Defective mitochondrial RNA processing due to PNPT1 variants causes Leigh syndrome

Abstract

Leigh syndrome is a severe infantile encephalopathy with an exceptionally variable genetic background. We studied the exome of a child manifesting with Leigh syndrome at one month of age and progressing to death by the age of 2.4 years, and identified novel compound heterozygous variants in PNPT1, encoding the polynucleotide phosphorylase (PNPase). Expression of the wild type PNPT1 in the subject’s myoblasts functionally complemented the defects, and the pathogenicity was further supported by structural predictions and protein and RNA analyses. PNPase is a key enzyme in mitochondrial RNA metabolism, with suggested roles in mitochondrial RNA import and degradation. The variants were predicted to locate in the PNPase active site and disturb the RNA processing activity of the enzyme. The PNPase trimer formation was not affected, but specific RNA processing intermediates derived from mitochondrial transcripts of the ND6 subunit of Complex I, as well as small mRNA fragments, accumulated in the subject’s myoblasts. Mitochondrial RNA processing mediated by the degradosome consisting of hSUV3 and PNPase is poorly characterized, and controversy on the role and location of PNPase within human mitochondria exists. Our evidence indicates that PNPase activity is essential for the correct maturation of the ND6 transcripts, and likely for the efficient removal of degradation intermediates. Loss of its activity will result in combined respiratory chain deficiency, and a classic respiratory chain-deficiency-associated disease, Leigh syndrome, indicating an essential role for the enzyme for normal function of the mitochondrial respiratory chain.

Introduction

Leigh syndrome, first described as subacute necrotizing encephalomyopathy in 1951 (1), manifests with a progressive encephalopathy, typically in infants. The diagnosis is based on consistent neuropathological features, symmetric lesions of the brainstem and basal ganglia, with characteristic histology and MRI-findings (2). Neurological manifestations commonly show signs of basal ganglia and brainstem dysfunction, such as motor and ophthalmological abnormalities, intellectual impairment, seizures, and often increased lactate levels in blood and CSF, although the disease onset, clinical presentation and disease course vary. The disease manifests typically between 3–12 months of age, most subjects presenting with symptoms by the age of two years, with a rapid disease course commonly leading to death by the age of three years (3). The genetic background of Leigh syndrome is diverse, with known causative variants in more than 75 mitochondrial metabolism associated mitochondrial and nuclear genes (4).

We report here whole-exome sequencing and functional characterization of a new genetic cause, PNPT1, underlying progressive, early-lethal mitochondrial Leigh encephalopathy. The results indicate that defective PNPT1, leading to defective mitochondrial RNA degradation, are a cause of Leigh syndrome.

Results

Clinical description

The subject was the first (1/1) child of healthy non-consanguineous parents of Vietnamese origin. She was born at term after an uneventful pregnancy and was small for gestational age (birth weight 2265 g, height 44cm, head circumference 31cm). Her cousin from her mother’s side has developmental delay of unknown cause, and mother’s younger sister has died of unknown cause at the age of one year. At the age of one month, when the subject was first investigated, she was irritable, hypotonic, had feeding problems, facial hypomimia, dystonic and athetotic movements of the body, mouth and tongue. Laboratory investigations showed slightly elevated blood lactate (2,8 mmol/l) and pyruvate (111mmol/l) with normal levels of creatine kinase, ammonia, liver and kidney enzymes, urine organic acids and plasma and urine amino acids. CSF lactate, protein, glucose and amino acids were normal. At the age of two months she was diagnosed with infantile spasms. Brain MRI at two months of age showed bilateral signal intensity of globus pallidus, suggestive of Leigh syndrome (Fig. 1A and B). Biochemical analysis of mitochondrial respiratory chain (RC) from muscle biopsy sample showed cytochrome c oxidase (COX) deficiency. She had no mitochondrial DNA (mtDNA) deletions, point mutations or depletion in muscle tissue. At six months of age she smiled and reacted to sounds, fixed her gaze to faces, but was extremely hypotonic and did not follow with eyes. At the age of ten months she had nystagmus and dystonic tongue and upper limb movements. She had episodes of vomiting and daily seizures, which were controlled with combined phenobarbital and vigabatrin treatment. She had very little development or active movements, and she died at the age of two years and five months during an infection.

