very rare (minor allele frequency, MAF, far below 1%), and these can be efficiently identified by traditional linkage-based approaches. On the other hand, risk factors of small effect-size (odds ratios <2) are common in the population (>5%) and most of these have been identified by genome- wide association studies in the past 5–10 years. However, there is a remaining category of variants of intermediate frequency and effect-size, that remained out of reach using traditional linkage or genome-wide association approaches. With NGS, we have the unprecedented opportunity to investigate every type of medically-relevant variants. Of course, the experimental design should be chosen based on the type of variants, the frequency of the disease of interest, its mode of inheritance, penetrance, and other features (Figure 1). In this introduction, we focus on the strategies for the identification of highly-penetrant, disease-causing variants, that are of high relevance for the clinical neurologists and neuro-geneticists. Mendelian mutations have been, and can still be successfully identified using linkage mapping and positional cloning (even, sometimes, candidate-gene) approaches [36]. Here, in principle, the power of NGS is not necessary, but it can be of great help. When multiple affected individuals from the same family are available, the genomic region of interest can be narrowed down dramatically by linkage mapping analysis (locus identification), and the few genes located within the locus can be quickly investigated in just one run of a NGS instrument (Figure 1A) [37]. Several disease-causing genes have already been identified with such a combined (linkage + NGS) approach: COQ2 in multiple system atrophy,[38] KCTD17 in autosomal dominant myoclonus-dystonia,[39] and RAB39B in X-linked intellectual disability with early-onset PD,[40] to name just a few. Similarly, in the case of autosomal recessive disorders, NGS can be complemented with homozygosity mapping, which is the traditional linkage method for the study of consanguineous pedigrees based on the analysis of homozygous regions identical-by-descent (Figure 1B) [41]. This combination has been successful when applied to movement disorders, allowing the identification of novel disease-causing genes for juvenile parkinsonism (DNAJC6 and SYNJ1),[42-44] dystonia (HPCA),[45] and many forms of cerebellar ataxia or hereditary spastic paraplegia [46,47]. NGS technologies can also speed up the analysis of candidate disease-causing or disease-predisposing genes, selected on

17 Chapter 1.1

Traditional strategies B NGS not required, but convenient A C Gene A

Gene B Homozygosity mapping Gene C

Linkage mapping Candidate gene ication ntif de s i cu o Novel strategies L NGS required D E Identification of novel disease gene Direct sequencing Double-hit approach of distant relatives (recessive inheritance)

F H

Trio approach Cases vs controls (de novo mutations) large NGS studies

Multiple unrelated patients or families

Figure 1. NGS strategies for gene finding.NGS-based strategies for the identification of novel disease-associated genes. The traditional mapping strategies and the novel NGS-based strategies are highlighted. Black symbols denote affected individuals; subjects to be sequenced using NGS are highlighted in yellow frame. the basis of their known biological or physiological roles (Figure 1C). Of note, this approach has not been very fruitful in the past, at least if compared to the unbiased genome-wide approaches [36]. Besides their utility in complementing the traditional linkage mapping methods, the NGS technologies allow today novel experimental designs that were impossible during the first-generation sequencing era. One of such approaches consists in sequencing the exome or the genome of two distant relatives (e.g., first-cousins) affected by the same autosomal dominant disease, looking for shared candidate variants selected according to given filtering criteria (see below; Figure 1D and Figure 2). This approach represents

18 Introduction and Scope of the Thesis a valid alternative to linkage mapping analysis for studying autosomal dominant traits, with the important advantage that it can be applied also when the number of affected relatives with DNA available is too small for a traditional linkage approach 1.1 [37]. Examples of successes using such direct-filtering strategies have been provided by the identification of VPS35 mutations in PD [48,49] and of PDGFRB and PDGFB mutations in primary familial brain calcifications [50,51]. Another, new experimental design made possible by NGS, is represented by the “double-hit strategy” for studying autosomal recessive traits (Figure 1E). Here the exome or genome of a patient with a suspected autosomal recessive disease is interrogated for all the genes carrying two novel or otherwise likely-pathogenic mutations. This approach is ideally conducted when the genome of both the (unaffected) parents can also be sequenced, providing thereby the possibility to directly filter the genes carrying two mutations that have each been transmitted by a different parent (in trans configuration in the affected offspring), as dictated by the autosomal recessive mode of inheritance. A third novel and important strategy is the search for mutations occurring de novo (i.e., a mutation present in the DNA of the patient but not detectable in the parents). This is undertaken by WES or WGS of family trio(s): the affected offspring and both unaffected parents (Figure 1F) [52]. De novo mutations are present by definition in patients who do not have a positive family history of the disease (sporadic presentation), and therefore cannot be studied by family-based linkage approaches [53]. The hypothesis of disease caused by de novo mutations is especially valuable when studying early-onset phenotypes that reduce the reproductive fitness of the affected individual, including for example, several severe neuropsychiatric phenotypes, such as autism spectrum disorders, intellectual disabilities, and schizophrenia. Indeed, recent large-scale NGS studies using the “trio” approach have been successful in the identification of de novo mutations in these disorders [52,54]. A similar approach yielded the identification of a de novo ADCY5 mutation as the cause of a movement disorder presenting with early-onset chorea and dystonia [55]. NGS can also be used to directly interrogate the exome or the genome of series of gene- alogically-unrelated patients, searching for variants occurring in the same gene (Figure 1G). This approach might be extremely powerful in finding the disease-causing gene when the phenotype of interest is very rare, because in this situation the underlying hypothesis is that only one (or very few) causative gene(s) are involved (limited genetic heterogeneity). In this scenario, it has been shown that even very few patients might be enough, because the number of genes carrying novel deleterious mutations, shared by unrelated patients is very small [56]. This strategy was instrumental for the identification of WDR45 as the disease-causing gene in a novel X-linked dominant form of

19 Chapter 1.1

70’000–90’000 WES variants

I) Rare in databases Va ~2’000–5’000

riants reduction (MAF<0.5%) ~500–700 II) Exonic or splicing

~300–500 III) Nonsynonymous

~110 IV) Predicted to be pathogenic

V) Mode of inheritance VI) Experimental strategies (see figure 1) VII) Functional studies

Disease-causing variant

Figure 2. Prioritization of NGS candidate variants. Variant prioritization progressively reduces the number of candidate variants identified by NGS. Schematic representation of a general filtering workflow for NGS variants. Of note, the number of candidate variants can be further reduced by filtering for variants that are totally absent from databases (instead of those that are present but rare, e.g. MAF <0.5%). neurodegeneration with brain iron accumulation [57]. The genetics of complex diseases is not the focus of this introduction. However, it is worth mentioning that the continuous drop of WES and WGS costs will make large- scale NGS case-control studies a reality in the near future (Figure 1H). This will offer the unprecedented opportunity to comprehensively explore the etiological role of every type of genetic variants (very rare, intermediate, and common) [58,59]. Our group recently reported the results of a first effort using this type of approach to identify novel genetic determinants of PD in an isolated population (Sardinia) [60]. Beside the high costs, large case-control exome or genome studies need novel statistical methods that consider the single gene, rather than the single variant, as the target for association with disease (“burden tests”) [61,62]. In addition to these tests, the study power can be further increased by differentially weighting distinct classes of genetic variants or testing larger collections of variants across functionally-related genes (pathways-based analysis)

20 Introduction and Scope of the Thesis

[62]. Similar gene- and pathway-based approaches have recently led to the identification of TUBA4A and TBK1 mutations in amyotrophic lateral sclerosis/fronto-temporal dementia [63-65] and 18 novel candidate genes for hereditary spastic paraplegia [66]. 1.1

Prioritizing Variants and Evaluating Pathogenicity The high throughput of NGS technologies is their main advantage but, at the same time, it creates technical challenges. The vast amount of data generated requires dedicated hardware, software, and trained personnel. More importantly, the interpretation of the large number of identified variants remains the biggest challenge in the NGS era. When searching for a highly-penetrant disease-causing mutation, a typical NGS approach usually identifies a large number of candidate variants. A substantial amount of benign polymorphisms can be excluded from the list of disease-causing candidates using the filters related to the experimental designs described previously (e.g., variants shared by affected cousins). However, this is usually not enough, and dozens of candidate variants remain, requiring additional filtering strategies (Figure 2). The common polymorphic variants that are frequently observed in the general population are usually removed first. This fundamental step relies on large repositories of NGS data from thousands of individuals originating from different human populations. A list of the currently available reference databases is reported in Table 2. Moreover, data from large cohorts of individuals sequenced in-house are often used for the same purposes. There are some important issues to consider during this step. First, the frequency of most genetic variants (both benign polymorphisms and disease-related variants) differs between ethnicities and some variants are exclusive of certain populations. Of note, individuals of different ethnic groups are included in several publicly available databases. Therefore, it is essential that controls used to remove disease-unrelated variants are ethnically-matched to the cases in the best way possible. Second, it is crucial to consider that some public databases intentionally include known pathogenic variants, which are labeled with clinical flags. More importantly, there are substantial chances that very large databases (tens of thousands of subjects) include also pathogenic alleles present in the general population at very low, but detectable frequencies. This fact can create problems particularly if NGS data are filtered in the search for autosomal recessive mutations, or for dominant disease-causing variants with reduced penetrance [67]. For all these reasons, the database and the MAF cut-offs must be carefully chosen. In order to facilitate the interpretation of candidate variants in known disease-causing genes, it is possible to exploit several well-curated repositories of known disease-causing variants and locus-specific databases (Table 2). The accuracy and completeness of these

21 Chapter 1.1

Table 2. Publicly-available population and disease-specific databases. Population databases Database of short variants, based on multiple dbSNP http://www.ncbi.nlm.nih.gov/snp submitters Database of structural variants, based on multiple dbVAR http://www.ncbi.nlm.nih.gov/dbvar submitters WGS of ~2,500 individuals from ~25 different 1000 Genomes Project http://browser.1000genomes.org populations WES data from ~6,500 individuals recruited for Exome Variant Server http://evs.gs.washington.edu/EVS several genomic studies A collection of WES data from ~60,000 Exome Aggregation unrelated individuals recruited for several http://exac.broadinstitute.org/ Consortium (ExAC) disease-specific and population genetic studies A collection of single nucleotide variants from KAVIAR ~70,000 individuals. It includes WGS data from http://db.systemsbiology.net/kaviar/ about ~8,000 individuals Disease-specific databases Catalog of human diseases and OMIM http://www.omim.org disease-related genes ClinVar Catalog of human pathogenic variants http://www.ncbi.nlm.nih.gov/clinvar Human Gene Mutation Catalog of published human pathogenic variants http://www.hgmd.org Database (fee required for certain content) Human Genome Variation It includes a list of locus-specific databases for http://www.hgvs.org/locus-specif- Society many genetic disorders ic-mutation-databases Locus-specific database for many genetic Leiden Open Variation disorders. It also includes a useful list of http://www.lovd.nl Database (LOVD) third-parties locus-specific databases Manually curated database of pathogenic and PDmutDB www.molgen.vib-ua.be/PDMutDB/ benign variations at PD-related genes Parkinson’s Disease Manually curated database of pathogenic and http://grenada.lumc.nl/LOVD2/TPI/ Mutation Database benign variations at PD-related genes PDgene Compendium of genetic association results in PD http://www.pdgene.org/ UMD-THAP1 THAP1 gene mutation database http://www.umd.be/THAP1/ CACNA1A CACNA1A gene mutation database http://chromium.lovd.nl/LOVD2 SETX SETX gene mutation database http://www.LOVD.nl/SETX SETX (UCLA) SETX gene mutation database http://149.142.212.78/LOVD/ ATM ATM gene mutation database http://www.LOVD.nl/ATM Human DNA POLG POLG gene mutation database http://tools.niehs.nih.gov/polg Mutation Database WES, whole-exome sequencing; WGS, whole-genome sequencing; PD, Parkinson’s disease

22 Introduction and Scope of the Thesis databases is a function of time, but it is anticipated that these tools will become more and more important in the future. The field of movement disorders would greatly benefit from the existence of well-curated disease-specific variant databases. 1.1 In addition, variants occurring deeply in intronic and in other non-coding genomic regions are usually removed from the candidate list, under the assumption that highly-penetrant, disease causing-mutations should affect the coding region of the genome. Moreover, to determine whether and how a non-coding variant is pathogenic might be very complex and time-consuming. Nevertheless, it is good to remember that there are some risks in removing the non-coding variants. Most of the disease-causing mutations known so far are in coding regions [68]. However, the non-coding regions have rarely been investigated in the era of Sanger sequencing, and this might have biased the figures in favor of disease-causing coding alleles. Additional strategies can be used to further prioritize the remaining candidate variants. One robust approach is the analysis of the evolutionary conservation of the nucleotide or amino-acid targeted by the variant (Table 3). The underlying principle is that novel deleterious variants tend to be removed by natural selection, while neutral variants are maintained in the genetic pool of the population during evolution. Therefore, highly conserved nucleotides (and amino acids) are usually very relevant for the gene (or protein) function [69]. Another common approach is based on the prediction of the functional damage caused by the mutated amino acid using in-silico tools (Table 3). A first group of these software relies on defined biological properties of deleterious mutations (often conservation) to classify variants in discrete classes of pathogenicity (first-principles approaches). A second group consists of classifiers that are trained to distinguish between a set of known deleterious mutations and a set of benign polymorphisms, using many potentially-relevant properties (trained classifiers). Both type of approaches have pros and cons, described elsewhere [70]. Last, the variants located in proximity to exon-intron boundaries (splice sites) should always be assessed using in-silico splicing prediction tools (Table 3) [71]. Overall, due to the incomplete reliability of both protein prediction (accuracy 60 to 80%)[72] and splicing prediction tools (accuracy 75 to 95%),[73] the results of these in-silico analyses should always be evaluated with caution. The results of multiple tools, preferably based on different methods, should be taken into account. Methods based on the functional predictions are particularly useful when looking for disease-causing variants acting by a loss of function. This is typically the case for recessive alleles, but also dominant alleles acting through haplo-insufficiency or dominant-negative effects.

23 Chapter 1.1

Table 3. A list of in-silico prediction tools. Nucleotide conservation analysis GERP++ http://mendel.stanford.edu/sidowlab/downloads/gerp/ PhastCons http://compgen.bscb.cornell.edu/phast/ PhyloP http://compgen.bscb.cornell.edu/phast/ SCONE http://genetics.bwh.harvard.edu/scone/ VISTA http://genome.lbl.gov/vista/ Damage prediction based on evolutionary conservation SIFT http://sift-dna.org FATHMM http://fathmm.biocompute.org.uk MutationAssessor http://mutationassessor.org PROVEAN http://provean.jcvi.org/ ConSurf http://consurftest.tau.ac.il PANTHER http://www.pantherdb.org/ PhD-SNP http://snps.biofold.org/phd-snp/phd-snp.html Damage prediction based on protein structure SNPs&GO http://snps-and-go.biocomp.unibo.it/snps-and-go Damage prediction based on protein structure and evolutionary conservation PolyPhen-2 http://genetics.bwh.harvard.edu/pph2 MutationTaster http://www.mutationtaster.org MutPred http://mutpred.mutdb.org MAPP http://mendel.stanford.edu/SidowLab/downloads/MAPP/ Align GVGD http://agvgd.iarc.fr/agvgd_input.php nsSNPAnalyzer http://snpanalyzer.uthsc.edu PMUT http://mmb2.pcb.ub.es/PMut/ SAPRED http://sapred.cbi.pku.edu.cn/ SNAP http://www.rostlab.org/services/SNAP/ SNPs3D http://www.snps3d.org/ Damage prediction based on other/mixed methods Condel http://bg.upf.edu/fannsdb/ CADD http://cadd.gs.washington.edu Splicing-prediction tools GeneSplicer http://www.cbcb.umd.edu/software/GeneSplicer/gene_spl.shtml Human Splicing Finder http://www.umd.be/HSF/ MaxEntScan http://genes.mit.edu/burgelab/maxent/Xmaxentscan_scoreseq.html NetGene2 http://www.cbs.dtu.dk/services/NetGene2 NNSplice http://www.fruitfly.org/seq_tools/splice.html (Continues)

24 Introduction and Scope of the Thesis

(Follows)

Splicing-prediction tools http://www.softberry.com/berry.phtml? FSPLICE 1.1 topic=fsplice&group=programs&subgroup=gfind GENSCAN http://genes.mit.edu/GENSCAN.html SpliceView http://bioinfo4.itb.cnr.it/~webgene/wwwspliceview.html SplicePredictor http://bioservices.usd.edu/splicepredictor/ ASSEDA http://splice.uwo.ca/ SplicePort http://spliceport.cbcb.umd.edu/ CRYP-SKIP http://cryp-skip.img.cas.cz/ SROOGLE http://sroogle.tau.ac.il/ Spliceman http://fairbrother.biomed.brown.edu/spliceman/ GERP, Genomic Evolutionary Rate Profiling; SCONE, Sequence CONservation Evaluation; SIFT, Sorting In- tolerant from Tolerant; FATHMM, Functional Analysis through Hidden Markov Models; PROVEAN, Protein Variation Effect Analyzer; PANTHER, Protein ANalysis THrough Evolutionary Relationships; MAPP, Multi- variate Analysis of Protein Polymorphism; PMUT, Pathogenic mutation prediction; SAPRED, Single Amino acid Polymorphisms disease-association PREdictor SNAP, screening for non-acceptable polymorphisms; Condel, CONsensus DELeteriousness score; CADD, Combined Annotation Dependent Depletion; ASSEDA, Automated Splice Site and Exon Definition Analyses; SROOGLE, Splicing RegulatiOn Online GraphicaL Engine.

Instead, in case of mutations with gain-of-function (usually dominant), the altered amino acids might introduce novel deleterious protein functions, and therefore they might not be predicted as deleterious by these tools. Here, filtering variants based on functional prediction is more risky. Some examples of highly-penetrant disease-causing mutations that are predicted as benign by the majority of the in-silico tools include the SNCA (PARK1) p.Ala53Thr, the PRKRA (DYT16) p.Pro222Leu, and the TARDBP p.Ala382Thr. All the previously mentioned methods reduce the number of candidate disease-causing variants. Additional criteria are important for conclusively claim the pathogenicity of mutations in novel genes. These include: independent genetic evidence (identification of independent patients/families with similar phenotype, carrying the same or other mutations in the same gene); biological plausibility (e.g., gene/protein should be expressed in the organ targeted by the disease; gene/protein might have a recognized role in a pathway already known to be involved in the disease); and functional studies (cell systems, model organisms). These traditional concepts remain and are even more valuable in the NGS era. Despite the use of all the above-mentioned approaches, some of the variants identified by NGS are of more difficult interpretation. For these variants, also termed variants of unknown significance (VUS), the available evidence is not conclusive to claim pathoge-

25 Chapter 1.1 nicity nor to exclude pathogenicity. Clear and universally valid guidelines on how to deal with this type of variants in the clinical practice are not available yet. Functional studies using patient-derived material, in-vitro systems, or animal models, can help in interpreting the role of these variants. Such approaches could evaluate expression, local- ization, and functionality of the protein(s). Very often, however, functional assays are not available simply because the role of the investigated protein-product is also unknown or unclear. The neurologist should keep in mind that the results of the functional tests might not always be conclusive. In these scenarios, the clinicians should carefully explain to the patients that the interpretation of these findings is currently unclear or inconclusive. Moreover, and very importantly, the patients should also be informed that the significance of the variant(s) might change in the future, as our clinical and genetic knowledge evolves over time.

Applications of NGS in the clinical practice NGS technologies have recently entered the genetic diagnostic field. Many laboratories are implementing diagnostic NGS methods, or they are offering NGS solutions (such as diagnostic gene-panels, or diagnostic WES, or even WGS). Of course, these new NGS-based methods must adhere to the existing professional and regulatory quality standards required for the genetic testing in diagnostic laboratories [74,75]. Further- more, the diagnostic application of NGS technologies raises novel important issues, among which, how to deal with incidental findings and long-term data storage for future reanalyses. Incidental findings are those that are unrelated to the disease that has generated the request for diagnostic sequencing, but that are nonetheless of medical value for the patient. This is an important and complex issue when dealing with WGS and WES. Local or country-specific regulations might apply,[76] and guidelines have been proposed by the relevant genetic organizations [77-79]. According to these, the clinician ordering the analyses must be sufficiently trained and must inform the patients about the risk of incidental findings. For example, according to the American College of Medical Genetics and Genomics (ACMG), the diagnostic laboratory should seek and report to the clinician certain variants in a minimum set of 56 genes associated with 24 disorders that are considered clinically actionable, regardless of the age of the patient. Of note, no genes related to movement disorders are currently included in this list. In addition to the ACMG, also the European Society of Human Genetics and the Canadian College of Medical Geneticists have released their own recommendations about management of incidental findings [80,81].

26 Introduction and Scope of the Thesis

The storage and management of WGS and WES data for the purpose of future reanalyses (e.g., VUS, disease-associated genes that are yet-to-be discovered) is another relevant issue. In theory, NGS data can be periodically reanalyzed to identify pathogenic variants 1.1 in novel genes, and/or re-assess the role of variants previously considered as VUS, but no clear recommendations for this reanalyses have been established. In addition to that, performing this work in a systematic way appears to be an untenable task for the diagnostic laboratories. According to the ACMG, the primary care physician would be responsible to inform the patient about the necessity to re-contact the clinical geneticist, who should discuss with the patient the possibility of reanalysis [75,82]. Although a debate about the clinical application of WGS or WES is ongoing, much of the current diagnostic work makes use of targeted resequencing panels [83]. These methods avoid the problem of incidental findings and are particularly cost- and time-effective when dealing with diseases associated with a high degree of allelic and locus heterogeneity (i.e., that might be caused by many types of mutations in different genes). This is the case of several movement disorders, but only very few studies have focused so far on the application of NGS technologies for diagnostic purposes. Because of the high number of causative genes (>40), the use of resequencing panels would be well suited for the diagnosis of cerebellar ataxias patients [84]. In 2010, a first pilot study reported the feasibility of this approach by screening seven ataxia-causing genes in five patients with known mutations [85]. The results showed that the identification of causative variants was possible, but the depth of coverage was crucial to achieve enough sensitivity. Recently, another study investigated 118 genes (including established and candidate disease-causing genes) in a cohort of 50 familial or early-onset patients with ataxia, in whom mutations in the most common ataxia-causing genes had been previously excluded. An overall diagnostic rate of 18% was achieved, which is comparable to the diagnostic rate obtained by the analysis of the know ataxia-causing genes using WES [86,87]. Similar results have been reported also for the hereditary spastic paraplegias, and for PD [88,89]. Collectively, these studies show that targeted resequencing using panels of genes is a good compromise between reliability and throughput [90]. Nevertheless, NGS has limitations in detecting complex mutations, such as repeat expansions or large insertions and deletions. Therefore, appropriate techniques should be used prior NGS for screening patients affected by such type of genetic conditions (e.g., ataxias due to DNA repeat expansions). For these reasons, and for the relatively-low costs, NGS-based methods represent a valid alternative to traditional gene testing by Sanger sequencing. However, Sanger is still considered the reference “gold standard” for the confirmation of NGS findings,

27 Chapter 1.1 because of its higher accuracy. Because NGS tests are expected to become much cheaper than other laboratory tests (e.g., imaging), their application in the near future might become more widespread, and the overall role of genetic testing in the clinical practice should increase. Targeted resequencing (WES in the short future and WGS later on) could become an important component in the differential diagnosis and personalized medicine.

New trends in data sharing and phenotyping The pace of discovery of novel disease-related genes has been substantially increased by the advent of WES and WGS. This revolutionizing wave of gene discovery holds the promise to solve the genetic basis of all human Mendelian phenotypes within a few years (an estimated 50% of these conditions still lack a cause). However, although NGS has increased the success rate of cloning projects, finding a novel disease-causing gene is still a challenging endeavor (current success rates are estimated at ~25–30%) [91]. An important and emerging issue is that the disease-genes that still remain to be discovered represent rare or extremely rare causes of the disease. Finding a second confirma- tory family that carries the same or a different mutation in a given gene might be very difficult. In this scenario, sharing of unpublished information between research groups or organizations is becoming a key factor. To boost this process, web-based platforms such as GeneMatcher[92] or DECIPHER[93] have been created. These types of tools are already playing a fundamental role in the current research, but they are expected to gain even more value in the near future. Of note, when sharing NGS individual data, guaranteeing the anonymity and protecting the individual privacy are crucial issues that remain to be fully addressed, and currently hampers the effective data sharing between researchers [94,95]. The heterogeneous clinical description and the existence of overlapping phenotypes is another critical issue that is hampering the identification of novel disease-causing genes, particularly in the case of rare disorders. The use of univocal medical terms is crucial for the rapid recognition, description, and ascertainment of patients worldwide. Recently, there have been efforts to harmonize the phenotypic description of Mendelian disorders by creating a list of standard terms to describe human phenotypes [96]. Moreover, several software have been developed to assist in the generation of medical information in a standardize way, as PhenoTips [97] and PhenoDB [92]. These tools should provide assistance in clustering the patients with the same phenotypes and in understanding the disorder pleiotropy, and they will consequently facilitate the identification of novel disease-causing genes [98].

28 Introduction and Scope of the Thesis

Table 4. Challenges of NGS that should be considered by the movement disorders specialist.

NGS pitfall Action/Solution

Incidental findings. Refer to country-specific regulations. Targeted 1.1 resequencing panels might be a preferable NGS choice. Variants of unknown significance. Consider functional studies, if possible.

Variants interpretation might change over time. Consider periodic reanalysis of the data.

Missing mutations due to regions with low depth of Increase the depth of coverage, and monitor the quality coverage. of the coverage (particularly in the first exons of genes, and in high guanine-cytosine-rich regions). Some type of mutations are currently not well detected Different laboratory techniques should be used to or undetectable (copy number variants, long insertions/ investigate such type of mutations. deletions, structural variants, aneuploidy).

A final word of caution There are great expectations about the opportunities offered by the NGS technologies. However, their use in both research and clinical settings comes with certain challenges and limitations that should be taken into account (Table 4). Despite continuous improvements, the overall accuracy and reproducibility of NGS methods remain sub-optimal, and the concordance between different platforms is still low, especially when analyzing WES or WGS data [99]. Furthermore, the quality of the data provided by different laboratories can be very variable. A recent study compared the performance of 30 different NGS laboratories in the identification and reporting of disease-causing mutations in three families with heritable genetic disorders. Despite the fact that all the laboratories started from the same data, only two of them were able to adequately complete the assigned task for all the families, highlighting the complexities inherent in the current analyses and report of NGS data [100]. Because the interpretation of the data remains a complex matter, genetic findings should be prudently reported. The analysis of WGS and WES data with the approaches described above can help to distinguish disease-causing variants from benign genomic variation, but the evidence is often not sufficient to conclusively declare the pathogenicity of novel variants, and several variants remains labeled as VUS. Erroneous interpretations might have serious consequences for the patient. To prevent such circum- stances, the interpretation and reporting of variants in clinical settings should follow the guidelines of relevant genetic organizations [101]. Besides the direct consequences for a patient, erroneous assignments have a negative impact on the research field: up

29 Chapter 1.1 to 27% of the genetic variants in the scientific literature have been estimated to have a misleading designation [102]. Claims of the identification of a novel disease-causing gene should be considered with caution, until confirmation is provided in independent datasets, or by independent, functional methods. This caveat is particularly important for claims of disease-causality based on findings in single patients or single pedigrees (examples are included in the table 1). To avoid the proliferation of erroneously annotated variants, researchers have set up recommendations to declare variants pathogenicity. According to these recommendations, assignments of variants pathogenicity should include the overall genetic, bioinformatic, and functional evidence, using most of the approaches already described in this chapter [103]. The application of these rules will hopefully help to prevent erroneous assignments in the future. These are exciting times for those interested in the genetics of movement disorders. It is hoped that the application of NGS technologies will result in a fuller understanding of the genetic basis of movement disorders. This knowledge will ultimately provide the rationale for novel clinical and therapeutic interventions.

Funding sources This work was supported by grants from the Stichting ParkinsonFonds (The Netherlands) to V.B.

Author Contributions 1) Research project: A. Conception, B. Organization, C. Execution; 2) Statistical Analysis: A. Design, B. Execution, C. Review and Critique; 3) Manuscript Preparation: A. Writing of the first draft, B. Review and Critique. SO: 1BC, 3A; MQ: 1BC, 3B; VB: 1A, 3B. All the authors approved the final version, and took responsibility for the conduct of the work.

30 Introduction and Scope of the Thesis

Scope of the thesis Movement disorders are a heterogeneous group of neurologic syndromes. In the past, a large number of genetic factors have been associated with Mendelian forms of movement 1.1 disorders. Many of these genetic factors were recently identified using next-generation sequencing (NGS) technologies (chapter 1.1). The scope of this thesis is the study of genetic factors involved in the etiology of movement disorders. We apply a combination of traditional positional cloning methods and novel NGS-based approaches to identify and characterize the genetic determinants of movement disorders. In part II of this thesis, we investigate the genetic factors associated with Parkinson’s disease (PD). For the first time, we identify pathogenic DNAJC6 mutations in families with early-onset PD (chapter 2.1). Our work delineate a novel form of autosomal recessive early-onset PD, caused by mutations in the DNAJC6 gene, and has important implications for the diagnostic work-up of early-onset PD patients. In chapter 2.2, we report a novel family with triplication of the SNCA locus, a very rare form of early-onset parkinsonism. Finally, in chapter 2.3, we report a large exome sequencing study of PD in the isolated population of Sardinia. This work identifies a list of novel candidate genetic determinants of PD, a resource for replication in follow-up studies. In part III, we use family-based strategies to study atypical forms of parkinsonism. First, by studying two siblings affected by early-onset atypical parkinsonism, we identify a novel form of parkinsonism caused by a homozygous mutation in the SYNJ1 gene (chapter 3.1). Then, in chapter 3.2, we report an additional family (the third known so far) with a milder clinical phenotype caused by the same SYNJ1 homozygous mutation and we suggest that the two mutations originated from different mutational events. Last, in chapter 3.3, we report a family with juvenile-onset atypical parkinsonism due to a homozygous mutation in the FBXO7 gene, contributing to the delineation of the genetic and clinical features of this rare form of parkinsonism. In part IV of this thesis, we focus our attention on movement disorders other than primary parkinsonisms. We identify ECHS1 compound heterozygous mutations as the disease cause in two siblings with different dystonic disorders. This work shows that the broad phenotype associated with ECHS1 mutations can also include isolated paroxysmal exercise-induced dystonia (chapter 4.1). Moreover, we identify a recurrent C19orf12 mutation in Turkish patients with neurodegeneration with brain iron accumulation, with implications for the diagnosis and genetic counseling of the patients with Turkish ancestry (chapter 4.2). Finally, in the last chapter (chapter 5.1), we discuss the results of this thesis in a broader

31 Chapter 1.1 context and examine how genetics studies are contributing to reshape our knowledge of the genetic and clinical aspects of movement disorders.

