Mutations in RAB39B Cause X-Linked Intellectual Disability and Early-Onset Parkinson Disease with α-Synuclein Pathology

Abstract

Advances in understanding the etiology of Parkinson disease have been driven by the identification of causative mutations in families. Genetic analysis of an Australian family with three males displaying clinical features of early-onset parkinsonism and intellectual disability identified a ∼45 kb deletion resulting in the complete loss of RAB39B. We subsequently identified a missense mutation (c.503C>A [p.Thr168Lys]) in RAB39B in an unrelated Wisconsin kindred affected by a similar clinical phenotype. In silico and in vitro studies demonstrated that the mutation destabilized the protein, consistent with loss of function. In vitro small-hairpin-RNA-mediated knockdown of Rab39b resulted in a reduction in the density of α-synuclein immunoreactive puncta in dendritic processes of cultured neurons. In addition, in multiple cell models, we demonstrated that knockdown of Rab39b was associated with reduced steady-state levels of α-synuclein. Post mortem studies demonstrated that loss of RAB39B resulted in pathologically confirmed Parkinson disease. There was extensive dopaminergic neuron loss in the substantia nigra and widespread classic Lewy body pathology. Additional pathological features included cortical Lewy bodies, brain iron accumulation, tau immunoreactivity, and axonal spheroids. Overall, we have shown that loss-of-function mutations in RAB39B cause intellectual disability and pathologically confirmed early-onset Parkinson disease. The loss of RAB39B results in dysregulation of α-synuclein homeostasis and a spectrum of neuropathological features that implicate RAB39B in the pathogenesis of Parkinson disease and potentially other neurodegenerative disorders.

Main Text

Parkinsonism is a neurological syndrome characterized by tremor, rigidity, balance problems, and a slowing of movement. The most common cause of parkinsonism is Parkinson disease (PD [MIM 168600]), which accounts for up to 70% of this syndrome. PD is a common progressive neurodegenerative disorder with motor symptoms due to the death of dopamine-generating cells, predominantly in the substantia nigra (SN). The pathological hallmark of PD is accumulation of α-synuclein in Lewy bodies and Lewy neurites, although additional pathology (such as neurofibrillary tangles [NFTs]) can be observed.¹ Recent genetic studies have driven advances in understanding the molecular pathogenesis of PD, and preclinical discovery projects have investigated compounds that target the identified proteins as a precursor to etiology-based therapeutics. To date, 18 PD-associated loci have been reported, and variants in 13 monogenic or susceptibility genes have been identified.² Common pathogenic mechanisms associated with these genes include protein turnover, mitochondrial function, and oxidative-stress pathways. However, approximately 90% of individuals with PD do not have a defined genetic etiology. Variants in known genes account for ∼10% of the variation in PD liability, suggesting that variants in additional genes and susceptibility loci remain to be identified.^3,4

We identified an Australian kindred with three brothers who presented in childhood with nonprogressive intellectual disability (ID), which included delayed developmental milestones, cognitive impairment, and macrocephaly (Figure 1; Table S1, available online). Subsequently, early-onset parkinsonism (onset prior to 45 years of age) was also apparent, although the clinical progression and presentation varied. The proband developed tremor in late childhood, but the symptoms did not progress to frank parkinsonism. In contrast, his male siblings developed tremor from their late 30s and were diagnosed with L-DOPA-responsive akinetic-rigid PD by their mid-40s. A complete description of the phenotype is presented in Table S1. We collected samples from the Australian family after receiving institutional ethics approval from Royal Childrens Hospital (Melbourne) and written informed consent from participants. Genomic DNA was isolated from whole blood, and primary fibroblast cultures were generated according to standard protocols. SNP array and linkage analysis using a recessive homozygous model did not demonstrate linkage to the autosomes but did identify two ∼10.6 Mb haplotypes shared by the affected brothers at Xp22.2 and Xq27.3–qter (chrX: 3,624,034–14,291,092 and chrX: 145,644,895–tel, respectively; GRCh38/hg38, UCSC Genome Browser; Table S2). Copy-number variation and subsequent PCR analysis identified a ∼45 kb deletion within the Xq haplotype (ClinVar accession number SCV00019029). The deletion segregated with the disease and resulted in the complete deletion of RAB39B (RAB39B, member RAS oncogene family [MIM 300774]) and the last three coding exons of CLIC2 (chloride intracellular channel 2 [MIM 300138]). To assess RAB39B expression, we extracted total RNA from fibroblasts by using the SV Total RNA Isolation System (Promega) and synthesized cDNA with the Transcriptor First Strand cDNA Synthesis Kit (Roche). Consistent with the genomic data, the RAB39B and CLIC2 transcripts were not detected by RT-PCR analysis of fibroblast cells derived from affected individuals (Figure S1).

