Mutations in ELAC2 associated with hypertrophic cardiomyopathy impair mitochondrial tRNA 3'-end processing.

Mutations in either the mitochondrial or nuclear genomes are associated with a diverse group of human disorders characterized by impaired mitochondrial respiration. Within this group, an increasing number of mutations have been identified in nuclear genes involved in mitochondrial RNA metabolism, including ELAC2. The ELAC2 gene codes for the mitochondrial RNase Z, responsible for endonucleolytic cleavage of the 3' ends of mitochondrial pre-tRNAs. Here, we report the identification of 16 novel ELAC2 variants in individuals presenting with mitochondrial respiratory chain deficiency, hypertrophic cardiomyopathy (HCM), and lactic acidosis. We provide evidence for the pathogenicity of the novel missense variants by studying the RNase Z activity in an in vitro system. We also modeled the residues affected by a missense mutation in solved RNase Z structures, providing insight into enzyme structure and function. Finally, we show that primary fibroblasts from the affected individuals have elevated levels of unprocessed mitochondrial RNA precursors. Our study thus broadly confirms the correlation of ELAC2 variants with severe infantile-onset forms of HCM and mitochondrial respiratory chain dysfunction. One rare missense variant associated with the occurrence of prostate cancer (p.Arg781His) impairs the mitochondrial RNase Z activity of ELAC2, suggesting a functional link between tumorigenesis and mitochondrial RNA metabolism.

variants by studying the RNase Z activity in an in vitro system. We also modeled the residues affected by a missense mutation in solved RNase Z structures, providing insight into enzyme structure and function. Finally, we show that primary fibroblasts from the affected individuals have elevated levels of unprocessed mitochondrial RNA precursors. Our study thus broadly confirms the correlation of ELAC2 variants with severe infantile-onset forms of HCM and mitochondrial respiratory chain dysfunction.
One rare missense variant associated with the occurrence of prostate cancer (p.Arg781His) impairs the mitochondrial RNase Z activity of ELAC2, suggesting a functional link between tumorigenesis and mitochondrial RNA metabolism tRNAs and rRNAs. All remaining protein components of the mitochondrial gene maintenance and expression machineries such as proteins responsible for mtDNA transcription, precursor RNA processing enzymes, the mitoribosomal proteins, mitochondrial aminoacyl tRNA synthetases, and others are encoded by the nuclear genes (Hallberg & Larsson, 2014;Rorbach & Minczuk, 2012). More than 50 nuclear-encoded mitochondrial proteins involved in mitochondrial gene expression are linked to heritable disorders (Frazier, Thorburn, & Compton, 2017;Rahman & Rahman, 2018;Stenton & Prokisch, 2018;Van Haute et al., 2015).
ELAC2 codes for a long form of RNase Z (also referred to as RNase Z L ). Alternative translation initiation of ELAC2 mRNA has been proposed to produce two ELAC2 protein isoforms; one targeted to the mitochondria, the other to the nucleus. Translation of the longer isoform initiates at the methionine codon 1 (M1) and the first 31 amino acids of this isoform are predicted to contain a mitochondrial targeting sequence (MTS; Rossmanith, 2011). This longer isoform of ELAC2 has a well-characterized role in mitochondrial pre-tRNA processing (Brzezniak et al., 2011;Dubrovsky, Dubrovskaya, Levinger, Schiffer, & Marchfelder, 2004;Lopez Sanchez et al., 2011;Rossmanith, 2011). If the translation is initiated at the methionine codon 16 (M16; relative to the longer mitochondrially-targeted form) ELAC2 is targeted to the nucleus. Recent data on the activity of ELAC2 in vivo, generated using a knockout mouse model, revealed that it has a nonredundant role in the processing of mtDNA-encoded tRNAs and that it contributes to the processing of nuclear-encoded tRNA, miRNA, and C/D box snoRNAs (Siira et al., 2018).
Thus far, seven different pathogenic mutations have been reported in ELAC2-associated mitochondrial dysfunction. In our previous work, we investigated the functional consequences of ELAC2 variants in patients presenting with a recessively inherited form of hypertrophic cardiomyopathy (HCM), hypotonia, lactic acidosis, and failure to thrive (Haack et al., 2013). This previous work identified a total of four disease alleles (coding for p.Phe154-Leu, p.Arg211*, p.Leu423Phe, and p.Thr520Ile) in three families. One of these, the variant coding for p.Phe154Leu, was also recently reported as prevalent in consanguineous Arabian families affected by infantile cardiomyopathy (Shinwari et al., 2017). In contrast, a homozygous splice site mutation (c.1423 + 2 T>A) in ELAC2 has been associated with developmental delay and minimal cardiac involvement in a consanguineous Pakistani family (Akawi et al., 2016). A single heterozygous ELAC2 variant coding for p.Pro32Arg was recently reported in an infant presenting with encephalopathy, epilepsy, and growth and developmental retardation (Kim, Kim, Lee, & Cheon, 2017). The patient also developed Tetralogy of Fallot, however, without evidence of cardiomyopathy. Since the transmission pattern of ELAC2-related disease in the families reported previously (Akawi et al., 2016;Haack et al., 2013;Shinwari et al., 2017) was consistent with recessive inheritance, the relevance of this variant remains unclear. Finally, an Assyrian patient presenting with chorea, psychosis, acanthocytosis, and displaying a prolonged survival has been identified to carry compound heterozygous ELAC2 variants coding for p.Gly132Arg and p.Ser347Phe. The patient had mild cardiac hypertrophy without evidence of pump failure. Muscle biopsy did not indicate any evidence of respiratory chain defect, despite the presence of cytochrome oxidase (COX)-negative and ragged-red fibers (Paucar et al., 2018).
In the present work, we report the identification of 16 additional ELAC2 variants (ten missense, two frameshift, and four splice mutations) in individuals who present with mitochondrial respiratory chain deficiency, HCM, and lactic acidosis. We provide further evidence for the pathogenicity of the two previously reported and new evidence for eight newly identified missense variants by studying the RNase Z activity in an in vitro system. Fibroblasts from the individuals with novel ELAC2 variants showed elevated levels of unprocessed mt-tRNA precursors. The combination of in vitro ELAC2 activity and mtRNA processing analysis provided the pathogenicity evidence for all patients harboring the previously unreported ELAC2 variants. Moreover, modeling of the missense substitutions provided additional insight into the effects of substitutions on enzyme structure.