References 1 Fahn S. Classification of movement disorders. Mov kinesigenic dyskinesia and infantile convulsions. Disord. 2011;26:947-57. Neurology. 2012;79:777-84. 2 Rothdach AJ, Trenkwalder C, Haberstock J, Keil 12 Cloarec R, Bruneau N, Rudolf G, Massacrier A, U, Berger K. Prevalence and risk factors of RLS in Salmi M, Bataillard M, et al. PRRT2 links infantile an elderly population: the MEMO study. Memory convulsions and paroxysmal dyskinesia with migraine. and Morbidity in Augsburg Elderly. Neurology. Neurology. 2012;79:2097-103. 2000;54:1064-8. 13 Sanger F, Nicklen S, Coulson AR. DNA sequencing 3 Louis ED, Ottman R, Hauser WA. How common is with chain-terminating inhibitors. Proc Natl Acad Sci the most common adult movement disorder? estimates U S A. 1977;74:5463-7. of the prevalence of essential tremor throughout the 14 McCormack JE, Hird SM, Zellmer AJ, Carstens world. Mov Disord. 1998;13:5-10. BC, Brumfield RT. Applications of next-generation 4 de Lau LM, Breteler MM. Epidemiology of Parkin- sequencing to phylogeography and phylogenetics. son’s disease. Lancet Neurol. 2006;5:525-35. Mol Phylogenet Evol. 2013;66:526-38. 5 Wolters E, Baumann C. Parkinson disease and other 15 Wang Z, Gerstein M, Snyder M. RNA-Seq: a revo- movement disorders: motor behavioral disorders and lutionary tool for transcriptomics. Nat Rev Genet. behavioral motor disorders. Amsterdam: VU Univer- 2009;10:57-63. sity Press; 2014. 16 Tsompana M, Buck MJ. Chromatin accessibility: a 6 Fahn S, Jankovic J, Hallet M. Principles and Practice window into the genome. Epigenetics Chromatin. of Movement Disorders. 2nd ed. Philadephia: Elsevier; 2014;7:33. 2011. 17 Park PJ. ChIP-seq: advantages and challenges of a 7 de Carvalho Aguiar P, Sweadner KJ, Penniston JT, maturing technology. Nat Rev Genet. 2009;10:669- Zaremba J, Liu L, Caton M, et al. Mutations in the 80. Na+/K+ -ATPase alpha3 gene ATP1A3 are associated 18 Petrosino JF, Highlander S, Luna RA, Gibbs RA, with rapid-onset dystonia parkinsonism. Neuron. Versalovic J. Metagenomic pyrosequencing and 2004;43:169-75. microbial identification. Clin Chem. 2009;55:856- 8 Rosewich H, Thiele H, Ohlenbusch A, Maschke 66. U, Altmuller J, Frommolt P, et al. Heterozygous 19 Mamanova L, Coffey AJ, Scott CE, Kozarewa I, de-novo mutations in ATP1A3 in patients with Turner EH, Kumar A, et al. Target-enrichment strat- alternating hemiplegia of childhood: a whole-exome egies for next-generation sequencing. Nat Methods. sequencing gene-identification study. Lancet Neurol. 2010;7:111-8. 2012;11:764-73. 20 Morey M, Fernandez-Marmiesse A, Castineiras D, 9 Heinzen EL, Swoboda KJ, Hitomi Y, Gurrieri F, Fraga JM, Couce ML, Cocho JA. A glimpse into past, Nicole S, de Vries B, et al. De novo mutations in present, and future DNA sequencing. Mol Genet ATP1A3 cause alternating hemiplegia of childhood. Metab. 2013;110:3-24. Nat Genet. 2012;44:1030-4. 21 Metzker ML. Sequencing technologies - the next 10 Marini C, Conti V, Mei D, Battaglia D, Lettori D, generation. Nat Rev Genet. 2010;11:31-46. Losito E, et al. PRRT2 mutations in familial infantile 22 Mardis ER. Next-generation DNA sequencing seizures, paroxysmal dyskinesia, and hemiplegic methods. Annu Rev Genomics Hum Genet. migraine. Neurology. 2012;79:2109-14. 2008;9:387-402. 11 van Vliet R, Breedveld G, de Rijk-van Andel J, Brilstra 23 Thompson JF, Milos PM. The properties and appli- E, Verbeek N, Verschuuren-Bemelmans C, et al. cations of single-molecule DNA sequencing. Genome PRRT2 phenotypes and penetrance of paroxysmal Biol. 2011;12:217.

32 Introduction and Scope of the Thesis

24 Valsesia A, Mace A, Jacquemont S, Beckmann JS, 36 Botstein DR, N. Discovering genotypes underlying Kutalik Z. The Growing Importance of CNVs: New human phenotypes: past successes for mendelian Insights for Detection and Clinical Interpretation. disease, future approaches for complex disease. Nat Front Genet. 2013;4:92. Genet. 2003;33:228-37. 1.1 25 Lam HY, Clark MJ, Chen R, Chen R, Natsoulis G, 37 Ott J, Wang J, Leal SM. Genetic linkage analysis in O’Huallachain M, et al. Performance comparison of the age of whole-genome sequencing. Nat Rev Genet. whole-genome sequencing platforms. Nat Biotechnol. 2015;16:275-84. 2012;30:78-82. 38 Multiple-System Atrophy Research C. Mutations 26 Loman NJ, Misra RV, Dallman TJ, Constantinidou in COQ2 in familial and sporadic multiple-system C, Gharbia SE, Wain J, et al. Performance comparison atrophy. N Engl J Med. 2013;369:233-44. of benchtop high-throughput sequencing platforms. 39 Mencacci NE, Rubio-Agusti I, Zdebik A, Asmus F, Nat Biotechnol. 2012;30:434-9. Ludtmann MH, Ryten M, et al. A Missense Mutation 27 Ross MG, Russ C, Costello M, Hollinger A, Lennon in KCTD17 Causes Autosomal Dominant Myoclo- NJ, Hegarty R, et al. Characterizing and measuring nus-Dystonia. Am J Hum Genet. 2015;96:938-47. bias in sequence data. Genome Biol. 2013;14:R51. 40 Wilson GR, Sim JC, McLean C, Giannandrea M, 28 Heron SE, Grinton BE, Kivity S, Afawi Z, Zuberi Galea CA, Riseley JR, et al. Mutations in RAB39B SM, Hughes JN, et al. PRRT2 mutations cause benign cause X-linked intellectual disability and early-onset familial infantile epilepsy and infantile convulsions Parkinson disease with alpha-synuclein pathology. Am with choreoathetosis syndrome. Am J Hum Genet. J Hum Genet. 2014;95:729-35. 2012;90:152-60. 41 Alkuraya FS. The application of next-generation 29 Schubert J, Paravidino R, Becker F, Berger A, Bebek sequencing in the autozygosity mapping of human N, Bianchi A, et al. PRRT2 mutations are the major recessive diseases. Hum Genet. 2013;132:1197-211. cause of benign familial infantile seizures. Hum 42 Edvardson S, Cinnamon Y, Ta-Shma A, Shaag A, Mutat. 2012;33:1439-43. Yim YI, Zenvirt S, et al. A deleterious mutation 30 Chen WJ, Lin Y, Xiong ZQ, Wei W, Ni W, Tan GH, et in DNAJC6 encoding the neuronal-specific clath- al. Exome sequencing identifies truncating mutations rin-uncoating co-chaperone auxilin, is associated with in PRRT2 that cause paroxysmal kinesigenic dyski- juvenile parkinsonism. PLoS One. 2012; 7:e36458. nesia. Nat Genet. 2011;43:1252-5. 43 Quadri M, Fang M, Picillo M, Olgiati S, Breedveld 31 Wang JL, Cao L, Li XH, Hu ZM, Li JD, Zhang GJ, Graafland J, et al. Mutation in the SYNJ1 gene JG, et al. Identification of PRRT2 as the causative associated with autosomal recessive, early-onset gene of paroxysmal kinesigenic dyskinesias. Brain. Parkinsonism. Hum Mutat. 2013;34:1208-15. 2011;134:3493-501. 44 Krebs CE, Karkheiran S, Powell JC, Cao M, Makarov 32 Li J, Zhu X, Wang X, Sun W, Feng B, Du T, et al. V, Darvish H, et al. The Sac1 Domain of SYNJ1 Iden- Targeted genomic sequencing identifies PRRT2 tified Mutated in a Family with Early-Onset Progres- mutations as a cause of paroxysmal kinesigenic sive Parkinsonism with Generalized Seizures. Hum choreoathetosis. J Med Genet. 2012;49:76-8. Mutat. 2013;34:1200-7. 33 Lee HY, Huang Y, Bruneau N, Roll P, Roberson ED, 45 Charlesworth G, Angelova PR, Bartolome-Robledo F, Hermann M, et al. Mutations in the gene PRRT2 Ryten M, Trabzuni D, Stamelou M, et al. Mutations cause paroxysmal kinesigenic dyskinesia with infantile in HPCA cause autosomal-recessive primary isolated convulsions. Cell Rep. 2012;1:2-12. dystonia. Am J Hum Genet. 2015;96:657-65. 34 Quadri M, Federico A, Zhao T, Breedveld GJ, Battisti 46 Smeets CJ, Verbeek DS. Cerebellar ataxia and func- C, Delnooz C, et al. Mutations in SLC30A10 cause tional genomics: Identifying the routes to cere- parkinsonism and dystonia with hypermanganesemia, bellar neurodegeneration. Biochim Biophys Acta. polycythemia, and chronic liver disease. Am J Hum 2014;1842:2030-8. Genet. 2012;90:467-77. 47 Tesson C, Koht J, Stevanin G. Delving into the 35 Manolio TA, Collins FS, Cox NJ, Goldstein complexity of hereditary spastic paraplegias: how DB, Hindorff LA, Hunter DJ, et al. Finding the unexpected phenotypes and inheritance modes missing heritability of complex diseases. Nature. are revolutionizing their nosology. Hum Genet. 2009;461:747-53. 2015;134:511-38.

33 Chapter 1.1

48 Vilarino-Guell C, Wider C, Ross OA, Dachsel JC, Neurogenetics. 2015;16:55-64. Kachergus JM, Lincoln SJ, et al. VPS35 mutations in 61 Bansal V, Libiger O, Torkamani A, Schork NJ. Statis- Parkinson disease. Am J Hum Genet. 2011;89:162-7. tical analysis strategies for association studies involving 49 Zimprich A, Benet-Pages A, Struhal W, Graf E, rare variants. Nat Rev Genet. 2010;11:773-85. Eck SH, Offman MN, et al. A mutation in VPS35, 62 Sham PC, Purcell SM. Statistical power and signifi- encoding a subunit of the retromer complex, causes cance testing in large-scale genetic studies. Nat Rev late-onset Parkinson disease. Am J Hum Genet. Genet. 2014;15:335-46. 2011;89:168-75. 63 Smith BN, Ticozzi N, Fallini C, Gkazi AS, Topp S, 50 Nicolas G, Pottier C, Maltete D, Coutant S, Rovelet- Kenna KP, et al. Exome-wide rare variant analysis Lecrux A, Legallic S, et al. Mutation of the PDGFRB identifies TUBA4A mutations associated with familial gene as a cause of idiopathic basal ganglia calcification. ALS. Neuron. 2014;84:324-31. Neurology. 2013;80:181-7. 64 Cirulli ET, Lasseigne BN, Petrovski S, Sapp PC, 51 Keller A, Westenberger A, Sobrido MJ, Garcia-Mu- Dion PA, Leblond CS, et al. Exome sequencing in rias M, Domingo A, Sears RL, et al. Mutations in the amyotrophic lateral sclerosis identifies risk genes and gene encoding PDGF-B cause brain calcifications in pathways. Science. 2015;347:1436-41. humans and mice. Nat Genet. 2013;45:1077-82. 65 Freischmidt A, Wieland T, Richter B, Ruf W, Schaeffer 52 Veltman JA, Brunner HG. De novo mutations in V, Muller K, et al. Haploinsufficiency of TBK1 causes human genetic disease. Nat Rev Genet. 2012;13:565- familial ALS and fronto-temporal dementia. Nat 75. Neurosci. 2015;18:631-6. 53 Roach JC, Glusman G, Smit AF, Huff CD, Hubley R, 66 Novarino G, Fenstermaker AG, Zaki MS, Hofree M, Shannon PT, et al. Analysis of genetic inheritance in a Silhavy JL, Heiberg AD, et al. Exome sequencing links family quartet by whole-genome sequencing. Science. corticospinal motor neuron disease to common neuro- 2010;328:636-9. degenerative disorders. Science. 2014;343:506-11. 54 Gratten J, Wray NR, Keller MC, Visscher PM. Large- 67 Bamshad MJ, Ng SB, Bigham AW, Tabor HK, Emond scale genomics unveils the genetic architecture of MJ, Nickerson DA, et al. Exome sequencing as a tool psychiatric disorders. Nat Neurosci. 2014;17:782-90. for Mendelian disease gene discovery. Nat Rev Genet. 55 Carapito R, Paul N, Untrau M, Le Gentil M, Ott L, 2011;12:745-55. Alsaleh G, et al. A de novo ADCY5 mutation causes 68 Stenson PD, Mort M, Ball EV, Shaw K, Phillips A, early-onset autosomal dominant chorea and dystonia. Cooper DN. The Human Gene Mutation Database: Mov Disord. 2015;30:423-7. building a comprehensive mutation repository for 56 Gilissen C, Hoischen A, Brunner HG, Veltman clinical and molecular genetics, diagnostic testing JA. Disease gene identification strategies for exome and personalized genomic medicine. Hum Genet. sequencing. Eur J Hum Genet. 2012;20:490-7. 2014;133:1-9. 57 Haack TB, Hogarth P, Kruer MC, Gregory A, 69 Cooper GM, Brown CD. Qualifying the relation- Wieland T, Schwarzmayr T, et al. Exome sequencing ship between sequence conservation and molecular reveals de novo WDR45 mutations causing a pheno- function. Genome Res. 2008;18:201-5. typically distinct, X-linked dominant form of NBIA. 70 Cooper GM, Shendure J. Needles in stacks of needles: Am J Hum Genet. 2012;91:1144-9. finding disease-causal variants in a wealth of genomic 58 Price AL, Kryukov GV, de Bakker PI, Purcell SM, data. Nat Rev Genet. 2011;12:628-40. Staples J, Wei LJ, et al. Pooled association tests for 71 Jian X, Boerwinkle E, Liu X. In silico tools for splicing rare variants in exon-resequencing studies. Am J Hum defect prediction: a survey from the viewpoint of end Genet. 2010;86:832-8. users. Genet Med. 2014;16:497-503. 59 Cirulli ET, Goldstein DB. Uncovering the roles of rare 72 Thusberg J, Olatubosun A, Vihinen M. Performance variants in common disease through whole-genome of mutation pathogenicity prediction methods on sequencing. Nat Rev Genet. 2010;11:415-25. missense variants. Hum Mutat. 2011;32:358-68. 60 Quadri M, Yang X, Cossu G, Olgiati S, Saddi VM, 73 Houdayer C, Caux-Moncoutier V, Krieger S, Barrois Breedveld GJ, et al. An exome study of Parkinson’s M, Bonnet F, Bourdon V, et al. Guidelines for splicing disease in Sardinia, a Mediterranean genetic isolate. analysis in molecular diagnosis derived from a set of

34 Introduction and Scope of the Thesis

327 combined in silico/in vitro studies on BRCA1 and Schnekenberg R, Becker EB, Bera KD, et al. Next BRCA2 variants. Hum Mutat. 2012;33:1228-38. generation sequencing for molecular diagnosis of 74 Gargis AS, Kalman L, Berry MW, Bick DP, Dimmock neurological disorders using ataxias as a model. Brain. DP, Hambuch T, et al. Assuring the quality of 2013;136:3106-18. 1.1 next-generation sequencing in clinical laboratory 87 Fogel BL, Lee H, Deignan JL, Strom SP, Kantarci practice. Nat Biotechnol. 2012;30:1033-6. S, Wang X, et al. Exome sequencing in the clinical 75 Rehm HL, Bale SJ, Bayrak-Toydemir P, Berg JS, diagnosis of sporadic or familial cerebellar ataxia. Brown KK, Deignan JL, et al. ACMG clinical labora- JAMA Neurol. 2014;71:1237-46. tory standards for next-generation sequencing. Genet 88 Kumar KR, Blair NF, Vandebona H, Liang C, Ng Med. 2013;15:733-47. K, Sharpe DM, et al. Targeted next generation 76 Knoppers BM, Zawati MH, Senecal K. Return of sequencing in SPAST-negative hereditary spastic para- genetic testing results in the era of whole-genome plegia. J Neurol. 2013;260:2516-22. sequencing. Nat Rev Genet. 2015;16:553-9. 89 Petersen MS, Guella I, Bech S, Gustavsson E, 77 Green RC, Berg JS, Grody WW, Kalia SS, Korf BR, Farrer MJ. Parkinson’s disease, genetic variability Martin CL, et al. ACMG recommendations for and the Faroe Islands. Parkinsonism Relat Disord. reporting of incidental findings in clinical exome and 2015;21:75-8. genome sequencing. Genet Med. 2013;15:565-74. 90 Sikkema-Raddatz B, Johansson LF, de Boer EN, 78 Directors ABo. ACMG policy statement: updated Almomani R, Boven LG, van den Berg MP, et al. recommendations regarding analysis and reporting Targeted next-generation sequencing can replace of secondary findings in clinical genome-scale Sanger sequencing in clinical diagnostics. Hum sequencing. Genet Med. 2015;17:68-9. Mutat. 2013;34:1035-42. 79 American College of Medical G, Genomics. Incidental 91 Chong JX, Buckingham KJ, Jhangiani SN, Boehm findings in clinical genomics: a clarification. Genet C, Sobreira N, Smith JD, et al. The Genetic Basis of Med. 2013;15:664-6. Mendelian Phenotypes: Discoveries, Challenges, and Opportunities. Am J Hum Genet. 2015;97:199-215. 80 van El CG, Cornel MC, Borry P, Hastings RJ, Fellmann F, Hodgson SV, et al. Whole-genome 92 Sobreira N, Schiettecatte F, Boehm C, Valle D, sequencing in health care: recommendations of the Hamosh A. New tools for Mendelian disease gene European Society of Human Genetics. Eur J Hum identification: PhenoDB variant analysis module; and Genet. 2013;21:580-4. GeneMatcher, a web-based tool for linking investiga- tors with an interest in the same gene. Hum Mutat. 81 Boycott K, Hartley T, Adam S, Bernier F, Chong 2015;36:425-31. K, Fernandez BA, et al. The clinical application of genome-wide sequencing for monogenic diseases in 93 Bragin E, Chatzimichali EA, Wright CF, Hurles ME, Canada: Position Statement of the Canadian College Firth HV, Bevan AP, et al. DECIPHER: database of Medical Geneticists. J Med Genet. 2015;52:431-7. for the interpretation of phenotype-linked plausibly pathogenic sequence and copy-number variation. 82 Hirschhorn K, Fleisher LD, Godmilow L, Howell Nucleic Acids Res. 2014;42:D993-D1000. RR, Lebel RR, McCabe ER, et al. Duty to re-contact. Genet Med. 1999;1:171-2. 94 Kaye J. The tension between data sharing and the protection of privacy in genomics research. Annu Rev 83 Rehm HL. Disease-targeted sequencing: a cornerstone Genomics Hum Genet. 2012;13:415-31. in the clinic. Nat Rev Genet. 2013;14:295-300. 95 Gymrek M, McGuire AL, Golan D, Halperin E, 84 Hersheson J, Haworth A, Houlden H. The inherited Erlich Y. Identifying personal genomes by surname ataxias: genetic heterogeneity, mutation databases, and inference. Science. 2013;339:321-4. future directions in research and clinical diagnostics. Hum Mutat. 2012;33:1324-32. 96 Kohler S, Doelken SC, Mungall CJ, Bauer S, Firth HV, Bailleul-Forestier I, et al. The Human Phenotype 85 Hoischen A, Gilissen C, Arts P, Wieskamp N, van der Ontology project: linking molecular biology and Vliet W, Vermeer S, et al. Massively parallel sequencing disease through phenotype data. Nucleic Acids Res. of ataxia genes after array-based enrichment. Hum 2014;42:D966-74. Mutat. 2010;31:494-9. 97 Girdea M, Dumitriu S, Fiume M, Bowdin S, Boycott 86 Nemeth AH, Kwasniewska AC, Lise S, Parolin KM, Chenier S, et al. PhenoTips: patient phenotyping

35 Chapter 1.1

software for clinical and research use. Hum Mutat. tremor and Parkinson disease. Proc Natl Acad Sci U S 2013;34:1057-65. A. 2014;111:18285-90. 98 Hennekam RC, Biesecker LG. Next-generation 109 Hor H, Francescatto L, Bartesaghi L, Ortega-Cubero sequencing demands next-generation phenotyping. S, Kousi M, Lorenzo-Betancor O, et al. Missense Hum Mutat. 2012;33:884-6. mutations in TENM4, a regulator of axon guidance 99 O’Rawe J, Jiang T, Sun G, Wu Y, Wang W, Hu J, et al. and central myelination, cause essential tremor. Hum Low concordance of multiple variant-calling pipelines: Mol Genet. 2015;24:5677-86. practical implications for exome and genome 110 Bergareche A, Bednarz M, Sanchez E, Krebs CE, sequencing. Genome Med. 2013;5:28. Ruiz-Martinez J, De La Riva P, et al. SCN4A pore 100 Brownstein CA, Beggs AH, Homer N, Merriman B, mutation pathogenetically contributes to autosomal Yu TW, Flannery KC, et al. An international effort dominant essential tremor and may increase suscepti- towards developing standards for best practices in bility to epilepsy. Hum Mol Genet. 2015;24:7111-20. analysis, interpretation and reporting of clinical 111 Dusi S, Valletta L, Haack TB, Tsuchiya Y, Venco P, genome sequencing results in the CLARITY Pasqualato S, et al. Exome sequence reveals mutations Challenge. Genome Biol. 2014;15:R53. in CoA synthase as a cause of neurodegeneration 101 Richards S, Aziz N, Bale S, Bick D, Das S, Gasti- with brain iron accumulation. Am J Hum Genet. er-Foster J, et al. Standards and guidelines for the 2014;94:11-22. interpretation of sequence variants: a joint consensus 112 Zhang X, Bogunovic D, Payelle-Brogard B, Fran- recommendation of the American College of Medical cois-Newton V, Speer SD, Yuan C, et al. Human Genetics and Genomics and the Association for intracellular ISG15 prevents interferon-alpha/beta Molecular Pathology. Genet Med. 2015;17:405-24. over-amplification and auto-inflammation. Nature. 102 Bell CJ, Dinwiddie DL, Miller NA, Hateley SL, 2015;517:89-93. Ganusova EE, Mudge J, et al. Carrier testing for 113 Legati A, Giovannini D, Nicolas G, Lopez-San- severe childhood recessive diseases by next-generation chez U, Quintans B, Oliveira JR, et al. Mutations sequencing. Sci Transl Med. 2011;3:65ra4. in XPR1 cause primary familial brain calcification 103 MacArthur DG, Manolio TA, Dimmock DP, Rehm associated with altered phosphate export. Nat Genet. HL, Shendure J, Abecasis GR, et al. Guidelines for 2015;47:579-81. investigating causality of sequence variants in human 114 Lohmann K, Wilcox RA, Winkler S, Ramirez A, disease. Nature. 2014;508:469-76. Rakovic A, Park JS, et al. Whispering dysphonia 104 Vilarino-Guell C, Rajput A, Milnerwood AJ, Shah (DYT4 dystonia) is caused by a mutation in the B, Szu-Tu C, Trinh J, et al. DNAJC13 mutations in TUBB4 gene. Ann Neurol. 2013;73:537-45. Parkinson disease. Hum Mol Genet. 2014;23:1794- 115 Hersheson J, Mencacci NE, Davis M, MacDonald N, 801. Trabzuni D, Ryten M, et al. Mutations in the auto- 105 Funayama M, Ohe K, Amo T, Furuya N, Yamaguchi regulatory domain of beta-tubulin 4a cause hereditary J, Saiki S, et al. CHCHD2 mutations in autosomal dystonia. Ann Neurol. 2013;73:546-53. dominant late-onset Parkinson’s disease: a genome- 116 Xiao J, Uitti RJ, Zhao Y, Vemula SR, Perlmutter wide linkage and sequencing study. Lancet Neurol. JS, Wszolek ZK, et al. Mutations in CIZ1 cause 2015;14:274-82. adult onset primary cervical dystonia. Ann Neurol. 106 Korvatska O, Strand NS, Berndt JD, Strovas T, Chen 2012;71:458-69. DH, Leverenz JB, et al. Altered splicing of ATP6AP2 117 Charlesworth G, Plagnol V, Holmstrom KM, Bras causes X-linked parkinsonism with spasticity (XPDS). J, Sheerin UM, Preza E, et al. Mutations in ANO3 Hum Mol Genet. 2013;22:3259-68. cause dominant craniocervical dystonia: ion channel 107 Merner ND, Girard SL, Catoire H, Bourassa CV, implicated in pathogenesis. Am J Hum Genet. Belzil VV, Riviere JB, et al. Exome sequencing identi- 2012;91:1041-50. fies FUS mutations as a cause of essential tremor. Am 118 Fuchs T, Saunders-Pullman R, Masuho I, Luciano J Hum Genet. 2012;91:313-9. MS, Raymond D, Factor S, et al. Mutations in 108 Unal Gulsuner H, Gulsuner S, Mercan FN, Onat GNAL cause primary torsion dystonia. Nat Genet. OE, Walsh T, Shahin H, et al. Mitochondrial serine 2013;45:88-92. protease HTRA2 p.G399S in a kindred with essential 119 Zech M, Lam DD, Francescatto L, Schormair B,

36 Introduction and Scope of the Thesis

Salminen AV, Jochim A, et al. Recessive mutations in the alpha3 (VI) collagen gene COL6A3 cause early-onset isolated dystonia. Am J Hum Genet. 2015;96:883-93. 1.1

PART II Parkinson’s Disease 40 2.1 DNAJC6 Mutations Associated with Early-Onset Parkinson’s Disease

S. Olgiati,*,1 M. Quadri,*,1 M. Fang,*,2 J.P.M.A. Rood,3 J.A. Saute,4 H. Fen Chien,5 C.G. Bouwkamp,1,6 J. Graafland,1 M. Minneboo,1 G.J. Breedveld,1 J. Zhang,2 The International Parkinsonism Genetics Network,† F.W. Verheijen,1 A.J.W. Boon,3 A.J.A. Kievit,1 L.B. Jardim,4,7 W. Mandemakers,1 E.R. Barbosa,5 C.R.M. Rieder,8 K.L. Leenders,9 J. Wang,2,10,11,12,# and V. Bonifati.1,#

Published in Annals of Neurology 2016; 79(2):244-56.

* These authors contributed equally to this work, as first authors; # These authors contributed equally to this work, as senior authors and corresponding authors; † Members of the network are listed in the Supplementary Material.

1) Department of Clinical Genetics, Erasmus MC, Rotterdam, The Netherlands; 2) BGI-Shenzhen, Shenzhen, China; 3) Department of Neurology, Erasmus MC, Rotterdam, The Netherlands; 4) Medical Genetics Service, Hospital de Clínicas de Porto Alegre, Porto Alegre, Brazil; 5) Department of Neurology, University of São Paulo, São Paulo, Brazil; 6) Department of Psychiatry, Erasmus MC, Rotterdam, The Netherlands; 7) Department of Internal Medicine, Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil; 8) Neurology Service, Hospital de Clínicas de Porto Alegre, Porto Alegre, Brazil; 9) Department of Neurology, University Medical Center Groningen, Groningen, The Netherlands; 10) Department of Biology, University of Copenhagen, Copenhagen, Denmark; 11) Princess Al Jawhara Center of Excellence in the Research of Hereditary Disorders, King Abdulaziz University, Jeddah, Saudi Arabia; 12) Macau University of Science and Technology, Macau, China.

41 Chapter 2.1

Abstract Objective DNAJC6 mutations were recently described in two families with autosomal recessive juvenile parkinsonism (onset age < 11), prominent atypical signs, poor or absent response to levodopa, and rapid progression (wheelchair-bound within ~10 years from onset). Here, for the first time, we report DNAJC6 mutations in early-onset Parkinson’s disease (PD). Methods The DNAJC6 open reading frame was analyzed in 274 patients with early- onset sporadic, or familial PD. Selected variants were followed up by cosegregation, homozygosity mapping, linkage analysis, whole-exome sequencing, and protein studies. Results We identified two families with different novel homozygous DNAJC6 mutations segregating with PD. In each family, the DNAJC6 mutation was flanked by long runs of homozygosity within highest linkage peaks. Exome sequencing did not detect additional pathogenic variants within the linkage regions. In both families, patients showed severely decreased steady-state levels of the auxilin protein in fibroblasts. We also identified a sporadic patient carrying two rare non-coding DNAJC6 variants possibly effecting RNA splicing. All these cases fulfilled the criteria for a clinical diagnosis of early-onset PD, had symptoms onset in the third-to-fifth decade, and slow disease progression. Response to dopaminergic therapies was prominent but, in some patients, limited by psychiatric side-effects. The phenotype overlaps that of other monogenic forms of early- onset PD. Interpretation Our findings delineate a novel form of hereditary early-onset PD. Screening of DNAJC6 is warranted in all patients with early-onset PD compatible with autosomal recessive inheritance. Our data provide further evidence for the involvement of synaptic vesicles endocytosis and trafficking in PD pathogenesis.

42 DNAJC6 Mutations Associated with Early-Onset Parkinson’s Disease

Introduction Parkinson’s disease (PD) is a progressive, neurodegenerative disorder manifesting with bradykinesia, resting tremor, muscular rigidity, postural instability, and a good, prolonged response to levodopa and dopaminergic agonists. The ongoing discovery of Mendelian forms of PD provides important clues to understand the disease pathogenesis. Mutations in Parkin, PINK1, and DJ-1 cause autosomal recessive forms of early- 2.1 onset PD (symptoms onset

43 Chapter 2.1 of the disease and have broad implications for understanding the disease pathogenesis, and for improving the diagnostic work-up and the genetic counselling of early-onset PD. Subjects and Methods Subjects Included in the Study We studied 274 unrelated probands with PD, including 92 cases with familial PD compatible with autosomal recessive inheritance (mean onset age: 54.65; standard deviation [SD], 12.34; range 19–84), and 182 patients affected by early-onset sporadic PD (mean onset age: 35.20; SD, 7.98; range 13–50). The diagnosis of clinically definite PD was made according to the criteria of the UK PD Society Brain Bank (with the exception that a positive family history of PD was not considered an exclusion criterion) [16]. The patients were examined and recruited in several movement disorder centers within the International Parkinsonism Genetics Network (members of the network are listed in the Supplementary Material). The patients originated from Italy (n=147), The Netherlands (n=78), Brazil (n=39), Portugal (n=7), Spain (n=2), and Turkey (n=1). Written informed consent was obtained from all the included individuals. This study was approved by appropriate Institutional Review Boards and complied with the legal requirements of the different jurisdictions involved. DNAJC6 Sequencing Genomic DNA was isolated from peripheral blood or saliva using standard protocols. DNA samples underwent Sanger sequencing of the entire DNAJC6 coding region and exons-intron boundaries (MIM 608375). Exons and intron-exon boundaries of the DNAJC6 gene were amplified using PCR (primers are reported in Suppl Table 1). Amplification reactions were performed in a total volume of 20 µl, containing 1× FastStart Taq DNA Polymerase buffer, 200 µM of each dNTP, 0.5 µM forward primer, 0.5 µM reverse primer, 0.5 units FastStart Taq DNA Polymerase (Roche, Basel, Switzerland) and 40 ng of genomic DNA. PCR conditions: 5 min 94°C initial denaturation followed by 30 or 28 cycles of 30 sec 94°C; 30 sec 60°C; 90 sec 72°C with a final extension for 5 min 72°C. PCR reactions were purified using 5 units ExoI and 1 unit Fast AP (Life Technologies, Carlsbad, California), 45 min 37°C, 15 min 80°C. Direct sequencing was performed using Big Dye Termi- nator chemistry ver. 3.1 (Life Technologies) as recommended by the manufacturer. Dye terminators were removed using SephadexG50 (GE Healthcare, Little Chalfont, United Kingdom) and loaded on an ABI 3130XL Genetic Analyzer (Life Technologies). For sequence analysis the software packages Seqscape v2.6 (Life Technologies) and Sequencing Analysis v5.1 (Life Tech- nologies) were used. Variants identified were annotated according to the GenBank accession numbers NM_001256864.1 and NP_001243793.1, corresponding to the longest protein isoform (auxilin isoform 1).