Figure 1

Brain imaging and histology findings consistent with Leigh syndrome. (A) Brain, magnetic resonance imaging; T2 weighed image, at two months of age. Bilateral signal intensity of medial globus pallidus and in pars intermedia (arrow). (B) Diffusion-weighted images. Restricted diffusion in the medial globus pallidus and in pars intermedia (arrow). (C) Nucleus lenticularis: severe atrophy, especially in putamen (left of midline), which is gliotic and practically devoid of neurons. Formalin-fixed, paraffin-embedded (FFPE) sections, Luxol fast blue – Cresyl violet (LFB) staining. Original magnification x10. (D) Midbrain. Substantia nigra: narrow with subtotal loss of dopaminergic neurons, narrowed crus cerebri. FFPE sections, LFB staining. x10. Arrow: n. oculomotorius. (E) Tegmentum (bottom of the 4th ventricle), medulla oblongata: IBA-1 immunohistochemistry to detect microglia. A striking proliferation of IBA-1-positive microglia (red), corresponding to a quite mild subacute necrotizing feature of the tegmentum. Asterisk: n. dorsalis of n. vagus; arrow n. hypoglossus. Original magnification x40. (F) Cerebellar cortical folium: atrophy, narrowed molecular layer (ML), total loss of Purkinje cells (arrows) and atrophic granular cell layer (GrL). FFPE sections, LFB staining x40. (G) Skeletal muscle, respiratory chain activity analysis. No signs of respiratory chain deficiency. Cytochrome c oxidase (COX, activity reflected as brown staining), succinate-dehydrogenase (SDH, potential increased activity reflected as blue staining) histochemical activity assay on frozen sections. No COX-negative, SDH-positive fibers are present. Some fibers show very mild pallor of COX-activity (arrows). x 200. (H) Liver, respiratory chain activity analysis. COX-SDH histochemical activity assay of slightly autolytic liver. No totally COX-negative hepatocytes, but slight pallor and some hepatocytes with increased SDH activity (arrows). x 400.

Neuropathology shows Leigh syndrome with pronounced striato-nigral-cerebellar features

The neuropathological examination revealed mainly subcortical, brainstem and cerebellar atrophic and degenerative features. The striatum/putamen, substantia nigra and cerebellar cortex were severely affected, but also a slight subacute necrotizing feature could be observed in the typical tegmental region in the medulla oblongata as well (Fig. 1C–F). The pathology was consistent with Leigh syndrome, with especially pronounced striato-nigral-cerebellar features. Histochemical activity analysis of cytochrome c oxidase (COX, respiratory chain enzyme complex IV) showed slight pallor in the liver and muscle, consistent with slightly decreased activity (Fig. 1G and H).