| Ethics statement
Informed consent for diagnostic and research-based studies was obtained for all subjects in accordance with the Declaration of Helsinki protocols and approved by local institutional review boards.

| Exome sequencing, variant prioritization, reevaluation, and verification
For reevaluation of the ELAC2 variants in P1 (previously reported as patient 27 in Taylor et al., 2014) and also for the analysis of the ELAC2 splice variants in P6 and P8, RNA purification from fibroblasts and cDNA retrotranscription was used to verify the identified variants, we used RNeasy mini kit (QIAGEN) and GoTaq 2-Step RTqPCR System (Promega), respectively, according to the manufacturers' protocols. Primers used for cDNA amplification are available upon request.
For P3, targeted NGS sequencing using a custom Ampliseq panel targeting 55 mitochondrial translation genes (IAD62266) and subsequent Ion Torrent PGM sequencing was performed essentially as described previously . Candidate gene sequencing was performed for all coding exons of the ELAC2 gene (including intron-exon boundaries) using M13-tagged amplicons and BigDye v3.1 sequencing kit (Life Technologies). Capillary electrophoresis was performed using an ABI3130xl (Life Technologies). NGS variant confirmation was performed by Sanger sequencing using oligonucleotides targeting the exons of interest.
For P4 and P8, genomic DNA from the individuals and their parents was isolated from whole blood using the chemagic DNA Blood Kit special (PerkinElmer,Waltham, MA), according to the manufacturer's protocol. Exome sequencing was performed as previously described (Kremer et al., 2017). Exonic regions were enriched using the SureSelect Human All Exon kit (50Mb_v5) from Agilent followed by sequencing as 100 bp paired-end runs on an Illumina HiSeq. 2500. Reads were aligned to the human reference genome (UCSC Genome Browser build hg19) using Burrows-Wheeler Aligner (v.0.7.5a). Identification of single-nucleotide variants and small insertions and deletions (indels) was performed with SAMtools (version 0.1.19). For analysis of rare bi-allelic variants, only variants with a minor allele frequency (MAF) of less than 1% in our internal Munich database of 14,000 exomes were considered. SAOURA ET AL.