44 DNAJC6 Mutations Associated with Early-Onset Parkinson’s Disease

Additional Genetic Studies in DNAJC6 Variant Carriers Variants were selected for further investigations if they were I) exonic or with a possible effect on splicing (< 100 base pairs [bp] from splice sites), II) rare in 1000 Genomes (allele frequency < 0.005), and III) segregating with PD (if additional family members were available). Patients carrying the selected variants were screened for mutations in genes known to cause autosomal recessive, early-onset PD (parkin, PINK1, and DJ-1) by DNA Sanger sequencing and copy number analysis using multiplex ligation-dependent probe amplification (MLPA) kits P051 and P052 2.1 (MRC Holland, Amsterdam, The Netherlands). Furthermore, in these patients we sequenced DNAJC6 5’UTR (untranslated region), 3’UTR, and promoter regions (1000 bp upstream of the translation initiation codon of NM_014787.3), and we performed DNAJC6 copy number analysis by real-time quantitative PCR (qPCR). Five qPCR assays targeting different exons (2, 5, 10, 15, and 19) of the DNAJC6 gene were carried out. qPCR reactions contained 200nM primer, 1× KAPA SYBR FAST qPCR Master Mix (Kapa Biosystems, Wilmington, Massachu- setts), and 20 ng genomic DNA. All reactions were performed in triplicates on a Bio-Rad CFX96 Real-Time System. Data was analysed using CFX Manager Software v3.0 (Bio-Rad Laboratories, Hercules, California). The assays SEL1L and RBM11 were used as reference targets assuming to detect two chromosomal copies in all the tested samples. PCR conditions: 3 min 94°C initial denaturation followed by 30 cycles of 5 sec 94°C; and 30 sec 60°C with relative fluorescence units (RFU) data collection. qPCR primers are reported in Supplementary Table 2. All assays were validated by testing linearity over 3 orders of magnitude and by observation of a single peak in RFU data collected during a melting curve as a function of temperature. Intensities were compared to at least two control DNA samples from unrelated or unaffected individuals and two genome reference targets without copy number variations. All copy number values were defined within the normal range for relative quantity values between 0.8 and 1.2. Genetic Studies of Families With DNAJC6 Biallelic Mutations We identified two familial and one sporadic PD probands carrying biallelic DNAJC6 variants. To further support the pathogenicity of these variants, all available and consenting affected and unaffected relatives were examined and genomic DNA samples were obtained. The two families with biallelic DNAJC6 variants were investigated using homozygosity mapping, linkage analysis, and whole-exome sequencing. Homozygosity Mapping DNA samples of the two families underwent genome-wide SNP array genotyping. For the GPS-0313 family we used Illumina Infinium CytoSNP-850K BeadChip (Illumina, San Diego, California; 851,274 SNPs at a median distance of 1.8 Kb); the DNA samples of the two parents and the three siblings were included. For the PAL-50 family we employed Illumina HumanOm- niExpress BeadChip (Illumina; 730,525 SNPs at a median distance of 2.1 kb); the DNA samples of four siblings were included. Analysis of homozygosity was performed using Nexus Copy Number, Discovery Edition, v7 (BioDiscovery, El Segundo, California). For the identification of regions of homozygosity shared between affected siblings, the minimum length for homozygous runs was set to 1 Mb.

45 Chapter 2.1

Linkage Analysis SNP array data were used to perform parametric multipoint linkage analysis with the software MERLIN (v1.1.2) [17]. We assumed an autosomal recessive model, with the following settings: markers allele frequencies set to equal; disease allele frequency set to 0.0001; and penetrances set to 0.0001/0.0001/1.0. Both the GPS-0313 and PAL-50 families are of white (Caucasian) ethnicity. In both families, the parents of the patients denied consanguinity within at least two previous generations. However, in both families the parents originate from very small villages in peripheral regions of The Neth- erlands and South Brazil, respectively. Their offspring show several extended (>3Mb) runs of homozygosity, supporting parental consanguinity (Supplementary Table 3). For these reasons, the two families were modeled as consanguineous with the two parents being third cousins. Since the parents of the patients denied consanguinity within at least two previous generations, we considered unlikely a closer consanguineous loop (first or second cousins), and assumed a third- cousins marriage as the closest relatedness. Of note, more distant relatedness between the parents would yield even higher logarithm of odds (LOD) scores at the linkage peak on chromosome 1. Whole-Exome Sequencing We performed whole-exome sequencing of one affected member of each family (GPS-0314 and PAL-54) at BGI Europe. Sequencing was performed using in-solution capturing (Agilent SureSelect Human All Exon V5 50Mb, Agilent Technologies, Santa Clara, CA) and Illumina HiSeq2000, 90 bases paired-end sequencing (Illumina). Reads were aligned to the human reference genome hg19 (build GRCh37.p13) using Burrows-Wheeler Alignment Tool and variants were called using Genome Analysis Toolkit (GATK) [18,19]. Variants were filtered using Cartagenia Bench Lab NGS 4.0 (Agilent Technologies). Genomic regions of interest were previously determined based on linkage and homozygosity mapping. We selected all the variants that were I) within the regions of interest, II) homozygous, III) nonsynonymous exonic or close to splice sites (< 8 bases), and IV) rare in databases (allele frequency lower than 0.01 in 1000 genomes, Exome Variant Server ESP6500SI-V2, and dbSNP141). Additional Linkage and Exome Data Analyses To exclude the presence of additional novel likely-deleterious variants in the rest of the genome, in both the GPS-0313 and PAL-50 families we also performed genome-wide linkage analysis under autosomal recessive inheritance using the parameters set as above, but assuming no parental relatedness. Regions with positive values of LOD scores were then inspected using whole-exome sequencing data (obtained in the GPS-0314 and the PAL-54 patient). We selected all the variants that were: I) within linkage regions; II) nonsynonymous exonic or close to splice sites (< 8 bases); III) absent from public databases (1000 genomes, Exome Variant Server ESP6500SI-V2, dbSNP141, Exome Aggregation Consortium); and IV) homozygous or double heterozygous (heterozygous variants located in genes with multiple heterozygous variants fulfilling criteria I, II, and III).

46 DNAJC6 Mutations Associated with Early-Onset Parkinson’s Disease

In Silico and Structural Analysis The effect of the missense variants identified in the study was investigated using five different protein damage prediction software: Sorting Intolerant From Tolerant (SIFT), PolyPhen-2, likelihood ratio test (LRT), MutationTaster, and MutationAssessor [20-24]. Moreover, variants with a possible effect on splicing were tested using five splicing prediction tools (SpliceSiteFinder- like, MaxEntScan, NNSPLICE, GeneSplicer, and Human Splicing Finder) integrated in Alamut Visual v4.2 (Interactive Biosoftware, Rouen, France) [25-28]. 2.1 The effect of the DNAJC6 missense mutation (p.Arg927Gly) was investigated using publicly-available X-ray crystallographic structures of the bovine auxilin J domain. The sequences of the auxilin protein homologs in Homo sapiens and Bos Taurus were aligned using the software T-Coffee [29]. The homologs were completely identical in the domain under study (J domain). Crystallographic structures of the auxilin J domain, and the complex between auxilin J domain and the ATPase domain of Hsc70 (AMPPNP form) were retrieved from the Protein Data Bank (PDB ID 1NZ6 and 2QWR) [30,31]. A modified structure with the mutant amino acid was generated with the UCSF Chimera package (v1.10.1). Chimera is developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco (supported by NIGMS P41-GM103311) [32]. Molecular graphics and analyses were performed with the Pymol Molecular Graphics System v1.5.0.5 (Schrödinger, New York, New York). Protein Expression Analysis To investigate further the pathogenicity of the DNAJC6 variants identified, we analyzed the expression of the auxilin protein in patient-derived fibroblasts. Primary fibroblasts were obtained from skin biopsies of both the Dutch patients (GPS-0313 and GPS-0314) and one of the two Brazilian patients with novel homozygous DNAJC6 mutations (PAL-54), one unaffected relative of the Dutch family, who was a heterozygous mutation carrier (GPS-0315), and four unrelated unaffected control donors. The expression of auxilin was studied in protein extracts from fibroblasts using Western blotting. Fibroblast cell lines were expanded in growth medium (Dulbecco’s modified Eagle’s medium, 15% fetal calf serum, penicillin, streptavidin), and at 90% confluence lysed in RIPA lysis buffer (150 mM NaCl, 1.0% IGEPAL CA-630, 0.5% sodium deoxycholate, 0.1% SDS, and 50 mM Tris, pH 8.0.) containing protease inhibitor Complete (Roche). Lysates were cleared by spinning 10,000g for 10 min. For Western blotting, proteins were precipitated using the TCA protein precipitation protocol, [33] separated on 4% to 15% Criterion TGX precast gels (Bio-Rad Labo- ratories), and transferred to nitrocellulose using the Trans-Blot® Turbo Transfer System (Bio-Rad Laboratories). Blots were blocked using 5% Protifar (Nutricia, Amsterdam, The Netherlands) in phosphate-buffered saline (PBS), 0.1% (v/v) TWEEN 20 (Sigma-Aldrich, St. Louis, Missouri). Primary antibodies used were: rabbit anti-Auxilin (1:1000; a gift from Dr. Pietro De Camilli, Yale University, New Haven, Connecticut; of note, this antibody is validated for the detection of endogenous auxilin in Western blotting) [34]; rabbit anti-gamma-tubulin (1:2000; Sigma-Al- drich T5192); mouse anti-vinculin (Clone V284; 1:5000; Santa Cruz Biotechnology, Dallas, Texas). After washing in PBS, 0.1% (v/v) TWEEN® 20, blots were incubated with fluorescently conjugated goat anti-mouse (IRDye 800) and goat anti-rabbit (IRDye 680) secondary antibodies

47 Chapter 2.1

(LI-COR Biosciences, Lincoln, Nebraska). After washing in PBS, 0.1% (v/v) TWEEN 20, the blots were imaged and analyzed using an Odyssey Imaging system (LI-COR). Images were analyzed by removing the median background from the total intensity of each band. Auxilin intensities were divided by intensities of the loading control vinculin and normalized using the sum of all data points in each replicate [35]. Four independent biological replicates of each individual were analyzed by signal quantification of multiple Western blots (12 technical replicates in total). The technical replicates belonging to the same biological replicate were averaged and used for the statistical analysis. We performed one-way analysis of variance (ANOVA) using eight groups, one for each subject (four unrelated healthy donors, GPS-0313, GPS-0314, GPS-0315, and PAL-54). Post hoc tests between the control subjects and the remaining individuals were conducted using pair-wise t tests and correcting for multiple comparisons (Bonferroni’s correction). Statistical analyses were conducted in R v3.2.1 (R Foundation for Statistical Computing, Vienna, Austria).

Results DNAJC6 Genetic Studies The results of the DNAJC6 sequencing analysis in the 274 PD probands are displayed in Table 1 and Figure 2C. We identified 14 samples with a total of 15 exonic or possible-splicing variants passing our filtering criteria. To exclude the possibility that any of these variants was in linkage disequilibrium with other variants in the gene, we also screened the entire DNAJC6 5’UTR, 3’UTR, and part of the promoter in these 14 samples, but no additional pathogenic variants were detected. Moreover, qPCR exper- iments could not detect any copy number variants in DNAJC6. Also, none of these 14 samples carried pathogenic variants in parkin, PINK1, and DJ-1. Three probands carriedDNAJC6 homozygous or double heterozygous variants (Fig 1). GPS-0313 and PAL-50, two patients with familial early-onset PD, carried novel homozygous mutations in DNAJC6, while BR-2652, a patient with early-onset sporadic PD, carried two heterozygous variants with possible effect on splicing. The clinical features in these patients are reported in Table 2.

Family GPS-0313 A novel homozygous DNAJC6 c.2779A>G variant (predicted protein effect: p.Ar- g927Gly) was present in GPS-0313, a Dutch proband with early-onset, familial PD compatible with autosomal recessive inheritance (Fig 2A). Sanger sequencing of his affected sister (GPS-0314) revealed the same homozygous mutation, while his parents and one healthy sibling were heterozygous mutation carriers (Fig 1A). Genome wide-linkage analysis revealed a single, nearly-significant peak on chromosome 1 (maximum LOD: 3.07) including the DNAJC6 locus. In this region, homozygosity

48 DNAJC6 Mutations Associated with Early-Onset Parkinson’s Disease N/A N/A N/A rs552001060 rs148204207 rs145329294 N/A rs61757223 N/A N/A N/A N/A rs555883956 N/A N/A rs number Absent 0.000017 Absent Absent 0.000305 0.001274 Absent 0.000281 Absent 0.000045 Absent Absent Absent Absent Absent Allele frequency ExAC 2.1 Absent Absent Absent 0.000400 0.001200 0.002803 Absent 0.000200 Absent Absent Absent Absent 0.004010 Absent Absent Allele frequency 1000 Genomes Het Het Het Het Het Het Het Het Het Het Het Comp Het Comp Het Hom Hom Zygosity c.2634+37A>G c.2108-21A>G c.995+86C>T c.801-84C>A c.513G>A (p.Lys171=) c.1492T>A (p.Cys498Ser) c.961A>G (p.Ile321Val) c.397A>T (p.Met133Leu) c.2517del (p.Phe839Leufs*22) c.1855C>T (p.Arg619Cys) c.626T>C (p.Leu209Pro) c.2038+3A>G c.1468+83del c.2223A>T (p.Thr741=) c.2779A>G (p.Arg927Gly) DNAJC6 variant a 26 2 6 35 16 9 2/11 35 19/15 3 21 24 20/15 27/15 Disease Disease duration (y.rs) a 44 30 64 24 35 42 65/65 27 54/58 33 29 33 42/31 21/29 Age at onset (y.rs) N/A N/A N/A N/A N/A N/A IT59-RM622 N/A IT53-RM603 N/A N/A N/A PAL-54 GPS-0314 Affected ID relative Sporadic Sporadic Familial Sporadic Sporadic Sporadic Familial Sporadic Familial Sporadic Sporadic Sporadic Familial Familial Presentation TOR-6A TH-20 PV-16 COMO-05 SAO-X35 GPS-107B IT59-RM613 GPS-60 IT53-RM590 FI-03 COMO-18 BR-2652 PAL-50 GPS-0313 Proband ID Proband For familial cases, the column reports data of proband/affected relative. data of proband/affected familial cases, the column reports For variants reported in the study. reported 1. PD patients and DNAJC6 variants Table bolded. to be deleterious are mutations and predicted Biallelic a Hom = homozygous; Het = heterozygous; Comp Het = compound heterozygous; N/A = data not available. = compound heterozygous; Comp Het = heterozygous; = homozygous; Het Hom

49 Chapter 2.1

ACB GPS-0313 Family PAL-50 Family BR-2652 Family c.2779A>G (p.Arg927Gly) c.2223A>T (p.Thr741=) c.1468+83del c.2038+3A>G

5 4 66 4 GPS GPS BR BR 0316 0315 2652 2652F M/R M/R c.1468+83del R/M R/R c.2038+3A>G M/R R/R 4 GPS GPS GPS PAL PAL PAL PAL BR 0313 0317 0314 50 55 56 54 2652A M/M M/R M/M M/M R/R M/R M/M c.1468+83del R/M c.2038+3A>G R/R Figure 1. Pedigrees of the PD patients with biallelic DNAJC6 mutations. The genotypes of the available family members are reported. To protect subjects’ privacy, the gender of some individuals has been disguised. Black symbols denote individuals affected by early-onset PD. PD = Parkinson’s disease; M = mutation; R = reference (normal sequence). mapping showed a 10.7-Mb run of homozygosity (from rs6660751 to rs10789400) present in both affected siblings but absent from the unaffected family members (Fig 3A,D). Exome sequencing in one of the affected siblings, GPS-0314, achieved an average coverage of 132×, with 91% of the target regions covered at least 10× (Supple- mentary Table 4). According to our filtering criteria, the p.Arg927Gly was the only likely-pathogenic variant within the linkage region. Also, this variant was absent from all the databases tested, including the Exome Aggregation Consortium (ExAC; 121,412 alleles), and was predicted to be pathogenic by all the prediction software (Supplemen- tary Table 5) [36]. Notably, the variant was also absent in the Genome of The Nether- lands (GoNL) database [37]. These two patients developed the first motor symptoms, bradykinesia and tremor, during the third decade of life. In one of them, GPS-0313, parkinsonism signs developed after six months of neuroleptic treatment, prescribed for psychotic disturbances since the age of 21. Over the years, the motor disorder progressed slowly. Their magnetic resonance imaging (MRI) and F-18 fluorodeoxyglucose/positron emission tomography 18( F-FDG PET) were unremarkable, whereas 18F-DOPA PET imaging revealed nigrostriatal abnormalities compatible with PD. Supplementary Video 1and 2 display these patients after 27 and 16 years of disease evolution. GPS-0313 had more severe parkinsonism and was wheelchair-bound, while, remarkably, his sister had very little postural instability and relatively preserved gait. Of note, both patients responded well to dopaminergic therapies, but the dosages had to be limited to control psychiatric side effects.

50 DNAJC6 Mutations Associated with Early-Onset Parkinson’s Disease

Table 2. Clinical features in PD patients with biallelic DNAJC6 mutations.

Patient code GPS-0313 GPS-0314 PAL-50 PAL-54 BR-2652

Gender male female male female male Age at last 48 44 62 46 57 examination (y.rs) Onset age (y.rs) ~21 29 42 31 33 2.1 Disease duration (y.rs) ~27 15 20 15 24 Symptoms at onset motor slownessa hand tremor motor slowness hand tremor hand tremor hand tremorsa Asymmetric onset - + + + + Signs at examination Hoehn-Yahr stage 5 3 3 3 4 Bradykinesia + + + + + Rest tremor - + + + + Rigidity + + + + + Postural instability + + + + + Levodopa response + + + + + Motor fluctuations - - + + + Levodopa-induced - - + + + dyskinesias Levodopa-induced + + + - - hallucinations Pyramidal signs - - - - - Cerebellar signs - - - - - Autonomic signs - - - - - Cognitive decline - - + - - Seizures - - - - - Brain MRI normal normal normal normal normal Brain 18F-Dopa PET abnormalb abnormalb N/P N/P N/P Brain FDG PET normal normal N/P N/P N/P Surgical therapies none none none STN-DBS pallidotomy + present; - absent; N/P = not performed a parkinsonism symptoms initially emerged after treatment with neuroleptics b severe striatal uptake deficit, particularly at putamen level, as seen in PD PD = Parkinson’s disease; MRI = magnetic resonance imaging; PET = positron emission tomography; FDG = F-18 fluorodeoxyglucose; STN-DBS - treated with deep-brain stimulation of the subthalamic nuclei.

51 Chapter 2.1

A B Arg927 c.2779A>G c.2223A>T p.Arg927Gly p.Thr741= H.sapiens PVGMADLVTPEQVKKVYRKAVLVVHPDKATGQPYE P.troglodytes PVGMADLVTPEQVKKVYRKAVLVVHPDKATGQPYE M.mulatta PVGMADLVTPEQVKKVYRKAVLVVHPDKATGQPYE C.lupus PVGMADLVTPEQVKKVYRRAVLVVHPDKATGQPYE B.taurus PVGMADLVTPEQVKKVYRKAVLVVHPDKATGQPYE M.musculus PVGMADLVTPEQVKKVYRRAVLVVHPDKATGQPYE R.norvegicus PVGMADLVTPEQVKKVYRRAVLVVHPDKATGQPYE G.gallus PVSMADLVTPEQVKKVYRRAVLVVHPDKATGQPYE D.rerio PVGMADLVTPEQVKKVYRKAVLVVHPDKATGQPYE G AAGGTGTA CGGGAAGGCTGT A CCTTGGGACT CTAGGTACAA X.tropicalis PVSMADLVTPEQVKKVYRKAVLVVHPDKATGQPYE

c.397A>T c.1855C>T c.2517del C p.Met133Leu p.Arg619Cys p.Phe839Leufs*22 c.1468+83del c.2038+3A>G c.626T>C c.2223A>T c.2779A>G p.Leu209Pro p.Thr741= p.Arg927Gly DNAJC6 5' 3' Isoform 1 NM_001256864.1 Isoform 2 NM_014787.3 Isoform 3 NM_001256865.1 1kb c.801-2A>G c.2371C>T p.Gln791*

Figure 2. DNAJC6 mutations. (A) Electropherograms of the two DNAJC6 homozygous mutations.(B) Alignment of the auxilin protein homologs in different species (protein accession numbers are reported in Supplementary Table 8). Protein homologs in different species were aligned using the software T-Coffee [29]. The arginine at position 927 is also indicated. (C) Schematic representation of the DNAJC6 transcripts with the mutations previously identified in the two PARK19 families (bottom) and the novel mutations identified here in patients with early-onset PD (top). Mutations detected in homozygous or compound heterozygous state are bolded. PD=Parkinson’s disease.

Family PAL-50 A different, novel DNAJC6 homozygous variant, c.2223A>T (p.Thr741=; Fig 2A), was present in PAL-50, a Brazilian proband with early-onset, familial PD compatible with autosomal recessive inheritance. This variant is located 5 bases before the end of exon 15. Sanger sequencing confirmed the mutation in homozygous state in the affected sister (PAL-54), while two unaffected siblings were heterozygous (PAL-56) or non-carriers (PAL-55; Fig 1B). Also in this family, linkage analysis identified a single, nearly-significant peak on chromosome 1 (LOD score 3.18), corresponding to a 16.1-Mb homozygous run (from rs3118027 to rs6672527; Fig 3B and D). Exome sequencing in PAL-54 achieved an average coverage of 141× with 91 % of the target regions covered at least 10× (Supplementary Table 4). Within this linkage region, exome sequencing revealed 2 additional homozygous variants besides the DNAJC6 c.2223A>T: a c.589A>G (p.Arg197Gly) in L1TD1 (NM_001164835.1) with an allele frequency of 0.003 in 1000 genomes and ExAC, and a novel c.2065G>A (p.Gly689ser) in HOOK1 (NM_015888.4). Both variants were excluded from further investigations because predicted to be benign by all the protein damage prediction tools, and because conservation analysis showed that species other than humans carry the mutated amino acids

52 DNAJC6 Mutations Associated with Early-Onset Parkinson’s Disease

A 100 cM Chromosome 13 15 17 19 21 123456789101112 14 16 18 20 22

C 2.1 LO D 0123 −1

B LO D 0246 LO D chr1 60 80 100 120 0123 Genomic position [cM] −1

DNAJC6 D 1 GPS 0313 0 1 GPS 0314 0 1 GPS 0315 0 1 B-allele frequenc GPS 0316 0 1 GPS 0317 0 1 PAL y 50 0 1 PAL 54 0 1 PAL 55 0 1 PAL 56 0 chr1 54 56 58 60 62 64 66 68 70 72 74 76 Genomic position [Mb] Figure 3. Genome-wide linkage analysis and homozygosity mapping. Genome-wide linkage analysis of the GPS-0313 (A) and PAL-50 (B) families. Values of LOD score (red lines) where calculated for all the autosomes applying a grid of 1 cM. (C) Linkage analysis of the chromosome 1 locus. Red line, GPS-0313 family; blue line, PAL-50 family; black line, cumulative LOD score. Dashed lines mark the LOD score threshold of 3.0. (D) Homozygosity mapping of GPS-0313 and PAL-50 families showing the overlapping regions of homozygosity on chromosome 1. For each available sample of the two families, and for each probe in the region, probe B-allele frequency (vertical axis) is represented as a function of its genomic position (horizontal axis). Dashed line indicates the position of the DNAJC6 locus; red probes and orange areas indicate regions of homozygosity longer that 1 Mb. LOD = logarithm of odds.

53 Chapter 2.1

(Supplementary Table 6). On the contrary, the DNAJC6 c.2223A>T variant was absent from all the available databases, including the ExAC, and three out of the five prediction software suggested a splicing effect: the introduction of an aberrant splicing acceptor site one base after the constitutive donor site, at position 2227+1 (Supplementary Table 7). These two patients developed PD symptoms at the age of 42 and 31, respectively. Their clinical course was typical for PD, including good response to levodopa and severe motor complications (wearing off and dyskinesia). Due to these, PAL-54 also underwent bilateral subthalamic nucleus DBS, with marked improvement. Supplementary Video 3 and 4 display the patients at the age of 62 and 46 years old.

Proband BR-2652 Two DNAJC6 heterozygous variants, a novel c.2038+3A>G and a c.1468+83del with allele frequency of 0.004 in 1000 genomes, were present in BR-2652, a sporadic Brazilian patient with early-onset PD (Fig 1C). Genotyping of available relatives was consistent with the patient being compound heterozygous. Four out of five splicing prediction software suggested that the c.2038+3A>G variant may cause loss of the splicing donor site at position 2038, while the same analysis for the c.1468+83del did not suggest an effect on splicing (Supplementary Table 7). This proband developed PD at the age of 33, with right hand tremor. The parkinsonism was responsive to levodopa, and displayed a prolonged course. He also received left side pallidotomy in 2006 and right side pallidotomy in 2010. Supplementary Video 5 displays the patient at the age of 44, after 11 years of disease.

Additional Linkage and Exome Data Analyses Our linkage analysis assuming no parental relatedness identified positive LOD scores in 33 genomic regions (total 671 Mb, maximum LOD 0.23) in the GPS-0313 family, and 32 regions (total 383 Mb, maximum LOD 0.24) in the PAL-50 family. The analysis of the whole exome sequencing data revealed the DNAJC6 c.2779A>G (p.Arg927Gly) as the only variant surpassing our filtering criteria in GPS-0314. In PAL-54, this analysis revealed only two variants, which were already identified by our homozygosity-based strategy, within the chromosome 1 linkage peak: the DNAJC6 c.2223A>T and the HOOK1 c.2065G>A (p.Gly689ser). Additional novel variants were not detected. Further inspection of the exome data revealed an heterozygous variant in the glucocerebrosidase gene (GBA) (NM_000157.3: c.1088T>C, p.Leu363Pro) in PAL-54, which was present also in PAL-50. GBA variants were not found in the exome data of GPS-0314. Heterozygous GBA variants are known to be risk factors for the development of PD. However, the p.Leu363Pro variant has never been reported in patients with this disease.

54 DNAJC6 Mutations Associated with Early-Onset Parkinson’s Disease

Table 3. Statistical analysis of DNAJC6 expression in subject-derived fibroblasts. GPS-0313 GPS-0314 GPS-0315 PAL-54 Control 1 6.6×10-06 1.3×10-05 0.02702 1.7×10-04 Control 2 3.0×10-04 5.8×10-04 0.40799 0.00715 Control 3 2.3×10-05 4.5×10-05 0.07662 6.0×10-04 Control 4 1.3×10-06 2.5×10-06 0.00600 3.1×10-05 The table displays the p values for pair-wise t test comparisons between controls 2.1 and subjects with DNAJC6 mutations. p values are not corrected for multiple testing. Values surpassing the Bonferroni-corrected threshold (p < 0.0125) are bolded. Expression and Structural Studies of the Auxilin Protein The results of our protein expression studies showed that, in primary fibroblasts, the steady-state levels of auxilin are markedly and significantly lower in patients with homozygous DNAJC6 mutations than in control subjects (ANOVA p= 1.08×10-06, Table 3, Fig 4A,B). The unaffected heterozygous mutation carrier (GPS-0315) had auxilin levels intermediate between those of controls and patients. Furthermore, our in-silico structural studies showed that the Arginine at position p.927 (replaced by Glycine in the patients of the Dutch family) is located in the J domain, which is a known crucial functional domain of auxilin, and it is facing the interacting partner, Hsc70. The mutation has a drastic steric effect and reduces the charge potential of a positively-charged patch at the protein surface, a region possibly involved in the interaction with Hsc70, or with other molecular partners (Fig 4C,D).

Discussion Our study reveals a novel role for DNAJC6 mutations in early-onset PD. Two out of the 92 probands (2.2%) with familial PD compatible with autosomal recessive inheritance carried novel homozygous mutations. Several lines of evidence support the contention that the two homozygous variants identified are disease-causing. First, each of these variants is located within the highest, and nearly-significant linkage peak obtained in the respective family (Fig 3A–C), and also supported by the results of the homozygosity mapping. Second, within these linkage regions, exome sequencing revealed no evidence of disease-causing variants in genes other than DNAJC6. Moreover, novel pathogenic variants were not found by exome sequencing in any other genomic loci detected in the two families under an autosomal recessive mode of inheritance, without assuming parental consanguinity. Third, both variants are absent from all the available databases (more than 120,000 alleles). Fourth, the two variants are predicted to be deleterious by our in-silico analyses. Last, the

55 Chapter 2.1

AB 150

4 KDa AL-5 GPS-0315GPS-0314GPS-0313ControlControl 1 Control 2 Control 3 P 4 100 130 Auxilin 110 50 Auxilin levels 130 Vinculin [% of controls average] 50 0 γ-Tubulin 3 1 4 40 0314 AL-54 S-0315P P ControlControl 2 Control 3 GPS-031GPS- G Control C D Wild type Mutant AAuxiliuxilin J domaindomain N

Arg927 Gly927 C

Hsc70Hsc70 N ATPasePase ddomaiomain Arg927Arg927 C Figure 4. Protein analyses. (A) Auxilin expression in human fibroblasts; a representative Western blot is shown; vinculin and γ-tubulin are included as loading controls. (B) Quantitative analysis of auxilin levels in patients’ fibroblasts, expressed as percentage of the average expression in four unrelated control individuals. All data are from four biological replicates. The plots of the patients with DNAJC6 homozygous mutations are in red; the heterozygous carrier of the Dutch missense mutation (GPS-0315) is in gray; the control subjects are in blue.(C) Crystal structure of the bovine auxilin J domain with visualization of the electrostatic potential surface for the wild type (left) and mutant (right) proteins (PDB accession n. 1NZ6). The replacement of the Arginine-927 with a Glycine modifies a large positively-charged patch on the protein surface, with drastic sterical and electrostatic consequences. The more positively charged regions are displayed in blue, while negatively charged regions are displayed in red; the position of the residue homolog to the human Arginine-927 is indicated. (D) Crystal structure of the bovine auxilin J domain complexed with the Hsc70 ATPase domain and ATP analog (PDB accession n. 2QWR). The residue homolog to the human Arginine-927 is located in the auxilin domain (red) that interacts with Hsc70 (blue). The Arginine-927 points towards Hsc70, suggesting a possible involvement in the interaction between the two proteins. steady-state levels of auxilin are markedly and significantly decreased in fibroblasts of patients from both families (Fig 4A,B), providing compelling experimental evidence for the pathogenicity of the DNAJC6 homozygous mutations. Of note, these results also delineate a loss-of-function mechanism for these mutations, in keeping with the recessive disease inheritance in these families. The c.2779A>G (p.Arg927Gly) mutation is classified as disease causing by all the prediction tools, it replaces an amino acid which has been highly conserved in evolution (Fig 2B), and is located within the J domain. Moreover, the mutation appears to cause a substantial modification of the biochemical properties of the domain (Fig 4C,D). This could affect the protein stability, or binding to Hsc70, or both, and would be in keeping

56 DNAJC6 Mutations Associated with Early-Onset Parkinson’s Disease with the observed decreased levels of auxilin protein in patients’ fibroblasts. The synonymous c.2223A>T is predicted to affect splicing by the majority of the prediction software tested. To investigate the effect of the c.2223A>T, we also undertook DNAJC6 mRNA expression analysis in fibroblasts, but the expression levels in control subjects were unre- liable (data not shown). This is in keeping with the known fact that DNAJC6 mRNA expression is essentially brain-specific and very low in any other tissues (EST profile in NCBI UniGene: Hs.647643) [8,38]. Furthermore, we cannot exclude the possibility 2.1 that this synonymous variant is in linkage disequilibrium with another causative variant located in introns or other non-coding regulatory elements of DNAJC6, which were not detectable by our screening. We also identified a sporadic patient with early-onset PD with two rare, compound heterozygous DNAJC6 variants. Due to its novelty and the results of the splicing prediction software, the c.2038+3A>G variant appear to be likely deleterious. Instead, the c.1468+83del is only moderately rare in databases, and the evidence of its splicing effect is less compelling. Despite being unable to conclusively prove their pathogenicity, the coexistence in the same patient of these two rare variants in trans is intriguing, and suggests a pathogenic nature. Furthermore, as said before, we cannot exclude the possibility that one or both of these variants are in linkage disequilibrium with other causative variants located in introns or other non-coding regulatory elements, which were not detectable by the methods used here. Unfortunately, fibroblasts from this patient were unavailable to perform protein expression studies. We also detected 11 patients with rare heterozygous variants in DNAJC6, but sequencing and gene dosage assays did not reveal additional mutations in these samples. Interestingly, at least four of the heterozygous variants identified (p.Phe839Leufs*22, p.Leu209Pro, p.Met133Leu, and p.Arg619Cys) are either novel or extremely rare in databases, and are predicted to have a deleterious effect for the protein function (Supplementary Table 5), which is dramatic in the case of the frameshift variant (p.Phe839Leufs*22). These variants might be a coincidental finding. However, we cannot exclude that additional variants in trans were missed because they are outside the sequenced regions or undetectable by our assays. Another hypothesis is that single heterozygous variants in DNAJC6 act as risk factors for PD. Similar hypotheses have been proposed for single heterozygous variants in other genes causing autosomal recessive PD [39]. It will be interesting to test this hypothesis for DNAJC6 in future studies. From the clinical standpoint, our patients with homozygous DNAJC6 mutations have a very different phenotype than the PARK19 patients reported earlier. The Palestinian and Turkish patients developed motor symptoms during childhood (before <11 years-old) and were all wheelchair-bound or bedridden within ~10 years from onset. Furthermore,

57 Chapter 2.1 they displayed atypical clinical signs, such as mental retardation, seizures, and dystonia, or they had poor or absent response to levodopa [8,9]. On the contrary, the patients identified in our study developed parkinsonism in their third to fifth decade of life, experienced a much slower disease progression, and they all responded well to dopaminergic therapies, including DBS in one case. These clinical features are compatible with the diagnosis of early-onset PD, and overlap the phenotype of other monogenic forms of PD caused by parkin, PINK1, or DJ-1 mutations. It is possible that the phenotypic variability is due to the fact that the novel mutations identified here might allow some residual activity to the auxilin protein. In other words, our data suggest a possible genotype-phenotype correlation in patients with DNAJC6 mutations, with complete loss-of- function variants causing the juvenile, atypical form (PARK19), while milder mutations (described in our study) cause early-onset PD. The observation of severely reduced but detectable residual levels of auxilin protein in the fibroblasts of our patients fits very well with this hypothesis. The established role of auxilin in the CME is crucial for neurons, that need CME for the formation of new vesicles at the presynaptic terminal and synaptic vesicles recycling [13]. Together with their molecular partner, Hsc70, auxilins are essential to complete the last step of the CME, the uncoating of the vesicles after their separation from the plasma membrane [13,40]. Our findings add to a growing body of evidence for the involvement of synaptic vesicles recycling in PD. First, recessive mutations in SYNJ1, encoding another close molecular partner of auxilin, cause early-onset parkinsonism (PARK20) [10,11]. The knockout mice models of DNAJC6 and SYNJ1 display similar severe neurological phenotypes, with defective synaptic vesicle recycling, and clathrin-coated vesicles accumulation [41,42]. Second, common variants in GAK (encoding auxilin-2, a different ubiquitously expressed form of auxilin) have been identified by GWAS studies as a risk factor for late-onset idiopathic PD [43,44]. Last, the protein products of other genes causing Mendelian PD have been linked to the recycling of synaptic vesicles. The LRRK2 and parkin protein interact with endophilinA [45,46]; the VPS35 protein is part of the retromer complex, that recycles membrane-associated protein to the trans-Golgi network [47]; and α-synuclein has been implicated in several ways to vesicles recycling and release [48,49]. In conclusion, the identification of early-onset PD patients with DNAJC6 mutations changes the current view of DNAJC6, from being the cause of a rare and atypical juvenile parkinsonism, to be the fourth gene involved in autosomal recessive, early-onset PD. Screening of DNAJC6 is therefore warranted in all patients with early-onset PD compatible with autosomal recessive inheritance, with broad implications for the diag-

58 DNAJC6 Mutations Associated with Early-Onset Parkinson’s Disease nostic work-up and genetic counseling of patients. Our data provide further support for disturbed synaptic vesicle endocytosis and trafficking in the pathogenesis of PD.

