Recessive mutations in the kinase ZAK cause a congenital myopathy with fibre type disproportion

Congenital myopathies are a heterogeneous group of neuromuscular diseases, in which the genetic basis is unknown in about half of patients. Vasli et al. report that homozygous mutations leading to premature stop codons in the ZAK gene, which encodes a MAP triple kinase, cause congenital myopathy with fibre type disproportion.

Congenital myopathies are a heterogeneous group of neuromuscular diseases, in which the genetic basis is unknown in about half of patients. Vasli et al. report that homozygous mutations leading to premature stop codons in the ZAK gene, which encodes a MAP triple kinase, cause congenital myopathy with fibre type disproportion.

Abstract

Congenital myopathies define a heterogeneous group of neuromuscular diseases with neonatal or childhood hypotonia and muscle weakness. The genetic cause is still unknown in many patients, precluding genetic counselling and better understanding of the physiopathology. To identify novel genetic causes of congenital myopathies, exome sequencing was performed in three consanguineous families. We identified two homozygous frameshift mutations and a homozygous nonsense mutation in the mitogen-activated protein triple kinase ZAK. In total, six affected patients carry these mutations. Reverse transcription polymerase chain reaction and transcriptome analyses suggested nonsense mRNA decay as a main impact of mutations. The patients demonstrated a generalized slowly progressive muscle weakness accompanied by decreased vital capacities. A combination of proximal contractures with distal joint hyperlaxity is a distinct feature in one family. The low endurance and compound muscle action potential amplitude were strongly ameliorated on treatment with anticholinesterase inhibitor in another patient. Common histopathological features encompassed fibre size variation, predominance of type 1 fibre and centralized nuclei. A peculiar subsarcolemmal accumulation of mitochondria pointing towards the centre of the fibre was a novel histological hallmark in one family. These findings will improve the molecular diagnosis of congenital myopathies and implicate the mitogen-activated protein kinase (MAPK) signalling as a novel pathway altered in these rare myopathies.

Introduction

Congenital myopathies form a heterogeneous group of genetic muscle diseases characterized by congenital or early childhood hypotonia, muscle weakness and structural abnormalities in muscle biopsies (Nance et al., 2012). Congenital myopathies are accompanied by chronic long-term disability, a strong decrease in the quality of life and a shorter lifespan (North et al., 2014). They are subclassified into seven distinct types based on clinical presentation and characteristic features on biopsy. Main classes encompass nemaline myopathies with protein aggregates, core myopathies with well-defined muscle fibre areas devoid of oxidative activity, centronuclear myopathy with predominance of fibres with internal and centralized nuclei, and congenital fibre type disproportion with a bias toward smaller and more abundant type 1 fibres (North et al., 2014). Extensive studies of large families and cohorts led to the identification of ∼30 genes (www.orpha.net) mutated in congenital myopathies; this number is even higher when considering the overlap with congenital muscular dystrophies (Bonnemann et al., 2014). Despite these advances, nearly half of patients still have no molecular diagnosis (Amburgey et al., 2011). This is due to the extensive clinical and genetic heterogeneity and to the limitations of gene-by-gene diagnostic sequencing. Therefore, each uncovered novel congenital myopathy gene probably explains an increasingly smaller subset of remaining unsolved patients. The use of next-generation sequencing (NGS) methods, such as gene panels or exome sequencing, will overcome some of the diagnostic challenges and help uncover novel congenital myopathy genes (Tetreault et al., 2015). The identification of the disease-causing gene allows better care and prognosis for the patients, and more accurate genetic counselling (Vasli and Laporte, 2013). Moreover, the identification of novel implicated genes leads to underlying pathological mechanisms, and highlights novel targets that may be more amenable for the development of treatments.

In this study, in an effort to identify additional genes implicated in congenital myopathies, three consanguineous families, from different ethnic backgrounds, were exome sequenced independently and were found to carry different homozygous mutations in the mitogen-activated protein triple kinase ZAK.

