Autosomal dominant sensory ataxia is a rare genetic condition that results in a progressive ataxia that is caused by degeneration of the posterior columns of the spinal cord. To date only two families have been clinically ascertained with this condition, both from Maritime Canada. We previously mapped both families to chromosome 8p12-8q12 and have now screened the majority of annotated protein-coding genes in the shared haplotype region by direct DNA sequencing. We have identified a putative pathogenic mutation in the gene encoding ring-finger protein RNF170, a potential ubiquitin ligase. This mutation is a rare non-synonymous change in a well-conserved residue and is predicted to be pathogenic by SIFT, PolyPhen, PANTHER and Align-GVD. Microinjection of wild-type or mutant orthologous messenger RNAs into zebrafish (Danio rerio) embryos confirmed that the mutation dominantly disrupts normal embryonic development. Together these results suggest that the mutation in RNF170 is causal for the sensory ataxia in these families.
The hereditary ataxias comprise a group of disorders characterized by a progressive incoordination of gait. Hereditary ataxias are heterogeneous and their associated symptoms may result from pyramidal, extrapyramidal or peripheral weakness. The hereditary neuropathies are a related group of inherited disorders that specifically affect the peripheral nervous system, which include the hereditary motor and sensory neuropathies, the hereditary motor neuropathies and hereditary sensory and autonomic neuropathies. Autosomal dominant sensory ataxia (ADSA) is a rare genetic disorder that shares several features of the hereditary ataxias and sensory neuropathies. To date, its scope is limited to two families from the Maritime region of Canada (Valdmanis et al., 2004, 2006). The presence of normal sensory nerve conductions, but absent sensory evoked potentials, suggests that the pathology is restricted to the posterior columns of the spinal cord. The result is that affected family members have a reduced ability to feel pain, temperature and vibration, particularly in the hands and feet. However, their most prominent feature is an ataxic gait resulting from a severe loss of proprioception. Thus, the patients rely on visual cues for maintaining proper body posture, such that they are unable to remain upright if their eyes are closed (Romberg sign). Autopsy results from a deceased patient demonstrated the presence of axonal spheroids within the dorsal columns of the spinal cord. Some of these spheroids stained positively with an anti-ubiquitin antibody (Moeller et al., 2008).
A microsatellite marker-based genome scan of the first ascertained ADSA family defined a linked locus on chromosome 8 bracketing the centromere, with a maximum LOD score of 5 and called SNAX1 for sensory ataxia locus 1 (Valdmanis et al., 2004). Genotypes of closely linked markers in the second ADSA family were also consistent with linkage to the SNAX1 region in that family, with a shared marker haplotype from 8p12-8q12.1, a genetic size of 7.3 cM and a molecular size of 21.5 Mb including the centromere (Valdmanis et al., 2006). We have now sequenced the protein-coding exons of all genes annotated in this chromosomal region. We have identified a single missense mutation in the gene encoding ring-finger protein RNF170, and provide evidence that the mutation is likely to be pathogenic and causal for the phenotype of ADSA in carriers.
Materials and methods
The study was approved by the ethics committee of the Centre Hospitalier de l’Université de Montréal (CHUM). Patients gave informed consent after which patient information and blood were collected. DNA was extracted from peripheral blood using standard protocols. Phenotypic details have been previously described (Valdmanis et al., 2004, 2006; Moeller et al., 2008).
Protein-coding exons of genes in the linked chromosomal interval were identified by visual inspection, amplified by polymerase chain reaction using standard methods and sequenced at the McGill University and Genome Quebec Centre for Innovation, using Sanger fluorescent sequencing and capillary electrophoresis. Sequence traces were analysed using MutationSurveyor (Soft Genetics Inc.) Specific primers and polymerase chain reaction conditions for amplification of RNF170 exons are provided in Supplementary Table 2.
Copy number variation analysis
DNA from two individuals, one from each ADSA family, was sent to Nimblegen Systems for detection of copy number variants using the chromosome 8 specific tiling array B3739001 that contains a median probe spacing of 341 nt. Patient DNA was compared with pooled control DNA that was provided by Nimblegen.
Protein sequence alignment
Cluster analysis and sequence alignment were performed using Clustal W. Species that were compared were: Homo sapiens (Uniprot Q96K19.2), Arabidopsis thaliana (Genbank NP_565037.1), Bos Taurus (Genbank XP_874720.1), Caenorhabditis elegans (Genbank NP_496760.1), Danio rerio (Genbank NP_999915.1), Gallus gallus (Genbank XP_424879.2), Monodelphis domestica (Genbank XP_001381940.1), Mus musculus (Genbank EDL24228.1), Nematostella vectensis (Genbank XP_001626594.1), Ornithorhynchus anatinus (Genbank XP_001509677.1), Oryza sativa (Genbank NP_001051839.1), Physcomitrella patens (Genbank XP_001769535.1), Rattus norvegicus (Genbank EDM09081.1) and Xenopus tropicalis (Genbank NP_001016872.1).
