Abstract
RNA polymerase III is essential for the transcription of non-coding RNAs, including tRNAs. Mutations in the genes encoding its largest subunits are known to cause hypomyelinating leukodystrophies (HLD7) with pathogenetic mechanisms hypothesised to involve impaired availability of tRNAs. We have identified a founder mutation in the POLR3A gene that leads to aberrant splicing, a premature termination codon and partial deficiency of the canonical full-length transcript. Our clinical and imaging data showed no evidence of the previously reported white matter or cerebellar involvement; instead the affected brain structures included the striatum and red nuclei with the ensuing clinical manifestations. Our transcriptome-wide investigations revealed an overall decrease in the levels of Pol III-transcribed tRNAs and an imbalance in the levels of regulatory ncRNAs such as small nuclear and nucleolar RNAs (snRNAs and snoRNAs). In addition, the Pol III mutation was found to exert complex downstream effects on the Pol II transcriptome, affecting the general regulation of RNA metabolism.
Introduction
Mutations in POLR3A, the gene encoding the largest subunit of RNA polymerase III (Pol III), have been identified as the cause of Hypomyelinating Leukodystrophy 7 (HLD7) (MIM607694) (reviewed in 1). The disorders classified under this heading show substantial inter- and intra-familial phenotypic variation and, in addition to the major manifestations of white matter involvement, may present with impaired dentition and hypogonadotropic hypogonadism. No cases of homozygosity for null mutations have been reported; all known defects are predicted to result in partial POLR3A deficiency.
Pol III is composed of 17 protein subunits and POLR3A and POLR3B are the largest subunits required for the catalytic activity of this enzyme complex. In the classical concept of eukaryotic transcription, Pol III is responsible for the constitutive transcription of house-keeping genes including transfer RNAs (tRNAs) and several non-coding RNAs (ncRNAs) (2). However, an increasing number of recent studies point to the complexity of the Pol III transcriptome, its multiple target ncRNA genes of unknown function, possible roles in genome organisation and epigenetic regulation, and interaction with Pol II (3–6). Hypotheses on the pathogenetic mechanisms in HDL7 include dysregulation of the expression of tRNAs leading to perturbed cytoplasmic protein synthesis in the brain (1), as well as impaired function of Pol III-transcribed ncRNAs important for myelin development and maintenance (7).
Here, we describe a founder mutation in the POLR3A gene in patients of Roma ethnicity who presented with an unusual clinical and neuroimaging phenotype. Our analysis of gene expression revealed wide-ranging changes that affect the balance of Pol III-transcribed tRNAs and ncRNAs, as well as the expression of some Pol II target genes involved in transcription, splicing and translation regulation.
Results
A unique disease phenotype caused by a new founder mutation in the POLR3A gene
Pedigree structures of families Sh and Iz with Sanger sequencing genotypes for the POLR3A c.1771-6C > G splicing mutation. Individuals with available whole exome sequencing (WES) data are indicated.
Pregnancy and delivery were uneventful in all cases. Early psychomotor development was normal. All patients completed year 8 at school. The individual clinical course of the disorder is described below. The findings at the most recent examination are summarised in Table 1.
Clinical features of the three patients
| Patients | ShII-1 | IzII-1 | IzII-2 |
|---|---|---|---|
| Year of birth | 1959 | 1985 | 1989 |
| Sex | M | M | M |
| Age at onset | 8 | 7 | 8 |
| Age at assessment | 52 | 27 | 23 |
| Upper motor neuron signs | |||
| Muscle weakness | − | − | − |
| Spasticity | − | − | − |
| Brisk tendon reflexes | Upper limbs | − | − |
| Babinsky sign | + | + | + |
| Extrapyramidal signs | |||
| Rigidity | LL, R wrist | LL | − |
| Dystonia | R hand, LL | LL predominantly R | Distal LL |
| Tremor | − | − | − |
| Ataxia (SARA scores)* | |||
| Gait (0–8) | 6 | 5 | 2 |
| Stance (0–6) | 5 | 4 | 1 |
| Sitting (0–4) | 3 | 1 | 0 |
| Speech (0–6)** | 5 | 5 | 3 |