Acknowledgments This work was supported by research grants from the Stichting ParkinsonFonds (The Netherlands) to AJWB and to VB, and by the Beijing Genomics Institute (BGI). We thank all the patients involved in the study and their families. We also would like to thank Dr. Pietro De Camilli for kindly providing the anti-auxilin antibody. 2.1 Author Contributions SO, MQ, MF, JG, GJB, and JZ acquired, analysed, and interpreted the genetic data. JPMAR, JAS, HFC, CGB, AJWB, AJAK, LBJ, ERB, CRMR, KLL, and members of the International Parkinsonism Genetics Network acquired, analysed, and interpreted the clinical data. MM, FWV, and WM acquired, analysed, and interpreted the protein expression data. JW and VB conceived and designed the study and interpreted the clinical and laboratory data. SO, MQ, JW, and VB drafted the manuscript. All the authors critically revised the article for important intellectual content and approved the version to be published.

Potential Conflicts of Interest Nothing to report.

References 1 Kitada T, Asakawa S, Hattori N, Matsumine H, mutations cause autosomal recessive, early-onset Yamamura Y, Minoshima S, et al. Mutations in the parkinsonian-pyramidal syndrome. Neurology. parkin gene cause autosomal recessive juvenile parkin- 2009;72:240-5. sonism. Nature. 1998;392:605-8. 8 Edvardson S, Cinnamon Y, Ta-Shma A, Shaag A, 2 Valente EM, Abou-Sleiman PM, Caputo V, Muqit Yim YI, Zenvirt S, et al. A deleterious mutation MM, Harvey K, Gispert S, et al. Hereditary early- in DNAJC6 encoding the neuronal-specific clath- onset Parkinson’s disease caused by mutations in rin-uncoating co-chaperone auxilin, is associated with PINK1. Science. 2004;304:1158-60. juvenile parkinsonism. PLoS One. 2012; 7:e36458. 3 Bonifati V, Rizzu P, van Baren MJ, Schaap O, 9 Koroglu C, Baysal L, Cetinkaya M, Karasoy H, Tolun Breedveld GJ, Krieger E, et al. Mutations in the DJ-1 A. DNAJC6 is responsible for juvenile parkinsonism gene associated with autosomal recessive early-onset with phenotypic variability. Parkinsonism Relat parkinsonism. Science. 2003;299:256-9. Disord. 2013;19:320-4. 4 Ramirez A, Heimbach A, Grundemann J, Stiller B, 10 Quadri M, Fang M, Picillo M, Olgiati S, Breedveld Hampshire D, Cid LP, et al. Hereditary parkinsonism GJ, Graafland J, et al. Mutation in the SYNJ1 gene with dementia is caused by mutations in ATP13A2, associated with autosomal recessive, early-onset encoding a lysosomal type 5 P-type ATPase. Nat Parkinsonism. Hum Mutat. 2013;34:1208-15. Genet. 2006;38:1184-91. 11 Krebs CE, Karkheiran S, Powell JC, Cao M, Makarov 5 Paisan-Ruiz C, Bhatia KP, Li A, Hernandez D, Davis V, Darvish H, et al. The Sac1 Domain of SYNJ1 Iden- M, Wood NW, et al. Characterization of PLA2G6 tified Mutated in a Family with Early-Onset Progres- as a locus for dystonia-parkinsonism. Ann Neurol. sive Parkinsonism with Generalized Seizures. Hum 2009;65:19-23. Mutat. 2013;34:1200-7. 6 Shojaee S, Sina F, Banihosseini SS, Kazemi MH, 12 Vauthier V, Jaillard S, Journel H, Dubourg C, Jockers Kalhor R, Shahidi GA, et al. Genome-wide linkage R, Dam J. Homozygous deletion of an 80 kb region analysis of a Parkinsonian-pyramidal syndrome comprising part of DNAJC6 and LEPR genes on pedigree by 500 K SNP arrays. Am J Hum Genet. chromosome 1P31.3 is associated with early onset 2008;82:1375-84. obesity, mental retardation and epilepsy. Mol Genet 7 Di Fonzo A, Dekker MC, Montagna P, Baruzzi Metab. 2012;106:345-50. A, Yonova EH, Correia Guedes L, et al. FBXO7 13 Kononenko NL, Haucke V. Molecular mechanisms of

59 Chapter 2.1

presynaptic membrane retrieval and synaptic vesicle Biol. 1997;4:311-23. reformation. Neuron. 2015;85:484-96. 27 Pertea M, Lin X, Salzberg SL. GeneSplicer: a new 14 Foo JN, Liany H, Tan LC, Au WL, Prakash KM, Liu computational method for splice site prediction. J, et al. DNAJ mutations are rare in Chinese Parkin- Nucleic Acids Res. 2001;29:1185-90. son’s disease patients and controls. Neurobiol Aging. 28 Desmet FO, Hamroun D, Lalande M, Collod-Beroud 2014; DOI:10.1016/j.neurobiolaging.2013.09.018. G, Claustres M, Beroud C. Human Splicing Finder: 15 Jesus S, Gomez-Garre P, Carrillo F, Caceres-Redondo an online bioinformatics tool to predict splicing MT, Huertas-Fernandez I, Bernal-Bernal I, et al. signals. Nucleic Acids Res. 2009;37:e67. Analysis of c.801-2A>G mutation in the DNAJC6 29 Notredame C, Higgins DG, Heringa J. T-Coffee: A gene in Parkinson’s disease in southern Spain. Parkin- novel method for fast and accurate multiple sequence sonism Relat Disord. 2014;20:248-9. alignment. J Mol Biol. 2000;302:205-17. 16 Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy 30 Jiang J, Taylor AB, Prasad K, Ishikawa-Brush Y, Hart of clinical diagnosis of idiopathic Parkinson’s disease: PJ, Lafer EM, et al. Structure-function analysis of the a clinico-pathological study of 100 cases. J Neurol auxilin J-domain reveals an extended Hsc70 interac- Neurosurg Psychiatry. 1992;55:181-4. tion interface. Biochemistry. 2003;42:5748-53. 17 Abecasis GR, Cherny SS, Cookson WO, Cardon LR. 31 Jiang J, Maes EG, Taylor AB, Wang L, Hinck AP, Lafer Merlin--rapid analysis of dense genetic maps using EM, et al. Structural basis of J cochaperone binding sparse gene flow trees. Nat Genet. 2002;30:97-101. and regulation of Hsp70. Mol Cell. 2007;28:422-33. 18 Li H, Durbin R. Fast and accurate short read 32 Pettersen EF, Goddard TD, Huang CC, Couch GS, alignment with Burrows-Wheeler transform. Bioinfor- Greenblatt DM, Meng EC, et al. UCSF Chimera--a matics. 2009;25:1754-60. visualization system for exploratory research and 19 McKenna A, Hanna M, Banks E, Sivachenko A, analysis. J Comput Chem. 2004;25:1605-12. Cibulskis K, Kernytsky A, et al. The Genome Analysis 33 Link AJ, LaBaer J. Trichloroacetic acid (TCA) Toolkit: a MapReduce framework for analyzing precipitation of proteins. Cold Spring Harb Protoc. next-generation DNA sequencing data. Genome Res. 2011;2011:993-4. 2010;20:1297-303. 34 Milosevic I, Giovedi S, Lou X, Raimondi A, Collesi 20 Kumar P, Henikoff S, Ng PC. Predicting the effects of C, Shen H, et al. Recruitment of endophilin to clath- coding non-synonymous variants on protein function rin-coated pit necks is required for efficient vesicle using the SIFT algorithm. Nat Protoc. 2009;4:1073- uncoating after fission. Neuron. 2011;72:587-601. 81. 35 Degasperi A, Birtwistle MR, Volinsky N, Rauch J, 21 Adzhubei IA, Schmidt S, Peshkin L, Ramensky VE, Kolch W, Kholodenko BN. Evaluating strategies to Gerasimova A, Bork P, et al. A method and server normalise biological replicates of Western blot data. for predicting damaging missense mutations. Nat PLoS One. 2014; e87293. Methods. 2010;7:248-9. 36 Exome Aggregation Consortium (ExAC), Cambridge, 22 Chun S, Fay JC. Identification of deleterious MA (URL: http://exac.broadinstitute.org) [accessed mutations within three human genomes. Genome August 31st, 2015]. Res. 2009;19:1553-61. 37 Genome of the Netherlands C. Whole-genome 23 Schwarz JM, Rodelsperger C, Schuelke M, Seelow D. sequence variation, population structure and demo- MutationTaster evaluates disease-causing potential of graphic history of the Dutch population. Nat Genet. sequence alterations. Nat Methods. 2010;7:575-6. 2014;46:818-25. 24 Reva B, Antipin Y, Sander C. Predicting the functional 38 Ishikawa K, Nagase T, Nakajima D, Seki N, Ohira M, impact of protein mutations: application to cancer Miyajima N, et al. Prediction of the coding sequences genomics. Nucleic Acids Res. 2011;39:e118. of unidentified human genes. VIII. 78 new cDNA 25 Yeo G, Burge CB. Maximum entropy modeling of clones from brain which code for large proteins in short sequence motifs with applications to RNA vitro. DNA Res. 1997;4:307-13. splicing signals. J Comput Biol. 2004;11:377-94. 39 Klein C, Lohmann-Hedrich K, Rogaeva E, Schloss- 26 Reese MG, Eeckman FH, Kulp D, Haussler D. macher MG, Lang AE. Deciphering the role of hetero- Improved splice site detection in Genie. J Comput zygous mutations in genes associated with parkin-

60 DNAJC6 Mutations Associated with Early-Onset Parkinson’s Disease

sonism. Lancet Neurol. 2007;6:652-62. 45 Matta S, Van Kolen K, da Cunha R, van den Bogaart 40 Ungewickell E, Ungewickell H, Holstein SE, G, Mandemakers W, Miskiewicz K, et al. LRRK2 Lindner R, Prasad K, Barouch W, et al. Role of controls an EndoA phosphorylation cycle in synaptic auxilin in uncoating clathrin-coated vesicles. Nature. endocytosis. Neuron. 2012;75:1008-21. 1995;378:632-5. 46 Trempe JF, Chen CX, Grenier K, Camacho EM, 41 Yim YI, Sun T, Wu LG, Raimondi A, De Camilli P, Kozlov G, McPherson PS, et al. SH3 domains from a Eisenberg E, et al. Endocytosis and clathrin-uncoating subset of BAR proteins define a Ubl-binding domain defects at synapses of auxilin knockout mice. Proc and implicate parkin in synaptic ubiquitination. Mol Natl Acad Sci U S A. 2010;107:4412-7. Cell. 2009;36:1034-47. 2.1 42 Cremona O, Di Paolo G, Wenk MR, Luthi A, Kim 47 Bonifacino JS, Hurley JH. Retromer. Curr Opin Cell WT, Takei K, et al. Essential role of phosphoinos- Biol. 2008;20:427-36. itide metabolism in synaptic vesicle recycling. Cell. 48 Nemani VM, Lu W, Berge V, Nakamura K, Onoa B, 1999;99:179-88. Lee MK, et al. Increased expression of alpha-synu- 43 Pankratz N, Wilk JB, Latourelle JC, DeStefano AL, clein reduces neurotransmitter release by inhibiting Halter C, Pugh EW, et al. Genomewide association synaptic vesicle reclustering after endocytosis. Neuron. study for susceptibility genes contributing to familial 2010;65:66-79. Parkinson disease. Hum Genet. 2009;124:593-605. 49 Burre J, Sharma M, Tsetsenis T, Buchman V, 44 Nalls MA, Pankratz N, Lill CM, Do CB, Hernandez Etherton MR, Sudhof TC. Alpha-synuclein promotes DG, Saad M, et al. Large-scale meta-analysis of SNARE-complex assembly in vivo and in vitro. genome-wide association data identifies six new risk Science. 2010;329:1663-7. loci for Parkinson’s disease. Nat Genet. 2014;46:989- 93.

Supplementary Materials

Members of The International Parkinsonism Genetics Network V. Bonifati, J.A. Kievit, J.P.M.A. Rood, and A.J.W. Boon, Erasmus MC, Rotterdam, The Netherlands; B. van de Warrenburg, C. Delnooz, A. Rietveld, and B. Bloem, Radboud University Medical Centre, Nijmegen, The Netherlands; J. Ferreira and L. Correia Guedes, Institute of Molecular Medicine, Lisbon, Portugal; E. Tolosa, Hospital Clínic de Barcelona, University of Barcelona, IDIBAPS, Centro de Investigacíon Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED), Catalonia, Spain; M.M. Kurtiz, Hospital Ruber Internacional, Madrid, Spain; J. Obeso, University of Navarra, Pamplona, Spain, and Centro de Investigación en Redes sobre Enfermedades Neurodegenerativas (CIBERNED), Madrid, Spain; M. Emre, H. Hanagasi, B. Bilgic, and Z. Tufekcioglu, Istanbul Faculty of Medicine, Istanbul, Turkey; B. Elibol, Hacettepe University, Ankara, Turkey; J.A. Saute, C. Rieder, and L. Jardim, Hospital de Clínicas de Porto Alegre, Brazil; Hsin F. Chien and Egberto R. Barbosa, University of São Paulo, Brazil; Chin-Song Lu, Yah-Huei Wu-Chou, and Tu-Hsueh Yeh, Chang Gung Memorial Hospital, Taipei, Taiwan; M. Atadzhanov and P. Kelly, University of Zambia, Lusaka, Zambia. Italian members: L. Lopiano, University of Torino; C. Tassorelli, C. Pacchetti, and G. Nappi, IRCCS “Mondino”, Pavia; G. Riboldazzi and G. Bono, Insubria University, Varese; C. Comi, University of Eastern Piedmont “Amedeo Avogadro”, Novara; A. Padovani and B. Borroni, University of Brescia; F. Raudino, “Valduce” Hospital, Como; E. Fincati and L. Bertolasi, University of Verona; M. Tinazzi and A. Bonizzato, Hospital “Borgo Trento”, Verona; C. Ferracci, Hospital of Belluno; A. Dalla Libera, “Boldrini” Hospital, Thiene; P. Marini and F. Massaro, University of Firenze; A. Federico, I. Taglia, and C. Battisti, University of Siena; R. Marconi, “Misericordia” Hospital, Grosseto; M. Onofrj and A. Thomas, University of Chieti-Pescara; N. Vanacore, National Institute of Health, Roma; G. Meco, G. Fabbrini, E. Fabrizio and A. Berardelli, “Sapienza” University, Roma; F. Stocchi, L. Vacca, IRCCS “San Raffaele Pisana”, Roma; M. Picillo, A. De Rosa, C. Criscuolo, G. De Michele, and A. Filla, Federico II University, Naples; P. Barone, CEMAND, University of Salerno; M. De Mari, C. Dell’Aquila, Hospital

61 Chapter 2.1 of Andria; G. Iliceto and P. Lamberti, University of Bari; V. Toni and G. Trianni, Hospital of Casarano; M. Gagliardi, G. Annesi, and A. Quattrone, National Research Council, Institute of Molecular Bioimaging and Physiology, Germaneto, and Institute of Neurology, Department of Medical Sciences, University Magna Graecia, Catanzaro; V. Saddi, Neurology Division, S. Francesco Hospital, Nuoro; G. Cossu, and M. Melis, Hospital S. Michele AOB “G. Brotzu”, Cagliari, Italy.

Supplementary videos are available on the journal website.

Supplementary Table 1. DNAJC6 sequencing primers.

Primer name Sequence Use PCR cycles DNAJC6_exon1A_F CGGTTCCATGCCATGTTTTG PCR 30 DNAJC6_exon1A_R GAAAGCTGAATGACACAATATTTGGAAAC PCR+Sequencing DNAJC6_exon1B_F GTGGCCGCTCCTTCTTTTCC PCR 30 DNAJC6_exon1B_R CCGGCTCTCTCGCAAACC PCR+Sequencing DNAJC6_exon2_F GAGGGACTGACCTGGGTCTCC PCR+Sequencing 28 DNAJC6_exon2_R TGAGCCAGACAAGCAGTCAGC PCR DNAJC6_exon3-4_F GGAGGGGAGTCGAAACATGC PCR+Sequencing 30 DNAJC6_exon3-4_R CACGTGCTTTAAGGGTATCTCTGAGG PCR DNAJC6_exon5_F TCACTTCCAGAAGATGTTGGTTGG PCR+Sequencing 30 DNAJC6_exon5_R CACACAAAAGTCAATGGCATTTTTAGC PCR DNAJC6_exon6_F AATGAAAAGCACCCACAGTTAAGAGG PCR DNAJC6_exon6_R CCAATTCAGTTTTTGCATAGTTCAGG PCR 30 DNAJC6_exon6_seq_R TTGCTCCTCCTTCCCCTTCC PCR+Sequencing DNAJC6_exon7_F CATTGGGGAAGTACAGCAATCTCAG PCR+Sequencing 30 DNAJC6_exon7_R TCATTTGAAAAAGGGCCATAACATC PCR DNAJC6_exon8_F CTGGGCGAACAGTCCTCTGG PCR+Sequencing 30 DNAJC6_exon8_R TGGGGAGAAAAGATTCAGAAATACTCC PCR DNAJC6_exon9_F GGGTTGGAACAGGGCATCC PCR+Sequencing 28 DNAJC6_exon9_R ACAGCCTGGTGCCATTGAGC PCR DNAJC6_exon10_F TGACTTCTGTGCCAAACATCATACC PCR+Sequencing 30 DNAJC6_exon10_R GAGGGATATCTATTGGAGCCTGTCC PCR DNAJC6_exon11_F TGGCCTTAGGAGGTATGAGTCACC PCR DNAJC6_exon11_R GGCCTTTTGCATCAGTCTAAGAGC PCR 30 DNAJC6_exon11_seq_R ACAGGTGGGAGTCCGTGTGC Sequencing DNAJC6_exon12A_F TGTTTGCTCAGGGCTGAATCC PCR+Sequencing 30 DNAJC6_exon12A_R ACCTCCACCCCCAAACAGG PCR DNAJC6_exon12B_F GTCAAGAAGCCCAGCAAAAAGC PCR 30 DNAJC6_exon12B_R AGCTCAAACGTCCACTCTTACACTACC PCR+Sequencing DNAJC6_exon13_F GAAATTCTTGGAGTCCACTGTTAGCC PCR+Sequencing 28 DNAJC6_exon13_R TGTTCTTTGGAATACCAATTTTGTGG PCR DNAJC6_exon14_F TGTCAAGCCTCATGTGAACAACC PCR+Sequencing 30 DNAJC6_exon14_R TCATCAAGTGCATTGGGAAAAGC PCR DNAJC6_exon15_F TGTTCAAGGCCACCATTCACC PCR+Sequencing 30 DNAJC6_exon15_R TTCATGCTACACATCCAGAATTAGGC PCR DNAJC6_exon16_F TTTTGCTCTCCTTCAGTGAAATGC PCR 30 DNAJC6_exon16_R TTCAGGCAGTATGACAACATAGACAGG PCR+Sequencing DNAJC6_exon17_F GCATGAGGGTTAAGGACTGAATGC PCR+Sequencing 30 DNAJC6_exon17_R TGTGTGCAACCTCCTCAGAGC PCR DNAJC6_exon18_F GATTGAGGAGGAGAAGCATTCCAG PCR+Sequencing 28 DNAJC6_exon18_R TTAAAAATAGGCCATATGAACCAGCAC PCR DNAJC6_exon19_F GAAATGGCCAAATAGAAAAACATTTCC PCR 30 DNAJC6_exon19_R TGTTACTGGAGTTTTTGGTTCATCTGC PCR+Sequencing

62 DNAJC6 Mutations Associated with Early-Onset Parkinson’s Disease

Supplementary Table 2. qPCR primers.

DNAJC6 assays Primer name Sequence DNAJC6_exon2_F ACCAGGTACGCACATTCTTCC DNAJC6_exon2_R GAGCCAGACAAGCAGTCAGC DNAJC6_exon5_F CAACCTTTTTGCTGTGTGTCG DNAJC6_exon5_R TGACCAAACCAAACCATCACC 2.1 DNAJC6_exon10-11_F GGTTTTGTTTTTCAGGATGAAGC DNAJC6_exon10-11_R ATGCCTGGGGTTGTCTGG DNAJC6_exon15_F TCCTTTTCAGGAGGCTTTGG DNAJC6_exon15_R GATGGGTGGGAGTACCATGC DNAJC6_exon19_F TCTACAGGCTACTGGGCAACC DNAJC6_exon19_R ATAAGGGCTTTTGGCCTTGG Reference target assays Primer name Sequence SEL1L_F TTAAGGCCTTTTTCTAGCCATCC SEL1L_R ACCAAGTTTCCCACTCTTAAGTCC RBM11_F TCTTCAAAGTGTTTTGTGGTTGG RBM11_R TGAGAAAATTTAACAATGACCAGAAGG

Supplementary Table 3. Homozygous runs in GPS-0313 and PAL-50 families.

Homozygous runs Sample ID Amount Total length (Mb)

GPS-0313 5 46.46

GPS-0314 5 24.39

GPS-0317 5 26.51

PAL-50 4 32.77

PAL-54 7 51.73

PAL-55 3 13.64

PAL-56 1 3.78 Only regions of homozygosity longer than 3 Mb are included. Sex chromosomes are excluded.

63 Chapter 2.1

Supplementary Table 4. Whole-exome sequencing statistics.

Patient ID GPS-0314 PAL-54 Bases in target regions 51,543,125 51,543,125 Reads mapped to genome 103,600,897 109,622,834 Reads uniquely mapped to genome 101,594,992 107,435,979 Reads uniquely mapped to target regions 78,991,916 84,134,111 Bases uniquely mapped to target regions(Mb) 6,679.86 7,138.77 Mean depth of target regions 131.79 140.92 Coverage of target regions (%) 96.81 96.67 Fraction of target regions covered >=4X (%) 94.27 93.96 Fraction of target regions covered >=10X (%) 91 90.54 Fraction of target regions covered >=20X (%) 86.02 85.58 Fraction of flanking regions covered >=4X (%) 50.45 48.15 Fraction of flanking regions covered >=10X (%) 35.91 34.32 Fraction of flanking regions covered >=20X (%) 26.08 25.2 Duplication rate (%) 12.21 12.76 Statistics regarding to depth and coverage in this table are calculated from de-duplicated data. Target region refers to the regions that are covered by the designed probes. Flanking region refers to a 200bp window on both sides of each target region. Duplication rate refers to the fraction of reads that are PCR duplicates, meaning all reads that have the same start and end for both mates.

Supplementary Table 5. Prediction of the protein effects caused by theDNAJC6 missense variants. Polyphen2 Polyphen2 Mutation Sample ID Variant SIFT LRT MutationTaster HDIV HVAR Assessor c.2779A>G Deleterious Probably Damaging Probably Damaging Deleterious Disease Causing High GPS-0313 p.Arg927Gly 0 1.0 1.0 5.16×10-9 1.000 4.17 c.397A>T Deleterious Probably Damaging Probably Damaging Deleterious Disease Causing Medium GPS-60 p.Met133Leu 0.05 0.975 0.979 1.42×10-7 1.000 2.595 c.626T>C Deleterious Probably Damaging Probably Damaging Deleterious Disease Causing High COMO-18 p.Leu209Pro 0 1.0 0.999 4.62×10-7 1.000 3.665 c.961A>G Tolerated Benign Benign Deleterious Disease Causing Neutral IT59-RM613 p.Ile321Val 1 0.025 0.067 2.26×10-9 1.000 0.025 c.1492T>A Tolerated Benign Benign Neutral Polymorphism Neutral GPS-107B p.Cys498Ser 0.69 0.001 0.0 5.61×10-2 1.000 -0.895 c.1855C>T Tolerated Probably Damaging Possibly Damaging Deleterious Disease Causing Low FI-03 p.Arg619Cys 0.06 0.994 0.784 1.77×10-4 1.000 1.39 For each variant, the results of the prediction software are reported, with predicted effect and score. Red, deleterious effects; green, benign effects.

64 DNAJC6 Mutations Associated with Early-Onset Parkinson’s Disease

Supplementary Table 6. Prediction of the protein effect for the missense variants identified in PAL-54 by whole-exome sequencing. Polyphen2 Polyphen2 Mutation- Mutation Gene Variant SIFT LRT HDIV HVAR Taster Assessor c.2065G>A Tolerated Benign Benign Neutral Polymorphism Low HOOK1 p.Gly689Ser 0.32 0.0 0.001 3.77×10-1 1.000 1.21 c.589A>G Tolerated Benign Benign Polymorphism Neutral L1TD1 p.Arg197Gly 0.51 0.0 0.0 N/A 1.000 0 For each variant, the results of the prediction software are reported, with predicted effect and score. Red, deleterious effects; green, 2.1 benign effects.

Supplementary Table 7. Results of the splicing prediction analysis. Splicing prediction method Variant Effect [Range], Threshold Type (Patient ID) position SSF MaxEnt NNSPLICE GeneSplicer HSF [0-100], ≥ 70 [0-12], ≥ 0 [0-1], ≥ 0.4 [0-15], ≥ 0 [0-100], ≥ 65 3.28 1.00 ► 0.96 71.00 c.2164 Donor = (-4.2%) = c.2223A>T 3.23 0.87 ► 0.88 74.71 c.2227 ‡ Donor (PAL-50) = (+1.6%) = 2.51 ► 5.10 77.47 ► 81.03 c.2227+1 Acceptor — ► 74.45 (+103.1%) (+4.6%) 86.16 ► 81.80 9.33 ► 7.61 0.99 ► 0.96 87.99 ► 87.33 c.2038 ‡ Donor c.2038+3A>G (-5.1%) (-18.4%) (-3.1%) (-0.7%) (BR-2652) 71.50 ► 70.60 c.2038+6 Acceptor (-1.3%) 82.32 4.78 0.68 4.19 ► 4.33 81.18 c.1468+61 Acceptor = = = (+3.2%) = c.1468+83del 73.48 ► 74.11 c.1468+92 Acceptor (BR-2652) (+0.9%) 75.89 ► 79.24 c.1468+95 Acceptor (+4.4%) c.395-2 Donor 69.26 ► — c.397A>T 74.72 8.87 ► 9.42 0.82 ► 0.80 7.10 ► 7.63 82.23 (GPS-60) c.395 ‡ Acceptor = (+6.2%) (-3.2%) (+7.4%) = 86.02 10.17 1.00 2.01 ► 1.62 93.21 c.543 ‡ Donor c.513G>A = = = (-19.6%) = (SAO-X35) 1.07 ► — 77.01 ► — c.514 Acceptor

71.83 ► 71.27 c.801-80 Acceptor c.801-84C>A (-0.8%) (COMO-05) 73.22 ► 70.65 c.801-77 Acceptor (-3.5%) — ► 73.14 — ► 4.07 — ► 0.79 — ► 82.05 c.995+84 Donor c.995+86C>T (PV-16) 76.99 1.63 ► 1.75 80.51 ► 79.85 c.995+90 Acceptor = (+7.3%) (-0.8%) (Continues)

65 Chapter 2.1

(Follows)

Splicing prediction method Variant Effect [Range], Threshold Type (Patient ID) position SSF MaxEnt NNSPLICE GeneSplicer HSF [0-100], ≥ 70 [0-12], ≥ 0 [0-1], ≥ 0.4 [0-15], ≥ 0 [0-100], ≥ 65 — ► 67.34 c.2108-22 Donor

3.28 1.00 ► 1.07 71.00 c.2164 Donor = (+7.6%) = c.2108-21A>G 76.09 6.23 2.71 ► 2.49 79.03 c.2108-91 Acceptor (TH-20) = = (-8.1%) = — ► 80.23 — ► 5.33 — ► 84.38 c.2108-20 Acceptor

97.07 11.45 0.98 ► 0.98 11.38 ► 10.20 96.31 c.2108 ‡ Acceptor = = (-0.0%) (-10.4%) = 95.33 10.47 1.00 14.38 ► 14.35 97.89 c.2634 ‡ Donor = = = (-0.2%) = c.2634+37A>G — ► 71.64 c.2634+36 Donor (TOR-6A) 70.24 ► 70.36 c.2634+46 Acceptor (+0.2%) For each variant, the list of genomic sites with altered scores in a ±100bp interval is displayed. For each prediction tool, the score in the reference and mutated sequences, as well as the percentage difference are reported. The complete range of possible values for each method is reported; only values passing the method-specific threshold are reported. ‡, Natural splice site; SSF, SpliceSiteFinder-like; MaxEnt, MaxEntScan; HSF, Human Splicing Finder. The interval between squared brackets indicates the range of values for each method; the cut-off value after the comma is the threshold applied for filtering the results.

