Musician’s dystonia (MD) is a task-specific movement disorder that is characterized by painless muscle incoordination or loss of voluntary motor control while the musician is playing the instrument [1
]. It occurs in professional instrumentalists with a prevalence of 1–2% [3
]. About 20% of MD patients report a positive family history, including MD or writer’s dystonia (WD) [4
]. WD is another form of a task-specific dystonia involving the fingers, hand, and/or forearm. Symptoms usually appear when a person is trying to do a task that requires fine motor movements, such as writing. Of note, MD may be accompanied by additional WD [5
]. Thus, MD and WD may share a molecular cause.
However, little is known about the etiology of MD and WD and both environmental and genetic factors have been discussed. Among the environmental risk factors for MD, extensively trained maximal fine-motor skills and high levels of anxiety and perfectionism have received increasing attention over the past few years [6
]. On the other hand, a considerable genetic contribution is suggested by its high heritability [4
]. In a recent genome-wide association study, an intronic variant (rs11655081) in the arylsulfatase G (ARSG
) gene was identified as a potential genetic risk factor [9
]. However, known monogenic causes of segmental and generalized dystonia including mutations in TOR1A
, and GNAL
have been excluded as a major cause in MD [4
RAB12, member RAS oncogene family, is part of a large family of small guanosine triphosphate (GTP)-hydrolyzing enzymes (GTPases) that play an important role in vesicle transport and trafficking within cells [12
]. RAB12 regulates the degradation of transmembrane proteins on the membranes of different cellular compartments including the Golgi complex, endosomes and lysosomes. One well characterized target of RAB12 degradation includes the transferrin receptor 1 (TFRC) [14
]. RAB12 is highly expressed in the human brain (The Human Protein Atlas, http://www.proteinatlas.org/ENSG00000206418-RAB12/tissue
) and many RAB genes have been linked to neurological disorders. For instance, mutations in RAB3
are associated with Warburg Micro syndrome [17
], and functional impairments in RAB7
have been linked to Charcot-Marie-Tooth disease Type 2B [18
]. More recently, mutations in RAB39B
were postulated as a rare cause of Parkinson’s disease [19
In the present study, we initially used next-generation sequencing (NGS) in three German families with autosomal dominantly inherited MD/WD to unravel the presumably monogenic disease cause. We identified a likely pathogenic variant in the RAB12 gene in two out of three families. Genetic screening of unrelated patients revealed two additional RAB12 mutation carriers among MD patients. To confirm a functional effect of the identified mutations, we analyzed the consequences of these RAB12 mutations on GTPase activity, its intracellular localization, lysosomal distribution, TFRC degradation, and autophagy. Additional RAB12 variants were found in other dystonia patients but were largely absent in the investigated controls as well as in publically available databases.
2. Materials and Methods
The study was approved by the ethics committee at the University of Lübeck (Lübeck, Germany, No 04-180 from 1 July 2005). All participants gave written informed consent. We included a total of 1906 subjects (Table 1
) comprising 241 professional musicians diagnosed with MD, 14 relatives from four MD families (Figure 1
), 74 WD patients, 604 other dystonia patients (Table S1
), 512 patients with Parkinson’s disease (PD), and 461 healthy controls from the population-based control cohort EPIPARK from Lübeck (Germany) [21
]. All subjects were of European origin (mainly German) with the exception of 86 dystonia patients from South Korea. Three of the MD patients and their families (Families A–C, Figure 1
) were previously reported [4
] and originated from Germany. The diagnostic work-up of MD patients included a complete neurological examination and visual inspection while patients were playing their instruments. All other patients were diagnosed by movement disorder specialists in Lübeck, Kiel, Tübingen (Germany), Seoul (South Korea), and Belgrade (Serbia).
