Recessive mutations in POLR1C cause a leukodystrophy by impairing biogenesis of RNA polymerase III

A small proportion of 4H (Hypomyelination, Hypodontia and Hypogonadotropic Hypogonadism) or RNA polymerase III (POLR3)-related leukodystrophy cases are negative for mutations in the previously identified causative genes POLR3A and POLR3B. Here we report eight of these cases carrying recessive mutations in POLR1C, a gene encoding a shared POLR1 and POLR3 subunit, also mutated in some Treacher Collins syndrome (TCS) cases. Using shotgun proteomics and ChIP sequencing, we demonstrate that leukodystrophy-causative mutations, but not TCS mutations, in POLR1C impair assembly and nuclear import of POLR3, but not POLR1, leading to decreased binding to POLR3 target genes. This study is the first to show that distinct mutations in a gene coding for a shared subunit of two RNA polymerases lead to selective modification of the enzymes' availability leading to two different clinical conditions and to shed some light on the pathophysiological mechanism of one of the most common hypomyelinating leukodystrophies, POLR3-related leukodystrophy.

L eukodystrophies are a heterogeneous group of genetically determined disorders characterized by abnormal white matter on brain imaging 1,2 . They are classified as hypomyelinating and non-hypomyelinating leukodystrophies based on magnetic resonance imaging (MRI) characteristics 2 , depending on whether the principal problem appears to be a lack of myelin deposition during development or altered myelin homeostasis.
RNA polymerase III (POLR3)-related leukodystrophy or 4H (Hypomyelination, Hypodontia and Hypogonadotropic Hypogonadism) leukodystrophy (MIM 607694, 614381) 3 was found to be caused by recessive mutations in POLR3A (MIM 614258) or POLR3B (MIM 614366) [4][5][6][7][8][9] and is characterized by an expanding spectrum of clinical 3,10 and radiological features [10][11][12] . POLR3A and POLR3B are, respectively, the largest and second largest of the 17 subunits that constitute POLR3. Together, they form the catalytic centre of the enzyme. POLR3 synthetizes small non-coding RNAs, including tRNAs, 5S RNA, 7SK RNA and U6 RNA, that are involved in the regulation of essential cellular processes, including transcription, RNA processing and translation 13 . A subset of patients (B5%) presenting with compatible clinical and/or radiological features of POLR3-related leukodystrophy have no detectible mutations in either POLR3A or POLR3B, suggesting that mutations in one or more additional genes may result in this presentation. We hypothesized that genes coding for other POLR3 subunits or for proteins interacting with POLR3 would be strong candidates in these cases.
In this study, we identified recessive mutations in POLR1C, a gene encoding for a subunit common to POLR1 and POLR3, which has thus far been known only to be associated with autosomal recessive Treacher Collins syndrome (TCS) 14 . We also demonstrated that the leukodystrophy-causing mutations affect POLR3, but not POLR1 assembly and nuclear import, leading to decrease binding to POLR3 target genes, whereas one TCS mutation leads to normal assembly of both polymerases, but rather affects POLR1 targeting to the nucleolus, the site of POLR1 gene transcription.

Results
Whole-exome and Sanger sequencing for gene identification. To investigate the genetic aetiology of these unexplained cases, we performed exome sequencing in three cases with typical clinical and/or radiological features of POLR3-related leukodystrophy negative for POLR3A or POLR3B mutation. Genome coverage for these three cases exceeded 56 Â (Supplementary Table 1) and yielded more than 4.6 million variants per genome. Review of the data using in-house filters revealed a homozygous mutation in POLR1C (NM_203290; GRCh37/hg19) in two patients (c.221A4G (p.Asn74Ser) and c.95A4T (p.Asn32Ile), respectively), and compound heterozygous mutations in one (c.436T4C (p.Cys146Arg) and c.883_885delAAG (p.Lys295del)), all consistent with autosomal recessive inheritance (Fig. 1, Supplementary Table 1). These four variants were validated using Sanger sequencing. Co-segregation analysis confirmed that the mutations were inherited from heterozygous carrier parents. These variants were not observed in over 6,500 individuals of European and African American ancestry in the NHLBI Exome Sequencing Project database, in the ExAc data set nor in our internal variant databases. Furthermore, these variants were not present in more than 300 Centre d'Etude du Polymorphisme Humain (CEPH) control chromosomes. In silico analysis predicted those to be disease-causing (Supplementary Table 2c). These variants were found to affect highly conserved amino-acid residues (Fig. 1). To identify additional mutations, we sequenced all exons, exon-intron boundaries and 3 0 and 5 0 untranslated repeat of POLR1C (see