Supplementary Table 8. Accession numbers used for the auxilin protein alignment.

Species DNAJC6 homolog accession number

H. sapiens NP_001243793.1

P. troglodytes XP_001161657.1

M. mulatta XP_001090057.1

C. lupus XP_546673.2

B. taurus NP_777261.1

M. musculus NP_001158055.1

R. norvegicus NP_001101419.1

G. gallus NP_001170886.1

D. rerio XP_005165940.1

X. tropicalis NP_001120147.1

66 2.2 Early-Onset Parkinsonism Caused by Alpha-Synuclein Gene Triplication: Clinical and Genetic Findings in a Novel Family

S. Olgiati,*,1 A. Thomas,*,2 M. Quadri,1 G.J. Breedveld,1 J. Graafland,1 H. Eussen,1 H. Douben,1 A. de Klein,1 M. Onofrj,2 and V. Bonifati.1,#

Published in Parkinsonism & Related Disorders 2015, 21(8):981-6.

* These authors contributed equally to this work, as first authors; # Corresponding author.

1) Department of Clinical Genetics, Erasmus MC, Rotterdam, The Netherlands; 2) Department of Neuroscience, Imaging, and Medical Sciences, G. d’Annunzio University, Chieti/Pescara, Italy; Aging Research Centre, Ce.S.I., G. d’Annunzio University Foundation, Chieti, Italy.

67 78 2.3 An Exome Study of Parkinson’s Disease in Sardinia, a Mediterranean Genetic Isolate

M. Quadri,*,1 X. Yang,*,2 G. Cossu,*,3 S. Olgiati,*,1 V.M. Saddi,4 G.J. Breedveld,1 L. Ouyang,2 J. Hu,2 N. Xu,2 J. Graafland,1 V. Ricchi,3 D. Murgia,3 L.C. Guedes,5 C. Mariani,6 M.J. Marti,7 P. Tarantino,8 R. Asselta,9 F. Valldeoriola,7 M. Gagliardi,8,10 G. Pezzoli,6 M. Ezquerra,11 A. Quattrone,8,10 J. Ferreira,5 G. Annesi,8 S. Goldwurm,6 E. Tolosa,7 B.A. Oostra,1 M. Melis,3 J. Wang,2,12,13,14,# V. Bonifati1,#

Published in Neurogenetics 2015, 16(1):55-64.

* These authors contributed equally to this work, as first authors; # These authors contributed equally to this work, as senior authors and corresponding authors.

1) Department of Clinical Genetics, Erasmus MC, Rotterdam, The Netherlands; 2) BGI-Shenzhen, Shenzhen, China; 3) Neurology Service and Stroke Unit, General Hospital S. Michele AOB “G. Brotzu”, Cagliari, Italy; 4) Department of Neurology and Stroke Unit, San Francesco Hospital , ASL No.3 Nuoro, Italy; 5) Clinical Pharmacology Unit, Instituto de Medicina Molecular, University of Lisbon, Lisbon, Portugal; 6) Parkinson Institute, Istituti Clinici di Perfezionamento, Milan, Italy; 7) Neurology Service, Hospital Clínic de Barcelona, Universitat de Barcelona, IDIBAPS, Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED), Barcelona, Catalonia, Spain; 8) Institute of Molecular Bioimaging and Physiology, National Research Council, Section of Germaneto, Catanzaro, Italy; 9) Dipartimento di Biotecnologie Mediche e Medicina Traslazionale (BIOMETRA), Università degli Studi di Milano, Milan, Italy; 10) Institute of Neurology, Department of Medical Sciences, University Magna Graecia, Catanzaro, Italy; 11) Laboratory of Neurodegenerative Disorders, Institut d’Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS)-Fundació Clinic, Barcelona, Spain; 12) Department of Biology, University of Copenhagen, Copenhagen, Denmark; 13) King Abdulaziz University, Jeddah, Saudi Arabia; 14) The Novo Nordisk Foundation Center for Basic Metabolic Research, University of Copenhagen, Copenhagen, Denmark.

PART III Atypical Parkinsonism 104 3.1 Mutation in the SYNJ1 Gene Associated With Autosomal Recessive, Early-Onset Parkinsonism

M. Quadri,*,1 M. Fang,*,2 M. Picillo,*,3 S. Olgiati,*,1 G.J. Breedveld,1 J. Graafland,1 B. Wu,2 F. Xu,2 R. Erro,3 M. Amboni,4,5 S. Pappatà,6 M. Quarantelli,6 G. Annesi,7 A. Quattrone,8 H.F. Chien,9 E.R. Barbosa,9 The International Parkinsonism Genetics Network,‡ B.A. Oostra,1 P. Barone,5,# J. Wang,2,10,11,12,# and V. Bonifati.1,#

Published in Human Mutation 2013, 34(9):1208-15.

* These authors contributed equally to this work, as first authors; # These authors contributed equally to this work, as senior authors and corresponding authors; ‡ Members of the network are listed in the Appendix of this chapter.

1) Department of Clinical Genetics, Erasmus MC, Rotterdam, The Netherlands; 2) BGI-Shenzhen, Shenzhen, China; 3) Department of Neurological Sciences, University of Naples “Federico II”, Naples, Italy; 4) IDC Hermitage-Capodimonte Institute, Naples, Italy; 5) Department of Medicine and Surgery, CEMAND, University of Salerno, Italy; 6) Biostructure and Bioimaging Institute, National Research Council, Naples, Italy; 7) Institute of Neurological Science, National Research Council, Cosenza, Italy; 8) Institute of Neurology, University Magna Græcia, Catanzaro, Italy; 9) Department of Neurology, University of São Paulo, Brazil; 10) Department of Biology, University of Copenhagen, Copenhagen, Denmark; 11) King Abdulaziz University, Jeddah, Saudi Arabia; 12) The Novo Nordisk Foundation Center for Basic Metabolic Research, University of Copenhagen, Copenhagen, Denmark.

105 Chapter 3.1

Abstract Autosomal recessive, early-onset parkinsonism is clinically and genetically heterogeneous. Here, we report the identification, by homozygosity mapping and exome sequencing, of a SYNJ1 homozygous mutation (p.Arg258Gln) segregating with disease in an Italian consanguineous family with parkinsonism, dystonia, and cognitive deterioration. Response to levodopa was poor, and limited by side effects. Neuroimaging revealed brain atrophy, nigrostriatal dopaminergic defects, and cerebral hypometabolism. SYNJ1 encodes synaptojanin 1, a phosphoinositide phosphatase protein with essential roles in the post-endocytic recycling of synaptic vesicles. The mutation is absent in variation databases and in ethnically-matched controls, is damaging according to all prediction programs, and replaces an amino acid that is extremely conserved in the synaptojanin 1 homologues and in SAC1-like domains of other proteins. Sequencing the SYNJ1 ORF in unrelated patients revealed another heterozygous mutation (p.Ser1422Arg), predicted as damaging, in a patient who also carries a heterozygous PINK1 truncating mutation. The SYNJ1 gene is a compelling candidate for parkinsonism; mutations in the functionally linked protein auxilin cause a similar early-onset phenotype, and other findings implicate endosomal dysfunctions in the pathogenesis. Our data delineate a novel form of human Mendelian parkinsonism, and provide further evidence for abnormal synaptic vesicle recycling as a central theme in the pathogenesis.

106 Mutation in the SYNJ1 Gene Associated With Autosomal Recessive, Early-Onset Parkinsonism

Introduction Parkinson’s disease (PD), one of the most common neurodegenerative disorders, is clinically characterized by the combination of slowness of movement (bradykinesia), muscular rigidity, resting tremor, and postural instability, and a beneficial response to levodopa or other dopaminergic therapies [1]. Brain pathology shows neuronal loss and gliosis in the substantia nigra and other areas, with pathological protein aggregates (Lewy bodies) in the surviving neurons. During the past few years, several Mendelian forms of human parkinsonism have been identified [2]. Mutations in each of the following three genes: parkin (PRKN, PARK2) [3], PTEN induced putative kinase 1 (PINK1, PARK6) [4], and Parkinson protein 7 (DJ-1, PARK7) [5] can cause autosomal recessive forms of early-onset parkinsonism, with a good and prolonged response to dopaminergic therapy without additional, atypical clinical signs, and clinically diagnosed as early- onset PD. Mutations in other genes, including ATPase type 13A2 (ATP13A2, PARK9) [6], phospholipase A2, group VI (PLA2G6, PARK14) [7], F-box only protein 7 (FBXO7, PARK15) [8,9], and DNAJC6 [10,11] cause early onset atypical parkinsonism, with poor or absent response to levodopa and variable combinations of additional clinical signs (such as dystonias, oculomotor disturbances, pyramidal signs, dementia). Here, 3.1 we report the identification, by homozygosity mapping and exome sequencing, of a homozygous mutation in the SYNJ1 gene in an Italian consanguineous family with early-onset atypical parkinsonism, delineating a novel disease form.

Subjects and Methods This study was approved by the appropriate institutional review boards, and written informed consent was obtained from all subjects. We studied one white consanguineous family originating from Sicily (Italy) with two siblings affected by early-onset atypical parkinsonism, and no history of neurological diseases in the previous generations; a pattern compatible with autosomal-recessive inheritance (Fig. 1A). Additional DNA samples from 118 unrelated patients with early-onset parkinsonism compatible with autosomal recessive inheritance, collected at several movement disorders centers contributing to the International Parkinsonism Genetics Network, were also analyzed. These patients originate from Italy (n=77), Brazil (n=31), The Netherlands (n=6), Portugal (n=2), Spain (n=1) and Belgium (n=1). According to the UK Brain Bank diagnostic criteria [12], most of these patients (n=92) had early-onset, typical PD; their average onset age was 30 ± 5.3 years (range 13–36). The remaining 26 patients had early-onset parkinsonism in combination with some atypical features (poor levodopa response, oculomotor disturbances, pyramidal signs, or dementia); their average onset age was 34.1 ± 16 years (range 6–49). Last, DNA samples from 180 unrelated control subjects from Southern Italy were analyzed.

107 Chapter 3.1

Figure 1. The Italian pedigree with SYNJ1 mutation and neuroimaging findings.Panel (A) shows the Italian pedigree with the SYNJ1 mutation. The black symbols denote individuals affected by early-onset, atypical parkinsonism. Imaging of the brain dopamine transporter density values (V˝3) using single photon emission computed tomography (DaTSCAN-SPECT) shows severe presynaptic nigrostriatal dopaminergic deficit in the NAPO-16 patient (B). A control subject (C) is also shown. Representative brain magnetic resonance images (MRI, 3 Tesla) are shown for the NAPO-16 (D,H,L) and NAPO-17 (E,I,M) patient. Axial images at the level of the basal ganglia acquired using a Fluid Attenuated Inversion Recovery (FLAIR) sequence (D,E) show cortical atrophy, more severe in NAPO-16. Mid-sagittal images from Magnetization Prepared Rapid Acquisition Gradient-Echo (MPRAGE) T1-weighted volumes (H,I) show thinning of the quadrigeminal plate in both patients (more prominent in NAPO-16) (white arrows). Coronal FLAIR images (L,M) show diffuse hyperintensity of the hyppocampi (white arrows) and enlarged lateral ventricles and sulci of the convexity, consistent with diffuse atrophy, more severe in NAPO-16. Imaging of the brain metabolic activity using [18F]fluorodeoxyglucose and positron emission tomography (PET) at the level of the basal ganglia (F,J,N) and of the centrum semiovale (G,K,O); representative axial images of the NAPO-16 patient (F,G), the NAPO-17 patient (J,K), and a control subject (N,O) are shown. Metabolic deficit are present in both patients, much more severe and widespread in NAPO-16, in several cortical areas and the caudate nucleus, with relative preservation of the putamen.

108 Mutation in the SYNJ1 Gene Associated With Autosomal Recessive, Early-Onset Parkinsonism

Neuroimaging Studies In the patients of the Italian family, brain MRI studies were performed at 3 Tesla (TRIO, Siemens Medical Solutions, Erlangen, Germany). Imaging sequences included a 3D T1-weighted Gradient-echo volume with isotropic voxels, axial diffusion-weighted images, and axial and coronal T2-weighted turbo spin-echo and fluid-attenuated inversion recovery (FLAIR) images. Brain single photon emission computed tomography (SPECT) studies were performed 4 h after intravenous administration of 185 MBq of [123I]FP-CIT (DaTSCAN; GE-Healthcare, Wauwatosa, WI) using a dual-headed camera (E.CAM; Siemens Medical Systems, Hoffman Estates, IL) equipped with low-energy high-resolution collimators. Outcome measures were the specific-to-nondisplaceable binding ratio, V˝3, for the putamen and caudate. [18F] FDG-PET brain images were acquired for 15 minutes in 3D mode between 45 and 60 minutes after intravenous administration of 213–250 MBq of [18F]fluorodeoxyglucose in resting state and eyes closed condition using a whole body PET/CT scanner (Discovery LS; GE Medical System, Milwaukee, WI). Images were reconstructed with iterative reconstruction (FORE-Iterative) and corrected for attenuation using the CT scan. Outcome measure was relative glucose metabolism. Both SPECT and PET images were spatially normalized in the Montreal Neurological Institute (MNI) space using SPM2 (Wellcome Department of Imaging Neuroscience, London, UK). Genetic Studies 3.1 Genomic DNA was isolated from peripheral blood with standard protocols. In the Italian family, our screening of known genes causing early-onset parkinsonism (Sanger sequencing of all exons and exon-intron boundaries of parkin, PINK1, DJ-1, ATP13A2, PLA2G6, FBXO7, DNAJC6, and exon dosage of SNCA, parkin, PINK1, DJ-1, and ATP13A2) revealed no mutations. A Filipin test for Niemann-Pick disease, type C, was also negative. We then performed genome- wide homozygosity mapping using Illumina (San Diego, CA) HumanOmniExpress BeadChip array data (730,525 SNPs at a median distance of 2.1 kb). Samples from the two affected siblings (NAPO-16 and NAPO-17), two unaffected siblings (NAPO-24 and NAPO-25), and the unaffected mother (NAPO-18) were included. Data were analysed using ALLEGRO (version 1.2c) through the graphical user interface EasyLinkage Plus v.5.08 [13]. SNPs with homozygous and identical genotypes in all tested individuals were removed. The marker spacing was set at 0.3 cM in a computation set of 100 markers. We specified an autosomal recessive-homozygous model (HOMOZ scoring function), 0.0001 disease allele frequency, and complete penetrance. Homo- zygosity within the linkage peaks was further analyzed with Nexus Copy Number, Discovery Edition, ver. 7 (BioDiscovery, El Segundo, CA). Exome sequencing was performed in both affected individuals (NAPO-16 and NAPO-17) at BGI-Shenzhen, using in-solution capturing (Agilent SureSelect Human All Exon 50Mb Kit; Agilent Technologies, Santa Clara, CA) and Illumina HiSeq2000, 90 paired-end sequencing. Reads were aligned to the human reference genome hg18 (build 36.3). SOAPaligner/SOAP2 aligner and SOAPsnp were used for SNP identification, while Burrows-Wheeler-Aligner and GATK (Genome-Analysis-Tool-Kit) were used for indels detection. For the selection of candidate variants, we first identified chromosomal regions of interest by homozygosity mapping. Within these regions, the variants detected by the exome sequencing, that fulfilled all the following

109 Chapter 3.1 criteria were selected: homozygous, coding non-synonymous or affecting splice-sites, absent in dbSNP132 and in 1000Genomes, present in both the affected individuals of the family. Last, we performed Sanger sequencing of the positional candidate gene SYNJ1 (MIM #604297) in order to confirm a mutation identified in the exome sequencing, and to screen additional patients. Primers and PCR protocols are reported in the Supp. Table S1. The SYNJ1 variants are annotated according to the GenBank accession numbers NM_003895.3 and NP_003886.3, corresponding to the longest isoform (synaptojanin-1, isoform a), and numbered at nucleotide level from the ‘‘A’’ of the ATG-translation initiation codon. The variants identified have also been submitted to a locus-specific database (http://www.lovd.nl/SYNJ1), and to the NCBI SNP database. Six different web-tools were used in order to predict the functional effects of novel missense variants: PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/), SIFT (http://sift.jcvi.org/www/ SIFT_enst_submit.html), PROVEAN (http://provean.jcvi.org/seq_submit.php), SNPs&GO (http://snps-and-go.biocomp.unibo.it/snps-and-go/index.html), SNAP (http://rostlab.org/ services/snap/submit), and Mutation Taster (http://www.mutationtaster.org/). Furthermore, for the analysis of the amino acid conservation, the closest evolutionary homologues of the human synaptojanin 1 protein, as well as additional human proteins containing a suppressor-of actin1 (SAC1)-like phosphatase domain, were identified by blasting the human sequence using the BLASTp program. Retrieved sequences were saved in FASTA format and aligned using the program T-Coffee. GenBank accession numbers are reported in the Supp. Table S2.

Results

Clinical Studies Two healthy first cousins from a small Sicilian village had four offspring, two of whom developed a progressive neurological disease. The first patient (NAPO-16) is a 47-year- old man with an uneventful birth and normal development until adulthood. At 22-years old he developed rapidly progressive slowness of movements, fatigue, leg stiffness, gait impairment, and involuntary movements in the upper arm. Within a few months he was unable to work, and within one year he needed assistance in his daily living activi- ties. Cognitive decline and severe dysarthria also developed, progressing to an anarthric state in about 3 years. Levodopa therapy, initiated within 2 years since the onset of symptoms, yielded unclear benefit on his motor disturbances, but had to be stopped due to the emergence of disabling side-effects, such as oro-mandibular and feet dystonias, that were further interfering with feeding and gait. On our first examination, twenty-five years after the onset of the disease, he presented with a stooped posture, and a severely disturbed, short-stepped, shuffling gait (Supp. Videos 1-2). There was severe axial and four-limb rigidity, more marked on the left side. The postural reflexes were also impaired. Dystonic postures were present in both hands

110 Mutation in the SYNJ1 Gene Associated With Autosomal Recessive, Early-Onset Parkinsonism

(left>right). Irregular, wide-amplitude tremor was present at rest and during action in the upper limbs; wide-amplitude jerky involuntary movements were also observed in the facial muscles. His gaze was markedly staring. There was eyelid apraxia, and a clear supranuclear vertical gaze limitation (downward more than upward). There was also mild dysphagia, allowing feeding with semiliquid food. The deep tendon reflexes were normal, and the Babinski sign was absent. There were no signs of cerebellar dysfunction. The score of the Unified Parkinson’s disease rating scale (UPDRS-III, motor scale) was 78 (max score is 108, with higher scores indicating more severe parkinsonism). An acute levodopa challenge (200 mg Melevodopa) yielded no relevant benefit, but induced marked oro-mandibular and limbs dystonias, and hypotension (blood pressure dropped from 110/80 to 80/60 mmHg). The Mini-Mental State Examination (MMSE) scale could not be administered, due to the severe motor and cognitive disability. Three years later, he remained untreated and his clinical picture was substantially unchanged. The second patient (NAPO-17) is a 31-year-old woman with normal birth and developmental milestones. At 28 years-old she developed slowness of movements, speech and gait difficulties, and involuntary movements in the right arm. Like in her brother, levodopa therapy had to be stopped because of marked side-effects, such as oro-mandibular and 3.1 feet dystonias, which were seriously interfering with her feeding and gait. On our first examination, three years after the onset of symptoms, she presented with a stooped posture, and a short-stepped, shuffling gait. Postural reflexes were markedly impaired. Her gaze was markedly staring and there was a moderate supranuclear vertical gaze limitation (Supp. Videos 3-4). There was no eyelid apraxia. Almost continuous, irregular, small-amplitude jerky involuntary movements were present in the lips and tongue. Intermittent, wide-amplitude tremor appeared in the right arm at rest and during action. There was moderate axial and four-limb rigidity (more marked on the left side), and dystonic postures were present in the hands and feet (again more severe on the left side). She also presented marked dysarthria, hypophonia, and mild dysphagia. She was fed with a semiliquid diet. The deep tendon reflexes were brisk, but the Babinski sign was absent. There were no signs of cerebellar dysfunction. The UPDRS-III score was 57. As seen in her brother, an acute levodopa challenge (200 mg Melevodopa) induced severe dystonias in both the oro-mandibular district and in the feet, and a drop of her blood pressure from 110/70 to 70/50 mmHg. The MMSE score was 26 (max 30, lower scores indicating lower cognitive performance). Three years later, she was still untreated by antiparkinsonian drugs; we noticed a worsening of the parkinsonism, of the trunk dystonia, and of the supranuclear gaze palsy. Her UPDRS-III score had increased to 68, while her MMSE score declined to 24/30. In both patients, blood chemistry was normal. Moreover, in both siblings, needle elec-

111 Chapter 3.1 tromyography, sensory and motor nerve conduction studies, somatosensory evoked potentials with median and tibial nerve stimulation were all unremarkable. Similarly, in both sibs, the motor evoked potentials measured in both the upper and lower limbs yielded normal amplitude and conduction time values. Brain MRI showed diffuse cerebral cortex atrophy, which was more severe in the brother (Fig. 1D,E). Diffuse hyperintensity of the hyppocampi and thinning of the midbrain quadrigeminal plate was appreciated in both patients (more prominent in NAPO-16), with no significant atrophy of the midbrain tegmentum (Fig. 1H,I,L,M). SPECT imaging was available for the NAPO-16 patient, and it showed a marked decrease of dopamine transporter density in the striatum (V˝3 reduced by 87% in the putamen and by 77% in the caudate nucleus), compared to control values (Fig. 1B). Moreover, FDG-PET imaging showed in both patients a relative cerebral cortex hypometabolism, more evident in the NAPO-16 patient, and in the frontal and parieto-occipital regions, and the caudate nuclei (Fig. 1F,G,J,K).

Genetic Studies Genome-wide homozygosity mapping revealed five interesting loci, with LOD>2, on chromosome 4, 7, 16, 19, and 21, that contained extended homozygous regions in the two patients and not in the unaffected relatives (Fig. 2A,B). The exome sequencing achieved an average coverage of 50× and 55× in the NAPO-16 and NAPO-17 patient, respectively, with 71% and 73% of the exome target region covered at least 20×. Within the five regions of extended homozygosity, the exome sequencing detected two homozygous exonic or splice site variants, shared by the two patients, which were not annotated in dbSNP132 or in the 1000 Genomes project. The first variant is a c.931C>T substitution in ZNF439, leading to p.Arg311Cys. This was confirmed by Sanger sequencing, and is not present in public databases. However, it is consistently considered as benign by the prediction programs (with the exception of SIFT, which yields a non-conclusive score) (Supp. Table S3). Although nothing is known about the function of the protein encoded by this gene, inspection of the ZNF439 variation in the NHLBI Exome Variant Server (EVS) (http://evs.gs.washington.edu/

Figure 2. Homozygosity mapping, SYNJ1 mutation, and in-silico protein analyses. (A) Genome-wide LOD score plot show five suggestive regions on chromosome 4, 7, 16, 19, and 21. (B) Homozygosity view of the chromosome 21 region. A large homozygous area is shared by the NAPO16 and NAPO-17 patients. The same region is heterozygous in the unaffected mother NAPO-18, and unaffected siblings (not shown). The position of the SYNJ1 gene is indicated. (C) Electropherograms showing the SYNJ1 c.773G>A (p.Arg258Gln) mutation in homozygous and heterozygous state, and a wild type, control sequence. (D) Schematic representation of the Synj1 protein domains (SAC – suppressor-of actin1-like phosphatase domain; INPP5c – inositol-5-phosphatase domain). (E) Aligmnent of the Synj1 protein homologs and of other human SAC1-like domain containing proteins.

112 Mutation in the SYNJ1 Gene Associated With Autosomal Recessive, Early-Onset Parkinsonism

3.1

113 Chapter 3.1

EVS/) reveals many frequent and rare deleterious variants (including truncating, splice- site, and frameshift mutations). Due to these evidences, we consider that ZNF439 is not a candidate gene for our phenotype of interest. The other homozygous variant was a c.773G>A substitution in the SYNJ1 gene (#MIM 604297), leading to the p.Arg258Gln missense change. The SYNJ1 gene encodes synaptojanin 1 (Synj1), a phosphoinositide phosphatase protein with a major, established role in the clathrin-mediated endocytosis in the adult brain [14]. Two main isoforms (synaptojanin 1, isoform a and b) of 170 and 145 kDa, containing 1612 and 1350 amino acids, are encoded by SYNJ1 as a result of alternative splicing. Both are ubiquitously expressed but the 145 kDa isoform (NM_203446.2, NP_982271.2) is the predominant isoform expressed in the presynaptic terminals in the brain [15,16]; it possesses a distinct and shorter C-terminus compared to the 170 kDa isoform (NM_003895.3, NP_003886.3). Two additional, shorter isoforms (synaptojanin 1, isoform c and d) of unknown functional relevance, (NM_001160302.1, NP_001153774.1; NM_001160306.1, NP_001153778.1) are also annotated in Genbank. SYNJ1 represents therefore an excellent candidate for the phenotype of interest in our study. Sanger sequencing confirmed the presence of the mutation in homozygous state in both patients and in heterozygous state in the unaffected siblings and the mother (Fig. 2C). Furthermore, sequencing of all SYNJ1 exons and exon-intron boundaries revealed no additional novel variants in these two patients. The p.Arg258Gln mutation is not present in dbSNP132, the 1000 Genomes, nor in the EVS database (>13,000 chromosomes). We did not detect any carrier in control chromosomes from Calabria (Southern Italy) (n=360). Furthermore, this mutation is damaging according to all prediction programs (Supp. Table S3). The affected amino acid is located in the SAC1-like phosphatase domain of Synj1 (Fig. 2D), and is extremely conserved in the evolution of the Synj1 homologues, and in the SAC1-like domain of heterologous proteins (Fig. 2E), suggesting a very important functional role for this phosphatase domain. In order to identify mutations in additional patients, we performed Sanger sequencing of the entire SYNJ1 coding region (32 exons) and exon-intron boundaries in 118 unrelated patients with early-onset parkinsonism compatible with autosomal recessive inheritance (26 of them with atypical clinical features, and 92 with early-onset typical PD). Novel variants, or known rare variants (MAF<1%), in homozygous or compound heterozygous state, were not detected in any of these patients. The complete list of the variants detected, and their allelic frequencies are reported in the Supp. Table S4. Only one novel heterozygous, non-synonymous variant was found, in one Brazilian patient (SAO-X20), a c.4264A>C substitution in the last 3’exon, leading to a missense change (p.Ser1422Arg), affecting only one of the two predominant Synj1 isoforms (170

114 Mutation in the SYNJ1 Gene Associated With Autosomal Recessive, Early-Onset Parkinsonism kDa isoform a). This mutation was not present in dbSNP132, the 1000 Genomes, nor in the EVS database, and is considered as damaging by several (though not all) prediction programs (Supp. Table S3). The amino acid replaced is incompletely conserved in the evolution (Supp. Figure S1). Of note, our previous screening (including Sanger sequencing and exon dosage) revealed that this patient carries also a heterozygous truncating PINK1 mutation (p.Trp437Stop). Clinical data concerning this patient are limited, and DNA samples from additional family members are not available. He was diagnosed with sporadic early-onset PD (onset age 35 years) and was treated with levodopa with satisfactory response. Atypical features are not reported. Another Brazilian patient (SAO-X41) with early-onset typical PD (onset age 35) carried a heterozygous c.3755G>A substitution, leading to a missense change (p.Arg1252Gln). This variant is rare but present in dbSNP (Supp. Table S4), and considered as damaging by some prediction programs (Supp. Table S3). In this patient, our previous screening of the other genes causing early-onset PD revealed no mutations. All the remaining non-synonymous variants detected in these 118 patients were either very frequent polymorphisms present in dbSNP, or rare known variants considered to be neutral by the prediction programs (Supp. Table S3 and Supp. Table S4). 3.1

Discussion Endocytosis is essential for the neurons in order to recycle membranes of synaptic vesicles. The phosphoinositides, especially the phosphatidylinositol-4,5-bisphosphate

[PI(4,5)P2] play important roles in the membrane traffic processes, such as endocytosis and exocytosis, among other cellular functions [17]. PI(4,5)P2 is concentrated in discrete regions on the cytosolic leaflet of the presynaptic membranes, where it functions in the early steps of the clathrin-mediated endocytosis by recruiting the clathrin adaptor proteins, which in turn, recruit clathrin, leading to the formation of the clathrin-coated pits (CCPs). The maturation of the CCPs and their fission from the plasma membrane is coordinated by the endophilins and the dynamin proteins. This chain of events generates the clathrin-coated vesicles (CCVs). On their transit to target sites the CCVs are uncoated by means of the coordinated action of Synj1, the chaperone Hsc70 and its co-chaperone auxilin. Hsc70 and auxilin mediate the clathrin disassembly [18], while

Synj1 dephosphorylates PI(4,5)P2 , thereby shedding the adaptor proteins [19,20]. The complete inactivation of the Synj1 gene in mice is associated with increased PI(4,5)

P2 levels, defective uncoating of clathrin, accumulation of CCVs, endocytic delays, and severe neurological phenotypes (weakness, ataxia), leading to perinatal lethality [19]. Strikingly similar phenotypes are seen in the auxilin and endophilin KO mice, confirming a close functional interaction between these proteins [21,22].