2.2. Mutation Screening
We initially performed exome or genome sequencing in 2011/2012 in six patients from three families (Figure 1
). Specifically, genome sequencing was carried out in three patients of Families B and C (Complete Genomics, Mountain View, AB, Canada) while three patients of Family A were exome sequenced on an Illumina Genome Analyzer (Atlas Biolabs, Berlin, Germany). Variant calling and annotation were performed by Complete Genomics and Atlas Biolabs, respectively. Detected variants were filtered (a) to be exonic or splicing; (b) to affect amino acid sequence (synonymous variants were discarded); (c) to be rare, with a known frequency < 1% in the database for single nucleotide polymorphisms (dbSNP132, http://www.ncbi.nlm.nih.gov/projects/SNP/ snp_summary.cgi?build_id=132
); and (d) to be shared among definitely affected members within a family (Table S2
). We further hypothesized that at least two of the families shared a mutation in the same gene. Candidate variants were validated by Sanger sequencing. All available family members were tested for segregation using Sanger sequencing.
We recently (in 2016) repeated exome sequencing in Family A (3 affected) and B (3 affected), and also included Family D (patient-parent trio) at Centogene, Rostock, Germany using an Illumina HiSeq 2000 machine and an in-house-annotation pipeline. In each of these families, we filtered for rare, protein-changing variants that were shared by all affected within a given family (Table S3
Based on the NGS analyses, a rare variant in RAB12
was the only plausible candidate. Next, we used Sanger sequencing to screen Exons 2 to 6 of RAB12
in 238 unrelated German MD patients, 54 WD patients, 378 patients with different forms of dystonia, and 461 unrelated healthy subjects. Exon 1, which has a high GC content of 78% and thus was difficult to amplify, was tested in all MD and WD patients as well as in 170 healthy controls. Primer sequences are shown in Table S4
Furthermore, all exons and exon/intron boundaries of RAB12 were included in an NGS-based gene panel analysis (Centogene, Rostock, Germany) that was performed for another 246 dystonia and 512 PD patients.
2.3. cDNA Analysis
A synonymous variant (c.276A>G, p.Arg92=) in Exon 2 of RAB12
that was predicted by MutationTaster to affect splicing of RAB12
was investigated on the RNA level. For this, RNA was extracted from a blood sample of a carrier using the QIAmp RNA Extraction Kit (QIAGEN, Germantown, MD, USA). Oligo-dT-Nucleotides of the Maxima First Strand cDNA Synthesis Kit (ThermoFisher, Waltham, MA, USA) served as primers to synthesize the complementary DNA (cDNA) by use of reverse transcriptase. PCR was performed with primers in Exons 1 and 4 (Table S4
) and the respective product was inspected for its size and Sanger sequenced.
2.4. Plasmid and Viral Construction
The complete coding sequence of RAB12
(RefSeq: NM_001025300.2) was subcloned into a pcDNA vector containing a FLAG TM
-tag upstream of RAB12
. The FLAG-tagged RAB12
sequence was then introduced into a lentiviral expression vector containing a puromycin resistance cassette. The mutations (c.38G>A, p.Gly13Asp and c.586A>G, p.Ile196Val) were introduced by site-directed mutagenesis (QuikChangeII, Stratagene, Cedar Creek, TX, USA, for primers see Table S4
). Sequences of all RAB12
constructs were verified by Sanger sequencing. For lentiviral production, Human Embryonic Kidney (HEK293T) cells were transfected with vectors containing vsv-g envelopes, pCMV delta R8.2, and expression vectors with the RAB12
constructs using FuGENE HD transfection reagent (Promega, Madison, WI, USA). The virus was harvested from the supernatant by ultracentrifugation and the titer was determined using the RETRO-TEK HIV p24 Antigen ELISA (enzyme-linked immunosorbent assay) (Zeptometrix, Buffalo, NY, USA).