| Total score (0–40) | 24 | 15 | 6 |
| Dysphagia | − | + | + |
| Gaze-evoked nystagmus | + | − | − |
| Abnormal Smooth Pursuits | + | − | − |
| IQ | 95 | 85 | 80 |
| Neuroimaging | |||
| Atrophy of striatal structures | + | + | + |
| Signal changes in red nucleus | + | + | + |
| Limb deformities | − | Bilateral pes cavus | Bilateral pes cavus |
| Hypodontia | − | − | − |
| Hypogonadotropic Hypogonadism | − | − | − |
| Patients | ShII-1 | IzII-1 | IzII-2 |
|---|---|---|---|
| Year of birth | 1959 | 1985 | 1989 |
| Sex | M | M | M |
| Age at onset | 8 | 7 | 8 |
| Age at assessment | 52 | 27 | 23 |
| Upper motor neuron signs | |||
| Muscle weakness | − | − | − |
| Spasticity | − | − | − |
| Brisk tendon reflexes | Upper limbs | − | − |
| Babinsky sign | + | + | + |
| Extrapyramidal signs | |||
| Rigidity | LL, R wrist | LL | − |
| Dystonia | R hand, LL | LL predominantly R | Distal LL |
| Tremor | − | − | − |
| Ataxia (SARA scores)* | |||
| Gait (0–8) | 6 | 5 | 2 |
| Stance (0–6) | 5 | 4 | 1 |
| Sitting (0–4) | 3 | 1 | 0 |
| Speech (0–6)** | 5 | 5 | 3 |
| Total score (0–40) | 24 | 15 | 6 |
| Dysphagia | − | + | + |
| Gaze-evoked nystagmus | + | − | − |
| Abnormal Smooth Pursuits | + | − | − |
| IQ | 95 | 85 | 80 |
| Neuroimaging | |||
| Atrophy of striatal structures | + | + | + |
| Signal changes in red nucleus | + | + | + |
| Limb deformities | − | Bilateral pes cavus | Bilateral pes cavus |
| Hypodontia | − | − | − |
| Hypogonadotropic Hypogonadism | − | − | − |
SARA: Scale for the Assessment and Rating of Ataxia (29): 0 indicates lack of impairment, and higher scores indicate increasing severity. The scores were low or normal for the remaining SARA items (data not shown).
The speech disturbance is due to a combination of dystonia and cerebellar dysarthria.
The following abbreviations are used: LL, lower limbs; R, right; L, left.
Clinical features of the three patients
| Patients | ShII-1 | IzII-1 | IzII-2 |
|---|---|---|---|
| Year of birth | 1959 | 1985 | 1989 |
| Sex | M | M | M |
| Age at onset | 8 | 7 | 8 |
| Age at assessment | 52 | 27 | 23 |
| Upper motor neuron signs | |||
| Muscle weakness | − | − | − |
| Spasticity | − | − | − |
| Brisk tendon reflexes | Upper limbs | − | − |
| Babinsky sign | + | + | + |
| Extrapyramidal signs | |||
| Rigidity | LL, R wrist | LL | − |
| Dystonia | R hand, LL | LL predominantly R | Distal LL |
| Tremor | − | − | − |
| Ataxia (SARA scores)* | |||
| Gait (0–8) | 6 | 5 | 2 |
| Stance (0–6) | 5 | 4 | 1 |
| Sitting (0–4) | 3 | 1 | 0 |
| Speech (0–6)** | 5 | 5 | 3 |
| Total score (0–40) | 24 | 15 | 6 |
| Dysphagia | − | + | + |
| Gaze-evoked nystagmus | + | − | − |
| Abnormal Smooth Pursuits | + | − | − |
| IQ | 95 | 85 | 80 |
| Neuroimaging | |||
| Atrophy of striatal structures | + | + | + |
| Signal changes in red nucleus | + | + | + |
| Limb deformities | − | Bilateral pes cavus | Bilateral pes cavus |
| Hypodontia | − | − | − |
| Hypogonadotropic Hypogonadism | − | − | − |
| Patients | ShII-1 | IzII-1 | IzII-2 |
|---|---|---|---|
| Year of birth | 1959 | 1985 | 1989 |
| Sex | M | M | M |
| Age at onset | 8 | 7 | 8 |
| Age at assessment | 52 | 27 | 23 |
| Upper motor neuron signs | |||
| Muscle weakness | − | − | − |
| Spasticity | − | − | − |
| Brisk tendon reflexes | Upper limbs | − | − |
| Babinsky sign | + | + | + |
| Extrapyramidal signs | |||
| Rigidity | LL, R wrist | LL | − |
| Dystonia | R hand, LL | LL predominantly R | Distal LL |
| Tremor | − | − | − |
| Ataxia (SARA scores)* | |||
| Gait (0–8) | 6 | 5 | 2 |
| Stance (0–6) | 5 | 4 | 1 |
| Sitting (0–4) | 3 | 1 | 0 |
| Speech (0–6)** | 5 | 5 | 3 |
| Total score (0–40) | 24 | 15 | 6 |
| Dysphagia | − | + | + |
| Gaze-evoked nystagmus | + | − | − |
| Abnormal Smooth Pursuits | + | − | − |
| IQ | 95 | 85 | 80 |
| Neuroimaging | |||
| Atrophy of striatal structures | + | + | + |
| Signal changes in red nucleus | + | + | + |
| Limb deformities | − | Bilateral pes cavus | Bilateral pes cavus |
| Hypodontia | − | − | − |
| Hypogonadotropic Hypogonadism | − | − | − |
SARA: Scale for the Assessment and Rating of Ataxia (29): 0 indicates lack of impairment, and higher scores indicate increasing severity. The scores were low or normal for the remaining SARA items (data not shown).