| 1733
For P5, WES was undertaken using previously described methodologies and bioinformatics variant filtering pipelines (Bonnen et al., 2013) For P6 and P9 a targeted custom panel (Nextera rapid capture; Illumina) containing genes responsible for mitochondrial disorders was used (Ardissone et al., 2018). Variant filtering was performed as described in Legati et al. (2016).
For P7, exome capture and massively parallel sequencing were outsourced (BGI, Shenzhen, China) using Sure Select Human All Exon V.4 Agilent and deep Illumina HiSeq technology (median reads depth = 50×). High-quality variants were filtered against the public (dbSNP146 and EXAC V.0.3) and in-house database, to retain private, rare (MAF less than 1%) and clinically pathogenic nucleotide changes. Variants prioritization in the context of their functional impact were performed using in silico programs for missense mutations (Sift and Poliphen2) and also taking into account changes potentially affecting splice sites. All variants identified by NGS were validated by Sanger Sequencing as well as the segregation in the family.

| RNA isolation and RNA Northern blot
RNA extraction and Northern blot were performed essentially as described previously (Pearce, Rorbach et al., 2017). Briefly, RNA was extracted from cells at 60-80% confluency using TRIzol reagent (Ambion), following the manufacturerʼs instructions. Gels run on the Bio-Rad Mini-Sub Cell GT were used for the separation of RNA samples (5 μg per sample) for northern blot. A half volume of Gel Loading Buffer II (Ambion) was added to samples before heating at 55°C for 10 min, chilled on ice for 2 min, and then loaded onto a 1.2% Agarose gel (1× MOPS; 0.7 M formaldehyde). Electrophoresis was carried out at 4°C in 1× MOPS, 0.3 M formaldehyde and 10 μg/ml ethidium bromide. Gels were semi-dry blotted onto a positively charged nylon membrane (Hybond-N+; GE Healthcare) for >12 hr, after which the membrane was cross-linked and hybridized with [ 32 P] labeled antisense RNA probes as described previously .

| Recombinant protein purification and mutagenesis
Wild type and mutant ELAC2 proteins (GenBank Accession number: NM_018127.6) were expressed from Gly50 to Gln826 using the baculovirus system and insect SF9 cells, and affinity purified as previously described (Yan, Zareen, & Levinger, 2006). Missense mutants were constructed by overlap extension PCR. Subcloning sites used for substitutions in the amino domain and linker were introduced from the BamHI site at the amino end to the internal natural EcoRI site (nt 1450 in NM_018127) and in the carboxy domain from the EcoRI site to the introduced XhoI site following the termination codon.
Sequences of the mutant constructs were confirmed by Sanger sequencing (Macrogen).

| RNase Z processing experiments
The mitochondrial pre-tRNA substrates for RNase Z reactions were prepared by runoff T7 transcription using cis-acting hammerheads to cleave at +1. The mt-tRNA Leu(UUR) substrate has a 29 nt 3′-end trailer with natural sequence ending with a SmaI runoff (-CCC) and mt-tRNA Ile has a 19 nt 3′-trailer, also ending with a SmaI runoff. Substrate 5′ ends were radiolabelled using [γ-32 P]-ATP and polynucleotide kinase. For processing experiments, the concentration of unlabeled substrate was varied, over the range from 4-100 nM, at a constant much lower concentration of labeled substrate used as a tracer. The enzyme was used at the lowest concentration that produces a quantifiable product band at the highest concentration of unlabeled substrate used in the experiment. Kinetic experiments were performed with wild type enzyme at 10 pM using mt-tRNA Leu(UUR) substrate, and at 50 pM using mt-tRNA Ile , as in previous experiments (Levinger & Serjanov, 2012). Mutant enzymes were used at a higher concentration than wild type depending on the impairment factor ( Figure S1). Variant processing experiments were performed in parallel with the wild-type on the same day. Reactions were performed using Processing Buffer (PB) consisting of 25 mM Tris-Cl pH 7.2, 1.5 mM CaCl 2 , 1 mM freshly prepared dithiothreitol, and 0.1 mg/ml BSA. The reason for the use of CaCl 2 , as opposed to for example, MgCl 2 , is related to the mitochondrial concentration of Ca 2+ , which is higher than that of Mg 2+ (Thiers & Vallee, 1957), with our previous studies indicating higher ELAC2 processing activity on mt-tRNA precursors on the presence of Ca 2+ (Yan et al., 2006). Reactions were sampled after 5, 10, and 15 min incubation at 37°C, electrophoresed on denaturing 6% polyacrylamide gels and images were obtained from dried gels