Figure 1.

Identification of Mutations in RAB39B

(A and B) Simplified pedigree structure of the Australian (A) and Wisconsin (B) kindreds. Asterisks indicate DNA samples analyzed.

(C) Immunoblot analysis (12162-1-AP, Proteintech; 1:1,000) of RAB39B in BE(2)-M17 neuroblastoma cells grown in the absence or presence of 10 μM MG-132. Abbreviations are as follows: Un, untransfected; WT, wild-type; and p.T168K, RAB39B p.Thr168Lys. β-actin was used for confirming equivalent protein loading (A5441, Sigma-Aldrich; 1:5,000).

(D) Representative images of mouse hippocampal neurons transduced with lentiviral vectors encoding GFP and either scramble (shScr) or shRab39b sequences. Images were captured, and Z-space slices (0.3–0.4 μm) were deconvolved and flattened by maximum projection. ImageJ analysis software (the “Gran filter” plug-in set the size from 1 to infinity) was used to measure α-synuclein density relative to the area of infected dendrites. Quantification showed a significant reduction (0.52 ± 0.03 versus 0.76 ± 0.03, mean ± SEM, p ≤ 0.0005) in the density of α-synuclein immunoreactive puncta in neurons downregulated for Rab39b. The experiment was performed in triplicate (n = 42 neurons scored). The scale bar represents 20 μm.

The phenotype of the Australian kindred is similar to the basal ganglia disorder (Waisman syndrome [MIM 311510]) reported for a Wisconsin kindred⁵ (Figure 1; Table S1). Members of the defining family included 13 affected males who presented with variable degrees of ID and early-onset parkinsonism. Multipoint linkage analysis of the Wisconsin family previously localized the disease-causing mutation to Xq27.3–qter with a maximum multipoint LOD of 6.75 at the genetic marker F8C (chrX:154,929,351–154,929,630; GRCh38/hg38). The minimal Xq linkage region in the Australian kindred is within the Xq27.3–qter interval and therefore defines the shared critical linkage interval.⁶ Genomic DNA from individuals of the Wisconsin kindred was kindly provided by Professor Ronald Gregg. Ethics approval was provided by the institutional review board (IRB) at the University of Wisconsin, and informed consent was obtained. Direct sequencing of RAB39B identified a missense mutation (c.503C>A [p.Thr168Lys]; RefSeq accession number NM_171998.2, ClinVar SCV000190018) that segregated with disease (Figure S2) and was not detected in 200 unrelated control individuals or public databases (dbSNP137 and ESP6500). The threonine residue is conserved in evolution (Figure S3), and the mutation is predicted to be damaging by PolyPhen-2 and SIFT. In contrast, no sequence variants were observed by direct sequencing of CLIC2 in the Wisconsin kindred.

The absence of mutations in CLIC2 in the Wisconsin kindred suggests that disruption of RAB39B is the cause of the shared phenotype in both families. It is possible that deletion of CLIC2 in the Australian kindred might act as a disease modifier, although the phenotypic features (seizures and cardiac anomalies) associated with a missense mutation in CLIC2 were not observed.⁷ X chromosome exome sequencing of affected males from both kindreds confirmed the deleterious RAB39B changes and did not identify any other candidate variants within the linkage regions (Table S3).