2.5. Cell Culture and Stable Transfection
Skin biopsies were collected from two MD patients carrying the Ile196Val mutation and two healthy controls. Fibroblasts and SH-SY5Y cells were cultivated in Dulbecco’s modified eagle medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin (all medium components were provided by PAA Laboratories, Pasching, Upper Austria, Austria). The cells were incubated at 37 °C and 5% CO2 in a humidified atmosphere. Fibroblasts of the MD patients were used to study effects of mutations in endogenously expressed RAB12. For overexpression studies, SH-SY5Y cells and fibroblasts from a healthy control subject were stably transfected with lentivirus (multiplicity of infection = 5) containing expression vectors with RAB12 wildtype (WT) and mutated cDNA sequences (c.38G>A, p.Gly13Asp and c.586A>G, p.Ile196Val). Selection of transfected cells was performed with 2 µg/mL puromycin (Sigma, St. Louis, MO, USA).
2.6. GTPase Assay
Proteins from SH-SY5Y cells stably expressing RAB12 WT and mutated proteins were extracted with an isotonic lysis buffer (15 mM Tris/HCl pH 7.4, 150 mM NaCl, 1% NP-40, 1 mM EDTA (Ethylendiamin-tetraacetat) protease inhibitor cocktail (Roche, Basel, Switzerland)) and subsequently centrifuged at 4 °C at 16,000× g
for 20 min. Protein concentration was measured using the Dc Protein Assay (BioRad, Hercules, CA, USA). Whole protein extracts were used for the measurement of GTPase activity because effector proteins of RAB12, except for the Guanine nucleotide exchange factor DENND3 (DENN domain containing 3) [23
], are largely unknown. The rate of inorganic phosphate that was newly produced within one hour of incubation was determined with the ATPase/GTPase Activity Assay Kit (Sigma), which was used according to the manufacturer’s protocol. Absorption was measured at 620 nm. The specific GTPase activity was calculated and data were normalized for the GTPase activity in RAB12 WT in each experiment. Means of four independent experiments were calculated.
2.7. RAB12 Protein Structure Modeling and Molecular Dynamics Simulations
The X-ray structure of inactive, guanosine diphosphate (GDP)-bound WT RAB12 (Protein Data Bank [PDB] ID: 2IL1, www.rcsb.org
) was prepared using the Protein Preparation workflow (Schrödinger Suite 2014-2 Protein Preparation Wizard; Epik version 2.8; Impact version 6.3; Prime version 3.6, Schrödinger, LLC, New York, NY, 2014) [24
]. Inactive p.Ile196Val-RAB12 was modeled with Schrödinger Prime (version 3.6, Schrödinger, LLC, New York, NY, 2014) based on the 2IL1-X-ray. Active, GTP-bound RAB12 models were built using the RAB1A X-ray structure (Protein Data Bank (PDB) ID: 3TKL [25
], chain A, resolution: 2.18 Å) and the homology-modeling module of Schrödinger Prime (version 3.6) with default settings. Non-template loops were refined by loop refinement (Prime, version 3.6). Additionally, the models were optimized employing the OPLS2005 [26
] force field for energy minimization. Validation of the homology models was done using Prosa2003 [27
] and PROCHECK. [28
] All molecular dynamics simulations were performed using the Desmond package (Desmond Molecular Dynamics System, version 3.8, D. E. Shaw Research, New York, NY, USA, 2014) [29
] and the OPLS 2005 force field [26
]. Prepared X-ray structures and homology models of RAB12 were used as starting structures; each system was solvated in an orthorhombic box of TIP3P (transferable intermolecular potential with 3 points)-modeled water molecules [30
]. The simulations were carried out with the default protocol provided in Desmond. Molecular dynamics simulations were performed at 300 K and 1bar for either 10 ns or 20 ns, if not stable after 10 ns simulation.