| Summary of clinical features of the investigated patient cohort
We investigated 13 families with a cohort of 13 infants, most presenting with early-onset, syndromic cardiomyopathy, with HCM being present in 10 subjects, dilated cardiomyopathy (DCM) in two subjects, and one subject being reported without any cardiac problems. All subjects with the exception of P10 also presented with lactic acidosis. These clinical features raised suspicion of mitochondrial disease. Indeed, a biochemical defect of the mitochondrial respiratory chain (MRC) complexes was detected in all investigated subjects (n = 10), with isolated Complex I deficiency prevailing in most of the patients (7/10) and the remaining subjects presenting with combined MRC deficiencies. The onset of symptoms was either from birth (P1, P2, and P11), neonatal (P5), infantile (P4, P6-P10, P12 and P13) or early childhood (P3). Most of the patients also displayed developmental delay. Evidence of brain involvement was found in P2 and P9. P7 and P10 were successfully treated by heart transplantation at age 3.8 years and 10 months, respectively (Parikh et al., 2016;Santorelli et al., 2002), whereas P12 underwent two failed cardiac transplants. The summary of genetic, biochemical, and clinical findings of all individuals is provided in Table 1. Pedigrees of investigated families and detailed case reports are provided in the Supporting Information Material.

| In vitro RNase Z activity of mutant ELAC2 enzymes
Despite the increased utility of genetic testing, providing proof of pathogenicity of novel variants remains challenging and follow up functional studies in vitro should therefore be included as an integral part of the evaluation. To provide evidence for the pathogenicity of identified ELAC2 variants, we set out to study the RNase Z activity of the enzyme in the presence of the missense substitutions or the truncating variants resulting from frameshift mutations. Substitutions were introduced into the human ELAC2 cDNA (Genbank Acc# NM_018127.6) and the mutant proteins were expressed in baculovirus using insect SF9 cells (Saoura, Pinnock, Pujantell-Graell, & Levinger, 2017). Affinity-purified recombinant mutant proteins were tested using precursor mt-tRNA LeuUUR or mt-tRNA Ile as substrates.
To assess the utility of the in vitro system to evaluate the pathogenicity of the novel ELAC2 variants, we first tested the three previously reported missense mutations (p.Phe154Leu, p.Leu423Phe, p.Thr520Ile), that have been extensively characterized in terms of pathogenicity in our previous paper (Haack et al., 2013)  Next, we expressed and tested 10 novel missense variants detected in our patients using pre-mt-tRNA Leu(UUR) as a substrate (  (Haack et al., 2013). With this in mind, we tested the p.Arg564Ser mutant (which did not show detectable impairment with the pre-mt-tRNA Leu(UUR) ) using a different substrate, pre-mt-tRNA Ile .
Moderate, however, not statistically significant, impairment of the k cat /K M values ratios (in the range of 20-80% of the WT enzyme) was observed for p.Arg564Ser with the pre-mt-tRNA Ile substrate ( Figure   2c; Table 4).
Two novel frameshift variants were detected in our patient cohort: p.Ile153Tyrfs*6 and p.Cys670Serfs*14, both resulting in premature stop codons (Table 1). The p.Ile153Tyrfs*6 mutation results in early truncation of the ELAC2 protein, eliminating all functional motifs, and was therefore considered a priori as loss of function; expression and assay of this mutant were not attempted.
On the other hand, p.Cys670Serfs*14 occurs closer to the carboxy terminus. We, therefore, tested p.Cys670Ser*14 for enzymatic activity. This mutant expressed poorly, however, and displayed no detectable enzyme activity, confirming the functional importance of domains beyond the truncation, including motif V and other functional elements ( Figure S3). Taken

| 1739
Assessing the clinical significance of sequence variants that may alter splicing, especially in the context of tissue-specific disease manifestation, can be challenging (Spurdle et al., 2008). In four patients from our cohort, P2, P3, P6, and P8, compound hetero-