In silico modeling of RAB39B was performed with the structure prediction programs MODELER⁸ and HHpred,⁹ and protein structures were visualized and superimposed with PyMOL. This analysis suggested that Thr168 is buried within the wild-type protein and interacts with Leu60 in the interswitch region, which undergoes conformational changes upon GTP-GDP exchange^10,11 (Figure 2). The mutation introduces a large, positively charged lysine residue that is predicted by multiple algorithms, including ERIS¹⁵ and PoPMuSiC,¹⁶ to destabilize the protein. We could not directly test this because endogenous RAB39B was not detectable in fibroblast cells; therefore, we generated stable BE(2)-M17 neuroblastoma lines overexpressing wild-type RAB39B and altered (p.Thr168 Lys) RAB39B. The complete RAB39B open reading frame was amplified from human brain cDNA and cloned into the mammalian expression vector pcDNA3.1 (Invitrogen). We used site-directed mutagenesis (QuickChangeII) to generate the p.Thr168Lys altered RAB39B construct and Sanger sequenced all clones to verify that no additional variants were present. RT-PCR analysis confirmed similar expression of the wild-type and altered constructs (data not shown). In contrast, immunoblot analysis revealed high steady-state levels of exogenous wild-type RAB39B but very low levels of exogenous altered RAB39B. Immunoblot and immunofluorescence analysis of the cells after treatment with the proteasome inhibitor MG-132 confirmed that the reduced steady-state level of altered RAB39B was due to rapid turnover of the protein by the ubiquitin proteasome system (Figure 1; Figure S4). These results confirm the in silico modeling suggesting that the altered protein is destabilized and collectively demonstrate that loss of function of RAB39B causes ID and parkinsonism. Previous studies have associated RAB39B mutations with ID^17–19 (pedigrees D-23 and MRX72; Table S1). The absence of parkinsonism in these additional families could be due to the individuals’ age at reporting, given that our data suggest that parkinsonism is likely to manifest after the second decade, albeit with some variability in both onset and clinical severity (Table S1). However, the lack of clinical data and inability to re-examine affected individuals mean that it is difficult to determine whether the phenotype associated with loss of RAB39B function represents an age-dependent progression of ID and parkinsonism or a spectrum of heterogeneous phenotypes extending from ID to ID with parkinsonism (see below).

Figure 2.

In Silico Modeling of the RAB39B Structure Predicts that p.Thr168Lys Is Destabilizing

(A) We generated a molecular model of the RAB39B structure (blue) by superimposing the structures of several known RAB proteins (gray) available in the Protein Data Bank (YTP1 [PDB 2BCG], RAB1A [PDB 3TKL], RAB11B [PDB 2F9L], RAB18 [PDB 1X3S], RAB23 [PDB 1Z2A], and RAB30 [PDB 2EW1]). GTP (green sticks) and Mg²⁺ (magenta sphere) are from the structure of RAB5A after superposition.

(B) We then used the predicted RAB39B structure (blue) to model the position of the native (Thr168, red) and altered (Lys168, yellow) amino acid. In the native form, Thr168 is predicted to interact with Leu60 within the interswitch region of RAB39B. The internal placement of the large positively charged residue in the p.Thr168Lys protein is predicted to destabilize the protein. The quality of the model was tested with PROCHECK,¹² ANOLEA,¹³ and Verify3D.¹⁴