Bioinformatic mutation severity prediction
The effect of amino acid substitution on protein function was predicted with SIFT, PolyPhen, PANTHER and Align-GVGD. Full details are provided in the Supplementary Materials.
Paraffin-embedded human tissue samples were detected using an anti-RNF170 antibody (Abcam, 1:200). Proteins were extracted from lymphoblast cell lines that had been treated with or without 20 µM MG132 (Calbiochem) for 6 h. Detection by an anti-RNF170 antibody (Abcam, 1:100) was carried out using 15 µg of extracted protein. Complete details for immunohistochemistry and western analysis are provided in the Supplementary Material.
Wild-type zebrafish (Danio rerio) were maintained according to established procedures (Westerfield, 1995). The putative zebrafish RNF170 orthologue (drRNF170) was identified by manual inspection of the D. rerio genome assembly using Ensembl BLAST Server. Antisense morpholinos targeting drRNF170 were designed by and purchased from Genetools LLC (Philomath, Oregon), based on the sequence of EST clone BC056330. Antisense morpholino injections were performed as previously described (Guernsey et al., 2009). For RNA-driven expression of drRNF170 in injected embryos, wild-type drRNF170 was amplified by polymerase chain reaction from zebrafish complementary DNA, verified by direct DNA sequencing, and cloned in the expression plasmid pCGlobin2 (Ro et al., 2004). The mutant clone was engineered by site-directed mutagenesis using standard methods. Both messenger RNAs were transcribed from linearized constructs using the T7 polymerase mMESSAGE mMACHINE kit (Ambion). The messenger RNAs were injected into one- to two-cell stage embryos using a Picospritzer. Volumes of ∼1.5 nl were injected at either 50 or 100 nM concentration of messenger RNA.
Gene screening and mutation analysis
The SNAX1 candidate interval spans ∼22.5 Mb in the human genome between outer recombinant markers D8S1769 and D8S601, including an estimated 3 Mb unsequenced centromeric gap (Valdmanis et al., 2006).
Array comparative genomic hybridization analysis was performed (NimbleGen Systems) in two affected individuals, one from each ADSA family, to check for copy-number variants. One deletion of ∼150 kb was present in both individuals in the candidate interval. However, this is a known copy number variation identified in several publications that study common human structural variation, including the chr8.33 variant detected from HapMap genotype analysis (McCarroll et al., 2006). The nominal genes ADAM3A (also known as TMDCI) and ADAM5P (also known as TMDCII) are included in the deletion, but they are now annotated as a likely pseudogenes or targets of non-sense-mediated decay. Furthermore, sequence analysis of single-nucleotide polymorphisms within the deletion itself indicates that the deletion is highly polymorphic and present in heterozygous or homozygous states among unaffected relatives in the two families, and thus is very unlikely to be pathogenic.
We proceeded to sequence the coding exons and intron/exon junctions of all annotated genes in the region based on the hg17 (NCBI35) draft. In the process of this sequencing, we were able to reduce the upper candidate interval based on the identification of several single-nucleotide polymorphisms, which did not segregate for all affected members of the two sensory ataxia families. This led to a slight reduction of the candidate region to 16.94 Mb with a new proximal disease haplotype boundary on chromosome 8 at rs6986249. In total, 67 annotated protein-coding genes were sequenced (refer to Supplementary Table 1 for complete list). We identified only one non-synonymous novel coding change. This was a c.595C>T transition in exon 7 of the gene-encoding RNF170 (NCBI accession number NM_030954) previously named the anonymous gene DKFZp564A022. This base pair change results in an arginine to cysteine substitution at amino acid 199 (p.R199C) (Fig. 1A and B). The variant was not detected in 90 Centre d'Etude du Polymorphisme Humain (CEPH) controls plus 140 ethnically matched local controls from Nova Scotia (460 total chromosomes, data not shown). In addition, we verified that all 15 affected individuals and none of the unaffected relatives in the two ADSA families had the mutation (Fig. 2A and B).
Multi-species alignment of putative orthologues of RNF170 shows that this residue is strictly conserved across the 16 species examined. The missense change is predicted to be pathogenic by SIFT, PolyPhen, Panther and Align-GVD (see Supplementary Material for details).