| In silico characterization of mutations in ELAC2
Having established the damaging nature of the detected ELAC2 missense mutations both in vitro and in living cells, we set out to develop a rationale for the observed biochemical impairment, as summarized in Table S1. To this end, we modeled all 13 substitutions  Figure S3).
Using the Trz1 structure, we have modeled the residues Arg68, Phe154, and Gln388 in the amino domain in which substitutions were found associated with HCM. Modeling using the Trz1 structure is less obviously effective in cases where the residues in the S. cerevisiae Trz1 sequence are not conserved ( Figure S3); in these cases, however, the overall sequence similarity in the region suggests structural relationships of inferred secondary structure elements. All  Figure S4). Phe154 is conserved and aligns with Trz1 Phe103 ( Figure S3). Across the N-and C-domain interface Phe154 closely approaches the motif II region, suggesting that reducing the size of the hydrophobic side chain in the case of p.Phe154Leu could affect the packing of a region which is critical for metal ion binding and catalysis ( Figure S5).
Contiguous residues Leu422 and Leu423 are located in the linker domain of the human ELAC2 at the apex of the loop between two twisted beta sheets in the amino domain. Substitutions at these positions may affect the structure due to subtle changes in regional hydrophobicity ( Figure S6).
We next set out to model residues in the carboxy domain (Pro493, Thr520, Arg564, Lys660, Ala680, Tyr729, His749, and Arg781). All of these but two could be effectively modeled using with the first histidine of motif II (His546). The model implies that subtle regional changes impair catalysis, transmitted to motif I ( Figure S7).
Arg564 is not conserved and is located in the long, overall poorly conserved region between motifs II and III, with no significant polar or hydrophobic contacts in this region of the Trz1 structure ( Figure S3). The mutated residue is reasonably close to motif II, implying that subtle regional changes may impair catalysis when transmitted to this motif.
The side chain of conserved Lys660 (Lys693 in the Trz1 structure) found in β24 is predicted to make polar contacts with the backbone of the preceding residues within 12 residues of the F I G U R E 4 ELAC2 substitutions mapped on the structure of Saccharomyces cerevisiae RNase Z. The structure of S. cerevisiae RNase Z (Trz1, PDB#5MTZ; Ma et al., 2017) is shown in cartoon using PyMol. The amino domain, inter-domain linker, and carboxy domain are colored green, blue, and pale green, respectively. Three views are shown to effectively visualize all the substitutions. All 13 ELAC2 substitutions (3 published previously (Haack et al., 2013), and 10 novel) are shown in all three views. Residues are labeled with S. cerevisiae RNase Z numbers and the numbers in brackets are for the H. sapiens ELAC2 residues. Some residues are not conserved between S. cerevisiae and H. sapiens RNase Z. Residues Arg68, Phe154, Gln388 localized in the amino domain are marked in red; Leu422 and Leu423 are in linker (also marked in red); Pro493, Thr520, Arg564, Lys660, Arg680, Tyr729, His749, Arg781 are in the carboxy domain and indicated in purple. (a) View with amino domain up, carboxy domain down and linker behind with the H. sapiens residues Phe154, Leu422, Leu423, Lys660, Ala680, Tyr729 and Arg781 labeled. (b) The ELAC2 model is rotated with linker on left with the residues Arg68, Gln388 and Arg564 labeled. (c) ELAC2 rotated with linker on right and the residues Pro493, Thr520 and His749 labeled. Note: the residues at Arg68, Phe154, and Gln388 in H. sapiens ELAC2 map to the domain interface motif IV aspartate ( Figures S4 and S8). Loss of these side chainspecific polar contacts with the p.Lys660Ala substitution could, therefore, affect the position of Asp666, the motif IV aspartate, which is critical for metal ion binding and catalysis ( Figure S8B).
Similarly, Ala680; Ser715 in the Trz1 structure) located below the base of β25 could, when replaced by the bulkier valine in the case of the Ala680Val mutation, indirectly affects the location of the Motif IV aspartate ( Figure S8B). Finally, His749 (Ser787 in the Trz1 structure), is predicted to be contiguous with the aspartate in the AxD loop and substitution at this position could perturb that residue.
The AxD loop is conserved and the aspartate in this loop makes polar contacts with the backbone of a conserved leucine at the start and a conserved asparagine at the end of the PxKxRN loop ( Figure S8B; cf, Wang et al., 2012).
The Arg781 residue is present on a long C-terminal α-helix.
Although this residue is not conserved, the region where it is found is generally highly polar. In metazoan ELAC2 enzymes this region consists of frequently interspersed acidic and basic residues while in Saccharomyces cerevisiae it is principally acidic ( Figure S3). In S.
cerevisiae RNase Z, the long α-helix is curved and approaches the predicted location where the substrate acceptor stem is clamped by polar contacts between Lys495 of the PxKxRN loop and nt + 1-2 of the pre-tRNA substrate and between Arg728 of the motif V loop and nt 71-72-73 of the substrate ( Figure 5; Figure S9). The region where Arg781 is found could thus modulate both substrate binding and catalysis, consistent with impairment of the p.Arg781His mutant which arises from the combination of a reduction in k cat and increased K M , (Table 3 and Table 4).