The speech disturbance is due to a combination of dystonia and cerebellar dysarthria.
The following abbreviations are used: LL, lower limbs; R, right; L, left.
Patient 1 (ShII-1)
A 56-year-old male patient with clinical onset at the age of 8 years, when dysarthria became obvious. Gait ataxia (wide-based gate with frequent falls) developed at age 11 years. Stiffness in the lower limbs developed a year later. These complaints progressed over time, leading to severely impaired ambulation at the age of 50 years. At the last examination, at age 52 years, ataxia of stance and gait was prominent (SARA 24). Gaze-evoked nystagmus and impaired smooth pusuits were present. Tendon reflexes were brisk at the arms, with bilaterally positive Babinsky sign. Extrapyramidal signs included symmetric rigidity and dystonic movements in the right hand and lower limbs, combined with cogwheel phenomenon in the right wrist. Severe speech disturbance, due to a combination of dystonia and cerebellar dysarthria, was present. Brain MRI showed a mild small brain, with bilateral symmetric atrophy of the caudate nucleus and putamen and associated increased signal. The region of the medial red nucleus and the intra-axial course of the third cranial nerve showed focal symmetric signal change. The white matter was of normal volume and signal. No evidence of oligodontia or hypogonadotropic hypogonadism was found. The IQ at 52 years was 95.
Patient 2 (IzII-1)
T2 FLAIR MRI axial (A) and coronal (B) demonstrating marked bilateral atrophy of the corpus striatum. (C) TSE T2 MRI axial illustrating the signal change (white arrow) in the region of the red nuclei and third cranial nerve tracts. (D) FDG PET-CT of subject Iz II-1 demonstrating dramatic bilateral hypometabolism in the striatum. (E) FDG PET-CT of a healthy age and sex matched control. The axial PET-CT images are acquired of the basal ganglia.
Patient 3 (IzII-2)
A 27-year-old male with clinical onset at 8 years, when dystonic movements in the distal parts of the lower limbs became apparent. At age 14 years, his speech became dysarthric and the patient started having swallowing problems. Gait ataxia was developed at 18 years. At the most recent examination at age 23 years, very mild ataxia of stance and gait were present (SARA 6). Normoreflexia was present in the four limbs, but an up going extensor plantar response was elicited. Symmetric dystonic movements in the distal parts of the lower limbs were observed. The dysarthria and dysphagia were mild, the pharyngeal reflexes were preserved. Bilateral pes cavus was observed. Upon brain MRI, the caudate nucleus and putamen showed mild changes. There was abnormal focal signal change in the medial red nucleus and the intra-axial course of the third cranial nerve. The white matter was normal. Dentition and function of gonadal axis were normal. The IQ was 80.
We performed whole exome sequencing (WES) using the TruSeq capture system and the HiSeq2000 platform on DNA samples from the three affected subjects and the parents from family Iz (Fig. 1). Based on pedigree structure, origin from a genetic isolate, and sharing of an identical rare phenotype, we assumed autosomal recessive inheritance with an ancestral deleterious mutation present in the homozygous state in all three patients and heterozygous in the parents. The WES data were analysed as described previously (8,9).
In a preliminary analysis, we used 64 973 polymorphisms at HapMap Phase II SNP positions, extracted from the WES data (10), to estimate inbreeding (11,12) and perform homozygosity mapping (13). The inbreeding coefficients ranged between 0.056 and 0.085, suggesting cryptic consanguinity and supporting our assumption of autosomal recessive inheritance. Homozygosity mapping identified 24 regions of homozygosity shared by the affected subjects.
