A homozygous missense mutation in ERAL1, encoding a mitochondrial rRNA chaperone, causes Perrault syndrome

Abstract Perrault syndrome (PS) is a rare recessive disorder characterized by ovarian dysgenesis and sensorineural deafness. It is clinically and genetically heterogeneous, and previously mutations have been described in different genes, mostly related to mitochondrial proteostasis. We diagnosed three unrelated females with PS and set out to identify the underlying genetic cause using exome sequencing. We excluded mutations in the known PS genes, but identified a single homozygous mutation in the ERAL1 gene (c.707A > T; p.Asn236Ile). Since ERAL1 protein binds to the mitochondrial 12S rRNA and is involved in the assembly of the small mitochondrial ribosomal subunit, the identified variant represented a likely candidate. In silico analysis of a 3D model for ERAL1 suggested that the mutated residue hinders protein-substrate interactions, potentially affecting its function. On a molecular basis, PS skin fibroblasts had reduced ERAL1 protein levels. Complexome profiling of the cells showed an overall decrease in the levels of assembled small ribosomal subunit, indicating that the ERAL1 variant affects mitochondrial ribosome assembly. Moreover, levels of the 12S rRNA were reduced in the patients, and were rescued by lentiviral expression of wild type ERAL1. At the physiological level, mitochondrial respiration was markedly decreased in PS fibroblasts, confirming disturbed mitochondrial function. Finally, knockdown of the C. elegans ERAL1 homologue E02H1.2 almost completely blocked egg production in worms, mimicking the compromised fertility in PS-affected women. Our cross-species data in patient cells and worms support the hypothesis that mutations in ERAL1 can cause PS and are associated with changes in mitochondrial metabolism.


Introduction
Perrault syndrome (PS) (MIM 233400) is a rare disorder that is inherited in an autosomal recessive manner. It is clinically characterized by sensorineural hearing loss in both male and female patients, while females also present with ovarian dysgenesis, which results in amenorrhea and infertility (1). Some patients also present with neurological manifestations, including ataxia, mild mental retardation and peripheral neuropathy (2)(3)(4). Because of the clinical heterogeneity, the disorder has been classified into type I, which is static and without neurological manifestations, and type II that includes progressive neurological symptoms (5).
The clinical heterogeneity of PS may partly be due to its genetic heterogeneity; to date, mutations in five different genes have been identified as disease-causing in different cases of PS. The first mutations were reported in HSD17B4, a gene encoding a peroxisomal enzyme involved in fatty acid b-oxidation and steroid metabolism (5). Later on, mutations in two genes encoding the mitochondrial aminoacyl-tRNA synthetases HARS2 and LARS2 (6,7) were identified, which are components of the mitochondrial translation machinery. Finally, two more genes encoding mitochondrial proteins-the peptidase CLPP involved in mitochondrial protein homeostasis (8,9) and the helicase Twinkle that is required for mitochondrial DNA (mtDNA) maintenance (10)-were found mutated in some PS patients. Altogether, four out of five PS causing genes are involved in mitochondrial gene expression and proteostasis, suggesting that mitochondria play a critical role in the development of the disease.
Here, we report a homozygous c.707A > T (p.Asn236Ile) missense mutation in the ERAL1 gene (NM_005702.2) identified by exome sequencing in two unrelated patients, and found a third PS patient with the same variant during the course of our study. The human ERAL1 protein (UniProtKB O75616) has been described as an rRNA chaperone indispensable for the assembly of the small 28S subunit of the mitochondrial ribosome (11,12). In line with this, patient skin fibroblasts and C. elegans with knockdown of the ERAL1 homologue display mitochondrial dysfunction, strongly suggesting that the identified mutation in ERAL1 is the cause of PS in our patients.