Materials and methods

Patients

One patient originated from France (Family 1), two affected siblings from Quebec (Family 2) and three affected siblings from the UK (Family 3) were recruited independently. All patients and family members underwent a detailed neurological examination by experienced neurologists. Sample collection was performed with written informed consent from the patients and all participating family members, according to the Declaration of Helsinki. Genomic DNA was extracted from peripheral blood lymphocytes using standard procedures.

Histology

Left deltoid muscle for Patient I-1 and deltoid muscle for Patient II-1 were analysed. For conventional histochemical techniques 10-μm thick cryostat sections were stained with haematoxylin and eosin, modified Gomori trichrome, periodic acid-Schiff technique, Oil red O, reduced nicotinamide adenine dinucleotide dehydrogenase-tetrazolium reductase, succinate dehydrogenase (SDH), cytochrome c oxidase, and adenosine triphosphatase (ATPase) preincubated at pH 9.4, 4.63, 4.35 and examined with a Zeiss Axioplan bright field microscope and photos processed with the AxioVision 4.4 software (Zeiss) (Malfatti et al., 2015). The fibre type pattern was determined by counting 1000 fibres from each patient in ATPase 9.4 and 4.35 reactions, and by calculating the percentage of type 1 and type 2 fibres.

Electron microscopy

Muscle specimens were fixed with glutaraldehyde (2.5%, pH 7.4), post-fixed with osmium tetroxide (2%), dehydrated and embedded in resin (EMBed-812, Electron Microscopy Sciences). Ultrathin sections were stained with uranyl acetate and lead citrate. The grids were observed using a Philips CM120 electron microscope (80 kV; Philips Electronics NV).

Genetic analyses

Family 1

MTM1, BIN1 and DNM2 were excluded for Patient I-1 by Sanger sequencing of exons and intron-exon boundaries. MTM1 level was normal by western blot analysis. RYR1 was excluded by cDNA sequencing from muscle, while no pathogenic variant was found in TTN in the exome sequencing analysis.

Homozygosity-by-descent analysis was performed after hybridization of fragmented genomic DNA on Affymetrix genomewide human SNP 6.0 arrays, following the manufacturer’s protocols (www.affymetrix.com).

Family 2

Genotyping analysis of microsatellite markers allowed us to exclude COL6A1 and COL6A2. Due to lack of informative markers in close proximity to COL6A3 we excluded the presence of mutations in this gene by sequencing all exon and exon-intron boundaries of the genomic DNA and the entire cDNA (Tetreault et al., 2006).

Homozygosity mapping was performed using the Illumina OmniExpress SNPs chip (700 K markers). Homozygous regions >1 Mb shared by both affected, heterozygous in parents and unaffected sibling were identified using Genome Studio (Illumina) and AutoSNPa (Carr et al., 2006).

Family 3

Direct sequencing of FKRP, DYSF and CAPN3 and copy number analysis of SMN1 excluded pathogenic mutation of these genes in Patient III-1. Haplotype analysis of all known limb-girdle muscular dystrophy (LGMD) loci in 2010 was performed in Patients III-1 and III-2, which excluded most loci, but was compatible with possible linkage to the loci encompassing DYSF (excluded by sequencing) and TTN (no putative pathogenic variants later identified by exome sequencing).

Exome sequencing

Exome sequencing was performed on the affected sibling and parents of Family 1, both affected siblings and the mother of Family 2, and on two of the three affected siblings (Patients III-1 and III-2) in Family 3. Exome sequencing for Family 1 was performed at BGI (China). Exome sequencing for Family 2 was performed at Perkin Elmer sequencing service centre (Perkin Elmer, USA). Exome sequencing for Family 3 was performed by deCODE genetics (Iceland). Fragmented genomic DNA was enriched with the Agilent SureSelect Human all Exon 50 Mb capture library (v4) and sequenced 90 nt paired-end on Illumina HiSeq2000 sequencer. Patient I-1 was also sequenced at CNG (Evry) following targeted enrichment on a custom library of 2500 genes including genes implicated in neuromuscular diseases and functional candidate genes for congenital myopathies based on data mining. Alignment to the reference genome was done with BWA or SOAP and variant calling with Samtools, SVA or GATK (Bao et al., 2011). Common variants (>1%) found in dbSNP, 1000 Genomes, Exome Variant Server, and internal exome databases were filtered out. Non-synonymous variants predicted to be damaging, putative splicing variants and coding indel found homozygous in affected and heterozygous in parents were selected, based on the known consanguinity of the families. Truncating variants in the ZAK gene (NM_133646) were identified in all sequencing experiments and in all three families and their presence and segregation confirmed by Sanger sequencing (primers in Supplementary Table 1; exome coverage metrics in Supplementary Table 2).