Our step-wise filtering strategy (8,9), driven by the inbreeding estimates, identified a single homozygous variant exome-wide that satisfied all criteria (Supplementary Material, Fig. S1). The variant, NC_000010.10: g.79769439G > C; NM_007055.3: c.1771-6C > G, was located in intron 13 of the POLR3A gene, in the largest region of homozygosity at 10q22.3 (hg19 chr10:73772999, 92.27cM to hg19 chr10:80704258, 98.32cM). Sanger sequencing (Supplementary Material, Table S1) confirmed the presence and expected segregation of the variant (Supplementary Material, Fig. S2).
Frequency in the general Roma population was analysed using a Taqman assay (Supplementary Material, Table S1) in a panel of 703 healthy population-matched controls, 202 representing Vlax Roma groups and 501 Balkan and Central/Western European Roma. We identified 10 heterozygotes, all confined to the Vlax Roma population, in agreement with the origin of the affected families. The overall carrier rate was estimated at 1.4%, and that in Vlax groups was close to 5%, compatible with previous observations on autosomal recessive disorders in the Roma (14).
The POLR3A mutation causes exon skipping
The POLR3A mutation causes a splicing defect. (A) Schematic representation of the predicted functional impact of the POLR3A c.1771-6C > G mutation. Vertical bars designate exons, exonic sequences are shown with capital letters, intronic with small letters (5' to 3' gene orientation, gene encoded on the minus strand). The SRp40 exonic splicing enhancer element (ESE) is highlighted. The codons of the predicted open reading frame of the transcript skipping exon 14 are indicated with brackets; the premature termination codon is boxed. (B) RT-PCR with primers in exons 13 and 15 reveals exon skipping in the POLR3A mRNA in patient fibroblasts carrying the POLR3A c.1771-6C > G mutation. Addition of cyclohexamide (CHX) stabilises the exon skipped mRNA, indicating that it is normally subject to non-sense-mediated decay (NMD). (C) Sanger sequencing of RT-PCR products confirms the identity of the exon skipped mRNA and the predicted translation reveals premature termination codons within the new reading frame.
The POLR3A mutation causes imbalance in pol III produced transcripts
The effects of the c.1771-6C > G mutation on gene expression were analysed by directional sequencing of RNA extracted from blood from the three patients and three healthy controls. We used small RNA sequencing (RNA-Seq) to capture Pol III transcripts such as tRNAs, and other short regulatory ncRNAs such as small nuclear RNAs (snRNAs) and small nucleolar RNAs (snoRNAs), and TruSeq libraries to investigate the changes in longer ncRNAs and in mRNAs as a downstream consequence of the POLR3A mutation. In the small RNA-Seq datasets the library size ranged from 21 to 30 million reads and 67–77% of total reads mapped to the human genome in control and patient datasets. The TruSeq library size ranged from ∼32 to 40 million reads and ∼76% of total reads mapped to the human genome in control and patient datasets. Basic alignment and mapping statistics from patients and controls are shown in Supplementary Material, Table S2.
Mutation in the POLR3A gene causes an imbalance in Pol III transcripts. (A) RNAseq was carried out on blood RNA from the three patients and three age matched healthy control and the datasets were analyzed for expression and represented as rmpk of the patients compared to the controls for tRNAs and other regulatory and ncRNA transcribed by Pol III. (B) The expression values for tRNAs was summarized by amino acid and anticodon. Normalized counts were summed for all expressed loci from each sample that encoded tRNAs of the same amino acid type or all copies of the same tRNA isotype. (C) Differential expression of non-tRNA Pol III transcripts is shown; for multi-copy genes, normalized counts for all expressed loci were summed prior to calculating log2 fold change. (D) RNA isolated from blood was analyzed by qRT-PCR for the Pol III produced transcripts, normalized to 18S rRNA expression and the statistical significance was calculated using the Student’s t-test (*P < 0.05 and **P < 0.01).