Results
Identification of a homozygous missense mutation in the ERAL1 gene in three PS patients Two women of Dutch ancestry, unrelated but both from the same village, presented at our clinic with symptoms of PS. In PS patient 1 (aged 66 years) hearing loss was diagnosed when she was 20 years old, but probably started much earlier and appeared to be progressive. She had normal menarche at 11 years of age, with irregular menses until the age of 27, when menopause occurred. One sister also had sensorineural hearing loss and premature ovarian failure, but did not participate in our study. PS patient 2 (aged 38 years) was diagnosed at 4 years of age with sensorineural hearing loss, which was more severe in the high frequencies and slowly progressive. At the age of 18 years she presented with primary amenorrhea and underdeveloped secondary sexual characteristics. Abdominal ultrasound revealed streak ovaries and a small uterus. An ovary biopsy showed fibrous tissue without primordial follicles. Her father had sensorineural hearing loss since childhood, but no fertility problems. Her mother and two sisters were healthy.
In order to identify the underlying genetic cause for the disease, we performed whole exome sequencing (WES), which excluded the presence of mutations in the known PS genes HSD17B4 (5), HARS2 (6), LARS2 (7), CLPP (8) and C10orf2 (10). Because both patients originate from a small village, a known genetic isolate, recessive inheritance and shared genetic cause were suspected. Subsequent filtering of all variants found in WES for homozygous variants that were shared in both patients, and with a minor allele frequency of <1% in public or in house databases yielded only one variant: a homozygous nucleotide substitution at position c.707A > T (p.Asn236Ile) in ERAL1 (NM_005702.2), a gene encoding the Era-like 12S mitochondrial rRNA chaperone 1.
The substituted residue is highly conserved in vertebrates as well as in fruit flies and in C. elegans (Fig. 1A). Moreover, the variant is predicted to be "probably damaging" by PolyPhen-2 (HumDiv score 1.000, HumVar score 0.992), SIFT (score 0) and MutationTaster (p-value 1). Subsequent sequencing of family members of patient 2 showed that the healthy females are heterozygous while the affected father is homozygous for this variant (Fig. 1B). The identified variant is not present in any public database. However, from the genotyping of 530 individuals of the same village as the patients, 49 were found heterozygous and 0 homozygous for the variant, demonstrating an allele frequency of 4.6% in this genetic isolate. This would suggest a relatively high prevalence (0.2%) of PS in this village. Indeed, during the course of our study a third PS patient, presenting with progressive sensorineural hearing loss since birth and primary amenorrhea, visited our outpatient clinic and was found to be homozygous for the same ERAL1 variant (Fig. 1C).
ERAL1 protein is indispensable for the assembly of the small 28S subunit of the mitochondrial ribosome through binding to its 12S rRNA component (11,12), and is therefore involved in the translation of mtDNA-encoded proteins. To predict in silico whether the p.Asn236Ile variant could affect the rRNA-ERAL1 interaction, we constructed a 3D model of the protein based on the crystal structure of the bacterial orthologue ERA, which has been characterized in the bacterium Aquifex aeolicus (13). We performed a 3D structural alignment between the bacterial ERA crystal and the human ERAL1 model in order to visualize the position of the residue found mutated in our patients in comparison to its bacterial counterpart (Fig. 1D, left panel). The alignment showed that the position of the asparagine 236 (N236) in human ERAL1 corresponds to the Aquifex aeolicus ERA asparagine 123 (N123), which has been described to directly interact with GTP through the formation of hydrogen bonds (13) (Fig. 1D, middle panel). Additionally, ERA N123 interacts with two amino acids, valine 14 (V14) and alanine 156 (A156) (corresponding to the ERAL1 residues A124 and A310), that are part of the loops flanking the substrate and, therefore, crucial for maintaining the active center's integrity (Fig. 1D, middle panel). Binding of ERA with GTP is necessary for its function, causing changes in its conformation, followed by further interaction of ERA-GTP with the bacterial 16S rRNA (13). Substitution of the polar amino acid asparagine with the hydrophobic isoleucine at position 236 in the human ERAL1 (Fig. 1D, right panel) is likely to impair interactions with GTP and the A124 and A310 residues, and predicted to disrupt protein conformation and its interaction with the human 12S rRNA.
Because previously described mutations causing PS are mostly in genes related to mitochondrial homeostasis-two of which are directly involved in mitochondrial translation (6,7), one in mtDNA maintenance (10) and one in mitochondrial proteostasis (8)-we hypothesized that the identified ERAL1 variant is likely to cause PS in our patients. To test this hypothesis, we set out to investigate the effects of the ERAL1 variant at a cellular as well as at an organismal level.