RNA studies

RNA-seq was performed on RNA extracted from muscle tissue of Patient II-1. TruSeq mRNA stranded library preparation was used and the sequencing was performed on an Illumina Hiseq 2000 at McGill University and Genome Quebec Innovation Center (Montreal, Canada). The reads were aligned using STAR (Dobin et al., 2013). Read counts were obtained using Featurecounts (Liao et al., 2014) and expression data using DESeq (Love et al., 2014). To obtain differential expression, the counts were compared to muscle from two samples with normal muscle pathology. Fold-change ≥2 for upregulated and ≤−2 for downregulated with a P-value ≤0.05 was considered significant. Gene Ontology (GO) term and pathway enrichment analysis was performed using DAVID (https://david.ncifcrf.gov). The lists of upregulated and downregulated genes were compared to a background list consisting of genes expressed in 11 muscles RNA-seq samples. Selected genes had a RPKM (reads per kilobase per million) ≥1 (Mortazavi et al., 2008) in more than five samples. For GO terms, biological process, molecular function and cellular compartment were considered. Clusters with an enrichment score >1.5 were considered significant. For KEGG analysis, pathways with a corrected Benjamini P-value ≤0.05 and a % false discovery rate (FDR) >20 were considered relevant (Paco et al., 2013).

For reverse transcription-polymerase chain reaction (RT-PCR), total RNA was extracted from cells and tissues with TRI Reagent® (Sigma), and cDNA was reverse transcribed using the SuperScript® II Reverse Transcriptase (Invitrogen) and random hexamer primers. Real-time quantitative RT-PCR was performed using a Lightcycler 480 (Roche Diagnostics) with the SYBR® Green 1 Master kit (Roche Diagnostics); results were standardized to GAPDH and 18S. Primers for ZAK fragments are in Supplementary Table 1.

Immunoblotting

Protein concentration of homogenized patient and control muscles was determined using the Bio-Rad DC^™ Protein Assay. Equal amounts were resolved by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) (4–12%) overnight and blotted onto nitrocellulose membranes. Membranes were blocked in 5% low-fat milk and following washing were labelled with primary antibody for 1 h at room temperature. The 4-23 antibody is a mouse monoclonal against the kinase domain common to both ZAK isoforms, a kind gift of Maria Ruggieri (The Feinstein Institute for Medical Research, NY). Following washing and incubation in horseradish peroxidase conjugated secondary antibody the protein bands were visualized using a chemiluminescent detection kit (SuperSignal® West Pico, Thermo Scientific).

MRI

Muscle MRI of upper and lower limbs was performed in Patient I-1 and of lower limbs in Patient III-1. Axial T₁-weighted axial images were reviewed.