Whole-exome sequencing identifies PNPT1 variants underlying Leigh syndrome

Whole-exome sequencing of the subject’s DNA identified altogether 106,524 single nucleotide variants (SNVs). Variants occurring at less than 1% frequency in the 1,000 genome database were selected forward and non-genetic variants were excluded. Biallelic variants were prioritized, resulting in 62 homozygous and 663 heterozygous variants. The SIFT genome tool predicted two (OR2T3, GPR123) of the homozygous SNVs to cause damaging amino acid changes and of the heterozygous changes, 10 genes (PNPT1, OTOP1, GFM2, SUN1, PABPC1, c11orf42, RIN3, AHNAK2, TPRM1, PRDM15) contained two or more damaging variants. Of these 12 candidate genes, we prioritized two (PNPT1, GFM2) that encoded a protein localizing in mitochondria (MitoCarta, mitochondrial protein database), whereas the other genes had no predicted functions in mitochondrial metabolism (5) (Fig. 2A). Parental samples were not available (Fig. 2B). However, cloning experiments showed the variants in GFM2 to occur in the same allele (data not shown), excluding it as a cause for a recessive disease. The variants in the PNPT1 gene encoding the polynucleotide phosphorylase (PNPase) were located in different alleles (Fig. 2C and D). The identified PNPT1 variants (g.chr2: 55910966: NM_033109.4: c.407G > A: p.Arg136His and g.chr2:55910954: c.419C > T: p.Pro140Leu) were extremely rare, with only two heterozygous carriers for each variant being observed in ExAC database. Several bioinformatics tools predicted the variants to be functionally deleterious for p.Arg136His (SIFT score: 0.0, prediction: deleterious; Polyphen score: 0.999, prediction: probably damaging; CADD score: 34.0) and p.Pro140Leu (SIFT score: 0.0, prediction: deleterious; Polyphen score: 0.992, prediction: probably damaging; CADD score: 35.0). The variants occurred in a highly conserved area of the protein (Fig. 2E).

Figure 2

Identification of PNPT1 variants by whole-exome sequencing and structural analysis of p.Arg136His and p.Pro140Leu variants. (A) Schematic representation of the whole-exome sequencing data analysis and filtering. (B) Pedigree of the family. (C) Integrative genomics viewer data showing variants located in different alleles. (D) PNPT1 sequence fragment. Variants locate in exon 5 of the PNPT1 gene and locate in different alleles. (E) Cross-species protein conservation of PNPase, flanking the altered amino acids. Positions Arg136 and Pro140 marked. (F) Domains of PNPase protein, with all known pathogenic variants marked, p.Arg136His and p.Pro140Leu reported here underlined. (G) Overall structure of human PNPase homotrimer (PDB id 3UIK). Monomers shown in different colors. Location of variants affecting the catalytic site, R136 and P140, marked in red. (H) Location of R136 and P140 within the catalytic site. Carbon atoms are colored in magenta. Catalytic residues probed by side-directed mutagenesis are shown with green colored carbon atoms. All of them are involved in phosphorolysis, but D135 and D544 additionally are essential for RNA import.

p.Arg136His and p.Pro140Leu disturb the active site

Functional PNPase is a homotrimer. Each subunit has a N-terminal mitochondrial targeting sequence, followed by two RNase PH (RPH) core domains separated by an α-helical PNPase domain, whereas C-terminal part of the protein contains two RNA-binding domains KH and S1 (6,7). Arg136 and Pro140 locate in the first RPH core domain (Fig. 2F).

To determine the structural and functional consequences of the PNPase variants, they were mapped onto the protein structure (Fig. 2G). Both Arg136 and Pro140 are located within the active site, close to catalytic residues Asp135, Arg445, Arg446, Ser484, Asp538 and Asp544 (Fig. 2H), which were identified by site-directed mutagenesis (7,8). All of these residues are involved in RNA degradation, but Asp135 and Asp544 are also essential for import or stabilization of the RNA component of RNase P (8). The exact modes of RNA binding for import and processive phosphorolysis are unknown. Structural analysis of the identified variants predicts that they affect both RNA import into mitochondria and RNA degradation within the organelle.

Tissue-specific pattern of RC defects in skeletal and cardiac muscle, distinct brain regions, liver, and myoblasts rescued by wild-type PNPT1 expression