115 Chapter 3.1

Several arguments support the contention that the SYNJ1 mutation is the cause of the disease in the Italian family. First, the identification was based on an unbiased strategy of genome-wide homozygosity mapping followed by exome sequencing. This is the only homozygous, damaging mutation that survived our strict filtering steps. Second, the mutation is absent from all variant databases and from our population-matched controls. Third, it replaces an extremely conserved amino acid, not only in the evolution of the synaptojanin 1, but also in other SAC1-like domains of different proteins, in keeping with a very important role for the phosphatase activity of this domain (Fig. 2E). Fourth, recessive mutations in the DNAJC6, encoding auxilin, the direct functional partner of Synj1 in the uncoating of CCVs, cause a very similar human phenotype of early-onset, atypical parkinsonism [10,11]. Fifth, the mice KO models of these two genes (DNAJC6 and SYNJ1) show nearly identical phenotypes of CCVs accumulation, defective synaptic vesicle recycling, and severe neurological phenotypes [19,22]. Sixth, common variants in the GAK gene, encoding auxilin 2 (another co-chaperone for Hsc70) were identified in GWAS studies and replicated with genome-wide significance as risk factor for late-onset PD [23,24]. Additional evidences point to defective recycling of synaptic vesicles in the pathogenesis of the more common forms of PD. The LRRK2 kinase protein, which is mutated in the most common, known form of autosomal dominant PD, was recently identified as a regulator of endophilinA functions in the synaptic vesicle endocytosis [25]. Moreover, parkin, an ubiquitin ligase mutated in the most common form of autosomal-recessive, early-onset PD, was also reported to interact with endophilinA [26]. The biology of alpha-synuclein, a pre-synaptic protein mutated in rare autosomal dominant forms of PD, and the major component of the aggregates found in all the patients with typical late-onset PD, has several ties with the traffic and recycling of the synaptic vesicles [27,28]. The VPS35 protein, mutated in a rare form of autosomal dominant parkinsonism [29,30], is part of the retromer complex involved in retrograde sorting steps on early endosomes [31,32]. The perinatal lethality in mice with complete ablation of Synj1, or of the directly interacting proteins auxilin or endophilins, indicates that these proteins are essential for postnatal life. Interestingly, however, the partial loss of endophilin causes severe neurological defects, including epilepsy and neurodegeneration [21]. In keeping with this evidence, the mutation identified in our Italian family is predicted to affect only the SAC1-like phosphatase domain, leaving the other 5’-phosphatase domain intact. It will be interesting to study the effects of this partial loss-of-function in model organisms. Our sequencing of the entire SYNJ1 ORF in 118 patients with early-onset parkinsonism has not identified additional homozygous or compound heterozygous mutations,

116 Mutation in the SYNJ1 Gene Associated With Autosomal Recessive, Early-Onset Parkinsonism suggesting that SYNJ1 mutations might be an uncommon cause of disease. However, only 26 of the 118 patients had atypical parkinsonism. Screening of additional patients with similar phenotypes is therefore warranted, in order to identify the same or different mutations and support our contention that SYNJ1 mutation is the cause of this novel form of parkinsonism. The interpretation of the heterozygous SYNJ1 mutation detected in the Brazilian patient (SAO-X20) is difficult. We did not identify a compound-heterozygous mutation, but did not screen for small heterozygous duplications or deletions. A compound-heterozygous mutation could also be present in the promoter or intronic regulatory elements. Although the mutation replaces a conserved residue and is predicted to be damaging in silico, it is located in the variable C-terminus region present in the 170 kDa, but not in the 145 kDa isoform, this last being the major synaptic Synj1 isoform. This mutation, could represent a coincidental finding, unrelated to the disease. However, its presence together with a single heterozygous mutation in another gene causing early- onset parkinsonism (PINK1) might suggest a possible additive or interactive effect in the pathogenesis of the disease in this case. The other, rare SYNJ1 heterozygous variant detected in the SAO-X41 patient might also be a coincidental finding. 3.1 The clinical phenotype in the Italian family is characterized by prominent, early-onset parkinsonism, dystonias, oculomotor disturbances and cognitive decline. Pyramidal or cerebellar signs were not present. The response of the motor symptoms to levodopa was poor and limited by the appearance of disabling side effects, such as involuntary movements (dystonias) and severe hypotension. The results of the neuroimaging investigations disclosed cerebral atrophy and hypometabolism, and severe nigrostriatal dopaminergic defects. The atrophy of the quadrigeminal plate of the midbrain might underlie the oculomotor disturbances, as this area is important for the control of eye movements. The widespread cerebral cortical atrophy, the hippocampal sclerosis, and the cortical and caudate nucleus hypometabolism are all in keeping with the cognitive decline present in these patients. Because of these clinical and neuroimaging features, this phenotype is clearly different from typical early-onset PD, and close to the atypical early-onset forms, caused by mutations in the PARK9, PARK14, and PARK15 [33]. The absence of pyramidal disturbances in the Italian family is of interest, as these are often present in the other above-mentioned forms. Of note, a very similar phenotype of early- onset parkinsonism and cognitive deterioration is associated with mutations in auxilin (encoded by DNAJC6), a direct functional interactor of Synj1 [10,11]. Response to levodopa and pyramidal disturbances are variable in the patients with auxilin mutations, and seizures are also described in some. Seizures have not been observed in our patients with Synj1 mutation. However, wide clinical variability is observed across all these forms of atypical parkinsonism [8,33,34], and the identification of additional families with

117 Chapter 3.1

SYNJ1 mutations is warranted to better define the phenotypic spectrum of this novel form.

Conclusions This work delineates a novel genetic form of early-onset parkinsonism associated with Synj1 mutation, pointing to defective post-endocytic uncoating and recycling of the synaptic vesicles in the pathogenesis. These findings might also have implications for understanding the pathogenesis of more common forms of neurodegenerative parkinsonisms.

Acknowledgments The Authors thank all the patients and relatives for their contributions.

Grant sponsors This study was supported by BGI-Shenzhen, China, and by research grants from the “Internationaal Parkinson Fonds” – Neth- erlands, and the Netherlands Organization for Scientific Research (NWO, VIDI grant, project n. 91786395) to V.B.

Disclosure of conflicts of interest The authors declare that they have no conflicts of interest.

Appendix

Members of the International Parkinsonism Genetics Network V. Bonifati, A. Maat-Kievit, J. Rood, and A. Boon, Erasmus MC, Rotterdam, The Netherlands; B. van de Warrenburg, C. Delnooz, A. Rietveld, and B. Bloem, Radboud University Medical Centre, Nijmegen, The Netherlands; J. Ferreira and L. Correia Guedes, Institute of Molecular Medicine, Lisbon, Portugal; E. Tolosa, Hospital Clínic de Barcelona, Universitat de Barcelona, IDIBAPS, Centro de Investigacíon Biomédicaen Red sobre Enfermedades Neurodegenerativas (CIBERNED), Catalonia, Spain; S. Janssens, Ghent University Hospital, Ghent, Belgium; M. Emre, H. Hanagasi, and B. Bilgic, Istanbul Faculty of Medicine, Istanbul, Turkey; B. Elibol, Hacettepe University, Ankara, Turkey; M. Socal and L. Jardim, Hospital de Clínicas de Porto Alegre, Brazil; Hsin F. Chien and Egberto R. Barbosa, University of São Paulo, Brazil; Chin-Song Lu, Yah-Huei Wu-Chou, and Tu-Hsueh Yeh, Chang Gung Memorial Hospital, Taipei, Taiwan; Masharip Atadzhanov and Paul Kelly, University of Zambia, Lusaka, Zambia. Italianmembers: L. Lopiano,University of Torino;C. Tassorelli, C. Pacchetti, andG.Nappi,IRCCS “Mondino”, Pavia; G. Riboldazzi and G. Bono, Insubria University, Varese; A.Padovani and B. Borroni, University of Brescia; F. Raudino, Hospital “Valduce”, Como; A. Di Fonzo, University of Milan, Italy; A. Volonte’, San Raffaele Institute, Milan; E. Fincati and L. Bertolasi, University of Verona; M. Tinazzi and A. Bonizzato, Hospital “Borgo Trento”, Verona; C. Ferracci, Hospital of Belluno; A. Dalla Libera, “Boldrini” Hospital, Thiene; P. Cortelli and S. Capellari, University of Bologna; P. Marini and F. Massaro, University of Firenze; A. Federico, I. Taglia, and C. Battisti, University of Siena; R. Marconi, “Misericordia” Hospital, Grosseto; M. Onofrj and A. Thomas, University of Chieti-Pescara; N. Vanacore, National Institute of Health, Roma; G.Meco, G. Fabbrini, E. Fabrizio, M. Manfredi, and A. Berardelli, “Sapienza” University, Roma; F. Stocchi, L. Vacca, IRCCS “San Raffaele Pisana”, Roma; G. De Michele, C. Criscuolo, L. Santoro, A. Filla, Federico II University, Naples; M. DeMari, C. Dell’Aquila,Hospital of Andria; G. Iliceto and P. Lamberti, University of Bari; V. Toni and G. Trianni, Hospital of Casarano; M. Gagliardi and G. Annesi, Institute of Neurological

118 Mutation in the SYNJ1 Gene Associated With Autosomal Recessive, Early-Onset Parkinsonism

Science, National Research Council, Cosenza; A. Quattrone, Institute of Neurology, University Magna Græcia, Catanzaro; V. Saddi, Neurology Division, S. Francesco Hospital, Nuoro; G. Cossu, and M. Melis, Hospital S. Michele AOB “G. Brotzu”, Cagliari, Italy.

References 1 Fahn S. Description of Parkinson’s disease as a clinical A. DNAJC6 is responsible for juvenile parkinsonism syndrome. Ann N Y Acad Sci. 2003;991:1-14. with phenotypic variability. Parkinsonism Relat 2 Singleton AB, Farrer MJ, Bonifati V. The genetics of Disord. 2013;19:320-4. Parkinson’s disease: progress and therapeutic implica- 12 Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy tions. Mov Disord. 2013;28:14-23. of clinical diagnosis of idiopathic Parkinson’s disease: 3 Kitada T, Asakawa S, Hattori N, Matsumine H, a clinico-pathological study of 100 cases. J Neurol Yamamura Y, Minoshima S, et al. Mutations in the Neurosurg Psychiatry. 1992;55:181-4. parkin gene cause autosomal recessive juvenile parkin- 13 Hoffmann K, Lindner TH. easyLINKAGE-Plus--au- sonism. Nature. 1998;392:605-8. tomated linkage analyses using large-scale SNP data. 4 Valente EM, Abou-Sleiman PM, Caputo V, Muqit Bioinformatics. 2005;21:3565-7. MM, Harvey K, Gispert S, et al. Hereditary early- 14 McPherson PS, Garcia EP, Slepnev VI, David onset Parkinson’s disease caused by mutations in C, Zhang X, Grabs D, et al. A presynaptic inosi- PINK1. Science. 2004;304:1158-60. tol-5-phosphatase. Nature. 1996;379:353-7. 5 Bonifati V, Rizzu P, van Baren MJ, Schaap O, 15 Perera RM, Zoncu R, Lucast L, De Camilli P, Toomre Breedveld GJ, Krieger E, et al. Mutations in the DJ-1 D. Two synaptojanin 1 isoforms are recruited to clath- 3.1 gene associated with autosomal recessive early-onset rin-coated pits at different stages. Proc Natl Acad Sci parkinsonism. Science. 2003;299:256-9. U S A. 2006;103:19332-7. 6 Ramirez A, Heimbach A, Grundemann J, Stiller B, 16 Ramjaun AR, McPherson PS. Tissue-specific alter- Hampshire D, Cid LP, et al. Hereditary parkinsonism native splicing generates two synaptojanin isoforms with dementia is caused by mutations in ATP13A2, with differential membrane binding properties. J Biol encoding a lysosomal type 5 P-type ATPase. Nat Chem. 1996;271:24856-61. Genet. 2006;38:1184-91. 17 Di Paolo G, De Camilli P. Phosphoinositides in 7 Paisan-Ruiz C, Bhatia KP, Li A, Hernandez D, Davis cell regulation and membrane dynamics. Nature. M, Wood NW, et al. Characterization of PLA2G6 2006;443:651-7. as a locus for dystonia-parkinsonism. Ann Neurol. 18 Ungewickell E, Ungewickell H, Holstein SE, 2009;65:19-23. Lindner R, Prasad K, Barouch W, et al. Role of 8 Di Fonzo A, Dekker MC, Montagna P, Baruzzi auxilin in uncoating clathrin-coated vesicles. Nature. A, Yonova EH, Correia Guedes L, et al. FBXO7 1995;378:632-5. mutations cause autosomal recessive, early-onset 19 Cremona O, Di Paolo G, Wenk MR, Luthi A, Kim parkinsonian-pyramidal syndrome. Neurology. WT, Takei K, et al. Essential role of phosphoinos- 2009;72:240-5. itide metabolism in synaptic vesicle recycling. Cell. 9 Shojaee S, Sina F, Banihosseini SS, Kazemi MH, 1999;99:179-88. Kalhor R, Shahidi GA, et al. Genome-wide linkage 20 Mani M, Lee SY, Lucast L, Cremona O, Di Paolo G, analysis of a Parkinsonian-pyramidal syndrome De Camilli P, et al. The dual phosphatase activity of pedigree by 500 K SNP arrays. Am J Hum Genet. synaptojanin1 is required for both efficient synaptic 2008;82:1375-84. vesicle endocytosis and reavailability at nerve 10 Edvardson S, Cinnamon Y, Ta-Shma A, Shaag A, terminals. Neuron. 2007;56:1004-18. Yim YI, Zenvirt S, et al. A deleterious mutation 21 Milosevic I, Giovedi S, Lou X, Raimondi A, Collesi in DNAJC6 encoding the neuronal-specific clath- C, Shen H, et al. Recruitment of endophilin to clathrin-uncoating co-chaperone auxilin, is associated with rin-coated pit necks is required for efficient vesicle juvenile parkinsonism. PLoS One. 2012;7:e36458. uncoating after fission. Neuron. 2011;72:587-601. 11 Koroglu C, Baysal L, Cetinkaya M, Karasoy H, Tolun 22 Yim YI, Sun T, Wu LG, Raimondi A, De Camilli P,

119 Chapter 3.1

Eisenberg E, et al. Endocytosis and clathrin-uncoating clein reduces neurotransmitter release by inhibiting defects at synapses of auxilin knockout mice. Proc synaptic vesicle reclustering after endocytosis. Neuron. Natl Acad Sci U S A. 2010;107:4412-7. 2010;65:66-79. 23 Nalls MA. Imputation of sequence variants for iden- 29 Vilarino-Guell C, Wider C, Ross OA, Dachsel JC, tification of genetic risks for Parkinson’s disease: a Kachergus JM, Lincoln SJ, et al. VPS35 mutations in meta-analysis of genome-wide association studies. Parkinson disease. Am J Hum Genet. 2011;89:162-7. International Parkinson Disease Genomics Consor- 30 Zimprich A, Benet-Pages A, Struhal W, Graf E, tium. Lancet. 2011;377:641-9. Eck SH, Offman MN, et al. A mutation in VPS35, 24 Pankratz N, Wilk JB, Latourelle JC, DeStefano AL, encoding a subunit of the retromer complex, causes Halter C, Pugh EW, et al. Genomewide association late-onset Parkinson disease. Am J Hum Genet. study for susceptibility genes contributing to familial 2011;89:168-75. Parkinson disease. Hum Genet. 2009;124:593-605. 31 Bonifacino JS, Hurley JH. Retromer. Curr Opin Cell 25 Matta S, Van Kolen K, da Cunha R, van den Bogaart Biol. 2008;20:427-36. G, Mandemakers W, Miskiewicz K, et al. LRRK2 32 Popoff V, Mardones GA, Tenza D, Rojas R, Lamaze controls an EndoA phosphorylation cycle in synaptic C, Bonifacino JS, et al. The retromer complex and endocytosis. Neuron. 2012;75:1008-21. clathrin define an early endosomal retrograde exit site. 26 Trempe JF, Chen CX, Grenier K, Camacho EM, J Cell Sci. 2007;120:2022-31. Kozlov G, McPherson PS, et al. SH3 domains from a 33 Bonifati V. Autosomal recessive parkinsonism. Parkin- subset of BAR proteins define a Ubl-binding domain sonism Relat Disord. 2012;18 Suppl 1:S4-6. and implicate parkin in synaptic ubiquitination. Mol 34 Santoro L, Breedveld GJ, Manganelli F, Iodice R, Cell. 2009;36:1034-47. Pisciotta C, Nolano M, et al. Novel ATP13A2 27 Ben Gedalya T, Loeb V, Israeli E, Altschuler Y, Selkoe (PARK9) homozygous mutation in a family with DJ, Sharon R. Alpha-synuclein and polyunsaturated marked phenotype variability. Neurogenetics. fatty acids promote clathrin-mediated endocytosis and 2011;12:33-9. synaptic vesicle recycling. Traffic. 2009;10:218-34. 28 Nemani VM, Lu W, Berge V, Nakamura K, Onoa B, Lee MK, et al. Increased expression of alpha-synu-

120 Mutation in the SYNJ1 Gene Associated With Autosomal Recessive, Early-Onset Parkinsonism

Supporting online material

PCR Protocol The amplification reactions were performed in a total volume of 20 µl, containing 1× FastStart Taq DNA Polymerase buffer, 200 µM of each dNTP, 0.5 µM forward primer, 0.5 µM reverse primer, 0.5 unit FastStart Taq DNA Polymerase (Roche Diagnostics) and 20 ng genomic DNA. SYNJ1_1F/R was amplified with addition of 1XGC-melt (Roche Diagnostics). The PCR conditions were as follows: initial denaturation, 7 min 30 sec at 96ºC, followed by 9 cycles of: 30 sec denaturation at 96ºC; 30 sec annealing (1st cycle at 70ºC, with 1ºC/cycle decrease); 1 min extension at 72ºC. Then, 25 cycles of: 30 sec denaturation at 96ºC; 30 sec annealing at 60ºC; 1 min extension at 72ºC. Final extension: 5 min at 72ºC. The PCR reactions were purified using 5 units ExoI (Fermentas) and 1 unit Fast AP (Fermentas), 45’ 37oC, 15’ 80oC. Direct sequencing was performed using Big Dye Terminator chemistry ver. 3.1 (Applied Biosystems) as recommended by the manufacturer. Dye terminators were removed using SephadexG50 (GE Healthcare) and loaded on an ABI 3130XL Genetic Analyzer (Applied Biosystems). For sequence analysis the software packages Sequence Analysis version 5.3 (Applid Biosystems) and Seqscape version 2.6 (Applied Biosystems) were used.

H. sapiens SSVSCMPTMPPIPARSQSQENMRSSPNPFIT 3.1 P. troglodytes SSVSCMPTMPPIPARSQSQENMRSSPNPFIT P. abelii XVCKCMPTMPPIQLES-SQEXYAKFSNLFIT C. lupus SSVSC---MPPIPARSKSQENMRCSPNPFIT O. cuniculus GSVSCMPTMPPIPARSKSQENMRSSPNPFIT M. musculus SSVSCMPTRPPGPEESKSQESMGSSANPFP- G. gallus NTVSCMPAMPPAPTLNISQEHTRSSPNPFVT X. tropicalis SGITCMATLPPVPARTGFQDSTRISPNPFVN Supp. Figure S1. Analysis of the conservation of Ser1422 in the Synj1 protein homologs. Note. In other species, such as R. norvegicus, D. rerio, D. melanogaster, B. taurus, and C. elegans, the Ser1422 residue and the flanking region are not present. The protein sequences from these species are therefore not shown.

Supplementary videos are available on the journal website.

121 Chapter 3.1

Supp. Table S1. PCR primers and protocol for the amplification and sequencing of SYNJ1 fragments. Amplification primersa Primer name Primer sequence Amplicon size SYNJ1_1F TGAAGAGCCCACGCAAAAGG 430 bp SYNJ1_1R CTGCGGAGGCAGAGACTGG SYNJ1_2F ATTAAACGGGAAACTCAACCAGACC 535 bp SYNJ1_2R GAGGAAATACAGTGAAAGGGTTGTCG SYNJ1_3F GAGTTCAGATCAGTGGACAGGATGG 502 bp SYNJ1_3R GCATGTAGCACGCCTTTTTGG SYNJ1_4F AGAGGTGCAAGAAGAAATGGAATAGC 615 bp SYNJ1_4R GGCATTCTTTCCAACACGTATACCAC SYNJ1_5F TCATAGATCACATGTTGAAGGATCTCG 567 bp SYNJ1_5R AAGGCCCATAAGTAACCAAGAACAATC SYNJ1_6F CAAACCTGTATAGTGGCCTGTTTTGG 397 bp SYNJ1_6R AACTCTGTACTCGTCTAGACCCTTATTTGC SYNJ1_7F GTTATTTGATTAACGACTGTGTGTTCAGC 367 bp SYNJ1_7R AAGACCCACACTTCCAGCACTACC SYNJ1_8F GCCTCCACAGCTCCAACAGG 541 bp SYNJ1_8R GAAAGCTTTTTAGAATCCCAATTTCAGC SYNJ1_9F CAGGGATAGTTTTTGAGGCTTTAGGG 843 bp SYNJ1_9R CACTGCCTGCTCCATAAAATGC SYNJ1_10F GGAGAGGGAAGGTGGATGCTACA 483 bp SYNJ1_10R TTTCCTCCAAAAGACAGGGAAAAA SYNJ1_11F GTTTTCAGAAGCTAGTCCCAAATAGTGC 448 bp SYNJ1_11R CCACTGTGAAAGGAAGTATCAAAGAAGG SYNJ1_12F GCTTTTTCATTTCTTTGTGCAGAGC 546 bp SYNJ1_12R AACAGACAAACTTCCCTTGGAGAATG SYNJ1_13F ACTGTAGGATTCTTCTTAGGTGCAGAAGG 470 bp SYNJ1_13R GCAATTGCTACAGGAAGACAAAGAGC SYNJ1_14F TTGCCTCATACCAGGCTTTGC 475 bp SYNJ1_14R TCATCAATTAATGCATCCTCTGAAGTG SYNJ1_15F AATTTCATTCCACTCAAGAGTGTTTGC 374 bp SYNJ1_15R TTTTACAGTTTTACGAGCACCCTTCC SYNJ1_16F TTAGATTTTGAAATGTGGTTTTCTTTTGG 534 bp SYNJ1_16R TGGTATGGTACTTTGAGGGCAAGC SYNJ1_17F ATGGGGTTTTGGGGGTTTTG 475 bp SYNJ1_17R TTCTAGGAACTATCTTAAATTTCCAAAAACATTG SYNJ1_18F CAGGACATGGGAGGAGACAAGC 376 bp SYNJ1_18R TGCATATTTAAGTACAAAAAGACCCCAAG SYNJ1_19F GGGACCCACCCACTTCTGC 472 bp SYNJ1_19R TTTGATATTATGTCATTCCCTGCTTCC SYNJ1_20F CCAGCCTCAAACCACTGAAGC 533 bp SYNJ1_20R AACTTCACCAAAACTTGCAAACTGC SYNJ1_21F CTCCAGGCACTTTGCTGCAC 459 bp SYNJ1_21R AATTCTCTCCTTCCCCAAATAAGGTG SYNJ1_22F TTTTTACTTCCTGGCCTCTTGTGG 296 bp SYNJ1_22R TCCCCTCCCATTTAGATAATCTTGC SYNJ1_23F AAACTTGCATCTCTGGGCATCC 515 bp SYNJ1_23R CCAAGAAGAAGAAGAAAGATGAACTTGG SYNJ1_24F TCATGCTTAATCCATGTGTGTTTGC 463 bp SYNJ1_24R GAGAAAACAGGGTGCACTTTATCACC (Continues)

122 Mutation in the SYNJ1 Gene Associated With Autosomal Recessive, Early-Onset Parkinsonism

(Follows)

Amplification primersa Primer name Primer sequence Amplicon size SYNJ1_25F TGAAAATACATGCCCTTTCTTTACACC 481 bp SYNJ1_25R TGAAAACACCTTAGAGGAATCTACTCTGC SYNJ1_26F CAAAATCCTCATTTCTGGATGATTTATAAGC 520 bp SYNJ1_26R CAGCCTACCTGGGAAATGAAAGC SYNJ1_27F GGATTCAGATTTGGGCCAAGC 374 bp SYNJ1_27R AAGGAGGGGAAAGAGAAGTTTACAGC SYNJ1_28F TCTCATGCCCCCATTTCTGC 303 bp SYNJ1_28R GCTTGAGGTGAAGGCTGACTGC SYNJ1_29F TGAATGTAGCTATTCACCCTTAAAACTGG 492 bp SYNJ1_29R AAATGATAAAAAGCTGCCCAAATCC SYNJ1_30F CAAAAGCAAATGTTTTCACAGAATTTTACC 566 bp SYNJ1_30R TGTAGAAGGTGGGAGATGACAGAAGG SYNJ1_31F TCTCTTTTTCCTTCTGTCATCTCC 511 bp SYNJ1_31R ACTTCTTATAACAGGCAATAATTAAAATAGG SYNJ1_32A_F AATTTACATGCTTGCCCCTAAAAGG 470 bp SYNJ1_32A_R TTGCCTCTGATTCTTCAGACTTGG SYNJ1_32B_F ACCTGTTCCCACCCCAGACC 488 bp SYNJ1_32B_R GCTGATTACCCAGTAAGTCTGAACAAGC SYNJ1_32C_F TCTACATTAAAAATTAGCAACCCGAAAGG 500 bp SYNJ1_32C_R CAATGAGAAGGGTTTTTCCTTATTTCC 3.1 Sanger sequencing primersb Primer name Primer sequence

SYNJ1_S4R CTGCTGAAAAGGGGCATCG SYNJ1_S9R GCCTGGGCAACAAAGTGAGG a primers used for PCR amplification and Sanger sequencing reactions. b primers used only for Sanger sequencing.

Supp. Table S2. Protein accession numbers Accession numbers for the Synj1 homologs |NP_003886.3| synaptojanin-1 isoform a [Homo sapiens] |XP_531429.2| PREDICTED: synaptojanin-1 isoform 4 [Pan troglodytes] |XP_002830685.2| PREDICTED: LOW QUALITY PROTEIN: synaptojanin-1 [Pongo abelii] |NP_776893.2| synaptojanin-1 [Bos taurus] |XP_535580.2| PREDICTED: synaptojanin-1 isoform 1 [Canis lupus familiaris] |XP_416706.3| PREDICTED: synaptojanin-1 [Gallus gallus] |NP_001164888.1| synaptojanin-1 [Oryctolagus cuniculus] |NP_445928.2| synaptojanin-1 [Rattus norvegicus] |NP_001157955.1| synaptojanin-1 isoform a [Mus musculus] |XP_002938495.1| PREDICTED: synaptojanin-1-like [Xenopus (Silurana) tropicalis] |NP_001007031.1| synaptojanin-1 [Danio rerio] |NP_726155.1| synaptojanin, isoform A [Drosophila melanogaster] |NP_001023266.1| Protein UNC-26, isoform b [Caenorhabditis elegans]

Accession numbers for different human proteins containing the SAC1-like phosphatase domain |NP_003889.1| synaptojanin-2 isoform 1 [Homo sapiens] |NP_054735.3| phosphatidylinositide phosphatase SAC1 [Homo sapiens] |NP_055660.1| polyphosphoinositide phosphatase [Homo sapiens] |NP_055752.1| phosphatidylinositide phosphatase SAC2 isoform 1 [Homo sapiens] |NP_054735.3| phosphatidylinositide phosphatase SAC1 [Homo sapiens]

123 Chapter 3.1

Supp. Table S3. In-silico prediction of functional effects of variants.

Synj1 p.Arg258Gln Synj1 p.Ser1422Arg ZNF439 p.Arg311Cys

PolyPhen-2 HumDiva PROBABLY DAMAGING PROBABLY DAMAGING BENIGN HumVarb POSSIBLY DAMAGING POSSIBLY DAMAGING BENIGN SIFT DAMAGING (score=0) DAMAGING (score=0,01) DAMAGING (score=0,05)c

PROVEAN DELETERIOUS NEUTRAL NEUTRAL

SNP-and-GO DISEASE NA NEUTRAL

SNAP NON-NEUTRAL NON-NEUTRAL NEUTRAL

MUTATION TASTING DISEASE-CAUSING POLYMORPHISM POLYMORPHISM

Synj1 p.Arg1252Gln Synj1 p.Arg1429Gln Synj1 p.Gly1453Glu

PolyPhen-2 HumDiva PROBABLY DAMAGING BENIGN PROBABLY DAMAGING HumVarb PROBABLY DAMAGING BENIGN PROBABLY DAMAGING SIFT TOLERATED (score=0,07) TOLERATED (score=0,22) DAMAGING (score=0,01)

PROVEAN NEUTRAL NEUTRAL NEUTRAL

SNP-and-GO NA NA NA SNAP NON-NEUTRAL NON-NEUTRAL NON-NEUTRAL

MUTATION TASTING DISEASE-CAUSING POLYMORPHISM DISEASE-CAUSING a preferred model for evaluating rare alleles, dense mapping of regions identified by GWAS and analysis of natural selection. b preferred model for the evaluation of Mendelian disease-causing variants, which requires distinguishing mutations with drastic effects from the remaining human variation, including abundant mildly deleterious alleles. c the substitution is predicted to be damaging if the score is ≤ 0.05, and tolerated if the score is > 0.05. NA = not available

124 Mutation in the SYNJ1 Gene Associated With Autosomal Recessive, Early-Onset Parkinsonism

Supp. Table S4. Variants detected in the SYNJ1 gene.

SYNJ1 positiona Ref SNPb Nucleotide change Protein change Frequencyc exon1 ss778182305# c.54C>T p.18Gly= 0.004 exon2 rs61750221 c.123G>A p.41Ala= 0.008 intron2 ss778182306# c.241+47T>C 0.004 intron2 rs74625247 c.242-148A>C 0.059 intron2 ss778182298# c.242-99A>G 0.004 intron3 rs184110321 c.328+134A>G 0.004 intron4 rs76502784 c.596+9A>G 0.017 intron7 rs73194279 c.968+43A>G 0.025 0.326 exon8 rs2254562 c.1001A>G p.Lys334Arg MAF, dbSNP137: 0.293 MAF, EVS: 0.297 intron8 rs844975 c.1065+81G>A 0.525 intron8 rs143229035 c.1066-38A>C 0.008 intron10 rs75924460 c.1317+55T>C 0.038 intron10 rs111884129 c.1317+69T>C 0.034 intron10 rs73194278 c.1317+273A>G 0.004 intron10 rs844988 c.1318-3A>T 0.076 intron11 rs80299581 c.1470+27C>T 0.004 intron12 rs116389986 c.1628-59T>A 0.004 intron13 ss778182299# c.1651+61G>A 0.004 intron15 rs114386610 c.1928+51A>G 0.004 intron15 rs844998 c.1929-100T>C 0.072 intron18 ss778182300# c.2421+55C>T 0.004 intron18 rs2833936 c.2422-110T>A 0.038 3.1 intron21 ss778182301# c.2913-23G>T 0.004 exon25 ss778182302# c.3486A>G p.1162 Pro= 0.004 intron25 rs2406 c.3508+99A>T 0.591d intron25 rs189532601 c.3509-129G>T 0.013 intron25 rs34001349 c.3509-96T>A 0.038 c.3509-202_3509- intron25 rs142688189 0.008 199delGTTT intron27 rs146753923 c.3596-29T>G 0.008 intron28 rs2833929 c.3635-7G>T 0.305 intron29 rs11702774 c.3705+14C>T 0.229 intron29 rs56238975 c.3705+128G>A 0.025 0.004 exon30 rs144048853* c.3755G>A p.Arg1252Gln MAF, EVS: 0.00007 intron30 rs147026277 c.3815-251G>C 0.004 exon31 rs2230766 c.3838C>T p.Leu1280= 0.085 c.4214_4215in- 0.555 exon32 rs71640263 p.Val1405delinsValAsnThr sAATACT MAF, dbSNP137: 0.511 exon32 ss778182303# c.4264 A>C p.Ser1422Arg 0.004 0.004 MAF, dbSNP137: 1 ID het exon32 rs112469776* c.4286G>A p.Arg1429Gln (Bushman); no additional IDs available 0.004 exon32 rs61750217* c.4358G>A p.Gly1453Glu MAF, dbSNP137: 0.008 MAF, EVS: 0.014 exon32 rs146425050 c.4581G>A p.Glu1527= 0.004 a References for annotation of variants: GenBank n. NM_003895.3 and NP_003886.3 b SNP reference number in dbSNP137 c Minor allele frequency in 92 early-onset Parkinson’s disease and 26 atypical parkinsonism patients d Minor allele frequency in 29 early-onset Parkinson’s disease and 26 atypical parkinsonism patients # Submitter SNP (ss) numbers provided by dbSNP Group NCBI/NLM/NIH (http://www.ncbi.nlm.nih.gov/projects/SNP/) * Web tools were used to predict the functional effects of these rare variants (See Supp. Table S3)

125 126 3.2 PARK20 Caused by SYNJ1 Homozygous Arg258Gln Mutation in a New Italian Family

S. Olgiati,1,* A. De Rosa,2,* M. Quadri,1 C. Criscuolo,2 G.J. Breedveld,1 M. Picillo,2 S. Pappatà,3 M. Quarantelli,3 P. Barone,4 G. De Michele,2,# and V. Bonifati.1,#

Published in Neurogenetics 2014,15(3):183-8.

* These authors contributed equally to this work, as first authors; # These authors contributed equally to this work, as senior authors and corresponding authors.

1) Department of Clinical Genetics, Erasmus MC, Rotterdam, The Netherlands; 2) Department of Neurosciences and Reproductive and Odontostomatological Sciences, “Federico II” University, Naples, Italy; 3) Biostructure and Bioimaging Institute, National Research Council, Naples, Italy; 4) Department of Medicine and Surgery, CEMAND, University of Salerno, Salerno, Italy.

127 140 3.3 A New Turkish Family With Homozygous FBXO7 Truncating Mutation and Juvenile Atypical Parkinsonism

G. Yalcin-Cakmakli,1 S. Olgiati,2 M. Quadri,2 G.J. Breedveld,2 P. Cortelli,3 V. Bonifati,2,# and B. Elibol.1,#

Published in Parkinsonism & Related Disorders 2014, 20(11):1248-52.

# Corresponding authors.

1) Department of Neurology, Hacettepe University, Ankara, Turkey; 2) Department of Clinical Genetics, Erasmus MC, Rotterdam, The Netherlands; 3) Department of Biomedical and NeuroMotor Sciences (DIBINEM), Alma Mater Studiorum - Università di Bologna, and IRCCS Istituto delle Scienze Neurologiche di Bologna, Italy.