ERAL1 protein levels and assembly of the 28S ribosomal subunit are compromised in PS patients
To test whether the sequence variant identified in our patients affects ERAL1 protein levels, we performed a Western blot on lysates from cultured skin fibroblasts of PS patients and control subjects. We observed that in both PS patients, ERAL1 protein levels were decreased when compared to fibroblasts from healthy controls ( Fig. 1E and F).
Since ERAL1 is involved in the assembly of the small 28S mitochondrial ribosomal subunit (11,12), we next investigated whether the PS patient cells showed an impaired assembly of the 28S subunit. We assessed the abundance of assembled mitochondrial ribosomal subunits using complexome profiling (14). With this technique migration profiles and relative abundance of protein complexes are identified using a combination of blue native electrophoresis and shotgun proteomics. Although the migration profiles of the proteins composing the small (28S) and large (39S) mitochondrial ribosomal subunits were similar between PS patients and controls ( Fig. 2A and B), we observed a remarkable decrease (30-40%) in the overall abundance of proteins composing the small 28S subunit in the PS patients compared to healthy controls ( Fig. 2A and C). In contrast, the abundance of proteins composing the large 39S subunit was comparable between cells from PS patients and controls ( Fig. 2B and D).
We also analyzed the steady state protein expression of MRPS22, a protein of the small 28S subunit, and found it to be decreased in PS patient skin fibroblasts compared to controls ( Fig. 2E and F). On the contrary, levels of the large 39S subunit protein MRPL54 were similar between patients and controls ( Fig.  2E and F).
These observations led us to conclude that the overall abundance of assembled small mitochondrial ribosomal subunits is lower in cells from the PS patients compared to those from healthy controls, suggesting that the identified ERAL1 variant perturbs proper assembly of the small 28S mitochondrial ribosomal subunit.
12S rRNA levels are low in PS patients, and rescued after ERAL1 lentiviral expression Because ERAL1 acts as a chaperone to the mitochondrial 12S rRNA of the small ribosomal subunit by protecting it from degradation (12), we hypothesized that levels of 12S rRNA are reduced in the patients, while the 16S rRNA of the large ribosomal subunit remains unaffected. We therefore employed qPCR and measured the 12S/16S rRNA ratio in patient and control fibroblasts. In support of our hypothesis, we found that this ratio was significantly reduced in the two patients compared to healthy controls (Fig. 3A).
Next, we asked whether the ectopic expression of wild type ERAL1 in the patient cells would be sufficient to reconstitute 12S rRNA levels. To address this question, we infected patient fibroblasts with lentiviral particles overexpressing either ERAL1 or GFP as a control. We found that overexpression of ERAL1 in both patients (Fig. 3B) rescued the 12S/16S rRNA ratio (Fig. 3C). These data demonstrate that the identified mutation leads to reduced 12S rRNA levels, a condition that is reversible after overexpression of wild type ERAL1.

Mitochondrial function is impaired in PS patients
Disturbed assembly of the mitochondrial ribosome may lead to inefficient translation of mitochondrial DNA (mtDNA)-encoded proteins, which form core subunits of the oxidative phosphorylation (OXPHOS) complexes. To test whether mitochondrial translation is disturbed, we quantified the ratio between the mtDNA-encoded cytochrome c oxidase subunit 1 (MT-CO1) and the nuclear DNA (nDNA)-encoded succinate dehydrogenase complex, subunit A (SDHA) (15) in the PS cells. Indeed, the abundance of MT-CO1 was found lower, while SDHA was unaffected ( Fig. 4A and B). We next measured mitochondrial activity by means of oxygen consumption rate (OCR) in the cells using Seahorse respirometry. Basal respiration was decreased in the PS cells when compared to controls (Fig. 4C). After injection of the uncoupler FCCP, maximal respiratory capacity of the PS fibroblasts was also impaired (Fig. 4C). These findings confirm that mitochondrial function is compromised in the PS cells.

Knockdown of the ERAL1 homologue in C. elegans compromises fertility
In order to test the phenotypic effects of low ERAL1 levels at an organismal level, we used the nematode C. elegans as a model organism and specifically knocked down the expression of the worm ERAL1 homologue E02H1.2. We used the RNAi-sensitive nematode strain rrf-3(pk1426) (16). Worms exposed to E02H1.2 RNAi had approx. 50% knockdown (Fig. 5A), developed normally and had a normal lifespan compared to control HT115 worms (data not shown). Strikingly, we observed that the animals undergoing E02H1.2 RNAi were not carrying eggs throughout adulthood (Fig. 5B). We quantified the number of eggs laid during the first four days of adulthood, which is the fertile period of C. elegans, and confirmed that the E02H1.2 RNAi fed animals hardly laid any eggs (Fig. 5C).
Because the PS skin fibroblasts presented with decreased mitochondrial respiration, we also measured OCR in the worms with E02H1.2 knockdown. Those worms demonstrated significantly decreased OCR (Fig. 5D). Collectively, these results demonstrate that low ERAL1 expression impairs mitochondrial function and compromises fecundity in C. elegans, mimicking the defects we observe in our PS patients.