Western blot protein analysis of RC subunits showed combined deficiency of respiratory chain complex amounts, which, however, varied in different tissues: complexes CI, CIII and CIV deficiency in the cerebral cortex, CI and CIV in the cerebellum and basal ganglia, isolated deficiency of CIV in the liver, mild reduction of CIV in the muscle and no deficiency in the cardiac muscle. Increased protein amount of other RC subunits in skeletal and cardiac muscle suggested increased mitochondrial biogenesis. An increase of CI, CIII and CIV relative to CII was seen in the liver, however, subunits of all complexes including CII were markedly decreased when compared to cytosolic GAPDH (Fig. 3A). We then analyzed the presence of the RC holocomplexes by blue-native-polyacrylamide gel electrophoresis (BN-PAGE) in cultured myoblasts, which showed a reduction in CI and CIV assembly (Supplementary Material, Fig. S1). Complementation of these cells with wild-type PNPT1 cDNA increased CIV and to a lesser extent CI in subject cells, compared to control cells, strongly supporting the causative role of the PNPT1 variants p.Arg136His and p.Pro140Leu in RC deficiency (Fig. 3B;Supplementary Material, Fig. S2). These results supported the role of the identified PNPase variants for the RC defects in the subject, and indicated that PNPase defect affects the RC complexes in a tissue-specific manner.

Figure 3

RC complex defect in PNPase deficiency. (A) Tissue-specific reduction of respiratory chain enzyme subunits. SDS-PAGE and western analysis (top), of respiratory chain complexes (C) I (NDUFA9-subunit), II (CII, SDHA, 70 kDa subunit), III (core II, 45 KDa subunit), IV (COX-II subunit) and V (ATP5a subunit). Middle and bottom: quantitation of protein levels from SDS-PAGE immunoblots: amount of CI, III, IV and CV (partially mtDNA-encoded), compared to CII (completely nuclear encoded, middle) and CI-CV compared to nuclear GAPDH (bottom) amount, expressed as average percentage and range of duplicates relative to age-matched control with non-mitochondrial disease. S, subject; C, control. (B) Myoblasts: Complementation assay for respiratory chain defects. Retroviral complementation expressing wild-type PNPT1 cDNA or empty virus (pBABE). Complex I and IV amount increased in subject myoblasts after PNPT1 expression compared to myoblasts of four controls with non-mitochondrial disease. Bottom: quantitation of protein levels from BN-PAGE immunoblots: amount of CI, CIII, CIV and CV relative to CII. Protein amount in myoblasts transfected with PNPT1 relative to cells transfected with empty vector, shown as mean and standard deviation from 3 experiments, each with 2–4 controls. S, subject; C, control. (C) PNPase trimerization. Blue-native-polyacrylamide gel electrophoresis analysis. CII used as loading control.

Compound heterozygosity for p.Arg136His and p.Pro140Leu does not affect PNPase trimer formation

The PNPase enzyme is a homotrimer, which can form dimers (8). In previous studies, variants affecting multimerization of the PNPase protein have been associated with decreased import of nuclear encoded RNA into mitochondria, resulting in RC defects (9). Our structural analysis, however, predicted no effect of the pathogenic variants of our subject on trimer assembly. BN-PAGE indicated that the PNPase trimer was formed correctly in the subject’s myoblasts, and the amount of trimer was not reduced (Fig. 3C), suggesting that the variants do not affect multimerization of the protein.

p.Arg136His and p.Pro140Leu variants affect transcript processing

The undefined role of PNPase in mitochondrial RNA metabolism prompted us to test the steady-state abundance of mitochondrial RNA species (mtRNAs). We designed a quantitative PCR (qPCR) approach to detect the unprocessed large precursor mt-transcripts of the polycistronic mtDNA transcript in a non-strand specific manner, testing a number of transcripts with different processing mechanisms (Fig. 4A–C). We found up to three-fold accumulation of unprocessed forms around ND6 subunit transcript of complex I in the patient’s cells (ND6 subunit together with antisense of ND5; ND6 plus antisense of Cyt b, separated by tRNA^Glu (Fig. 4B, top; Fig. 4C)). However correct processing was found to occur for transcripts of COXI and COXII (cytochrome-c-oxidase subunits I and II), separated by tRNA^Ser and tRNA^Asp, (Fig. 4B middle), and adjacent ATPase6 and COXIII transcripts (ATP synthase subunit 6, cytochrome-c-oxidase subunit III) without adjacent tRNAs (Fig. 4B bottom; Fig. 4C). The amount of COXI and COXIII transcripts was decreased (Fig. 4C). These results suggested that PNPase processes specific sites and transcripts, and is not involving processing of all mtRNAs from the polycistronic transcript. The finding of increased aberrant ND6 transcript and decreased COXI transcript was confirmed in mRNA analysis by Northern blotting. The abundance of processed ND6 was remarkably diminished, while the processing intermediate of larger size accumulated in high levels (Fig. 4D). The processed transcripts for ND5 and COXI were diminished, but to a lesser extent when compared to ND6. For both ND5 and COXI no over-accumulation of large unprocessed variants could be observed, but accumulation of smaller fragments for both transcripts was observed, possibly representing degradation intermediates.