In addition to tRNAs, Pol III transcribes 5S rRNA, and various non-coding RNAs including 7SL and 7SK RNAs, RNase P RNA, vault RNAs and Y RNAs (2). Therefore, we considered the effect of the mutation on these transcripts. In contrast to the general decrease in tRNAs, we observed an imbalance in levels of different Pol III-transcribed ncRNA in the patient dataset (Fig. 4C). For example, we found decreased levels of the snoRNA U14 involved in pre-rRNA processing and in the RNA component of the signal recognition particle (SRP), RN7SL1 and increased levels of the transcriptional regulator 7SK RNA and the processing regulatory RNAs RMRP and RPPH1 in the patients (Fig. 4C). We used qRT-PCR to confirm these changes in the blood from patients and controls (Fig. 4D). The decrease in 7SL RNA, a component of the signal recognition particle that mediates co-translational translocation across membranes, and increase in 7SK RNA, a component of the 7SK RNP that couples transcriptional elongation with alternate splicing, supports the presence of both translational and transcriptional consequences of the POLR3A mutation. Changes in the levels of Pol III transcripts may exert downstream effects on mRNAs, for example an increase in 7SK RNA may have a negative effect on Pol II-mediated transcription, by repressing the positive Pol II transcription elongation factor P-TEFb via its sequestration within the 7SK RNP (16). In addition, the increased levels in 7SK RNA may affect the expression of different genes that undergo alternative splicing in the patients. RPPH1 encodes the catalytic RNA component of the ribonucleoprotein complex, RNase P, responsible for the cleavage and maturation of the 5′ termini of pre-tRNA molecules (17). RNase P has been shown to influence Pol I and III transcription (18), suggesting that in the patients the increased levels of RPPH1 may be in response to the decrease in tRNA levels.
We observed an increase in 5S rRNA expression (Fig. 4C and D) that may be a compensatory response to decreased tRNA levels and preferential recruitment of Pol III to the promoters of the 5S rRNA, possibly complemented by perturbed translation regulation resulting from the reduced levels of 7SL RNA. Nucleosome positioning and chromatin context are known to affect the transcription of Pol III genes (5), in particular U6 sRNA and 7SK RNA (19,20) and may explain the increased expression of specific Pol III transcripts by preferential recruitment of Pol III to their promoters.
Transcriptome-wide changes caused by the POLR3A mutation
The POLR3A mutation leads to specific changes in genes involved in transcription regulation, splicing, translation and proteostasis. (A) Summary of the differentially expressed genes in the TruSeq datasets that are affected by the Pol III mutation; the affected cellular processes and their relevance to neurological diseases are also shown. (B) The levels of regulatory ncRNAs involved in splicing were measured by qRT-PCR in RNA from the patients’ and controls’ blood to corroborate the changes identified by RNAseq. Amplification of the ncRNA transcripts is shown, normalized to 18S rRNA as means ± SD of three independent biological experiments. (C) RNA isolated from blood was analyzed by qRT-PCR for differentially expressed mRNAs involved in protein folding, degradation and cell death. The data are normalized to 18S rRNA as means ± SD of three independent biological experiments; the statistical significance was calculated using the Student’s t-test (*P < 0.05 and **P < 0.01).
Next, we investigated changes in the expression of genes involved in ncRNA-mediated processes in the central nervous system such as snoRNAs and regulatory RNAs that are components of spliceosomes and RNP complexes involved in transcription and translation regulation (Fig. 5A) (21). Two groups of dysregulated genes in c.1771-6C > G homozygotes included small nuclear RNAs (snRNAs) and small nucleolar RNAs (snoRNAs) that have been implicated in a range of human diseases, including neurodegenerative disorders (22,23). SnRNAs that are components of the spliceosomal complexes required for RNA splicing and their levels are imbalanced in the patients (Fig. 5A and B) as observed for the Pol III-generated transcripts that are part of spliceosomal complexes (Fig. 4C and D). Although Pol III has been observed to transcribe snoRNAs in C. elegans, D. melanogaster and S. cerevisiae, no such Pol III-dependent snoRNAs have been described in humans to date (2). We observed that the POLR3A mutation significantly affected the levels of several different snoRNAs involved in pre-rRNA splicing and maturation (Fig. 5A and B), suggesting that some of them may be transcribed by Pol III or their levels are affected indirectly by the changes in the ribonucleoprotein complexes that snoRNAs form with other recently identified Pol III-transcribed RNAs including vault RNA1-2, 7SK and 7SL RNAs (24). The altered levels of snoRNAs may contribute to the neurological phenotype observed in the patients, similarly to other snoRNA-mediated neurological disease (25). Furthermore, the decrease of the splicing factor HNRNPH2 (Fig. 5A and B) which binds G-rich sequence motifs and regulates pre-mRNA splicing (26) points to potential effects on splicing in the patients.