Discussion
Three women from three different families presented with signs of PS, including deafness and ovarian dysgenesis. We employed exome sequencing and identified a homozygous missense mutation in the ERAL1 gene, c.707A > T. The mutation leads to a damaging p.Asn236Ile substitution that is likely to interfere with the GTP binding capacity of the ERAL1 protein.
Specifically, the observation that the assembly of the small mitochondrial ribosomal subunit-for which ERAL1 is known to play a pivotal role (11,12)-is compromised in the PS cells, implies that the identified mutation affects the proper function of ERAL1 protein. Importantly, the impaired 12S rRNA levels in PS fibroblasts and their rescue upon expression of wild type ERAL1 underlines the causal link between the mutation and mitochondrial function. Additionally, the decreased respiration that was noted in the PS fibroblasts demonstrates disturbed mitochondrial function, confirming that the identified mutation affects mitochondria and is pathogenic. Finally, we performed RNAi experiments in the nematode C. elegans to demonstrate the role of ERAL1 in fertility. Knockdown of the worm ERAL1 homologue E02H1.2 indeed impaired egg production presenting a crucial role of this protein in the nematode's fecundity. Our cross-species findings on both patient fibroblasts and the model organism C. elegans identified the ERAL1 variant as the cause for the clinical symptoms in our three PS patients.
The identification of ERAL1 as a causative gene for PS extends the mutational spectrum for this disease. So far, mutations in five different genes have been described to cause mut/mut wt/mut wt/mut wt/mut mut/mut mut/mut mut/mut   GADD45GIP1  ICT1  MRPL53  MRPL12  MRPL10  MRPL14  MRPL32  MRPL34  MRPL42  MRPL33  MRPL17  MRPL13  MRPL22  MRPL49  MRPL39  MRPL20  MRPL41  MRPL24  MRPL28  MRPL47  MRPL23  MRPL2  MRPL9  MRPL44  MRPL1  MRPL4  MRPL50  MRPL21  MRPL15  MRPL45  MRPL52  MRPL35  MRPL43  MRPL51  MRPL18  MRPL30  MRPL38  MRPL27  MRPL3  MRPL19  MRPL37  MRPL46  MRPL55  MRPL40  MRPL48  MRPL16  MRPL54  MRPL11   relative  Perrault syndrome. These include the HARS2 (6) and LARS2 (7) genes, both encoding mitochondrial tRNA synthetases, the CLPP (8) that codes for a protease of the mitochondrial matrix and the C10orf2 (10) encoding the mitochondrial helicase Twinkle. These findings imply that mitochondria are involved in the development of the disease. Interestingly, recent work on CLPP deficient mice (17) elegantly links ERAL1 mechanistically to CLPP by demonstrating that removal of ERAL1 from the assembled 28S ribosomal subunit is essential for the final maturation of the whole 55S mitochondrial ribosome, and that this removal requires CLPP. Our identification of ERAL1 as a Perrault syndrome gene further ties CLPP and ERAL1 together, as mutations in either of them lead to similar pathology in humans. The involvement of mitochondrial dysfunction in development of deafness is already well established (18)(19)(20). On the other hand, the role of mitochondria in infertility is less known and is only recently starting to emerge. Studies in mice have shown that perturbations of mitochondrial function can affect oocyte maturation (21,22). In humans, age-related decline in oocyte quality and quantity has been associated with mitochondrial dysfunction and impaired OXPHOS, while dietary supplementation with the OXPHOS enhancer CoQ10 improved ovulation in aged women (23). In C. elegans, knockdown of the mitochondrial tRNA synthetase hars-1 led to smaller gonads and to loss of fertility, which was partly due to germ cell apoptosis (6). Similarly, the lars-2 mutant worm had underdeveloped gonads and was completely sterile (7). Also, sub-lethal disruptions in OXPHOS subunits cause sterility in worms (24-26). Our findings of compromised fertility and mitochondrial respiration after knockdown of the worm ERAL1 homologue serve as additional proof for the importance of mitochondria in fertility. The exact mechanisms through which mitochondria regulate fertility, however, remain to be resolved.
Collectively, our results indicate that mutations in another mitochondrial gene, ERAL1, can cause sensorineural deafness and ovarian dysgenesis of PS, strengthening the notion that this syndrome is caused by dysfunctional mitochondria. Detailed functional mitochondrial assays with cells from other unresolved cases of PS may shed more light on the molecular mechanisms underlying the disease and reveal other PS-causing genes.

Materials and Methods
All three PS patients visited the outpatient clinic of the AMC and signed informed written consent for this study. Information regarding the medical history was obtained by interviewing the patients and from their medical files.