For immunostaining, patient-derived fibroblasts, control fibroblasts, and fibroblasts stably expressing WT and mutant forms of RAB12 were cultured on glass cover slips in a 24-well plate. Cells were fixed with 4% paraformaldehyde (Sigma) and FLAG-tagged RAB12 proteins were detected with a primary antibody raised against FLAG (1:1000, rabbit, Cell Signaling, Cambridge, UK). For detection of lysosomes and TFRC, monoclonal antibodies raised against LAMP-1 (lysosomal associated membrane protein 1) as a lysosomal marker (1:200, mouse, Santa Cruz, Dallas, TX, USA) and TFRC (1:300, mouse, Invitrogen, Carlsbad, CA, USA) were used. The Endoplasmic reticulum (ER) and the Golgi apparatus were detected with monoclonal antibodies raised against PDI (1:500, mouse, Abcam, Cambridge, UK) and 58K Golgi protein (1:50,000, mouse, Abcam), respectively. Alexa fluor 488 and Alexa fluor 568 coupled secondary antibodies (1:400, goat, Life Technologies, Carlsbad, CA, USA) were utilized for visualization. Cell nuclei were stained with 1 µg/mL DAPI in DAPI Fluoromount G mounting medium (Southern Biotech, Birmingham, AL, USA). Images were analyzed with a Zeiss Axiovert 200 M microscope (Zeiss, Oberkochen, Germany) with ApoTome and Axiovision Rel 4.8 software (Zeiss, Oberkochen, Germany). For each of the three constructs (1 WT, 2 mutants), we randomly chose areas for analysis. Within these areas, all cells with intact nuclei (in total 110 cells per construct) were analyzed and assigned to one of three groups according to their properties: (a) RAB12 and lysosomes were uniformly distributed in the cytosol; (b) RAB12 and lysosomes were located in the perinuclear region and in large parts of the cytosol; (c) RAB12 and lysosomes accumulated exclusively in the perinuclear region.
2.9. Autophagy Inhibition and TFRC Degradation
SH-SY5Y cells stably expressing mutant and WT RAB12 were treated with the lysosomal inhibitor Bafilomycin A1 (3 nM, Sigma) for 24 h. Cells were detached using Accutase (PAA laboratories, Pasching, Austria) and proteins were extracted with RIPA buffer (50 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1% deoxycholate (DOC), 1% NP-40 and 0.1% sodium dodecyl sulfate (SDS)). After centrifugation at 4 °C at 16,000× g for 20 min, the proteins of the supernatant were used for Western blot analysis. Protein concentrations were determined utilizing Dc Protein Assay (BioRad) and 10 µg of the proteins were loaded on NuPAGE 4–12% Bis-Tris protein gels (Life technologies). Proteins were then transferred to a nitrocellulose membrane (Protran, AnalytikJena, Jena, Germany) and primary antibodies raised against FLAG (1:1 × 107, mouse, Sigma), TFRC (1:20,000, mouse, Invitrogen), p62 (1:1000, rabbit, Cell Signaling), LC3 (1:4000, rabbit, Novus, Wiesbaden Nordenstadt, Germany), and β-actin (1:1 × 106, mouse, Cell Signaling) were used. The signal intensities of TFRC, p62, β-actin, and LC3 II (light chain 3, also known as microtubule associated protein 1 light chain 3 alpha) were analyzed densitometrically with the Totallab 100 v2006 software. For calculations of TFRC degradation, the protein bands of bafilomycin-treated cells were set to 100%.
2.10. Endogenous TFRC Levels in Patients’ Blood
To test for the concentration of soluble TFRC levels in the blood of available patients from Family A (L-2283, L-2381, L-2276), we used routine blood count measurements from local medical laboratories.
2.11. Statistical Analysis
To compare frequencies of diseased and healthy RAB12
variant carriers, Chi-square test (comparison with frequencies in about 60,000 individuals from the Exome Aggregation Consortium (ExAC) at http://exac.broadinstitute.org/
) and Fisher’s exact test (comparison with our control individuals) were performed. All statistical tests related to the functional assays were performed with Graph Pad Prism 6. One-way analysis of variance (ANOVA) with Bonferroni’s post-hoc test was carried out to analyze the effect of RAB12
mutations on the GTPase activity, TFRC degradation, and the activation of autophagy. A Chi-square test was used for comparison of the cellular distribution of RAB12 and LAMP-1 in fibroblasts stably expressing WT or mutant RAB12
. Likewise, for comparison of LAMP-1 distribution in patient and control fibroblasts, a Chi-square test was utilized.