The POLR3A mutation also leads to decreased expression of mRNAs coding for proteins involved in protein folding (Fig. 5A and C). Hsp90 involved in protein folding and stress, encoded by HSP90AA1 is decreased in the patients and reduced phosphorylation of this protein in the brain has been correlated with frontotemporal lobe degeneration (27). The heat shock proteins required for protein folding and import and clearance of aggregated tau (28), such as DNAJA1 and the heat shock protein DNAJB2, which is expressed primarily in the neuronal layers of the brain (29) and are both decreased in the patients. These proteins have protective neuronal role by suppressing the aggregation of proteins that cause neurological diseases such as ALS (30) and Huntington’s disease (31) or regulating their degradation such as that of ataxin-3 (32).
Decreased expression of other genes in the patients may contribute to the neurodegenerative phenotype, such as decrease in ubiquitin (Fig. 5), which is required for protein degradation as its loss has been implicated in hypothalamic neurodegeneration (33). Decreased ubiquitin that regulates the transcriptional activity of FOXO4 (34), a forkhead box transcription factor required for cell cycle regulation and neural differentiation (35) may also contribute to the observed FOXO4 decrease in the patients. We observe a significant decrease in the iron-binding glycoprotein lactoferrin that protects from prion protein-induced cell death (36) (Fig. 5), and its decrease may contribute to the neurodegeneration observed in the patients. These data indicate that perturbed Pol III RNA metabolism can have downstream effects on proteostasis and protein quality control, processes that are known to be commonly altered in neurodegenerative diseases.
Discussion
Our findings characterize c.1771-6C > G in POLR3A as a pathogenic mutation leading to aberrant splicing, reduced levels of the full-length transcript and expected partial protein deficiency. Mutations in POLR3A and POLR3B, the genes encoding the largest subunits of DNA-directed RNA polymerase III, cause several allelic autosomal recessive disorders (1), now grouped under the heading Hypomyelinating Leukodystrophy 7 (MIM607694). Phenotypic heterogeneity in this group of disorders is manifested not only by the presence or absence of defects in dentition and hormonal regulation, but also by the extent of white matter versus cerebellar involvement (1,7). Our patients showed no MRI evidence of white matter involvement, and did not replicate previous findings of thinned corpus callosum, vermian cerebellar atrophy, prominent cerebellar folia and brainstem atrophy. Instead, they presented with striatal and red nucleus involvement and clinical manifestations corresponding to the unusual localization of the pathological changes (Table 1). While some clinical features seen in our patients are also present in Hypomyelinating Leukodystrophy 7, their origins and pathogenesis are different. Dysphagia and dysarthria are thought to result from bulbar weakness in HLD7 (37), whereas the preserved pharyngeal reflexes in our patients point to dystonia of the pharyngeal and laryngeal muscles as a likely cause. Similarly, the ataxia in HDL7 is explained by the cerebellar atrophy and white matter changes (1,37–39) whereas in our patients it appears to be due to dorsal midbrain involvement, as suggested by the imaging data. The differences between the phenotype described here and those reported previously suggest that the clinical spectrum of POLR3 deficiency is broader than currently appreciated. Its better understanding will depend on expanding the clinical indications for genetic analysis beyond the leukodystrophies.
Our study of gene expression aimed to examine the effects of the mutation on both Pol III target genes and Pol II-mediated expression. Wide-ranging between-tissue differences in overall gene expression, as well as in Pol III transcript levels are well known (40). In our experiments, we used whole blood RNA as previous studies have seen correlation between brain and blood in the expression of specific gene clusters involved in transcription regulation, protein synthesis, synapse formation and energy metabolism (41). These same clusters showed significant changes in our patients. Understanding the brain-specific effects of POLR3 mutations will await analyses of gene expression in affected brains, with studies in patients with divergent pathologies being of particular interest. While individual gene changes in the brain may not exactly mirror those described here, our findings already point to significant derangement in Pol III transcription leading to an imbalance, not just a decrease, in tRNAs and ncRNAs. Such changes are likely to be present in the brain, supporting the previously hypothesized altered availability of tRNAs and ncRNAs in patients with POLR3 mutations (1,7). Moreover, our data suggest that imbalance in Pol III-transcribed RNAs as a result of the Pol III mutation has downstream effects on the promoter recognition of other Pol III transcripts, as it has been shown recently (42) or on Pol II target genes involved in the regulation of RNA metabolism (transcription, splicing and translation) and proteostasis, processes that are required for brain development and have been implicated in neurodegeneration. The dysregulation in these pathways likely contributes to the pathology of the disorder and provides a resource for elucidating the regulatory mechanisms underlying Pol III transcription.