Rab GTPases belong to the Ras superfamily of small GTPases and act as essential regulators of vesicular trafficking. They dynamically localize to distinct intracellular membranes and regulate vesicular transport by recruiting effector proteins.²⁰ The precise localization and function of RAB39B is unknown, but the protein is thought to play a role in synapse formation and maintenance.^17,18,21 In support of this, we observed colocalization of endogenous RAB39B with markers of the vesicular-transport pathway, particularly the early endosome in mouse and human neuroblastoma cells (Figure S5). Given the postmortem results (below) and the association between α-synuclein and vesicular-trafficking pathways,²² we tested the effect of downregulation of RAB39B on α-synuclein localization. Mouse hippocampal neurons were prepared and transduced with lentivirus expressing validated Rab39b small hairpin RNA (shRNA) as previously described.¹⁷ Fourteen days after transduction, we observed that the density of α-synuclein immunoreactive puncta in the dendritic processes was 30% lower than in the cells transduced with the scramble control shRNA (p ≤ 0.0005; Figure 1; Figure S6). Immunoblot analysis confirmed ∼40% reduction of RAB39B but also demonstrated that α-synuclein levels (detected with the anti-α-synuclein antibody 97/8,²³ 1:1,000) were reduced by ∼50% (p ≤ 0.05; Figure S6). Similarly, in P19 mouse neuroblastoma cells, immunoblot analysis demonstrated that shRNA-mediated knockdown of Rab39b resulted in ∼50% reduction in α-synuclein steady-state levels (p ≤ 0.005; Figure S6). Although the mechanism remains to be fully defined, these results suggest that downregulation of RAB39B results in dysregulation of α-synuclein homeostasis. We sequenced RAB39B in a cohort of 187 individuals with early-onset PD; they had been previously sequenced and shown not to have mutations in known PD-associated genes, including SNCA (MIM 163890), PARK2 (MIM 602544), DJ1 (MIM 602533), PINK1 (MIM 608309), and LRRK2 (MIM 609007).²⁴ This analysis did not identify any additional variants, suggesting that mutations in RAB39B are not a common cause of early-onset PD.

To determine whether the parkinsonism observed in the affected individuals resulted from PD, we investigated the neuropathology associated with the loss of RAB39B. Individual II:1 died at age 48 years from positional asphyxia. Postmortem neuropathological studies on the brain of II:1 were consistent with PD. The macroscopic findings were unremarkable, and serial coronal sections showed normal cortex, white matter, and ventricles with normal-appearing basal ganglia and thalamus. Cross section of the brain stem showed pallor of the SN and locus coeruleus. SN sections stained with haematoxylin and eosin (H&E) revealed hallmark neuropathological PD features, including loss of pigmented neurons and Lewy bodies in surviving neurons (Figure 3). Immunoreactive staining revealed the presence of α-synuclein-positive Lewy bodies and Lewy neurites in >10% of the surviving neurons. Additional neuropathological features included an abundance of cortical Lewy bodies, which are a pathological feature characteristic of dementia with Lewy bodies (DLB [MIM 127750]; reviewed in^27,28). Tau-immunoreactive NFTs were also observed in a small proportion of the surviving pigmented SN neurons (Figure 3). Tau pathology has previously been observed in familial and idiopathic PD, and the tau-encoding gene (microtubule-associated protein tau [MAPT (MIM 157140)]) exists within a PD susceptibility locus. Tau plays a role in iron homeostasis,^29–31 and Perl staining revealing a modest accumulation of iron in the SN (Figure 3) was consistent with the slight reduction in T2 signal intensity observed in individual II:1 (Table S1). In addition, analysis of the basal ganglia identified rare axonal spheroids in the white-matter tracts (data not shown), similar to the Wallerian-like degeneration observed in neurodegenerative diseases with impaired axonal transport.³² The additional pathological and clinical features share similarities with other neurodegenerative disorders, the most similar of which was neurodegeneration with brain iron accumulation (NBIA [MIM 234200]). Notably, in rare cases, NBIA can manifest with developmental delay and subsequent early-onset parkinsonism.³³ Although MRI of II:3 was normal and II:1 did not show symptoms typical of NBIA,³⁴ genomic DNA from individuals with NBIA was analyzed. Ethics approval was provided by the IRB at Oregon Health & Science University, and informed consent was obtained. Sequence analysis of RAB39B in a cohort of 48 male individuals with NBIA of unknown etiology did not identify any variants.