Figure 4

RNA processing defect in PNPase deficiency. (A) Circular mtDNA, transcribed to polycistronic mtRNA. Studied gene regions marked in color. (B) Strategy for PCR-quantification of fused transcripts. Top: ND6+ND5 (CI subunits; brown) indicates primer sets generating amplification products only, if the 3’ end of ND6 is unprocessed, and the antisense-strand of ND5 is not excised from the precursor transcript. Cyt b (CIII subunit)+ND6 (dark gray) indicates primer sets generating products when the 5’ end of ND6 is unprocessed and tRNA^Glu and the antisense strand of Cyt b are not excised from the precursor transcript. Middle: COXI+COXII (Complex IV subunits; blue) indicates fragments generated when tRNA^Asp and tRNA^Ser are unprocessed and remain between COXI and COXII transcripts. Bottom: COXIII+ATPase6 (green) indicates fragments when the adjacent COXIII and ATPase6 transcripts are unprocessed. Heavy and light strand encoded transcripts indicated. (C) Unprocessed polycistronic transcripts in subject myoblasts. Cyt b-ND6: dark gray bar; ND6+ND5: brown bar: subject sample compared to age-matched control with non-mitochondrial disease; expression related to nuclear single-copy gene expression (beta-actin). COXI+COXII: blue bar; COXIII+ATPase6: green bar, subject sample compared to age-matched control with non-mitochondrial disease and quantitation of expression relative to nuclear control (GAPDH (glyceraldehyde-3-phosphate dehydrogenase)). Mean and variation between triplicate samples shown. (D) Unprocessed mitochondrial transcripts and degradation intermediates accumulate in subject myoblasts. Northern analysis of mtDNA-encoded ribosomal and mRNAs using oligonucleotide probes. Loading control: cytoplasmic 18S rRNA. Arrows: immature precursor transcript of ND6 (imND6) and the mature ND6 mRNA below. Asterisk: smearing below the mature heavy-strand transcripts ND5 and COX1, indicating accumulation of degradation intermediates. S, subject; C, control.

Discussion

Here, we present evidence that pathogenic PNPT1 variants cause a neonatal-onset fatal mitochondrial Leigh syndrome. The functional defect of the protein PNPase affects processing of the polycistronic mtDNA transcript, and leads to accumulation of specific unprocessed mitochondrial RNA transcripts and combined tissue-specific defects of respiratory chain complexes. The evidence suggests that deficient mtRNA metabolism is a cause for the devastating mitochondrial encephalopathy, Leigh syndrome.

The PNPT1 p.Arg136His and p.Pro140Leu variants in our subject were not previously reported as disease-causing. However, 1) the mutated amino acids are highly conserved and our structural analysis supported them to be functionally deleterious; 2) The carrier frequencies of the variants are very low in populations (1:60.000), with no homozygous carriers found; 3) The RC complex amounts in the subject’s myoblasts were functionally rescued by expression of wild-type PNPT1. These data strongly support the pathogenic role of the variants in the subject.