Materials and Methods
Methods and subjects
Two families with three affected individuals and four unaffected relatives participated in the study (Fig. 1). The two families resided in different geographical areas of Bulgaria and were unaware of each other’s existence. Both belonged to a young population sub-isolate, a Roma group known as the Bowlmakers, characterized by small founding size, limited genetic diversity and a high prevalence of autosomal recessive disorders (43,44). Carrier rates were investigated in a panel of 703 healthy population controls from diverse Roma/Gypsy sub-isolates. Written informed consent was obtained from all participants. The study complied with the guidelines of the institutions involved.
Clinical investigations
The affected subjects underwent neurological, ophthalmological, endocrinological, dental and radiological examinations at the University Hospitals in Sofia and Varna. Clinical data were also collected from. Information on the evolution of the disease was obtained from previous hospital records and interviews with care-providers.
Ataxia severity was evaluated with the Scale for the Assessment and Rating of Ataxia (SARA) (45). Cognitive performance was evaluated using the Mini-Mental State Examination (MMSE), and formal neuropsychological assessment of general intelligence, memory, word fluency and executive function. The endocrinological investigations included basal hormonal levels and stimulation tests of the gonadal axis with gonadotropin releasing hormone and clomiphene citrate (Supplementary Methods On-line). The dental status was assessed through detailed examinations and orthopantomograms (OPG).
Nerve conduction studies (NCS) and electromyography (EMG) using a Dantec–Keypoint portable electromyograph (Natus, Copenhagen, Denmark) were performed in all patients.
Magnetic resonance imaging (MRI) of the brain was performed on a 1.5T MR imager (MR Signa HDxt, GE Healthcare, Milwaukee, WI, USA). Patient IzII-1 also had positron emission tomography–computed tomography with fluorodeoxyglucose (FDG PET-CT) of the brain, using Gemmini TF 16, Philips equipment.
Exome sequencing and data analysis
Whole exome sequencing (WES) was performed on DNA samples from the three affected subjects and the parents in the Iz family (Fig. 1). The analysis was done by Axeq Technologies (Seoul, Korea) using the TruSeq capture system and the HiSeq2000 platform (Illumina, San Diego, CA, USA). The initial data processing included alignment to the hg19 reference genome (46), variant calling in SAMtools (47) using default parameters, and identification of variants in dbSNP135 (http://www.ncbi.nlm.nih.gov/projects/SNP/). Variants were annotated to the UCSC Known Genes by ANNOVAR (48) version 23 Oct 2012, and ANNOVAR-formatted databases based on the UCSC Known Gene (hg19_knownGene), the 1000 Genomes project (hg19_ALL.sites.2012_02) and the NHLBI Exome Sequencing Project (http://evs.gs.washington.edu/EVS/) (hg19_esp6500_all).
The WES data were used to extract 64,973 polymorphisms at HapMap Phase II SNP positions (8,10). These were used for homozygosity mapping in AutoSNPa (13) with a cut-off ≥10 consecutive SNPs. A selected subset of 5,534 markers (in approximate linkage equilibrium, 0.15 cM, average heterozygosity 0.42) was used to calculate inbreeding coefficients in FEstim (11), with allele frequencies from HapMap-CEU data. Relatedness between the two families was estimated in PLINK (12) with allele frequencies obtained from an in-house dataset of 28 Roma exomes.
The search for the disease-causing mutation focused on variants with a quality score ≥20 and coverage ≥4X, which were located outside of segmental duplications and simple repeats. Based on the unusual phenotype, the endogamous nature of the population and pedigree structures, the search was based on the assumption of a rare deleterious variant for which all three patients are homozygous, the parents are heterozygous and no homozygotes are present among population controls. We used a step-wise filtering strategy (Supplementary Material, Fig. S1) including: (a) removal of variants with allele frequency ≥1% in the 1000Genomes (http://www.1000genomes.org/) or NHLBI Exome Sequencing (http://evs.gs.washington.edu/EVS/) projects; (b) ‘deleteriousness’ predictions, retaining non-synonymous variants with Polyphen2 (49) scores >0.8 and SIFT (50) scores ≤0.05, splice-site, non-sense, non-stop and small in-frame or frame-shift in/dels (including exonic and splicing variants in ncRNAs); (c) homozygosity in all affected individuals; (d) heterozygosity in the parents; (e) no homozygoous subjects in a larger in-house database of 45 Roma exomes.