RNF170 expression analysis
We were able to detect prominent expression of the RNF170 protein in cross-sections of upper cervical and lumbar spinal cord of a deceased patient with ADSA that was specific to an antibody against RNF170 (Supplementary Figure 1A and B). Control human spinal cord sections also displayed immunoreactivity to anti-RNF170, though with a more diffuse signal (Supplementary Fig. 1C and D). Nonetheless, this shows that the RNF170 protein is expressed in disease-relevant tissue. To determine if RNF170-protein levels are perturbed in ADSA, lymphoblast cell lines were generated from patients with RNF170 mutations and their unaffected relatives. Importantly, there was no substantial difference in RNF170 protein expression between patients and controls (Supplementary Fig. 2). Thus the p.R199C mutation does not affect the level of protein concentration at least in this tissue, suggesting that the mutation does not result in haploinsufficiency, but instead may cause a toxic gain-of-function; this result was confirmed when we examined a zebrafish model of the disease (detailed below). Protein aggregates are often a hallmark of neurodegenerative disease and ubiquitin staining structures have been seen in brain tissue of the patients with ADSA. These aggregates can accumulate in detergent-insoluble fractions of cell-free extracts (Schlehe et al., 2008). We treated the lymphoblast cells with MG132, a potent proteasome inhibitor, and separated the results into soluble and insoluble fractions. However, this did not lead to an accumulation of aggregate protein in the insoluble fraction in the conditions we tested (Supplementary Fig. 2A–D).
In order to test for a biological role for this gene, we sought to knock down the expression of the protein in a vertebrate model of development. We found that the zebrafish genome contains a single putative orthologue of RFN170 with a sequence conservation of 64% at the amino acid level when compared to the human protein (Supplementary Fig. 3). We designed a morpholino anti-sense oligomer targeting the zebrafish orthologue of RNF170. Injection of this morpholino caused a dose dependent increase in abnormalities and cell death prior to one day of life, and eventually death of the embryos. These results suggest a vital role for this protein in early development. Next, we used the zebrafish to develop a model for the mutation observed in the human patients with ADSA. We injected synthetic messenger RNA derived from the endogenous drRNF170 gene into zebrafish embryos, containing either the wild-type fish-coding sequence or else a mutation engineered to replicate the variant seen in the human RNF170 gene in the patients with ADSA, at the equivalent position in the orthologous fish gene (Supplementary Material). Injection of the mutant messenger RNA caused moderate to severe developmental disruptions (including death) in the majority of injected embryos (79%, n = 75; Fig. 3), whereas injection of wild-type messenger RNA at the same concentrations causes developmental disruptions in a minority of embryos (18%, n = 68; P < 0.0001 versus wild-type messenger RNA by χ2-squared test). Co-injection of the two messenger RNAs together, where each is used at half concentration, led to an intermediate proportion of disturbed embryos (45%, n = 103). These results indicate that the mutation has a toxic effect even in the presence of the endogenous gene function, and that this effect is dependent of the relative dose of the mutant version. Together with the genetic and bioinformatic studies, the zebrafish experiments support the interpretation that the RNF170 p.R199C mutation acts in a dominant fashion and is causal for the ataxia observed in human patients.
By direct DNA sequencing of protein-coding exons of genes in the chromosomal interval previously linked to the ADSA phenotype in two Maritime Canadian families, we have identified the likely underlying causal mutation, an arginine-to-cysteine variant at residue 199 in the gene encoding ring finger RNF170.
Several lines of evidence support the interpretation of this variant as pathogenic and causal. First, no other protein-coding change was detected in any sequenced genes in the linked interval, other than those also found in the National Centre for Biotechnology Information single-nucleotide polymorphism database or in control individuals. Given the high penetrance of the phenotype as transmitted dominantly in our families, it is very unlikely that an individual sequenced or genotyped as part of the HapMap or genome project consortia (including the 1000 Genomes) would be suffering from this severe neurological condition, thereby ruling out known single-nucleotide polymorphisms as causal for the phenotype. We have been unable to ascertain any other cases with similar clinical presentation, despite extensive canvassing of ataxia experts. This specific phenotype appears particularly rare, even among monogenic disorders. In fact, we sequenced 58 unrelated patients with ataxia from our clinical cohorts, whose presentation while different in detail, could conceivably overlap with ADSA. None showed any variation in the RNF170 gene. The R199C variant in RNF170 occurs at a highly conserved position, and the mutation is predicted to be pathogenic by all bioinformatic tools tested.