Early-onset progressive Leigh disease has not been previously associated with PNPT1 variants. The gene has previously been associated with a non-progressive encephalomyopathy (MIM 614932) (9), encephalopathy combined with visual impairment and sensorineural hearing impairment (10,11), or isolated hearing impairment (MIM 614934) (12) (Table 1 summarizes findings in all up-to-date reported subjects). The clinical phenotype of our subject lacked signs of hearing impairment, and showed a rapidly progressive disease course leading to early death. Based on the previous and current results, PNPT1 disorders range from isolated hearing impairment to severe Leigh encephalopathy with combined RC deficiency.

Table 1

Clinical and variant data in our subject and all reported subjects with PNPT1 variants

Reference Subjects	Verdenne et al. (9) (S1)	Verdenne et al. (9) (S2)	Slavotinek et al. (10) (one subject)	Alodaib et al. (11) (2 subjects)	von Ameln et al. (12) (3 subjects)	reported here (one subject)
Nucleotide change	c.1160A>G	c.1160A>G	c.401-1G>A	c.760C>A	c.1424A>G	c.407G>A
			c.1519G>T	c.1528G>C		c.419C>T
Amino acid change	p.Gln387Arg	p.Gln387Arg	splice site	p.Gln254Lys	p.Glu475Gly	p.Arg136His
			p.Ala507Ser	p.Ala510Pro		p.Pro140Leu
Allele frequency in ExAC	NA	NA	NA	NA	NA	2/121116
			27/121342	NA		2/121202
Subject background and disease progression
Ethnic origin	Moroccan	Moroccan	NA	NA	Moroccan	Vietnamese
Pregnancy/birth	Normal	Normal	IUGR	NA	NA	IUGR
Sex	f	m	m	m, m	m, m, f	f
Age at onset	9mo	6mo	NA	NA	childhood	1mo
Outcome	Alive at 14y	Alive at 3.5y	Alive at 13mo	NA	Alive at 24y	Died 2.4y
Clinical, imaging and laboratory findings and symptoms
Developmental delay		+	+	+		+
Hypotonia	+	+	+	+		+
Muscle weakness	+			+		–
Choreoathetosis	+	+				+
Dystonia	+	+				+
Buccofacial dyskinesia	+					+
Dysphagia	+ gastrostomy			bulbar palsy		–
Nystagmus	–			+		+
Ophthalmological abnormalities	–	–	OA, VI, CD	OA	NA	VI
Hearing	NA	NA	impaired	impaired	impaired	normal
Neuropathy	+			+		NA
Epilepsy	Abnormal EEG	Abnormal EEG	IS, myoclonic epilepsy			IS
Muscle histology and histochemistry	Abnormal histology	NA	NA	NA	NA	Normal
Neuroimaging (MRI)	age 14y	age 3.5y	age 6mo	NA	NA	age 2mo
Basal ganglia	+	+	–			+
White-matter	–	–	+			–
Brainstem	NA	NA	NA			–
Blood/CSF lactate	Blood+, CSF+	Blood+, CSF+	NA	Blood-, CSF-	NA	Blood+, CSF-
Observed PNPase defect, mitochondrial DNA, respiratory chain and translation defects
Affected domain	Core 2	Core 2	Core 1/OFD	OFD	Core 2	Core 1
PNPase trimer	Absent	NA	NA	Absent	Absent	Unaffected
PNPase protein	Reduced	NA	NA	Reduced	Unaffected	Unaffected
mtDNA amount	Normal	Normal	NA	NA	NA	Normal
Respiratory chain defects
Fibroblasts	NA	Normal	NA	CI,CIII,CIV↓	NA	NA
Myoblasts	NA	NA	NA	NA	NA	CI, CIV↓
Muscle	Normal	NA	NA	Normal	NA	CIV↓
Liver	CIII,CIV↓	NA	NA	Normal	NA	CIV↓
Brain	NA	NA	NA	NA	NA	CI,CIII,CIV↓
Translation	Affected	NA	NA	Affected	NA	NA

NA, not available; IUGR, intrauterine growth retardation; f, female; m, male; y, years; mo, months; IS, infantile spasms; OA, optic atrophy; VI, visual impairment; CD, chorioretinal defect; OFD, Outside functional domains; mtDNA, mitochondrial DNA; +, abnormal finding; -, normal finding.