We here report an enrichment of rare missense variants in dystonia patients (12/919, 1.3%), particularly in patients with MD (5/242, 2.1%) compared to healthy controls (1/461, 0.2%) and PD patients (0/512). Specifically, we detected the p.Ile196Val substitution in 3 of 241 patients of our specifically recruited MD patients and in two relatives with WD as well as in one unrelated WD patient. Rare missense variants in two additional MD patients included p.Gly13Asp and p.Ala174Thr. The latter change was found in a patient of South Korean origin who was initially grouped among the WD patients but co-incidentally also suffered from MD. Ala174Thr was also found in a patient with segmental dystonia including WD. A total of 5 additional carriers of rare variants were found among 604 patients with other forms of dystonia (cervical dystonia) including two carriers of p.Ile196Val, as well as one carrier each of a p.Ala174Thr, p.Ala148Thr, or p.Arg181Gln substitution. There were no family members available to test for segregation but all missense changes received a CADD score > 15 and were extremely rare (<0.0006) in public databases, including GnomAD (genome aggregation database, Table 2
For statistical analysis, we compared the number of missense variant carriers in dystonia patients (10/916, not including the initial families) and the non-dystonic subjects (1/973) which yielded a significant p
-value of 0.005 (two-tailed Fisher´s exact test). Although this included three South Korean dystonia patients for whom we did not have ethnically matched controls, the difference for our overall study remains significant even when focusing on Caucasians by taking out the 86 South Korean patients (7/830 vs. 1/973, p
= 0.0278 [Fisher´s exact test]). Of note, the Ala174Thr that we found in three South Korean patients is found almost exclusively in the East Asian population with a carrier frequency of 120/18864 in GnomAD (http://gnomad.broadinstitute.org/variant/18-8636254-G-A
) and seems to be enriched in dystonia patients (albeit not statistically significant; 3.5% vs. 1.3%).
is located on chromosome 18p11.22 and the encoded protein belongs to a large family of small GTPases, which play an important role in vesicle transport and trafficking within cells [12
]. RAB12 is reported to be located on the membranes of different cellular compartments, including the Golgi complex, endosomes, and lysosomes where it regulates degradation of transmembrane proteins [14
]. Of note, we here demonstrated colocalization with a lysosomal marker but not with the ER or the Golgi complex.
Our in vitro studies focused on the two initially identified missense variants (p.Gly13Asp and p.Ile196Val) in RAB12 among MD patients and revealed functional alterations, indicating several lines of support for a possible pathogenic role of these substitutions in RAB12. This includes an increased GTPase activity that was observed for RAB12 mutant cells despite the localization of the mutations outside of the reported GTP-binding sites of RAB12. This elevated enzymatic activity could not be explained by 3D modeling and simulations of p.Ile196Val. Of note, the Gly13Asp variant that showed a >2-fold increase in GTPase activity, could not be modeled due to lack of a suitable template. Considering the same genetic background in RAB12 overexpressing SH-SY5Y cells, an indirect effect of RAB12 mutations on GTP hydrolysis is also possible. Interestingly, we observed perinuclear accumulation of RAB12 and lysosomes in cells with RAB12 mutations, which may be related to the altered GTPase activity. This idea is supported by recently published findings showing perinuclear clustering of constitutively active RAB12 in RBL-2H3 (rat basophilic leukemia) cells [34
Furthermore, RAB12 is thought to be involved in iron metabolism by regulating the degradation of TFRC [15
]. Cells take up transferrin-bound TFRC via receptor-mediated endocytosis. This mechanism is essential for iron uptake in neural tissue [35
]. Of note, iron deficiency is implicated in some types of neurodegeneration [36
]. Furthermore, in several genetic forms of neurodegeneration with brain iron accumulation (NBIA), dystonia (DYT) is a prominent feature of the disease such as in NBIA/DYT-PANK2 [37
], NBIA/DYT/PARK-PLA2G6 [38
], NBIA/DYT/PARK-CP [39
], and NBIA/DYT-DCAF17 [31
]. Interestingly, in all three tested patients from Family A, the levels of soluble blood TFRC were reduced.