| Genotype-phenotype correlation
Clinical syndromes associated with defects in mtRNA metabolism are characterized by the variable combination of encephalopathy, myopathy, sideroblastic anemia, cardiomyopathy, hearing loss, optic atrophy, and renal or liver dysfunction (Boczonadi, Ricci, & Horvath, 2018;D'Souza & Minczuk, 2018). HCM and lactic acidosis are frequent presentations in some mitochondrial diseases related to dysfunctional mt-tRNA maturation, such as those caused by biallelic variants in MTO1 (MIM# 614667), GTPBP3 (MIM# 608536), AARS2 (MIM# 612035) and RARS2 (MIM# 611524; Ghezzi et al., 2012;Gotz et al., 2011;Kopajtich et al., 2014;Lax et al., 2015).  (Akawi et al., 2016). Both of these variants affect the same consensus donor sequence (exon 15), and it would be interesting to further explore the features of pre-mRNA splice site selection of this particular donor site in cardiac tissue. Interestingly, mtDNA depletion was detected in the hypertrophic heart and not the skeletal muscle of P7 (Santorelli et al., 2002). To the best of our knowledge, quantitative abnormalities of the mtDNA have so far never been associated with defects in mtRNA processing, and additional observations are necessary to confirm this association.

| Enzymatic mechanism of ELAC2
The results of our in vitro analysis of the ELAC2 mutants could be used to design further experiments aimed at better understanding the enzymology of this mitochondrial RNase Z. In particular, the p.Pro493Leu and p.Tyr729Cys substitutions most severely impaired enzyme function, resulting in~1% of WT enzyme activity.
No mutations were found which interfere directly with metal ion binding or catalysis, although numerous such substitutions, constructed by site-directed mutagenesis, greatly impair catalysis, between 500-10,000-fold relative to WT (Karkashon, Hopkinson, & Levinger, 2007;Zareen, Yan, Hopkinson, & Levinger, 2005 P4, and P10, which is exceedingly rare in the general population (MAF = 0.0005165 in gnomAD), is here shown to be significant in the context of mitochondrial tRNA metabolism. In previously published data, no differences in catalysis were observed between wild type and prostate cancer-associated mutants of ELAC2 (the missense substitution and two much more frequent polymorphisms) using nuclear-encoded pre-tRNA substrates (Minagawa, Takaku, Takagi, & Nashimoto, 2005;Takaku, Minagawa, Takagi, & Nashimoto, 2003;Yan and Levinger, unpublished observations). The p.Arg781His substitution, which here reemerged in three described cases of HCM (patients 3, 4, and 10), clearly impairs processing of mitochondrial pre-tRNA substrates both in vitro and in living cells, suggesting that the associated phenotypes (possibly including prostate cancer susceptibility) are mitochondrially-based.
Catalytic efficiencies with WT enzyme are generally lower using mitochondrial substrates than with nuclear-encoded substrates and the lower catalytic efficiency is more pronounced with mt-tRNA Ile than with mt-tRNA Leu(UUR) . Two levels of catalytic activity were thus observed, between nuclear versus mitochondrial and further among different mitochondrial tRNAs (Levinger & Serjanov, 2012;Yan et al., 2006). To explain the first level distinction, mitochondrial tRNAs generally have a weaker and noncanonical secondary and tertiary structure (reviewed in Florentz, Sohm, Tryoen-Toth, Putz, & Sissler, 2003). For example, mt-tRNA Leu(UUR) displays a heterogeneous secondary structure and among mitochondrial tRNAs, it is closer to canonical than mt-tRNA Ile , which displays AU rich stems including an A/C mismatch in the T-stem (Levinger, Morl, & Florentz, 2004). Wild type substrate structures that reduce wild type ELAC2 catalytic efficiencies may thus lead to more pronounced impairment with missense substitutions which otherwise were not observed with more canonical nuclear-encoded substrates.

| Conclusions
Pathogenic variants in ELAC2 impair the RNase Z activity of this critical mitochondrial enzyme. Decreased ELAC2 activity leads to a disturbance of proper mitochondrial gene expression by increasing the amounts of incorrectly processed mtRNA. The consequence of perturbed ELAC2 function is manifested by multiple mitochondrial respiratory chain deficiencies, HCM, and lactic acidosis. Therefore, the ELAC2 gene should be included in gene panels to screen infantile-