The different functional effects of the identified PNPT1 variants likely explain the variable clinical manifestations. Our subject, with the most severe disease reported up-to-date, carried two deleterious variants affecting the first PNPase RPH core domain, whereas other reported variants, causing more moderately progressing or non-progressive disease, affected either one allele of the first core domain, two in the second core domain, or resided in regions outside the core areas (Fig. 2F). These findings suggest that the first core domain is functionally most crucial for the protein function. The RNA-degradation function of PNPase was still preserved in cultured cells carrying experimental deletion mutants incapable of trimer formation (6), but this has not been assessed in patient samples with impaired PNPase trimer formation. Our results indicate that the PNPase trimer with the mutant first core domain is unable to perform specific mtRNA-processing functions.

The early embryonal lethality of the total inactivation of PNPT1 in mice indicates that it is essential for viability (8), but the specific functions of PNPase in RNA metabolism and its localization in human cells are still debated. PNPase has been proposed to function as an RNA import factor in the intermembrane space, mediating the matrix translocation of the 5S-RNA and an RNA component of ribonuclease P (RNAseP). RNAseP is essential for the processing of polycistronic transcripts from tRNA-mRNA junctions (8,13). While 5S-rRNA is not part of the mitochondrial ribosome, and the mitochondrial RNAseP is fully functional in vitro without the RNA component (14), recent evidence suggests that at least a fraction of PNPase is localized in RNA granules (15) of the mitochondrial matrix (16,17), and contributes to RNA-degradation co-operatively with hSUV3 (17,18). Disruption of the PNPase-mediated RNA metabolism pathway should therefore result in increased RNA stability and accumulation of degradation intermediates or polycistronic transcripts, which were found to exist in our subject. These findings strongly support the RNA-processing function of PNPase for ND6 transcript processing from the polycistronic transcript and a role in mitochondrial RNA degradation. Normal processing of polycistronic transcripts flanked by tRNA genes, suggests that the PNPase as an RNA nuclease does not contribute to RNaseP mediated tRNA excision.

Together, our data indicate that a matrix-localized pool of PNPase contributes to mitochondrial transcript processing and quality control, targeting especially ND6 subunit of Complex I, but also COXI and cytb. Defects in this PNPase-dependent RNA processing challenges respiratory chain maintenance, leading to a variable combined respiratory chain complex defect, causing progressive Leigh encephalopathy.

Materials and Methods

Standard protocol approvals, registrations, and patient consents

All subjects’ samples were taken for diagnostic purposes, and used for research with informed consent from the parents. The study was approved by the Ethics committee for Paediatrics, Adolescents and Psychiatry of Helsinki University Hospital.

Whole-exome sequencing

The exome targets of subject muscle DNA were captured with NimbleGen 2.1M Human Exome V.2.0 array and sequenced with the Illumina Genome Analyzer-IIx platform with 2x100bp paired end reads. 72% of the target region was sequenced at a minimum depth of 10-fold. Variant calling was done as described previously (19).

DNA sequence analysis

Sanger sequencing was done using standard procedures. Exon 5 of PNPT1 and the exon-intron boundaries were amplified using PCR, using intronic primers (Supplementary Material, Table S1) and DNA sequences compared with the NCBI reference sequence NG_033012.1.

Blunt-end PCR was done using intronic primers for exon 5 (Supplementary Material, Table S1) and products were purified using QUIquick PCR purification –kit (Cat# 28106, Qiagen, Hilden, Germany) and cloned in E. coli using the Zero Blunt® TOPO® PCR Cloning –kit (Cat.# K2830-20, Thermo Scientific, Waltham, MA, USA). Exon-intron boundaries from the products were amplified by PCR, using intronic primers (Supplementary Material, Table S1). The DNA sequences were analyzed using the BigDye terminator Ready Reaction kit v 3.1 and ABI3730xl DNA Analyzer (Applied Biosystems, Foster City, California, USA) and the sequences processed with the Sequencer 4.5 software (GeneCodes, Ann Arbor, Michigan, USA) and compared to the NCBI reference sequence NG_033012.1.