Besides the reported regulatory impact of RAB12 on TFRC degradation, RAB12 is thought to play a role in the initiation of autophagy [16
]. We speculated that elevation of the GTPase activity and alteration of the subcellular localization of lysosomes in the mutants may have an impact on autophagy. We found a slight increase in relative LC3II protein levels in RAB12 mutants but p62 levels were not affected, indicating a rather minor effect of the RAB12 mutations on activation of autophagy.
Despite RAB12’s important role and plausible link between its function and a neurological disease, screening of additional patient cohorts is warranted to confirm the pathogenicity of RAB12 mutations in dystonia patients. The incomplete segregation in Families C and D as well as the relatively high number (n
= 83) of carriers of the p.Ile196Val mutation among approximately 140,000 seemingly unaffected individuals in GnomAD are at first glance not compatible with the hypothesis of a pathogenic role of RAB12 mutations in MD. However, it is conceivable that the affected father of Patient D who carries two RAB12 WT alleles and has WD without other neurological symptoms, presents a phenocopy with a different disease cause. Phenocopies are not an infrequent observation, especially in the context of hereditary movement disorders [41
]. Furthermore, there is the possibility that the missing disease phenotype in both the mother of Patient D—in contrast to the maternal grandmother - and mother of Family C could be explained by reduced penetrance, a phenomenon often seen in dystonia and other disorders [42
]. For instance, penetrance of the GAG deletion in TOR1A
is reduced to only 30% [43
] and 30 presumably unaffected carriers are included in GnomAD (http://gnomad.broadinstitute.org/variant/9-132576340-TCTC-T
). Protective genetic variants as well as environmental triggering or other tutelary factors are under discussion in the development of MD [8
]. The argument of reduced penetrance can also be applied to the 83 variant carriers in GnomAD. Supporting this idea, having or developing a focal dystonia in the future would not exclude individuals from being included into GnomAD. Similarly, in our about 1000 non-dystonia samples, we also detected only one carrier of a rare missense RAB12
variant. Of note, the age at onset in our patients was up to 76 years, i.e., late-onset. In theory, developing MD requires extensive training and professional performance in playing an instrument, e.g., in a pianist training especially the fingers of the right hand for ~26,000 h before the average onset of MD [44
]. Thus, an accumulating pathological effect (e.g., on TFRC degradation or autophagy) in specific brain regions is conceivable as disease mechanism but cannot be investigated in a cellular model.
Several of our results support the idea of a possible pathogenic role of RAB12 mutations in MD and probably also other dystonias including increased GTPase activities, altered lysosomal distribution, reduced levels of soluble TFRC in patients, and the finding that RAB12
mutations were found with a higher frequency in dystonia patients than in controls. Of note, RAB12
seems to be a highly invariable gene. Except for the six listed missense and two synonymous variants (Table 2
) we did not detect any additional variant in RAB12
among the almost 2000 screened individuals. The invariance of the 244-amino acid protein RAB12 is also underlined by the presence of only 55 missense and a single loss-of-function variant in ExAC (http://exac.broadinstitute.org/gene/ENSG00000206418
). The ExAC z-score of 1.63 for missense variants indicates that the number of observed variants in RAB12
was lower than the expected number reflecting decreased tolerance to missense variations in the RAB12
coding sequence [45
]. The intolerance to loss-of-function changes is relatively high with a pLI score (probability of loss-of-function intolerance) of 0.72 [45
], which also points to an important functional role of the RAB12 protein in humans. For comparison, in another, similarly sized dystonia-linked protein (THAP1 [THAP domain containing 1], 213 amino acids), ExAC reports a missense z-score of 1.35 and a pLI score of 0.90 while in CDKN1A, a protein of only 164 amino acids and not (yet) linked to any disease, more missense and loss-of-function variants were observed than expected resulting in a lower missense z-score and pLI score, respectively (z = 0.01, pLI = 0.03).