Figure 3.

Neuropathology Associated with Loss of RAB39B

CNS postmortem tissue from II:1 was collected at autopsy, and microscopic examination was performed on H&E-stained sections. To detect PD-associated proteins, we applied immunohistochemistry to 5 μm formalin-fixed paraffin-embedded sections as previously described.²⁵ Primary antibodies utilized were anti-α-synuclein (97/8; 1:1,000)²³ and anti-tau (A0024, Dako; 1:500). CNS nonhaem iron (Fe²⁺ and Fe³⁺) was detected with a previously described Perl’s methodology.²⁶ The scale bar represents 50 μm (A) or 20 μm (B–F).

(A–E) SN sections show (A) neuronal loss and pigment incontinence (H&E), (B) intraneuronal Lewy bodies (H&E), (C) α-synuclein-reactive Lewy bodies and neurites, (D) tau-immunoreactive intraneuronal NFTs, and (E) extracellular iron deposition (Perl’s stain).

(F) Neocortical Lewy bodies were identified by H&E staining.

In conclusion, genetic studies have demonstrated that loss of RAB39B causes pathologically defined PD, and functional studies have provided additional evidence for pathogenicity. Our results link the loss of a single gene involved in neuronal organization and synaptic function to the early manifestation of both ID and neurodegeneration and suggest that the loss of RAB39B dysregulates α-synuclein. For the two families we ascertained and clinically characterized, there appears to be a canonical age-dependent progression, namely ID first and then a slowly progressive basal ganglia disorder that advances after puberty. This was observed or reported in all 16 affected males in the Australian and Wisconsin families. However, given the lack of clinical data for the other two families previously described to be affected by RAB39B mutations,¹⁷ it is unclear whether this age-dependent phenotype predominates or whether heterogeneous phenotypes extending from ID to ID with parkinsonism are associated with loss of RAB39B function. This issue will be resolved by future studies of additional individuals with mutations in RAB39B.

The proposed role of RAB39B in vesicular trafficking identifies a potential disease mechanism that is distinct from pathways associated with genes in which mutations are currently known to cause familial early-onset PD. Previous in vitro studies have demonstrated that α-synuclein-mediated deficits in vesicular trafficking can be ameliorated by the overexpression of several RAB proteins^35,36 but have not shown that loss of a specific RAB can cause PD. Current studies are further investigating how loss of RAB39B might cause the observed in vitro deficits in localization and reduced steady-state levels of α-synuclein but in vivo accumulation of significant α-synuclein pathology at end-stage disease. It is possible that in simple cell models with efficient protein-metabolism pathways, the “mislocalized” α-synuclein is rapidly turned over and thus leads to reduced steady-state levels. However, protein-turnover pathways are compromised in individuals with PD;³⁷ therefore, mislocalized α-synuclein might not be turned over efficiently, and as the disease progresses, the protein could accumulate and be incorporated into the protein aggregates that define PD.

The broader pathology of iron accumulation, NFTs, and axonal spheroids is similar to that reported for a range of neurodegenerative conditions. However, the abundance of both brainstem and cortical Lewy bodies suggests that RAB39B and/or associated pathways might directly contribute to the pathogenic mechanisms underlying dementia disorders such as DLB. Further studies, including the development of animal models, will be important for understanding the underlying pathogenic mechanism(s) of RAB39B dysfunction and identifying potential targeted therapeutic interventions.

Web Resources

The URLs for data presented herein are as follows:

• ClinVar, http://www.ncbi.nlm.nih.gov/clinvar/

• Online Mendelian Inheritance in Man (OMIM), http://www.omim.org/

• PyMOL, http://www.pymol.org/

• RefSeq, http://www.ncbi.nlm.nih.gov/refseq/

Accession Numbers

The ClinVar accession numbers for the RAB39B variants reported in this paper are SCV000190018 and SCV000190929.