Results

Clinical features

Family 1

The male patient from Family 1 is French of Caucasian origin. Parents are first-degree cousins. He was affected from 28 months and is presently aged 27 years (Table 1). Muscle weakness was proximal in upper limbs and diffuse, proximal and distal in lower limbs. Winged scapula was present. The patient could walk up to 200 m, had difficulties climbing stairs and was not able to run. Fatigability and diurnal cramps were noted. Clinical examination did not reveal hyperlaxity or contractures. Hyperlordosis was present, and ventilatory capacity slightly decreased from 85% at 17 years to 80% at 21 years. Apart from a slightly elongated face, no facial weakness, ptosis or opthalmoplegia were present and tongue movement, speech and swallowing were normal. Cardiac examination by ECG and cardiac ultrasound was normal. Creatine kinase was mildly elevated (777, 604, 1334 U/l). EMG was myopathic in the four limbs. We observed a decremental response on 3 Hz repetitive nerve stimulation ranging from 10 to 12% in the most wasted muscles. The compound muscle action potential (CMAP) was 1.7 mV (normal >4 mV) for the trapezius, 2.7 mV (normal >4 mV) for the tibialis anterior, and 4 mV (normal >4 mV) for the right anconeus. We treated this patient with anticholinesterase inhibitor (Mytelase) and the patient reported a better functional performance, which was confirmed by clinical tests showing a better muscular endurance (walking, upper limbs muscle endurance). It was also confirmed by the nerve conduction studies and repetitive nerve stimulation at 3 Hz, as CMAP was 4.5 mV for trapezius, 7.44 mV for tibialis anterior and 11.7 mV for the right anconeus after treatment.

Table 1.

Summary of clinical, histopathological and molecular features

	Patient I-1	Patient II-1	Patient II-2	Patient III-1	Patient III-2	Patient III-3
Origin	French Caucasian	French Canadian	French Canadian	British Pakistani	British Pakistani	British Pakistani
Clinic
Age at diagnosis	28 months	4 years	2 years	4–5 years (floppy baby and delayed motor development)	2 years (floppy baby and delayed motor development)	early childhood (floppy baby and delayed motor development)
Muscle atrophy	Proximal and distal muscles	Thighs	Thighs	Thighs	Quadriceps
Muscle weakness	Proximal and distal, upper and lower limbs	+	+	Proximal weakness, upper and lower limbs	Proximal and distal weakness lower limbs	Proximal weakness lower limbs (waddling gait)
Joint laxity	-	++ (elbow, shoulder)	++	-	+	-
Contractures	-	+ (finger, ankle)	-	-	-	-
Scoliosis	+	+	+	+	+	+
Calf hypertrophy	-	-	-	+	+	+
Vital capacity	85% (2004); 80% (2008)	63%	86%	98% (sitting); 86% (lying)	-	-
Facial weakness, ptosis	-	-	-	-	-	-
EMG	Myopathic	Myopathic	-	Myopathic	Myopathic with spontaneous activity	Normal
ECG, cardiac ultrasound	Normal	Normal	Normal	Normal	-	-
CPK	777–604–1334 U/l	Normal (32 U/I)	Normal (90 U/I)	360–439 U/l	709–1526 U/l	Normal / mildly elevated
Histology			No biopsy	No biopsy
Endomysial connective tissue	Normal	Increased			Increased	Normal
Fibre size variation	++	+			++	+
Rounded fibres	+	+			++	+
Type 1 predominance	++	++			+	++
Splitted fibres	+	-			-	-
Central nuclei	++	+			+	+
Rimmed vacuoles	-	++			+	-
Sarcoplasmic disorganization	+	-			+	+
Subsarcolemma material	++	-			+	-
Genetic
ZAK mutation (nt) NM_133646	c.490_491delAT	c.515G>A	c.515G>A	c.280_281insT	c.280_281insT	c.280_281insT
Predicted effect on protein	p.Met164fs*24	p.Trp172*	p.Trp172*	p.Asn95*	p.Asn95*	p.Asn95*

CPK = creatine phospshokinase.

Family 2

Family 2 is of French-Canadian origin from the southwestern part of Quebec. Parents are third degree cousins and do not present any myopathy phenotype, although distal laxity is observed in parents and an unaffected sister. No variants in genes associated with collagenopathies or distal laxity segregating with the hyperlaxity phenotype were identified. Age at diagnosis was 4 years and 2 years for female Patients II-1 and II-2. Patient II-1 passed away at age 37 and the youngest is now aged 33 years (Table 1) (Tetreault et al., 2006). They were hypotonic with contractures at birth. They demonstrated a generalized slowly progressive muscle weakness. Motor delay was observed in both patients, Patient II-1 walked at 3 years and Patient II-2 at 17 months of age. Patient II-2 is still ambulant but Patient II-1 was wheelchair-bound following a car accident and a period of prolonged immobilization at 32 years. They both have hyperlaxity, cervical spine hypermotility and contractures; Patient II-1 showed elbow laxity and repetitive shoulder dislocation due to the hyperlaxity in combination with ankle and finger contractures. Patient II-1 also had a mild scoliosis, which did not necessitate any surgery. Vital capacity was decreased to 63% and 86% in Patients II-1 and II-2, respectively. Cardiac examination by ECG and ultrasound was normal, and there was no CNS involvement. Creatine kinase level were normal for both siblings (32 U/I for Patient II-1 and 90 U/I for Patient II-2) (Table 1).