Mutation verification
The presence of the POLR3A mutation was confirmed with bidirectional Sanger sequencing (Australian Genome Research Facility, AGRF, Perth Node) of amplified fragments (Primers listed in Supplementary Material, Table S1). Evolutionary constraint was assessed using Genomic Evolutionary Rate Profiling (GERP) scores (51,52). The functional impact on splicing was predicted using the Human Splicing Finder (HSF) (15), the Expasy translate tool (http://web.expasy.org/translate/) and the ESEfinder 3.0 (53). Mutation screening in the panel of healthy, ethnically matched controls was performed with a custom designed TaqMan assay (Applied Biosystems), following the manufacturer’s protocol (Supplementary Material, Table S1). Carrier status was confirmed by Sanger sequencing.
Cell culture
Fibroblasts were established from Sh II-1 and a healthy control from skin biopsies as previously described (54) and cultured at 37°C under humidified 95% air/5% CO2 in Dulbecco’s modified Eagle’s medium (DMEM, Gibco, Life Technologies) containing glucose (4.5 g l−1), 2 mM glutamine, penicillin (100 U ml−1), streptomycin sulfate (100 µg ml−1) and 10% fetal bovine serum (FBS). Fibroblasts were grown in the presence and absence of 100 ng/µl cycloheximide (CHX) for 24 h to inhibit non-sense-mediated decay before RNA was isolated and analyzed by qRT-PCR.
RNA isolation and quantitative reverse transcription PCR
Total RNA was extracted from primary fibroblast cultures (patient Sh II-1 and a healthy control), and from whole blood (from Iz II-1 and Iz II-2 and two healthy controls), using the miRNeasy Mini kit (Qiagen) and the PAXgene Blood RNA system that included a DNase treatment step, respectively, following the manufacturer’s protocols. cDNA was prepared with the QuantiTect Reverse Transcription kit (Qiagen), used as a template in the subsequent PCR for mRNAs and ncRNAs. Reactions were performed using a Corbett Rotorgene 6000 using SensiMix SYBR mix (Bioline) and normalised to 18S rRNA.
The effect of the mutation on POLR3A mRNA processing was analysed by PCR amplification and product separation by 2% agarose/TAE gel electrophoresis. Visible bands were excised, extracted from gel pieces using the GeneJET PCR Purification Kit (Fermentas) and Sanger sequenced. The abundance of normal and aberrant transcripts was assessed by qRT-PCR as described above.
RNA sequencing
RNA sequencing libraries were prepared using 4 µg of total RNA from blood of the three patients and three, aged matched healthy controls, using the TruSeq Small RNA Prep kit, and the small RNA-seq kit (Illumina) for which we size selected fragments ranging in size from 50 to 200 bp. RNA library preparation and strand-specific RNA sequencing was carried out by the Cologne Genome Sequencing Centre, Germany on a Hi-Seq Illumina sequencer.
Alignment and differential expression analysis
Small RNA library
Adapter sequences were trimmed from the 3′ end with cutadapt 1.7.1 (55) (-e 0.1 -n 2 -O 6 -m 16 –match-read-wildcards). Successfully trimmed reads were aligned to the human genome (hg19) with bowtie2 v2.2.5 (56) (-N 1) and repeat expression analysed with HOMER v4.7 (57) (makeTagDirectory -keepOne -sspe; analyzeRepeats.pl repeats hg19 -rpkm).
Total RNA library
Quality and adapter trimming of raw reads was performed with Trimmomatic 0.32 (ILLUMINACLIP:TruSeq3-PE-2.fa:2:30:10 LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:36). Paired and unpaired trimmed reads were aligned with TopHat 2.0.12 (–b2-very-sensitive –library-type fr-firststrand –read-realign-edit-dist 0 –mate-inner-dist -37 –mate-std-dev 56 -G genes.gtf) against the human genome (hg19). Gene-specific fragment counts were generated with HTSeq 0.6.1p1 (-s yes) and the Ensembl 75 gene annotation, and differential expression was tested with DESeq2 1.6.2 using standard parameters (parametric dispersion estimates, Wald statistic and Benjamini-Hochberg multiple testing correction).
Supplementary Material
Supplementary Material is available at HMG online.
Acknowledgements
We thank the affected families for participating in this study and Alistair Forrest for advice on bioinformatics analyses.
Conflict of Interest statement. None declared.
Funding
National Health and Medical Research Council (APP1058442, APP1045677, APP1041582, APP1023460, APP1005030 to A.F. and O.R. and APP1078273 to A.F., D.A. and L.K.), the Australian Research Council (FT0991008, FT0991113 to A.F. and O.R.), the University of Western Australia, APA scholarship from the Australian Government (to S.S.).
References
Author notes
The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors.