141

PART IV Other Movement Disorders 154 4.1 Paroxysmal Exercise-Induced Dystonia Within the Phenotypic Spectrum of ECHS1 Deficiency

S. Olgiati,*,1 M. Skorvanek,*,2,3 M. Quadri,1 M. Minneboo,1 J. Graafland,1 G.J. Breedveld,1 R. Bonte,1 Z. Ozgur,4 M.C.G.N. van den Hout,4 K. Schoonderwoerd,1 F.W. Verheijen,1 W.F.J. van IJcken,4 H. Fen Chien,5 E.R. Barbosa,5 H.C. Chang,6 S.C. Lai,6 T.H. Yeh,6 C.S. Lu,6 Y.H. Wu-Chou,7 A.J.A. Kievit,1 V. Han,2,3 Z. Gdovinova,2,3 R. Jech,8 R.M.W. Hofstra,1 G.J.G. Ruijter,1 W. Mandemakers,1 and V. Bonifati.1,#

Manuscript in press, Movement Disorders 2016

* These authors contributed equally to this work, as first authors; # Corresponding author.

1) Department of Clinical Genetics, Erasmus MC, Rotterdam, The Netherlands; 2) Department of Neurology, Safarik University, Kosice, Slovakia; 3) Department of Neurology, University Hospital L. Pasteur, Kosice, Slovakia; 4) Center for Biomics, Erasmus MC, Rotterdam, The Netherlands; 5) Department of Neurology, University of São Paulo, São Paulo, Brazil; 6) Neuroscience Research Center, Division of Movement Disorders, Department of Neurology, Chang Gung Memorial Hospital and Chang Gung University, Taoyuan, Taiwan; 7) Human Molecular Genetics Laboratory, Department of Medical Research, Chang Gung Memorial Hospital and Chang Gung University, Taoyuan, Taiwan; 8) Department of Neurology, Charles University in Prague, First Faculty of Medicine, Prague, Czech Republic.

155 176 4.2 A Recurrent Mutation in the C19orf12 Gene Causes MPAN in Turkey

S. Olgiati,1 O. Doğu,2 H.A. Hanagasi,3 Y. Diler,4 M. Gultekin,5 D. Kuipers,1 M. Quadri,1 G.J. Breedveld,1 J. Graafland,1 R. Sürmeli,4 G. Sünter,4 A.D. Yalçin,4 Z. Tufekcioglu,3 B. Bilgiç,3 B. Elibol,6 M. Emre,3 and V. Bonifati.1,#

Manuscript in preparation

# Corresponding author.

1) Department of Clinical Genetics, Erasmus MC, Rotterdam, The Netherlands; 2) Department of Neurology, Mersin University, Mersin, Turkey; 3) Department of Neurology, Istanbul Faculty of Medicine, Istanbul University, Istanbul, Turkey; 4) Department of Neurology, Istanbul Education and Research Hospital, Istanbul, Turkey; 5) Department of Neurology, School of Medicine, Erciyes university, Kayseri, Turkey; 6) Department of Neurology, Hacettepe University, Ankara, Turkey.

177

PART V General Discussion 190 5.1 General Discussion

191 Chapter 5.1

During the last three decades, linkage analysis and positional cloning have been largely applied to human diseases, including movement disorders, allowing the identification of many highly penetrant genetic factors. For many disorders, these findings have laid the foundation for understanding the pathogenetic mechanisms of the disease. More recently, genome-wide association studies (GWAS) have been very successful in identifying risk factors for complex diseases. However, for most disorders, GWAS have not explained large proportions of the disease burden so far. Now, the field of human genetics is undergoing a third technological revolution. Next-generation sequencing technologies (NGS) are fueling a new wave of gene discovery, holding great promises for the movement disorders field. The applications of these tools to movement disorders are discussed in chapter 1.1 of this thesis. The aim of this thesis was to investigate the genetic factors involved in the etiology of movement disorders. To achieve this goal, we applied different genetic approaches, including NGS-based strategies, but also traditional linkage and positional-cloning methods, and candidate gene analyses. This work allowed us to expand our current knowledge about movement disorders by I) identifying novel disease-causing genes, II) expanding the phenotype associated with known disease genes, III) proposing novel candidate risk genes, and IV) contributing to the clinical characterization of previously-described forms. These findings provide insights into the mechanisms of these disorders by supporting novel or previously-postulated hypotheses. Moreover, our results have direct clinical relevance, and will be important for the diagnostic work-up and the genetic counseling of the patients and their relatives.

The Genetic Architecture of Movement Disorders During the last twenty years, a large number of genetics factors have been implicated in the etiology of movement disorders, including rare Mendelian factors, moderately-rare factors with moderate effect size, and common variants with low effect size. The identification of such genetic factors has been very important, and it has allowed the scientific community to formulate novel hypotheses about the disease mechanisms. However, with few notable exceptions, these findings have not been translated into novel therapies for these disorders so far. Moreover, despite the successful identification of genetic factors, the genetic architecture of many movement disorders is still incompletely understood. In movement disorders with a clear monogenic etiology (especially rare, autosomal recessive disorders), many of the disease genes are now known. Moreover, it is anticipated that the application of NGS technologies and DNA arrays methods to family-based studies will resolve most of the remaining conditions in the near future. A good example

192 General Discussion of this type of disorders is neurodegeneration with brain iron accumulation (NBIA). This group of genetically heterogeneous disorders is now associated with mutations in ten genes, accounting for the majority of the cases (~70%) [1-3]. The identification and characterization of these different NBIA forms have been instrumental for the ongoing development of specific diagnostic algorithms that take clinical signs and brain imaging features into account. Our work described in chapter 4.2 contributes to the characterization of one of these forms, caused by mutations in the C19orf12 gene. For movement disorders with a more complex genetic architecture, our genetic knowledge is still quite limited. Parkinson’s disease (PD) is a perfect example. The disorder is characterized by a strong familial aggregation (the relative risk of PD for having a ﬁrst-degree relative with PD versus having no ﬁrst-degree relative with PD is 2.9), suggesting a genetic component in PD [4]. Consistently, GWAS studies have identified common variants associated with the disease in 24 loci [5]. However, these variants have only small effect size (0.7>OR<2.3) and they explain only a small proportion of the heritable component of PD [6]. In addition to variants with a small effect size, moderate risk factors for PD have also been identified. Mutations in GBA and one mutation in LRRK2 (p.Gly2019Ser) are moderate-to-strong risk factors for PD (OR ~5–10), and their prevalence is especially high in some populations [7,8]. Moreover, family-based studies have been successful in the identification of Mendelian factors for PD. Mutations in seven genes have been confirmed to cause clinically typical late-onset or early-onset PD: SNCA [9], LRRK2 [10,11], VPS35 [12,13], parkin [14], PINK1 [15], DJ-1 [16], and DNAJC6 [17]. Mutations in these genes are Mendelian causes for PD, but they are extremely rare in the general population and they explain only a small proportion of the patients. Of note, mutations in one of these genes, DNAJC6, were first identified in association with early-onset PD by the work described in chapter 2.1. In addition, mutations in multiple genes (ATP13A2 [18], PLA2G6 [19], FBXO7 [20,21], DNAJC6 [22,23], SYNJ1 [24,25], and RAB39B [26]) have been associated to juvenile forms of parkinsonism with additional, atypical clinical signs. Despite some clinical similarities with the typical PD, the presentation of these forms usually includes other clinical features that are not observed in typical PD, and they are likely repre- senting different disease entities. One of these forms, caused by a mutation in SYNJ1, was identified for the first time and characterized by the work reported in chapter 3.1 and 3.2. 5.1 Despite the identification of several disease-causing genes for PD and juvenile atypical parkinsonism, mutations in these genes are not found in most of the PD cases, including some large families with multiple affected individuals. This suggests that additional genes are likely to be discovered in the future.

193 Chapter 5.1

Parkinson’s Disease: One or Many? The heritable component of complex movement disorders, like PD, seems to be the result of a multitude of different genetic factors, covering the entire spectrum of risk, from low to high. Small effect-size factors are very common, but have limited clinical relevance. On the other side of the spectrum, high risk alleles are often good diagnostic and prognostic markers, but they explain only small minorities of patients. Moreover, the fundamental question arises of whether these genetic forms and the so called “idiopathic” patients share the same disease mechanisms, and ultimately, whether they are the same disease. To answer this question, we should first consider how movement disorders are currently defined. For most movement disorders, including PD, the disease is usually diagnosed by a combination of many criteria, including clinical presentation, response to therapies, pathological signs, and brain imaging patterns. The problem with this definition is that most of these features are not specific to one disease. Instead, they are usually observed also in other neurological disorders or even in healthy individuals. Taking PD as example, the disease definition is based on the clinical presentation (presence of parkinsonism), the response to dopaminergic therapies, and the neuronal loss in the substantia nigra pars compacta and in other brain areas, with the presence of Lewy bodies and Lewy neurites in the remaining neurons [27,28]. However, parkinsonism can be observed also in other lesions or dysfunctions of the basal ganglia; response to dopaminergic therapies is also a feature of other disorders (e.g., dopa-responsive dystonia); and Lewy bodies are present also in other disorders, including Alzheimer’s disease, dementia with Lewy bodies, and some forms of NBIA, but also in elderly individuals without neurological disorders (i.e., incidental Lewy bodies) [29,30]. Based on these observations, it is reasonable to say that the genetic heterogeneity of PD might reflect an intrinsic heterogeneity of the disease. In other words, using etiology to define the disorder, one could say that PD is likely made of different disease entities having different etiologies, but similar clinical and pathological outcomes. At the same time, some of the different disease entities that are diagnosed as PD are clinically indistinguishable. For example, some of the patients with Mendelian forms, like LRRK2 mutation carriers, are clinically indistinguishable from the rest of the patients (with no mutations in known genes). Therefore, one could argue thatLRRK2 mutation carriers do not have a different disease entity, and instead, they also have PD. In this view, the disorder can be seen as the result of many different factors, or a combination of them, all acting through one, or more likely few, molecular pathways. These factors can be genetic (e.g., LRRK2 mutations), but also aging, and most likely environmental and stochastic factors. Similarly to PD, other movement disorders can be described using

194 General Discussion these two opposite approaches: etiologic or pathogenetic. In my view, these two interpretations of the complexity of PD must coexist. For unknown reasons, multiple etiologic factors produce similar, clinical outcomes. Both the etiologic and the pathogenetic approach can be used to describe this complex disease entity. The etiologic approach helps us to better understand the complexity of this disorder by allowing the study of the disease mechanisms. At the same time, because the different etiologic factors involved in PD might alter only a limited number of molecular pathways, a pathogenetic approach might be more successful in the development of novel therapeutic strategies broadly applicable to the patients.

The Contribution of This Thesis In the complex scenario of the genetics of movement disorders, the identification and characterization of the genetic factors associated with these diseases is paving the way to a better understanding of these conditions. In this thesis we provide an original contribution to the current knowledge of the Mendelian factors for Parkinson’s disease (Part II), atypical parkinsonisms (Part III), paroxysmal exercise-induced dystonia, and NBIA (Part IV). Our findings provide novel insights into the pathological mechanisms of these disorders. The identification of DNAJC6 mutations in early-onset Parkinson’s disease (chapter 2.1) provide strong evidence for the involvement of synaptic vesicles recycling in PD. Previously, mutations in this gene were known to cause an atypical form of parkinsonism (PARK19) with juvenile onset (<11 years) [22,23]. We reported, for the first time, DNAJC6 mutations in patients with onset in the third-to-fifth decade, slow disease progression, absence of atypical clinical signs, and prominent response to dopaminergic therapies. This phenotype is similar to other monogenic forms of early-onset PD caused by mutations in parkin, PINK1, and DJ-1. Interestingly, the DNAJC6 mutations reported by us might allow some residual protein activity, while the mutations causing juvenile, atypical parkinsonism are predicted to cause a complete loss-of-function. This allowed us to propose a genotype-phenotype correlation in patients with DNAJC6 mutations, with complete loss-of-function variants causing the juvenile, atypical form (PARK19), whereas milder mutations (described in our study) cause early-onset PD. Our hypothesis has been partially confirmed by another group that reported a new patient with atypical juvenile parkinsonism caused by a novel homozygous truncating 5.1 mutation in DNAJC6 [31,32]. DNAJC6 encodes for the brain-specific isoform of auxilin, a protein involved in the clathrin-mediated endocytosis (CME). In neurons, this process is an important

195 Chapter 5.1 mechanism for the formation of clathrin-coated vesicles, and for the recycling of synaptic vesicles at the presynaptic terminal [33]. Consistently, another finding reported in this thesis points to the involvement of the CME pathway in the pathogenesis of parkinsonisms. In chapter 3.1 and 3.2, we identified and characterized a novel form of early-onset, atypical parkinsonism caused by a mutation in SYNJ1. The protein encoded by this gene, synaptojanin 1, is a close molecular partner of auxilin. Both proteins act in the last step of the CME, the vesicles uncoating. The uncoating process requires the coordinated actions of three proteins: the chaperone Hsc70, its co-chaperone auxilin, and the phosphatase Synj1. Hsc70 and auxilin operate in the clathrin disassembly [34], while Synj1 dephosphorylates

PI(4,5)P2, a molecule acting in the early-stages of CME to recruit the clathrin adaptor proteins [35,36]. The identification of mutations in DNAJC6 and SYNJ1 in patients with parkinsonism strongly supports the involvement of the synaptic vesicles recycling in the pathogenesis of the disorder. These findings add to a growing body of evidence supporting the involvement of this pathway. Common variants in GAK (encoding a ubiquitously expressed form of auxilin, called auxilin-2) were identified by GWAS as low-risk factors for PD [5,37]. The Lrrk2 kinase was recently described as a modulator of the activity of endophilin-A [38], a protein involved in vesicle formation and uncoating [39-42]. In addition, endophilin-A was also reported to interact with parkin [43]. Moreover, Vps35 is part of the retromer, a protein complex that mediates the transport of transmembrane protein from endosomes to the trans-Golgi network [44]. Finally, α-synuclein has been involved in the several step of the CME process [45-47]. In light of the findings reported in this thesis, and the previously existing evidence, the dynamic of vesicle endocytosis and trafficking at the synaptic terminal appears to have a central role in the pathogenesis of parkinsonism. In order to identify novel risk factors for PD, we also performed a large exome sequencing study in the genetically isolated population of Sardinia (chapter 2.3). Taking advantage of the unique genetic architecture of the isolated populations, we aimed at the identification of moderately rare variants with strong effect size for PD. To our knowledge, this was the first exome-wide attempt to identify moderately rare variants of moderate effect size for PD. Although no variants achieved genome-wide significance after Bonfer- roni correction, we reported a catalog of variants exclusively present in PD patients or enriched among them. These variants might point to genetic determinants of PD with moderate-to-strong effect size. However, these findings are not conclusive, and they should be further validated by independent studies. Of note, similar inconclusive results were achieved by another, recently published, large exome sequencing project in PD

196 General Discussion

[48]. Regarding the genomic regions investigated, the results of our study support the high degree of genetic heterogeneity of PD, even in the isolated population of Sardinia. This suggests that the identification of moderately-rare risk factors for PD using large exome sequencing projects requires much larger cohorts of patients and controls, or the application of gene-based or pathway-based analyses. To conclude the discussion of our work on parkinsonisms, part of this thesis was also devoted to the clinical and genetic characterization of the known forms of parkinsonism associated to SNCA triplications (chapter 2.2) and FBXO7 mutations (chapter 3.3). Both SNCA triplications and FBXO7 mutations are very rare causes of parkinsonism, with only few families described so far. Our identification and description of these families contributes to the ongoing characterization of these rare Mendelian disorders. In addition to that, our study on SNCA triplication emphasizes the possible role of copy number variants in the pathogenesis of movement disorders and shows the importance of gene dosage assays, also in the NGS era. In part IV of this thesis, we focused on to the study of movement disorders other than primary parkinsonisms. In chapter 4.1, we suggested ECHS1 mutations as one of the causes of paroxysmal exercise-induced dystonia (PED). PED are rare forms of paroxysmal dyskinesias in which attacks of abnormal involuntary movements are triggered by prolonged or sustained exercise. Mutations in ECHS1 were recently reported as a novel cause of Leigh syndrome (subacute, necrotizing encephalopathy), or of atypical forms with later onset (Leigh-like syndromes), with a broad range of clinical presentations [49,50]. In our study, we showed that the phenotype associated with ECHS1 mutations might be milder than reported earlier, and includes isolated paroxysmal exercise-induced dystonia. Therefore, the screening of this gene should be considered in the patients with otherwise unexplained PED. Most importantly, our findings have potential therapeutic implications. Indeed, diet regimens and detoxifying agents represent potential therapeutic strategies for ECHS1 deficiency, but first, their efficacy and safety should be investigated. Finally, in chapter 4.2 we studied the role of C19orf12 mutations in Turkish patients with NBIA. Mutations in C19orf12 have been recently associated with an autosomal recessive form of NBIA called mitochondrial-membrane protein associated neurodegeneration (MPAN) [51]. Our work identified a single C19orf12 mutation in eight unrelated Turkish patients with NBIA (40% of our cohort). MPAN is a rare disease, 5.1 and complete neurological evaluations have been reported only in a limited number of patients. Therefore, our work represents a novel contribution to the clinical characterization of this form of NBIA. Moreover, we identified a recurrent mutation in C19orf12 in Turkish MPAN patients. This finding has important consequences for the genetic

197 Chapter 5.1 counseling of NBIA patients with Turkish origins.

Genetics of Movement Disorders: the Road Ahead The genetic findings of the past years helped us to better understand the pathogenetic mechanisms of movement disorders but, in order to more fully understand the genetic architecture of these conditions, further work remains ahead. In the coming years, novel genetic factors will likely be identified and the previously-identified factors will be better characterized. For the most common movement disorders, like PD, GWAS have identified several loci associated with the disease. However, GWAS results usually nominate large genomic regions associated with the disease, not single genes. Moreover, the disease-associated variants are often not located in the coding regions. As a result, for most GWAS loci, it is difficult to pinpoint the gene responsible for the association. In the future, the GWAS loci identified should be characterized in order to understand the biological bases for the association. This process poses major challenges and will likely require the integration of different genomics and transcriptomics technologies. Despite the successes of GWAS, these studies are limited by the possibility to investigate only common variants (minor allele frequency, MAF > ~1% in the general population). The application of NGS technologies to large association studies will solve this major limitation, allowing the study of variants that are too rare to be investigated by GWAS [52]. In chapter 3.2, we reported the first example of this type of studies in PD. As discussed above, our results have been so far inconclusive, suggesting that PD has a high degree of genetic heterogeneity. For this reason, future studies should rely on larger cohorts of patients and controls and should apply gene-based or pathway-based analyses in order to increase the statistical power. Several methods for conducting such types of analyses have been already proposed [53]. In addition, the power of the study could be increased by evaluating rare variants pathogenicity, using the same tools described in the introduction of this thesis [54]. Last, another way to possibly increase the power of large NGS studies in PD would be the inclusion of patients with more extreme phenotypes, like cases with earlier onset or more aggressive disease progression. In disorders with a high level of genetic heterogeneity, the study of families with multiple affected individuals can represent a more effective experimental design. For this reason, the identification of novel genetic factors in movement disorders might also rely on family based studies. In chapters 3.1 and 4.1, we have applied a combination of classical positional cloning methods and NGS technologies in two families with movement disorders. In both studies, this approach successfully identified the disease-

198 General Discussion causing mutations. Moreover, these findings have direct clinical relevance. In the future, both population studies and family-based studies might still be applied to movement disorders, and should serve as complementary strategies to explore the whole spectrum of genetic factors associated with these conditions. Finally, it is important to remember that the identification of disease-causing genes should also rely on the important biological information that can be gathered by studying the effect of candidate mutations at the RNA and protein levels, and by using in vitro and in vivo models. Most importantly, these studies are fundamental to understand the pathogenetic mechanism of the disease and to translate the genetic findings into novel therapeutic strategies.

Conclusions Despite the great progresses of the last 20 years, much is still unknown about the etiology of movement disorders. Genetic studies are leading the way towards the identification of the molecular mechanisms of these disorders, but most likely, many genetic factors are still to be found. A more complete knowledge about the genetic architecture of these disorders could be the cornerstone for mechanistic studies of the involved proteins and molecular pathways. A clear understanding of the molecular mechanisms should provide hints for the development of novel and personalized therapies for these disorders.

References 1 Rouault TA. Iron metabolism in the CNS: impli- 6 Keller MF, Saad M, Bras J, Bettella F, Nicolaou N, cations for neurodegenerative diseases. Nat Rev Simon-Sanchez J, et al. Using genome-wide complex Neurosci. 2013;14:551-64. trait analysis to quantify ‘missing heritability’ in 2 Dusi S, Valletta L, Haack TB, Tsuchiya Y, Venco P, Parkinson’s disease. Hum Mol Genet. 2012;21:4996- Pasqualato S, et al. Exome sequence reveals mutations 5009. in CoA synthase as a cause of neurodegeneration 7 Healy DG, Falchi M, O’Sullivan SS, Bonifati V, with brain iron accumulation. Am J Hum Genet. Durr A, Bressman S, et al. Phenotype, genotype, 2014;94:11-22. and worldwide genetic penetrance of LRRK2-associ- 3 Gregory A, Hayflick S. Neurodegeneration with Brain ated Parkinson’s disease: a case-control study. Lancet Iron Accumulation Disorders Overview. GeneRe- Neurol. 2008;7:583-90. views®. 1993-2015. 8 Sidransky E, Nalls MA, Aasly JO, Aharon-Peretz J, 4 Thacker EL, Ascherio A. Familial aggregation of Annesi G, Barbosa ER, et al. Multicenter analysis of Parkinson’s disease: a meta-analysis. Mov Disord. glucocerebrosidase mutations in Parkinson’s disease. N 2008;23:1174-83. Engl J Med. 2009;361:1651-61. 5.1 5 Nalls MA, Pankratz N, Lill CM, Do CB, Hernandez 9 Polymeropoulos MH, Lavedan C, Leroy E, Ide SE, DG, Saad M, et al. Large-scale meta-analysis of Dehejia A, Dutra A, et al. Mutation in the alpha-sy- genome-wide association data identifies six new risk nuclein gene identified in families with Parkinson’s loci for Parkinson’s disease. Nat Genet. 2014;46:989- disease. Science. 1997;276:2045-7. 93. 10 Zimprich A, Biskup S, Leitner P, Lichtner P, Farrer M,

199 Chapter 5.1

Lincoln S, et al. Mutations in LRRK2 Cause Auto- 2009;72:240-5. somal-Dominant Parkinsonism with Pleomorphic 22 Edvardson S, Cinnamon Y, Ta-Shma A, Shaag A, Pathology. Neuron. 2004;44:601-7. Yim YI, Zenvirt S, et al. A deleterious mutation 11 Paisan-Ruiz C, Jain S, Evans EW, Gilks WP, Simon J, in DNAJC6 encoding the neuronal-specific clath- van der Brug M, et al. Cloning of the Gene Containing rin-uncoating co-chaperone auxilin, is associated with Mutations that Cause PARK8-Linked Parkinson’s juvenile parkinsonism. PLoS One. 2012; 7:e36458. Disease. Neuron. 2004;44:595-600. 23 Koroglu C, Baysal L, Cetinkaya M, Karasoy H, Tolun 12 Vilarino-Guell C, Wider C, Ross OA, Dachsel JC, A. DNAJC6 is responsible for juvenile parkinsonism Kachergus JM, Lincoln SJ, et al. VPS35 mutations in with phenotypic variability. Parkinsonism Relat Parkinson disease. Am J Hum Genet. 2011;89:162-7. Disord. 2013;19:320-4. 13 Zimprich A, Benet-Pages A, Struhal W, Graf E, 24 Quadri M, Fang M, Picillo M, Olgiati S, Breedveld Eck SH, Offman MN, et al. A mutation in VPS35, GJ, Graafland J, et al. Mutation in the SYNJ1 gene encoding a subunit of the retromer complex, causes associated with autosomal recessive, early-onset late-onset Parkinson disease. Am J Hum Genet. Parkinsonism. Hum Mutat. 2013;34:1208-15. 2011;89:168-75. 25 Krebs CE, Karkheiran S, Powell JC, Cao M, Makarov 14 Kitada T, Asakawa S, Hattori N, Matsumine H, V, Darvish H, et al. The Sac1 Domain of SYNJ1 Iden- Yamamura Y, Minoshima S, et al. Mutations in the tified Mutated in a Family with Early-Onset Progres- parkin gene cause autosomal recessive juvenile parkin- sive Parkinsonism with Generalized Seizures. Hum sonism. Nature. 1998;392:605-8. Mutat. 2013;34:1200-7. 15 Valente EM, Abou-Sleiman PM, Caputo V, Muqit 26 Wilson GR, Sim JC, McLean C, Giannandrea M, MM, Harvey K, Gispert S, et al. Hereditary early- Galea CA, Riseley JR, et al. Mutations in RAB39B onset Parkinson’s disease caused by mutations in cause X-linked intellectual disability and early-onset PINK1. Science. 2004;304:1158-60. Parkinson disease with alpha-synuclein pathology. Am 16 Bonifati V, Rizzu P, van Baren MJ, Schaap O, J Hum Genet. 2014;95:729-35. Breedveld GJ, Krieger E, et al. Mutations in the DJ-1 27 Tolosa E, Wenning G, Poewe W. The diagnosis of gene associated with autosomal recessive early-onset Parkinson’s disease. Lancet Neurol. 2006;5:75-86. parkinsonism. Science. 2003;299:256-9. 28 Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy 17 Olgiati S, Quadri M, Fang M, Rood JP, Saute JA, of clinical diagnosis of idiopathic Parkinson’s disease: Fen Chien H, et al. DNAJC6 mutations associated a clinico-pathological study of 100 cases. J Neurol with early-onset Parkinson’s disease. Ann Neurol. Neurosurg Psychiatry. 1992;55:181-4. 2016;79:244-56. 29 Gibb WR, Lees AJ. The relevance of the Lewy body 18 Ramirez A, Heimbach A, Grundemann J, Stiller B, to the pathogenesis of idiopathic Parkinson’s disease. J Hampshire D, Cid LP, et al. Hereditary parkinsonism Neurol Neurosurg Psychiatry. 1988;51:745-52. with dementia is caused by mutations in ATP13A2, 30 Wakabayashi K, Tanji K, Odagiri S, Miki Y, Mori encoding a lysosomal type 5 P-type ATPase. Nat F, Takahashi H. The Lewy body in Parkinson’s Genet. 2006;38:1184-91. disease and related neurodegenerative disorders. Mol 19 Paisan-Ruiz C, Bhatia KP, Li A, Hernandez D, Davis Neurobiol. 2013;47:495-508. M, Wood NW, et al. Characterization of PLA2G6 31 Elsayed LE, Drouet V, Usenko T, Mohammed IN, as a locus for dystonia-parkinsonism. Ann Neurol. Hamed AA, Elseed MA, et al. A novel nonsense 2009;65:19-23. mutation in DNAJC6 expands the phenotype of 20 Shojaee S, Sina F, Banihosseini SS, Kazemi MH, autosomal recessive juvenile-onset Parkinson disease. Kalhor R, Shahidi GA, et al. Genome-wide linkage Ann Neurol. 2016;79(2):335-7. analysis of a Parkinsonian-pyramidal syndrome 32 Olgiati S, Quadri M, Mandemakers W, Bonifati V. pedigree by 500 K SNP arrays. Am J Hum Genet. Reply to Letter by Dr. Elsayed and colleagues. Ann 2008;82:1375-84. Neurol. 2016;79:337-8. 21 Di Fonzo A, Dekker MC, Montagna P, Baruzzi 33 Kononenko NL, Haucke V. Molecular mechanisms of A, Yonova EH, Correia Guedes L, et al. FBXO7 presynaptic membrane retrieval and synaptic vesicle mutations cause autosomal recessive, early-onset reformation. Neuron. 2015;85:484-96. parkinsonian-pyramidal syndrome. Neurology.