Family 3

Family 3 is of British Pakistani origin. Parents are first-degree cousins and are unaffected. They have six offspring, of whom three are affected. Patient III-1 is female, and developed pain in her legs and walking difficulties at age 4–5 years, and was noted to have scoliosis at that time. Since then she has experienced slowly progressive weakness. On examination at age 32 years she had waddling gait, hyperlordosis, and prominent proximal lower limb with mild foot extensor weakness (Table 1). There was mild proximal upper limb weakness with scapular winging. There was no facial weakness, contractures or hyperlaxity. Forced vital capacity was 98% in sitting and 86% in lying. Recent echocardiography and ECG were normal. Creatine kinase was mildly elevated (439 U/l). Patient III-2 is male and was noted to be floppy from birth with slow motor development. He had calf hypertrophy and pain in legs while walking in childhood. On examination at age 26 years he had striae, asymmetric calf hypertrophy, quadriceps wasting, a waddling gait and scapular winging (Table 1). Creatine kinase was mild to moderately elevated (709–1526 U/l). Patient III-3 is male and was reported to have floppiness from birth and delayed motor development and scoliosis. He has proximal upper and lower limb weakness with calf hypertrophy and scapular winging. Creatine kinase was normal to mildly elevated.

Muscle pathology and MRI

Affected cases from all families shared the histopathological features (Fig. 1) of fibre size variation with hypo and hypertrophic fibres and numerous rounded fibres. Type 1 fibres were predominant and common to all patients; type 2 fibres were atrophic for Patient II-1 (Fig. 1F) and normal in size for Patient I-1. Patients I-1 and II-1 also displayed regional replacement of fibres by adipocytes but no inflammatory infiltrate. Some split fibres were noted for Patient I-1 (Fig. 1A–C), while an increase in endomysial connective tissue was specific for Patient II-1 (Fig. 1D and E). Hyaline bodies were absent. The muscle biopsy of Patient I-1 was classified as centronuclear myopathy with rimmed sarcolemma based on the predominance of fibres with central nuclei and abnormal accumulation of mitochondria at the periphery. Centralized and internalized nuclei were also a common feature for Patient II-1, sometimes with clump of nuclei. The main difference between the biopsies was the presence of rimmed vacuoles (Fig. 1E) sometimes containing eosinophilic or basophilic masses in Patient II-1, while fibres in Patient I-1 had a spicular aspect due to the peculiar subsarcolemmal accumulation of mitochondria clearly noted on SDH staining (Fig. 1B). The presence of rimmed vacuoles was also an important feature in Patient III-2 (Fig. 1G and H). Both Patients III-2 and III-3 had evidence of sarcoplasmic disorganization and Patient III-2 also had accumulation of subsarcolemmal material, in keeping with findings in Patient I-1 (Fig. 1I).

Figure 1.