We used an animal-model system to validate the functional significance of the observed variant. While over-expressing RNF170 in a zebrafish embryo-model system does not help address the neuronal specificity and late-onset features of ADSA, it does provide evidence for the mechanism of action of the mutation and its relative severity. We were able to identify a single zebrafish orthologous gene to human RNF170 by examination of genomic and EST databases. Due to a relatively recent whole-genome duplication, many human genes exist in two haploid copies in the zebrafish genome, although a process of gene reduction is ongoing. In the case of RNF170, only a single-functional gene was detected in the zebrafish genome sequence. Antisense morphant embryonic zebrafish showed a relatively widespread disruption of development. The extreme rarity of the human clinical phenotype suggested to us that the missense mutation in RNF170 might be a gain-of-function allele. If so, then injection of a synthetic messenger RNA carrying the mutant sequence might show damaging effects on embryonic development in an otherwise wild-type genetic background. This indeed was our observation. The sequence similarity between the human and zebrafish genes was sufficient (Fig. 1B) that we could engineer the equivalent amino acid missense into the orthologous fish gene at the same coding position. Injection of wild-type messenger RNA had little effect (other than on a small fraction of embryos, typically seen in any microinjection experiment when sufficient foreign material is injected), whereas injection of even modest amounts of mutant messenger RNA severely disrupted a large proportion of developing embryos. This result strongly supports the interpretation that the p.R199C variant is pathogenic and causal for the ADSA phenotype.
Little is known regarding the function of the protein encoded by RNF170. There are several alternative spliced isoforms of the transcript. According to BioGPS (SymAtlas), it is expressed fairly ubiquitously, with a higher level in thyroid. RNF170 contains one C3HC4 type RING-type zinc-finger domain from amino acids 87–130. The InterPRO program also predicts a ‘protein of unknown function DUF1232 family’ domain from amino acids 195–248, a region that includes the p.R199C mutation. In addition, three transmembrane regions are predicted between amino acids 25–45, 202–222 and 224–244.
Several ring-finger proteins have been identified in the human genome. Typically a zinc atom binds to a Ring HC finger formed by a CXXCX3-39CX1-3HX2-3CXXCX4-48CXXC amino acid sequence motif. Ring H2-finger proteins are very similar with the exception that the fourth cysteine is replaced by a histidine residue. The RNF170 protein contains the following consensus sequence from amino acids 87–130: CXXCX11CXHXXCXXCX15CXXC. Given the mutation that was identified results in the introduction of a cysteine residue in the protein sequence, this may impair the function or proper formation of the zinc binding site. Alternatively, the possibility exists that a novel intramolecular disulphide bridge is inadvertently formed.
Many ring-finger proteins act as E3-ubiquitin ligases. This is particularly suggestive given the observation of ubiquitin-staining aggregates in brain cells of one autopsied deceased patient with ADSA (Moeller et al., 2008). The most notable ring-finger protein that has been identified in connection with a neurological disease is the PARK2 gene, mutated in patients with juvenile recessive Parkinson’s disease (Kitada et al., 1998). The PARK2 protein is an E3-ubiquitin ligase that promotes the addition of ubiquitin residues to proteins destined for degradation by the ubiquitin-proteasome system. Of note, mutations in PARK2 appear to cause the accumulation of a detergent-insoluble protein isoforms (Schlehe et al., 2008). The aggregates may be of a similar nature to the aggregates observed in the spheroids of patients with ADSA (Moeller et al., 2008). We did not observe an insoluble accumulation of RNF170 in lymphoblast cell lines in the two patients that were analysed (Supplementary Fig. 2), although the mutation in RNF170 may cause other cellular proteins to accumulate inappropriately in such aggregates. Mutations in potential E3-ubiquitin ligases have also been associated with Charcot-Marie-Tooth disease, another neurological condition involving axonal defects (Saifi et al. 2005; Guernsey et al. 2010).
From a pathological standpoint, it remains to be seen what causes the particular problem of axons in the patients with ADSA. Are the axons degenerating? Are the accumulating spheroids interfering with proper function? The ultimate goal will be to determine how to alleviate the axonopathy that is observed, both specifically in patients with ADSA and generally in people with sensory complaints.
Canadian Institutes of Health Research (to P.N.V., V.V.B., I.T., N.D. and G.A.R.); Genome Atlantic; Genome Canada; the Nova Scotia Health Research Foundation; the Nova Scotia Research and Innovation Trust; the Capital Health Research Foundation; the IWK Health Centre Foundation (to M.E.S.).
Supplementary material is available at Brain online.
We graciously thank the families participating in this study; David Roquis, Francois Bacot and Pierre Lepage at the Genome Quebec Centre for Innovation for their sequencing expertise and would like to thank Dr Robert Macaulay (Dalhousie University) for kindly providing tissue for immunohistochemistry.
autosomal dominant sensory ataxia