200 General Discussion

34 Ungewickell E, Ungewickell H, Holstein SE, 44 Bonifacino JS, Hurley JH. Retromer. Curr Opin Cell Lindner R, Prasad K, Barouch W, et al. Role of Biol. 2008;20:427-36. auxilin in uncoating clathrin-coated vesicles. Nature. 45 Nemani VM, Lu W, Berge V, Nakamura K, Onoa B, 1995;378:632-5. Lee MK, et al. Increased expression of alpha-synu- 35 Cremona O, Di Paolo G, Wenk MR, Luthi A, Kim clein reduces neurotransmitter release by inhibiting WT, Takei K, et al. Essential role of phosphoinos- synaptic vesicle reclustering after endocytosis. Neuron. itide metabolism in synaptic vesicle recycling. Cell. 2010;65:66-79. 1999;99:179-88. 46 Burre J, Sharma M, Tsetsenis T, Buchman V, 36 Mani M, Lee SY, Lucast L, Cremona O, Di Paolo G, Etherton MR, Sudhof TC. Alpha-synuclein promotes De Camilli P, et al. The dual phosphatase activity of SNARE-complex assembly in vivo and in vitro. synaptojanin1 is required for both efficient synaptic Science. 2010;329:1663-7. vesicle endocytosis and reavailability at nerve 47 Ben Gedalya T, Loeb V, Israeli E, Altschuler Y, Selkoe terminals. Neuron. 2007;56:1004-18. DJ, Sharon R. Alpha-synuclein and polyunsaturated 37 Pankratz N, Wilk JB, Latourelle JC, DeStefano AL, fatty acids promote clathrin-mediated endocytosis and Halter C, Pugh EW, et al. Genomewide association synaptic vesicle recycling. Traffic. 2009;10:218-34. study for susceptibility genes contributing to familial 48 Farlow JL, Robak LA, Hetrick K, Bowling K, Boer- Parkinson disease. Hum Genet. 2009;124:593-605. winkle E, Coban-Akdemir ZH, et al. Whole-Exome 38 Matta S, Van Kolen K, da Cunha R, van den Bogaart Sequencing in Familial Parkinson Disease. JAMA G, Mandemakers W, Miskiewicz K, et al. LRRK2 Neurol. 2016;73:68-75. controls an EndoA phosphorylation cycle in synaptic 49 Peters H, Buck N, Wanders R, Ruiter J, Waterham endocytosis. Neuron. 2012;75:1008-21. H, Koster J, et al. ECHS1 mutations in Leigh disease: 39 Gallop JL, Jao CC, Kent HM, Butler PJ, Evans PR, a new inborn error of metabolism affecting valine Langen R, et al. Mechanism of endophilin N-BAR metabolism. Brain. 2014;137:2903-8. domain-mediated membrane curvature. EMBO J. 50 Haack TB, Jackson CB, Murayama K, Kremer LS, 2006;25:2898-910. Schaller A, Kotzaeridou U, et al. Deficiency of ECHS1 40 Masuda M, Takeda S, Sone M, Ohki T, Mori H, causes mitochondrial encephalopathy with cardiac Kamioka Y, et al. Endophilin BAR domain drives involvement. Ann Clin Transl Neurol. 2015;2:492- membrane curvature by two newly identified struc- 509. ture-based mechanisms. EMBO J. 2006;25:2889-97. 51 Hartig MB, Iuso A, Haack T, Kmiec T, Jurkiewicz E, 41 Milosevic I, Giovedi S, Lou X, Raimondi A, Collesi Heim K, et al. Absence of an orphan mitochondrial C, Shen H, et al. Recruitment of endophilin to clath- protein, c19orf12, causes a distinct clinical subtype of rin-coated pit necks is required for efficient vesicle neurodegeneration with brain iron accumulation. Am uncoating after fission. Neuron. 2011;72:587-601. J Hum Genet. 2011;89:543-50. 42 Verstreken P, Kjaerulff O, Lloyd TE, Atkinson52 Cirulli ET, Goldstein DB. Uncovering the roles of rare R, Zhou Y, Meinertzhagen IA, et al. Endophilin variants in common disease through whole-genome mutations block clathrin-mediated endocytosis but sequencing. Nat Rev Genet. 2010;11:415-25. not neurotransmitter release. Cell. 2002;109:101-12. 53 Lee S, Abecasis GR, Boehnke M, Lin X. Rare-variant 43 Trempe JF, Chen CX, Grenier K, Camacho EM, association analysis: study designs and statistical tests. Kozlov G, McPherson PS, et al. SH3 domains from a Am J Hum Genet. 2014;95:5-23. subset of BAR proteins define a Ubl-binding domain 54 Sham PC, Purcell SM. Statistical power and signifi- and implicate parkin in synaptic ubiquitination. Mol cance testing in large-scale genetic studies. Nat Rev Cell. 2009;36:1034-47. Genet. 2014;15:335-46. 5.1

201

Appendix Summary Movement disorders are neurologic syndromes in which there is either an excess of movements or a paucity of voluntary and automatic movements, unrelated to weakness or spasticity. These disorders are associated with dysfunctions of the basal ganglia and cerebellum, brain structures that together orchestrate the control of movements. During the last two decades, genetic studies of movement disorders identified many genetics factors associated with these diseases, including rare Mendelian factors, moderately-rare factors with moderate effect size, and common variants with low effect size. In the coming years, the application of novel DNA sequencing technologies to the study of movement disorders promises to unveil even more genetic factors, possibly leading to a full understanding of the genetic architecture of these disorders. The introduction (Part I) of this thesis reviews the modern strategies for the identification of genetic factors associated with Mendelian forms of movement disorders. After a brief definition of these disorders, we focus on the use of the next-generation sequencing technologies to identify Mendelian mutations in movement disorders (chapter 1.1). Our experimental work is presented in the next three parts of this thesis. Part II is focused on the genetic studies of Parkinson’s disease (PD). In chapter 2.1, for the first time, we report the identification of DNAJC6 mutations in families with early-onset PD. Our work delineate a novel form of autosomal recessive early-onset PD, caused by mutations in the DNAJC6 gene, and has important implications for the diagnostic work-up and genetic counseling of early-onset PD patients. In chapter 2.2, we report genetic and clinical findings in a novel family with early-onset parkinsonism associated with triplication of the SNCA locus. This study contributes to the genetic and clinical characterization of this very rare form. In chapter 2.3, we perform a large exome sequencing project in the isolated population of Sardinia to identify novel risk factors for PD with intermediate or strong effect size. In this study we identify a list of candidate variants highly enriched or exclusively present in PD, a resource of candidates for future studies. Notably, this was the first study analyzing a large cohort of PD patients at exome-wide level. Part III of this thesis focuses on the study of atypical forms of parkinsonisms. In chapter 3.1, we report the identification of a novel form of early-onset atypical parkinsonism caused by a homozygous mutation in SYNJ1. This study was the first to report the association between a SYNJ1 mutation and parkinsonism. In chapter 3.2, we identify a family with a milder phenotype caused by the same homozygous mutation in a different haplotype, suggesting that the two SYNJ1 mutations originated from different mutational events. Interestingly, the proteins encoded by SYNJ1 and DNAJC6 both act in the molecular pathway of the clathrin-mediated endocytosis, suggesting an involvement of

204 Appendix this pathway in the pathogenesis of early-onset parkinsonisms. Chapter 3.3 is focused on another form of parkinsonism, caused by mutations in the FBXO7 gene. In this study, we describe a novel family with juvenile atypical parkinsonism due to a homozygous mutation in the FBXO7 gene, contributing to the clinical and genetic characterization of this rare form. Part IV of the thesis contains our work in movement disorders other than parkinsonian syndromes. In chapter 4.1, we present clinical, genetics, functional, and metabolic findings in two siblings with dystonic disorders caused by compound heterozygous ECHS1 mutations. This work expands the phenotype associated with ECHS1 mutations, showing that mutations in this gene can also cause primary paroxysmal exercised-induced dystonia. In chapter 4.2, we report a recurrent C19orf12 mutation in Turkish patients with neurodegeneration with brain iron accumulation. Our work contributes to the phenotypic characterization of this recently identified form of neurodegeneration, and has important implications for the genetic testing of the patients with Turkish origins. Finally, in the last part of the thesis (chapter 5.1), we discuss the genetic architecture of movement disorders and the contribution of this thesis in the characterization of these disorders and in understanding their pathogenetic mechanisms. Appendix

205 Samenvatting Bewegingsstoornissen zijn neurologische syndromen waarbij er een overschot dan wel een tekort is aan vrijwillige en automatische bewegingen die niet gerelateerd zijn aan spierzwakte of spasticiteit. Deze bewegingsstoornissen zijn geassocieerd met een gestoorde functie van de basale ganglia en het cerebellum. Dit zijn de hersenstructuren die samen de controle over bewegingen orkestreren. In de afgelopen twee decennia hebben genetische studies naar bewegingsstoornissen onderliggende erfelijke factoren geïdentificeerd, variërend van zeldzame Mendeliaanse factoren met een groot effect tot veelvoorkomende varianten met een kleine effectgrootte. Met het toepassen van nieuwe DNA sequencing technieken zullen in de komende jaren nog veel meer genetische factoren geïdentificeerd worden en dit zal mogelijk leiden tot het volledig in kaart brengen van de genetische architectuur van deze stoornissen. In de introductie van dit proefschrift (deel I) worden de moderne strategieën voor de identificatie van genetische factoren besproken die zijn geassocieerd met Mendeliaanse vormen van bewegingsstoornissen. Na een korte beschrijving van deze stoornissen focussen we op het gebruik van next-generation sequencing technologieën voor de identificatie van Mendeliaanse mutaties in bewegingsstoornissen (hoofdstuk 1.1). De volgende drie delen van dit proefschrift beschrijven ons experimentele werk. In Deel II focussen we op het genetische onderzoek naar de ziekte van Parkinson (PD). In hoofdstuk 2.1 beschrijven wij de identificatie van DNAJC6 mutaties in families met vroeg opgetreden PD. Ons werk beschrijft een nieuwe vorm van autosomaal recessieve vroeg opgetreden PD veroorzaakt door mutaties in DNAJC6 en dat deze mutaties belangrijke consequenties hebben voor de genetische diagnostiek en counseling van patiënten met deze vorm van PD. In hoofdstuk 2.2 beschrijven we de genetische en klinische bevindingen in een nieuwe familie met vroeg opgetreden parkinsonisme welke geassocieerd is met een triplicatie van de SNCA locus. Deze studie draagt bij aan de genetische en klinische karakterisatie van deze zeldzame vorm van parkinsonisme. In hoofdstuk 2.3 beschrijven we de resultaten van een groot exome sequencing project in de geïsoleerde populatie van het eiland Sardinië met het doel nieuwe risicofactoren te identificeren met gematigde tot grote effectgrootte. In deze studie hebben we kandidaat-varianten geïdentificeerd die verrijkt dan wel exclusief aanwezig zijn in patiënten met PD. Deze varianten zijn een belangrijke bron voor toekomstig onderzoek. Opmerkelijk is dat dit de eerste studie is waarin een groot cohort patiënten met PD op exome-breed niveau wordt geanalyseerd. Deel III van dit proefschrift is gefocust op de studie van atypische vormen van parkinsonismen. In hoofdstuk 3.1 rapporteren we identificatie van een nieuwe vroeg opgetreden vorm van atypisch parkinsonisme veroorzaakt door een homozygote

206 Appendix mutatie in het gen SYNJ1. Dit was de eerste studie waarin de associatie tussen een SYNJ1 mutatie en parkinsonisme werd gerapporteerd. In hoofdstuk 3.2 beschrijven we een familie met een milder fenotype veroorzaakt door dezelfde homozygote mutatie in een ander haplotype. Dit suggereert dat de twee SYNJ1 mutaties voortkomen uit twee verschillende mutationele gebeurtenissen. Interessant is dat het eiwitten die worden gecodeerd door SYNJ1 en DNAJC6 beiden een rol spelen in clathrin-gemedieerde endocytose. Dit suggereert dat deze cascade een rol speelt in de pathogenese van vroeg opgetreden parkinsonismen. In hoofdstuk 3.3 focussen we op een andere vorm van parkinsonisme, veroorzaakt door mutaties in het gen FBXO7. In deze studie beschrijven we een nieuwe familie met een juveniele atypische vorm van parkinsonisme veroorzaakt door een homozygote mutatie in FBXO7. Deze studie draagt bij aan de genetische en klinische karakterisatie van deze zeldzame vorm van parkinsonisme. Deel IV van dit proefschrift bevat ons werk in het veld van bewegingsstoornissen, anders dan parkinsonistische syndromen. In hoofdstuk 4.1 presenteren we klinische, genetische, functionele en metabole bevindingen van de studie van twee broers met dystone stoornissen veroorzaakt door een compound-heterozygote mutatie in ECHS1. Dit werk breidt het fenotype uit dat geassocieerd kan worden met mutaties in ECHS1 en laat zien dat mutaties in dit gen ook paroxysmale inspannings-geïnduceerde dystonie kan veroorzaken. In hoofdstuk 4.2 beschrijven we een vaak voorkomende C19orf12 mutatie in Turkse patiënten met neurodegeneratie met ijzerstapeling in de hersenen. Ons werk draagt bij aan de fenotypische karakterisatie van deze recent geïdentificeerde vorm van neurodegeneratie en heeft belangrijke implicaties voor de genetische diagnostiek van patiënten met een Turkse achtergrond. Ten slotte, in het laatste deel van dit proefschrift (hoofdstuk 5.1) bespreken we de genetische architectuur van bewegingsstoornissen en de bijdrage van dit proefschrift aan de karakterisatie en het begrijpen van de onderliggende pathogenetische mechanismen van deze stoornissen. Appendix

207 Riassunto I disordini del movimento sono sindromi neurologiche che presentano un eccesso di movimenti o una scarsezza di movimenti sia volontari che automatici, in assenza di debolezza o spasticità. Queste malattie sono associate a disfunzioni dei gangli della base e del cervelletto, le strutture cerebrali che si occupano di orchestrare il controllo dei movimenti. Vent’anni di studi hanno portato all’identificazione di molti fattori genetici associati ai disturbi del movimento. Questi includono sia fattori Mendeliani, che fattori comuni nella popolazione che modificano di poco la probabilità di sviluppare la malattia, ma anche fattori con frequenza intermedia ed effetto intermedio. Nei prossimi anni, grazie all’applicazione delle nuove tecnologie di sequenziamento del DNA, un numero crescente di fattori verrà probabilmente identificato nel campo dei disturbi del movimento. Questo ci permetterà di comprendere appieno l’architettura genetica di queste malattie. L’introduzione (Parte I) di questa tesi tratta le strategie moderne per l’identificazione di fattori genetici associati a forme Mendeliane di disordini del movimento. Dopo una breve definizione di queste malattie, approfondiamo l’uso delle cosiddette tecnologie di sequenziamento di prossima generazione (next-generation sequencing) per l’identificazione di mutazioni Mendeliane nei disordini del movimento (capitolo 1.1). Le successive tre parti di questa tesi illustrano il lavoro sperimentale svolto. La seconda parte è dedicata agli studi genetici sulla malattia di Parkinson (MP). Nel capitolo 2.1 riportiamo, per la prima volta, l’identificazione di mutazioni nel gene DNAJC6 in famiglie con MP. Il nostro lavoro delinea una nuova forma autosomica recessiva di MP ad esordio precoce causata da mutazioni in DNAJC6. Questa scoperta ha delle importanti implicazioni per la diagnosi e la consulenza genetica di pazienti con MP ad esordio precoce. Nel capitolo 2.2 riportiamo studi genetici e clinici in una nuova famiglia affetta da parkinsonismo ad esordio precoce associata ad una triplicazione del locus del gene alpha sinucleina (SNCA). Questo studio contribuisce alla caratterizzazione genetica e clinica di questa forma rara di parkinsonismo. Nel capitolo 2.3, al fine d’identificare nuovi fattori di rischio per la MP con effetto intermedio o pronunciato, applichiamo su larga scala la tecnica di exome sequencing alla popolazione sarda, un isolato genetico. Il nostro lavoro ha portato all’identificazione di una serie di varianti che, in questo studio, sono più frequenti in individui con la MP. Questi dati forniscono una risorsa di candidati genetici che potranno essere validati in studi futuri. Questo è stato il primo studio di popolazione a riportare l’analisi dell’intero esoma in una larga coorte di pazienti con la MP. La terza parte di questa tesi è focalizzata sullo studio di diverse forme di parkinsonismi atipici. Nel capitolo 3.1 riportiamo l’identificazione di una nuova forma genetica di

208 Appendix parkinsonismo atipico ad esordio precoce causata da una mutazione omozigote del gene sinaptogianina (SYNJ1). Questo studio è stato il primo ad associare una mutazione in SYNJ1 a questo tipo di disordine. Nel capitolo 3.2 presentiamo una famiglia con un quadro clinico meno severo, causato però dalla stessa mutazione in omozigosi. Dopo aver analizzato la serie di alleli presenti in questa regione genomica nei pazienti delle due famiglie (geneticamente definito come “aplotipo”), suggeriamo che la mutazione sia probabilmente originata da due distinti eventi mutazionali. È interessante notare che sinaptogianina e auxilina, ovvero le proteine codificate dai geni SYNJ1 e DNAJC6, hanno entrambe un ruolo essenziale nel processo cellulare di endocitosi mediato da clatrina. Da qui l’importante scoperta che questo processo sia direttamente coinvolto nella patogenesi dei parkinsonismi ad esordio precoce. Il capitolo 3.3 è focalizzato su un’altra forma di parkinsonismo, quella causata da mutazioni nel gene FBXO7. In questo studio contribuiamo alla caratterizzazione clinica e genetica di questa forma rara descrivendo una nuova famiglia con parkinsonismo giovanile causato da una mutazione omozigote nel gene FBXO7. La quarta parte della tesi contiene il nostro lavoro sui disturbi del movimento diversi dalle sindromi parkinsoniane. Nel capitolo 4.1 descriviamo due fratelli con disturbi distonici differenti causati da mutazioni eterozigote composte del gene ECHS1 e supportiamo questa scoperta con dati clinici, genetici, funzionali e metabolici. Questo lavoro espande il fenotipo associato a mutazioni in ECHS1 e mostra come mutazioni in questo gene possano anche causare distonia parossistica primaria indotta da esercizio fisico (primary paroxysmal exercised-induced dystonia). Nel capitolo 4.2 riportiamo una mutazione ricorrente nel gene C19orf12 in pazienti turchi affetti da neurodegenerazione con accumulo cerebrale di ferro (neurodegeneration with brain iron accumulation). Il nostro lavoro contribuisce alla caratterizzazione fenotipica di questa forma di neurodegenerazione recentemente identificata e ha importanti implicazioni per le analisi genetiche di pazienti con origini turche. Per finire, nell’ultima parte (capitolo 5.1), discutiamo l’architettura genetica dei disordini del movimento ed il contributo di questa tesi nella caratterizzazione di queste malattie e nella comprensione dei loro meccanismi patogenetici. Appendix

209

Appendix

List of Publications 1) Olgiati S,* Skorvanek M,* Quadri M, Minneboo M, Graafland J, Breedveld GJ, Bonte R, Ozgur Z, van den Hout MCGN, Schoonderwoerd K, Verheijen FW, van Ijcken WFJ, Fen Chien H, Barbosa ER, Chang HC, Lai SC, Yeh TH, Lu CS, Wu-Chou YH, Kievit AJA, Han V, Gdovinova Z, Jech R, Hofstra RMW, Ruijter GJG, Mandemakers W, Bonifati V. Paroxysmal exercise-induced dystonia within the phenotypic spectrum of ECHS1 deficiency. Mov Disord. 2016. In press 2) Quadri M, Olgiati S, Sensi M, Gualandi F, Groppo E, Rispoli V, Graafland J, BSc,1 Breedveld GJ, Fabbrini G, Berardelli A, Bonifati V. PRKRA founder mutation causing childhood-onset generalized dystonia-parkinsonism (DYT16) in an Italian family. Mov Disord. 2016. In press 3) Olgiati S, Quadri M, Bonifati V. Genetics of movement disorders in the next-generation sequencing era. Mov Disord. 2016; doi: 10.1002/mds.26521. Epub ahead of print 4) Olgiati S, Quadri M, Mandemakers W, Bonifati V. Reply to Letter by Dr. Elsayed and colleagues. Ann Neurol. 2016 Feb;79(2):337-8. 5) Olgiati S,* Quadri M,* Fang M,* Rood JP, Saute JA, Fen Chien H, Bouwkamp CG, Graafland J, Minneboo M, Breedveld GJ, Zhang J; International Parkinsonism Genetics Network, Verheijen FW, Boon AJ, Kievit AJ, Jardim LB, Mandemakers W, Barbosa ER, Rieder CR, Leenders KL, Wang J,# Bonifati V.# DNAJC6 mutations associated with early-onset Parkinson’s disease. Ann Neurol. 2016 Feb;79(2):244-56. 6) Olgiati S,* Thomas A,* Quadri M, Breedveld GJ, Graafland J, Eussen H, Douben H, de Klein A, Onofrj M, Bonifati V. Early-onset parkinsonism caused by alpha-synuclein gene triplication: Clinical and genetic findings in a novel family. Parkinsonism Relat Disord. 2015 Aug;21(8):981-6. 7) Quadri M,* Kamate M,* Sharma S,* Olgiati S, Graafland J, Breedveld GJ, Kori I, Hattiholi V, Jain P, Aneja S, Kumar A, Gulati P, Goel M, Talukdar B, Bonifati V. Manganese transport disorder: novel SLC30A10 mutations and early phenotypes. Mov Disord. 2015 Jun;30(7):996-1001. 8) Lewthwaite AJ,* Lambert TD,* Rolfe EB, Olgiati S, Quadri M, Simons EJ, Morrison KE, Bonifati V, Nicholl DJ. Novel GCH1 variant in Dopa-responsive dystonia and Parkinson’s disease. Parkinsonism Relat Disord. 2015 Apr;21(4):394-7. 9) Quadri M,* Yang X,* Cossu G,* Olgiati S,* Saddi VM, Breedveld GJ, Ouyang L, Hu J, Xu N, Graafland J, Ricchi V, Murgia D, Guedes LC, Mariani C, Marti MJ, Tarantino P, Asselta R, Valldeoriola F, Gagliardi M, Pezzoli G, Ezquerra M, Quattrone A, Ferreira J, Annesi G, Goldwurm S, Tolosa E, Oostra BA, Melis M, Wang J, # Bonifati V.# An exome study of Parkinson’s disease in Sardinia, a Mediterranean genetic isolate. Neurogenetics. 2015 Jan;16(1):55-64. Appendix

211 10) Taglia I,* Mignarri A,* Olgiati S, Menci E, Petrocelli PL, Breedveld GJ, Scaglione C, Martinelli P, Federico A, Bonifati V, Dotti MT. Primary familial brain calcification: Genetic analysis and clinical spectrum. Mov Disord. 2014 Nov;29(13):1691-5. 11) Yalcin-Cakmakli G, Olgiati S, Quadri M, Breedveld GJ, Cortelli P, Bonifati V, Elibol B. A new Turkish family with homozygous FBXO7 truncating mutation and juvenile atypical parkinsonism. Parkinsonism Relat Disord. 2014 Nov;20(11):1248-52. 12) Olgiati S,* De Rosa A,* Quadri M, Criscuolo C, Breedveld GJ, Picillo M, Pappatà S, Quarantelli M, Barone P, De Michele G, # Bonifati V.# PARK20 caused by SYNJ1 homozygous Arg258Gln mutation in a new Italian family. Neurogenetics. 2014 Aug;15(3):183-8. 13) Verhoeven WM, Egger JI, Koolen DA, Yntema H, Olgiati S, Breedveld GJ, Bonifati V, van de Warrenburg BP. Beta-propeller protein-associated neurodegeneration (BPAN), a rare form of NBIA: novel mutations and neuropsychiatric phenotype in three adult patients. Parkinsonism Relat Disord. 2014 Mar;20(3):332-6. 14) Quadri M,* Fang M,* Picillo M,* Olgiati S,* Breedveld GJ, Graafland J, Wu B, Xu F, Erro R, Amboni M, Pappatà S, Quarantelli M, Annesi G, Quattrone A, Chien HF, Barbosa ER; International Parkinsonism Genetics Network, Oostra BA, Barone P, Wang J,# Bonifati V. # Mutation in the SYNJ1 gene associated with autosomal recessive, early-onset Parkinsonism. Hum Mutat. 2013 Sep;34(9):1208-15. 15) Alberio T, Pippione AC, Zibetti M, Olgiati S, Cecconi D, Comi C, Lopiano L, Fasano M. Discovery and verification of panels of T-lymphocyte proteins as biomarkers of Parkinson’s disease. Sci Rep. 2012;2:953. 16) Alberio T, Pippione AC, Comi C, Olgiati S, Cecconi D, Zibetti M, Lopiano L, Fasano M. Dopaminergic therapies modulate the T-CELL proteome of patients with Parkinson’s disease. IUBMB Life. 2012 Oct;64(10):846-52.

* These authors contributed equally, as first authors # These authors contributed equally, as senior authors

212 Appendix

PhD Portfolio Summary of PhD training and teaching Name: Simone Olgiati Department: Clinical Genetics, Erasmus MC Research school: MGC, Medisch-Genetisch Centrum zuid-west Nederland PhD period: February 2012 – March 2016 Promotors: Prof.dr. V. Bonifati, and Prof.dr. R.M.W. Hofstra

Workload 1) Training Year (ECTS/duration) Courses PhD Teaching Program of Biomedical Science: Genetics 2012 3 Biochemistry and Biophysics 2012 3 Cell and Developmental Biology 2013 3 Literature 2013 2

Safely Working in the Lab 2012 0.25 Next-Generation Sequencing Data Analysis 2012 1.4 Statistics 2013 2 Biomedical English Writing and Communication 2014 3 Browsing Genes and Genomes with UCSC – advanced 2015 0.2 Seminars and workshops Population Genomics and Complex Traits and Disease – Rotterdam, The Netherlands 2013 0.9 Sophia Research Days – Rotterdam, The Netherlands 2014 0.4 Masterclass prof. dr. Eric Lander – Amsterdam, The Netherlands 2015 0.25

2) Congresses & presentations

Oral Presentations World Congress of Parkinsonism and Related Disorders – Geneve, Switzerland 2013 1 MGC symposium – Rotterdam, The Netherlands 2014 1 American Society of Human Genetics Annual Meeting – Baltimore, USA 2015 1 World Congress of Parkinsonism and Related Disorders – Milan, Italy 2015 1 Dept. Clinical Genetics Research Meeting 2012-2015 4

International and national conferences World Congress of Parkinson’s Disease and Movement Disorders – Dublin, Ireland 2012 1 MGC PhD Workshop – Düsseldorf, Germany 2012 1 American Society of Human Genetics Annual Meeting – Boston, USA 2013 1 NVHG Autumn symposium – Arnhem, The Netherlands 2013 1 World Congress of Parkinsonism and Related Disorders – Geneve, Switzerland 2013 1 World Congress of Parkinson’s Disease and Movement Disorders – Stockholm, Sweden 2014 1 MGC symposium – Rotterdam, The Netherlands 2014 1 American Society of Human Genetics Annual Meeting – Baltimore, USA 2015 1 World Congress of Parkinsonism and Related Disorders – Milan, Italy 2015 1

(Continues) Appendix

213 (Follows) Workload 3) Teaching Year (ECTS/duration)

Lecturing Module MNEU-2.10, Genetics and Neurological Diseases, MSc Program Neuroscience 2013-2016 0.5/year

Supervising practicals Overerving in de praktijk, Thema Ba.3.A 2014-2016 0.25/year

Supervising students Ilaria Taglia’s internship (visiting PhD student, University of Siena, Italy) 2013 9 months Demy Kuipers’s internship (student, Hogeschool Rotterdam, The Netherlands) 2014 6 months

214 Appendix

Acknowledgments Many colleagues and friends contributed to this thesis with their work, their inspiring passion, and their support. I wish to thank them all! Some people were particularly important and I would like to mention them. Dear Prof. Bonifati, dear Vincenzo, thank you for your essential contribution to this thesis and for everything you have done for me. During these years, your contagious passion for science and your rigorous experimental approach inspired me to become a better scientist. Most importantly, you believed in my potential since the beginning, and you gave me a great opportunity that changed my life for good. I will never forget that, grazie! Dear Prof. Hofstra, dear Robert, thank you for the nice scientific discussions and for your comments on this thesis. I really appreciated the time you invested in being my promotor. Your suggestions have improved this book. I also would like to thank Prof. Reiko Krüger, Prof. Caroline Klaver, Prof. John van Swieten, Prof. Berry Kremer, and Dr. Anneke Kievit for their comments on this thesis and for taking part in my dissertation committee. It is an honor to have you all in my PhD committee. A great thank goes to all the patients and family members who were involved in our projects. Moreover, I thank the Stichting ParkinsonFonds and its president, Mr. Rene Kruijff, for their generous financial support to our research over the years. I would like to thank my paranymphs for their fundamental help in arranging everything for the defence, but also for their great support during these years. Dear Dr. Quadri, dear Marialuisa, your contribution to this thesis goes much beyond your role as paranymph. I cannot thank you enough. You started helping me even before I arrived in the Netherlands and you initiated me to the lab (and to the Dutch habits!). During the last four years you have been both a great colleague, and one of my best friends. Thank you for being such a nice person! Dear Christian Bouwkamp, you really are a terrific colleague and friend, thanks for everything you have done for me! During our discussions about science and life, you always showed me novel and interesting perspec- tives. I will miss your brightness, your revealing opinions, and your scientific jokes. I wish to thank the many researchers and clinicians who collaborated with us during these years. A special thank goes to the colleagues directly involved in the projects presented here, and especially those involved in multiple studies of this thesis. Among these, I wish to thank Dr. Anneke Kievit, Dr. Frans Verheijen, and Prof. Ben Oostra (Department of Clinical Genetics, Erasmus MC); Prof. Egberto Reis Barbosa and Dr. Hsin Fen Appendix

215 Chien (University of São Paulo, Brazil); Prof. Bülent Elibol and co-workers (Hacettepe University, Ankara, Turkey); Prof. Aldo Quattrone, Dr. Grazia Annesi, and co-workers (University Magna Græcia, Catanzaro, Italy). Moreover, for the fruitful collaborations with the BGI-Shenzen (China), I wish to thank Prof. Jun Wang, Mingyan Fang, Xu Yang, and associates. For our study on DNAJC6 mutations in Parkinson’s disease, I wish to thank Dr. Agnita Boon and Dr. Janneke Rood (Department of Neurology, Erasmus MC); Prof. Klaus Leenders (University Medical Center Groningen, The Netherlands); Prof. Laura Bannach Jardim, Prof. Carlos Rieder, and Dr. Jonas Saute (Hospital de Clinícas de Porto Alegre, Brazil). For our study on a family with alpha-synuclein triplication, I wish to thank Prof. Marco Onofrj and Prof. Astrid Thomas (G. d’Annunzio Univer- sity, Chieti/Pescara, Italy); Dr. Annelies de Klein, Hannie Douben, and Bert Eussen (Department of Clinical Genetics, Erasmus MC). For the large exome sequencing project in Sardinia, I wish to thank Dr. Giovanni Cossu, Dr. Maurizio Melis, and co-workers (General Hospital S. Michele, AOB “G. Brotzu”, Cagliari, Italy); Dr. Valeria Saddi (San Francesco Hospital, Nuoro, Italy); Dr. Stefano Goldwurm, Prof. Gianni Pezzoli, and their associates (Istituti Clinici di Perfezionamento, Milan, Italy); Dr. Mario Ezquerra, Prof. Eduardo Tolosa, and co-workers (Hospital Clínic de Barcelona, Spain); Prof. Joaquim Ferreira and Dr. Leonor Correia Guedes (University of Lisbon, Portugal). For their contribution to the SYNJ1 studies, I would like to thank Prof. Paolo Barone, Dr. Marina Picillo, and co-workers (University of Salerno, Italy); Prof. Mario Quarantelli and co-workers (National Research Council, Naples, Italy); Prof. Giuseppe De Michele, Dr. Anna De Rosa, and co-workers (University of Naples “Federico II”, Italy). For their contribution to the identification ofECHS1 mutations in patients with dystonia, I would like to thank Dr. Matej Skorvanek, Prof. Zuzana Gdovinova, and co-workers (Safarik University and University Hospital L. Pasteur, Kosice, Slovakia); Prof. Robert Jech (Charles University in Prague, Czech Republic); Dr. George Rujiter and colleagues (Department of Clinical Genetics, Erasmus MC); Dr. Wilfred van IJcken and associates (Center for Biomics, Erasmus MC); Prof. Chin-Song Lu, Prof. Yah-Huei Wu-Chou, and co-workers (Chang Gung Memorial Hospital and Chang Gung University, Taoyuan, Taiwan). For their work on Turkish patients with neurodegeneration with brain iron accumulation, I wish to thank Prof. Murat Emre, Prof. Hasmet A. Hanagasi, and colleagues (Istanbul University, Turkey); Prof. Okan Doğu (Mersin University, Turkey); Dr. Reyhan Sürmeli and colleagues (Istanbul Education and Research Hospital, Turkey); and Prof. Murat Gultekin (Erciyes university, Kayseri, Turkey). Finally, for his contribution in the FBXO7 study, I would like to thank Prof. Pietro Cortelli (University of Bologna, Italy). In addition, I would like to thank Prof. Steven Kushner (Department of Psychiatry,

216 Appendix

Erasmus MC) for the projects on the psychiatric families. Moreover, I would like to thank Prof. Mauro Fasano (University of Insubria, Varese, Italy), not only for his crucial role during my undergraduate education, but especially for his great support in finding a position as PhD student and for suggesting Vincenzo’s lab to me. Grazie Mauro! Scientific research is a team effort and I was lucky to work in a group of nice and talented people. Therefore, I wish to thank all the members of the PD group: Guido, Josja, Isa, Christian, Wim, Michelle, Roy, Tianna, Demy, Victor, Stephanie, Martyna, Ilaria, and Monica. Your work was essential for this thesis and I cannot thank you enough! I learned a lot from you all, both scientifically and personally and your dedication to science inspires me every day. Thank you for being always so nice and helpful with me! I also would like to thank the roommate of the office Ee-914: Agnese, Friedemann, Renske, Ruben, Cristina, Joachim, Gert, and Gert-Jan. Thank you for being always very kind with me, and for listening to my (often silly) questions. I also would like to thank all the members of the Department of Clinical Genetics (especially the Ee-930 and Ee-922 labs) for creating a nice working environment. In particular, I would like to thank the people who helped me with scientific comments and key questions on my work, especially Prof. Rob Willemsen, Dr. Alan Burns, Dr. Pim Pijnappel, Dr. Grazia Mancini, Dr. Tjakko van Ham, and Dr. Renate Hukema. A special thank goes to Bep Hoogstraten, Jeannette Lokker, and Cindy Jongen for their secretarial support, to Sjozef van Baal, Mario Redeker, Ton Verkerk, and Nils Wijgers for the IT support, and to Ruud Koppenol and Tom de Vries Lentsch for the photographic support. Moreover, I would like to thank the members of the Department of Developmental Biology for being such friendly office neighbors. Finishing a PhD is much easier if you have a good balance between work and social life. Therefore, I am grateful to all my friends, especially the members of the international community in Rotterdam. Thank you guys for all the dinners and parties, especially the great evenings at Franco’s. Un ringraziamento speciale va infine a tutta la mia famiglia (includendo nonni, zii, cugini e famiglia Niglio!) ed in particolare ai miei genitori. La vostra spinta é stata fondamentale per convincermi ad iniziare un dottorato in Olanda (grazie mamma!). Il vostro entusiasmo e il vostro amore incondizionato mi hanno permesso di proseguire questo percorso fino alla fine. Grazie per tutto quello che avete fatto per me. L’ultimo ringraziamento va alla mia compagna. Cara Flavia, grazie per esserti lanciata senza dubbi in questa avventura olandese con me. Insieme abbiamo condiviso gioie e dolori di questa scelta complicata. Senza le tue attenzioni e il tuo supporto quotidiano non ce l’avrei

mai fatta. Questo libro é per entrambi. Appendix

217