Muscle pathology. (A–C) Haematoxylin and eosin, SDH and Gomori trichrome staining from Patient I-1 muscle biopsy. Arrows point to centralized nuclei and arrowheads to subsarcolemmal accumulation of mitochondria. (D–F) Muscle pathology from Patient II-1 biopsy. (D) Islands of muscle fibres are lost and replaced by adipocytes. (E) Scattered atrophic or rounded fibres are mixed with normal sized and hypertrophied fibres. Several muscle fibres (arrows) have typical rimmed vacuoles. Endomysial connective tissue is increased in some areas. A few muscle fibres have centrally situated myonuclei. (haematoxylin and eosin, ×350). (F) The vast majority of muscle fibres are of histochemical type I. The remaining few type II fibres are atrophic (myofibrillar ATPase, preincubation pH 10.2; ×350). (G–I) Muscle pathology from Patient III-2 biopsy. (G) Muscle fibres with linear cracks and rimmed vacuoles. When displaced, nuclei tend to be in a central position. (H) Scattered fibres with multiple nuclei. (I) Numerous ring fibres (Desmin staining). (J and K) Electron microscopy on Patient I-1 biopsy. (J) Ultrastructure of a central nucleus tightly surrounded by myofibrils. (K) Peculiar subsarcolemmal accumulation of mitochondria pointing inside the fibres. Asterisk indicates a central nucleus (J) and the peculiar subsarcolemmal accumulation of mitochondria (K).

Electron microscopy analysis revealed that central nuclei are tightly surrounded by myofibrils but not by a halo of membranes and organelles as for classical centronuclear myopathies with mutations in MTM1, BIN1 or DNM2 (Fig. 1J). In Patient I-1, it confirmed the spicular aspect is due to subsarcolemmal accumulations of mitochondria that point towards the centre of the fibre, separating the myofibrils (Fig. 1K). In Patient II-2, it showed the rimmed vacuoles contain whorls of cytomembranes, suggesting an autophagosomal origin (data not shown).

Muscle MRI in two affected individuals (Patients I-1 and III-2) demonstrated a symmetrical selective pattern of muscle involvement (Supplementary Fig. 2). The glutei were affected in both individuals. In the thigh there was widespread fatty infiltration of muscle in both anterior and posterior compartments with preservation of the sartorius and rectus femoris. Both also demonstrated fatty infiltration of the gracilis with a similar clear delineation between affected and unaffected muscle. Imaging in Patient I-1 showed that in the distal lower limbs the posterior compartment was more severely affected than anterior, and there was involvement of the deltoids, trapezius and anterior compartment of the proximal upper limbs.

Identification of recessive ZAK mutations

To identify the genetic basis of this myopathy, next-generation sequencing was performed on all families independently (Supplementary Table 2). Targeted sequencing of 2500 candidate genes followed by exome sequencing from Patient I-1 and parents was performed. After filtering following a recessive inheritance mode of transmission based on family structure and known consanguinity, the only gene that displayed pathogenic variants on both alleles with the expected segregation was ZAK on chromosome 2q31.1, encoding for the sterile alpha motif and leucine zipper containing kinase (Fig. 2). ZAK mapped to one of the largest homozygous regions, based on DNA microarray analysis and confirmed by exome data (Supplementary Fig. 1A). For Family 2, an exome approach was used on both affected siblings and their mother. After filtering for rare homozygous variants [<1% minor allele frequency (MAF)], four candidate genes were identified: RHBG, TMEM177, ZAK and LOXHD1. All of these genes were present in a homozygous region previously identified by homozygosity mapping (Supplementary Fig. 1B). Segregation analysis with the entire family allowed the exclusion of three of these genes, leaving ZAK as the only candidate. In Family 3 Patients III-1 and III-2 underwent exome sequencing and the resulting variants were filtered to include rare (<1% MAF) homozygous variants. Single nucleotide polymorphism data extracted from exome sequencing was applied in Homozygosity Mapper to define regions of autozygosity (Supplementary Fig. 1C). Only two genes had rare homozygous variants and these were both within regions of autozygosity (ZAK, CAMK2D). Segregation analysis in Patient III-3, both parents and one unaffected sibling excluded CAMK2D and confirmed ZAK as the only remaining candidate gene. Candidate ZAK homozygous variants were: a deletion c.490_491delAT in exon 7 leading to a frameshift and a predicted premature stop codon p.Met164fs*24 in Patient I-1, a nonsense variant (c.515G>A; p.Trp172*) also in exon 7 in Patients II-1 and II-2, and a T insertion in exon 4 leading to a premature stop codon (c.280_281insT p.Asn95*) in sibling Patients III-1, III-2 and III-3 (Fig. 2A–D).