Whole exome sequencing
Whole exome sequencing (WES) of PS patients was conducted using the SeqCap EZ Human Exome Library v3.0 (Roche NimbleGen) and a 5500 SOLiD TM instrument (Life Technologies). Samples were prepared using standard SolID 75x35 paired end sequencing protocols. Alignment of sequence reads to the human reference genome (hg19) was done using Lifescope 2.5.1, and variants were called using the GATK2.5 software package. Mean target coverage was 86x for PS patient 1 and 66x for PS patient 2. Coverage of targeted regions at 10x read depth (after removal of duplicate reads) was 87% for patient 1 and 84% for patient 2.

Sanger sequencing of ERAL1
Confirmation of the mutation in the PS patients and genotyping of family members of PS patient 2 was done by Sanger sequencing. Primers were designed to amplify exon 6 of ERAL1 using the Primer3 software (http://bioinfo.ut.ee/primer3-0.4.0/; date last accessed April 27, 2017). Amplification was performed with M13-tagged primers using HOT FIREPol TM DNA polymerase (Solys Biodyne) and a touchdown PCR program. PCR fragments were sequenced using the Bigdye kit v1.1 (Applied Biosystems). Reactions were run on an ABI3700 or ABI3730XL genetic analyzer (Applied Biosystems) and sequences were analyzed using Sequence Pilot (JSI Medical systems) or CodonCode aligner (CodonCode Corporation).

Genotyping TaqMan
Genotyping of 530 individuals from the same village as the PS patients was done by a TaqMan assay on a Roche LC480 lightcycler. For the SNPs, Sanger sequencing confirmed homozygous reference, heterozygous and homozygous mutant samples were used as standards for SNP calling.

ERAL1 3D model
The crystallized structure of Aquifex aeolicus ERA (PDB code: 3IEV) was selected as a template to build a model for human ERAL1 using SWISS-MODEL server (http://swissmodel.expasy.     (14). Protein identification and data analysis was done essentially as previously described (30) using MaxQuant (version 1.5.0.25 (31)). Protein abundancies were determined by label-free quantitation using the composite iBAQ values determined by MaxQuant (31) and normalized considering multiple migration profiles of individual proteins, that is taking into account iBAQ values from all 180 gel slices (60 slices per sample). The profiles were hierarchically clustered by distance measures based on Pearson correlation coefficient (uncentered) and the average linkage method and further analyzed by manual correlation profiling. The clustering and the visualization and analysis of the heat maps were done with the NOVA software v0.5 (32).

Respiration assays
Oxygen consumption was measured using the Seahorse XF96 analyzer (Seahorse Bioscience). Primary skin fibroblasts were plated in 96-well Seahorse plates at a density of 10,000 cells per well and incubated overnight under normal cell culture conditions. The following day, medium was replaced by DMEM (Sigma, D5030) containing 25 mM glucose (Sigma), 1 mM sodium pyruvate (Lonza), and 2 mM L-Glutamine (Life technologies). Basal respiration was measured three times, followed by measurements after addition of 1.5 mM oligomycin, 1 mM FCCP and 2.5 mM antimycin A and 1.25 mM rotenone. For respiration assays in worms, animals were transferred in 96-well Seahorse plates (20 worms per well) and basal oxygen consumption was measured six times.

C. elegans culturing and RNAi experiments
The C. elegans RNAi-sensitive strain rrf-3(pk1426) was obtained from The Caenorhabditis Genetics Center (CGC, University of Minnesota). Worms were cultured and maintained as described previously (33) in 20 C incubators and fed with E. coli OP50 strain, also obtained from the CGC. RNAi experiments were carried out as previously described (34). The RNAi clone targeting the ERAL1 worm orthologue (E02H1.2) is an E. coli HT115 strain and was a kind gift of Dr. Yelena Budovskaya (SILS, Science Park, University of Amsterdam).
For the fertility assays, hermaphrodite worms were treated with E02H1.2 RNAi from L4 stage of development, and allowed to lay eggs. Their progeny (F1) was continuously exposed to the RNAi treatment, and observed throughout development. The number of F2 progeny from individual F1 worms (n ¼ 30) was counted daily for the 4 first days of adulthood. Images of F1 worms before and after reaching adulthood were taken with a Leica DFC320 camera, using an Axio Observer.A1 (Zeiss) microscope and a 10x magnification objective. For imaging, worms were placed on top of a 2% agarose pad on slides, and anesthetized using 1 mM Levamisole in M9 buffer. For the respiration assays, F1 worms at day 1 of adulthood were used.