Supplementary Table 3 for primers) in 16 other individuals selected on the basis of clinical (compatible neurological features with at least one non-neurological feature such as dental abnormalities or hypogonadotropic hypogonadism, together with hypomyelination on MRI) and/or radiological characteristics (typical MRI features of 4H leukodystrophy) 3,10,11 but negative for POLR3A and POLR3B mutations and uncovered five additional compound heterozygous or homozygous cases (Supplementary Table 2  to confirm segregation. In total, 13 POLR1C mutations were detected in eight cases (Supplementary Impact of POLR1C recessive mutations on POLR1 and POLR3. To investigate the potential pathogenic role of these mutations, we evaluated the impact of the two homozygous mutations ( Table 1) on the function of nuclear POLR1 and POLR3. FLAGtagged versions of the wild-type (WT) form of POLR1C and its variants having the p.Asn32Ile (N32I) or p.Asn74Ser (N74S) substitution were expressed in HeLa cells. Anti-FLAG affinity purification was performed on cell extracts and the purified proteins were analysed using shotgun proteomics. These experiments were performed in triplicate. The expression level of the various forms of POLR1C (that is, WT and mutants) were equivalent and comparable (see Supplementary Fig. 2), and protein expression levels were normalized by the expression level of the bait in each purification (Supplementary Table 6). The tagged WT POLR1C pulled down all subunits of both POLR1 and POLR3 (Fig. 3a), a finding that was expected since POLR1C is a shared subunit of both polymerases (see Fig. 3c for a schematic representation). Both tagged mutated POLR1C (N32I and N74S) pulled down amounts of POLR1-specific (POLR1A, 1B, 1E, CD3EAP, TWISTNB and ZNRD1) and POLR1/POLR3-shared subunits (POLR2E, 2F, 2H, 2K, 2L, 1C and 1D) that were not significantly different from those pulled down by the WT (this is especially true for the POLR1-specific subunits). However, both mutated subunits pulled down lower amounts of POLR3 (most specific subunits) relative to WT. This finding suggests that the mutations lead to a selective defect in POLR3 assembly, and not in POLR1. Indeed, assembly of nuclear RNA polymerases has previously been shown to occur in the cell's cytoplasm and defects in RNA polymerase assembly caused by functional disruption of the RNA Polymerase-Associated proteins was previously shown to lead to cytoplasmic accumulation of polymerase subunits [15][16][17][18][19][20][21] .
Notably, the position of mutated residues in the proposed structure of POLR3 is compatible with defects in enzyme assembly and/or folding (see Supplementary Fig. 1). To further confirm that mutated POLR1C variants (N32I and N74S) are impaired in supporting enzyme assembly and nuclear import, immunofluorescence studies were performed using anti-FLAG antibodies. The results reveal an accumulation of both mutated POLR1C subunits, but not the WT subunit, in the cytoplasm (Fig. 3d). We then performed chromatin immunoprecipitation (ChIP) of FLAG-tagged POLR1C followed by high-throughput sequencing (ChIP-Seq), as a proxy of gene transcription activity [22][23][24] , to investigate the impact of POLR1C mutations on gene occupancy by POLR1 and POLR3. After alignment of the reads to the human reference genome (hg19), we compared occupancy of WT and mutated POLR1C variants over 659 POLR3-transcribed genes, including all transfer RNA (tRNA) and 5S ribosomal RNA genes (Supplementary Table 7). As expected, mutated POLR1C variants displayed reduced binding to POLR3transcribed genes compared with WT POLR1C for all three classes of POLR3-transcribed genes (classified according to their regulatory elements; see legend to Fig. 4a). In contrast, there were no differences in WT and mutated POLR1C occupancy over the ribosomal RNA gene transcribed by POLR1 (Fig. 4c). Together, these results indicate that the N32I and N74S substitutions in the POLR1/POLR3-shared subunit POLR1C specifically interfere with assembly, nuclear import and chromatin association of POLR3. To compare the roles of leukodystrophy versus TCScausing mutations in the biogenesis of POLR1 and POLR3, we expressed FLAG-tagged POLR1C with the p.Arg279Gln (R279Q) mutation in HeLa cells, affinity-purified the tagged subunit and identified the purified interactors using mass spectrometry. Contrary to POLR1C (N32I) and POLR1C (N74S), none of the subunits of POLR1 and POLR3 were pulled down by tagged POLR1C (R279Q) in amounts that were statistically significantly different from the WT (Fig. 5a,b, Supplementary Table 6), suggesting that this TCS-causing mutation does not affect the assembly of these polymerases. Notably, however, immunofluorescence results indicate that POLR1C (R279Q) targeting to the nucleolus is impaired as compared with the WT subunit (Fig. 5c) and the N32I-and N74S-mutated subunits (see Fig. 3d).