The variants identified in this study have been submitted to the Leiden Open Variation Database (http://www.lovd.nl/3.0/home; date last accessed June 17, 2017).

Protein structure modeling

p.Arg136His and p.Pro140Leu variants were mapped within the resolved crystal structure of human PNPase (PDB id 3UIK)(20). Structural analysis was performed in Accelrys Discovery Studio v4.1 (provided by Center for Scientific Computing (CSC) the IT center for science (www.csc.fi; date last accessed June 17, 2017)).

Blue-native (BN) and sodium dodecyl sulphate (SDS) -polyacrylamide gel electrophoresis (PAGE) and immunoblotting

BN-PAGE was performed as previously described. (21) For SDS-PAGE, protein was extracted from autopsy-derived samples and Western blot was performed by standard methods. The antibodies used in this study are listed in Supplementary Material, Table S1.

Myoblast culture

Myoblasts were cultured according to standard culturing conditions and immortalized with retrovirus vector containing E6/E7 gene of the human papilloma virus.

Complementation of subject myoblasts with wild-type PNPT1

An expression vector containing full-length human PNPT1 ORF DNA (Source BioScience, Nottingham, UK, 6062060) was Gateway cloned (Invitrogen, Carlsbad, CA USA) into pBABEpuro retroviral expression vector. The pBABE-PNPT1 plasmid or empty pBABE were transfected into amphotrophic Phoenix packaging cells by jetPRIME reagent (Polyplus transfection). Retroviral supernatants were collected, filtered, and used to infect myoblasts. Stable mixed cell lines were maintained by puromycin selection.

qPCR

Total cellular RNA was isolated using Trizol reagent (cat #15596-018, Invitrogen). The RNA samples were treated with RQ1 DNAse (M610A, Promega, Madison, WI, USA) and reverse transcribed using Maxima first strand cDNA kit (cat #K1671, Thermo Scientific, Waltham, MA, USA). Primers used are shown in Supplementary Material, Table S1. QPCR was done with iQ Sybr green (cat#1708880, Biorad, Hercules, CA, USA,). Relative expression was quantified using the delta delta CT method (22).

Northern blot

RNA was isolated from cells using Trizol (Invitrogen). Five μg of total RNA was separated by 1.2% agarose-formaldehyde gel and transferred to HybondTM-N+ membrane (GE Healthcare, Chicago, IL USA) by neutral transfer. T4 Polynucleotide Kinase (NEB) 5' radiolabeled oligonucleotides were used for detection of mitochondrial transcripts (Supplementary Material, Table S1). Hybridization (25% Formamide, 7% SDS, 1% BSA, 0.25M sodium phosphate pH 7.2, 1mM EDTA pH 8.0, 0.25M NaCl2) was performed for 16–20 h at 37 °C. Membranes were washed (2x SSC/0.1% SDS) and then dried for exposure to a Phosphor screen (GE Healthcare, Chicago, IL USA) and scanned with a Typhoon 9400 (GE Healthcare, Chicago, IL USA).

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

Markus Innilä, Anu Harju, Tuula Manninen and Babette Hollman are thanked for technical expertise and Brendan Battersby for providing critical comments on the manuscript.

Conflict of Interest statement. None declared.

Funding

Sigrid Jusélius Foundation (A.S.), Jane and Aatos Erkko Foundation (A.S.), Foundation for Pediatric Research (P.I.), Finnish Cultural Foundation (S.M.), European Research Council (A.S.), Academy of Finland (A.S.; U.R.) and University of Helsinki (A.S. and C.J.C.).

Author notes

The variants identified in this study have been submitted to the Leiden Open Variation Database.