Discussion
With the advent of exome sequencing, it is becoming increasingly apparent that allelic heterogeneity in genes encoding essential  14 . TCS caused by mutations in POLR1D or POLR1C has been proposed to arise as a consequence of a decreased quantity of functional ribosomes in the neuroepithelium and the neuronal crest cells during critical points of embryogenesis 14,25 . We assessed the role of a TCScausing mutation (R279Q) in the biogenesis of POLR1 and POLR3. Our results (see Fig. 5, Supplementary Table 6) indicate that this mutation does not impair polymerase assembly, as opposed to leukodystrophy-causing mutations (see Fig. 3), but affect targeting to the nucleolus, the site for Pol I transcription.
Our findings suggest that improper assembly and nuclear import of POLR3 resulting from leukodystrophy-causative mutations lead to decreased availability of the complex at the chromatin. As POLR3 binding is well correlated to tRNA expression [22][23][24] , this decreased POLR3 occupancy is likely to cause reduced transcription of tRNAs and other essential small non-coding RNAs. One hypothesis is that mutations in POLR3A, POLR3B or POLR1C lead to decreased levels of certain tRNAs crucial for the synthesis of proteins essential for central nervous system myelin development. tRNA function has also been suggested to be impaired in other white matter disorders [26][27][28][29][30] caused by mutations in tRNA-aminoacyl synthetases, including hypomyelinating leukodystrophies such as RARS-associated hypomyelination 30 and Hypomyelination with Brain Stem and spinal cord involvement and Leg spasticity 29 . Of note, tRNA synthetases have not been found in our POLR3 purifications, suggesting that tRNA aminoacylation is not coupled with POLR3 transcription. An alternative hypothesis would involve changes in the expression of other essential small non-coding RNAs synthesized by POLR3.
In conclusion, our sequencing study of 18 cases with compatible clinical and/or radiological features with 4H or POLR3-related leukodystrophy identified 13 different POLR1C mutations in eight cases. POLR1C joins an emerging group of genes with dual roles in pathogenesis of human diseases 31     . In all cases, filtering queries were created as specific presets that allowed in silico reduction of variant lists down to candidates with correlation to phenotype, transmission mode of inheritance and alteration classification of pathogenicity. Sanger sequencing and co-segregation analysis were performed on genomic DNA using primer pairs designed with the primer3 software package and the genomic sequence of POLR1C (NM_203290; GRCh37/hg19). PCR products were forward-and reverse-sequenced at the McGill University and Genome Quebec Innovation Centre using an ABI 3730xl DNA Analyzer (ABI; Applied Biosystems, Foster City, CA, USA). Sequences were analysed using SeqMan 4.03 (DNAStar, Wisconsin, USA) and Chromas 1.62 (Technelysium Pty, Ltd, Australia).
Immunofluorescence and western blot analyses. Transfection experiments for generating stable HeLa cell lines expressing FLAG-tagged versions of POLR1C variants used lipofectamine, as described by the supplier (Invitrogen, Carlsbad, CA, USA) 18  Protein affinity purification coupled to mass spectrometry. Generation of cell lines stably expressing FLAG-tagged POLR1C subunits (WT and mutated) and affinity purification from the soluble fraction were performed using standard procedures 38,39 . The eluates were digested with trypsin and the resulting tryptic peptides were purified and identified with tandem mass spectrometry (LC-MS/MS) using a microcapillary reversed-phase high-pressure liquid chromatographycoupled LTQ-Orbitrap (ThermoElectron) quadrupole ion trap mass spectrometer with a nanospray interface, as recently described 40 . Protein database searching and protein spectral count quantification were performed with Mascot (version 2.3.02) 41  The rDNA gene encodes a 45S pre-rRNA precursor that will generate the 5.8S, 18S and 28S rRNAs. There are B400 copies of the rDNA gene arranged in tandem repeats in the human genome. rDNA repeats are not present in the reference genome assemblies; therefore, unique reads were aligned directly to the human rDNA reference sequence (NCBI accession number: HSU13369) 56 . No differences were observed in POLR1C occupancy between wild-type and mutant variants. A schematic of a rDNA repeat is included below the graph.
specificity of our approach, a protein is deemed to show a level of differential interaction with POLR1C that is statistically significant when its adjusted P value is o0.05 and that its average spectral count fold-change (MUT/WT) is o0. 6 or 41.5.
ChIP-sequencing and data analysis. Stable HeLa cell lines expressing FLAGtagged POLR1C (WT or mutated) were cultured to 80% confluence and crosslinked with 1% formaldehyde directly in the cell medium for 5 min followed by a 5-min quenching in 125 mM glycine. For ChIP experiments 43 , nuclei from 3 Â 10 6 cells were lysed and re-suspended in sonication buffer (