Figure 2.

Genetic analysis and identification of ZAK mutations. (A–C) Pedigrees of the three families and chromatopherograms showing the mutations observed in each family, in relation to the reference sequence NM_133646. Black symbol indicates affected individuals and the dot indicates heterozygous carriers. (A) Family 1; (B) Family 2; and (C) Family 3. (D) The ZAK protein encompasses a kinase domain (aa 16–277) with an ATP binding site (aa 45), a proton acceptor site (aa 133) and autophosphorylation sites (aa 161, 165), a leucine zipper (LZ; aa 287–308), and a sterile alpha motif (SAM; aa 336–410) in isoform 1. Two isoforms (iso 1 and iso 2) are known and differ in their C-terminus. Red arrows indicate positions of the frameshift in family 1, the nonsense in Family 2 and the frameshift in Family 3. Blue arrows indicate the positions of mutations linked to limb defects and reported in Spielmann et al. (2016).

The mutations were confirmed using Sanger sequencing, and parents were all heterozygous for the changes in accordance with a recessive inheritance. These mutations were not reported in dbSNP, 1000 Genomes, NHLBI Exome Variant Server, or the ExAC database. In addition, we did not observe these variants in two internal French and Canadian exome databases (1550 and 2000 samples including ethnically matched individuals) and by Sanger sequencing of controls (200 French and 65 French-Canadians). Of note, there are no homozygous nonsense or frameshift variants for the ZAK gene in any of these databases, supporting the notion that truncations or loss of function in this protein are deleterious. The sequencing of a cohort of 184 unrelated patients classified as centronuclear myopathy and 26 French-Canadian patients affected by a congenital myopathy with hyperlaxity did not reveal additional mutations.

ZAK, also called MLTK, MRK or MLK7, is a member of the mitogen-activated protein triple kinase (MAPKKK) family implicated in signal transduction and encompasses an N-terminal kinase domain encoded by exons 2 to 9, followed by dimerization domains: a leucine zipper motif and a sterile-alpha motif (SAM) (Fig. 2D) (Liu et al., 2000; Bloem et al., 2001; Gotoh et al., 2001). An alternate splice isoform 2 differs in the 3’UTR and does not encode the SAM domain.

As all uncovered ZAK mutations are predicted to lead to loss of expression of the full-length protein we explored if RNA sequencing would give us some evidence to support mRNA decay. To assess the impact of the variants on RNA level, transcriptome analysis on Patients I-1 and II-1 were performed. Whole transcript microarrays of cells from Patient I-1 detected an important decrease in ZAK mRNA level (down to 43% of controls). This observation was confirmed by RT-PCR analysis using two different primer pairs in Patient I-1 (Fig. 3A). RT-PCR using the same primer pairs in Patient III-2 showed a reduction in ZAK transcripts compared to controls (Fig. 3A). Sequencing of the RT-PCR product in Patient III-2 confirmed the presence of the homozygous mutation. RNA sequencing on skeletal muscle of Patient II-1 also demonstrated a significant decrease of ZAK (−3.5-fold change, P-value: 4.17 × 10⁻¹⁰) compared to control samples (Fig. 3B). Western blot analysis on skeletal muscle proteins from Patient III-2, using an antibody specific to the kinase domain, which is present in both ZAK isoforms, showed a marked reduction in intensity of the 52 kDa protein that corresponds to the shorter isoform 2 of ZAK in comparison to two controls (Fig. 3C). The longer ZAK isoform 1, expected to be ∼91 kDa, was not detected in keeping with muscle specific expression of the isoform 2 (Gross et al., 2002). RNA-seq data on Patient II-1 confirms the absence of expression of isoform 2 in skeletal muscle. Together, these data strongly support nonsense mRNA decay as the main impact of ZAK mutations.

Figure 3.

ZAK expression and impact of mutations. (A) RT-PCR from Patient I-1 (PI-1) (left) and Patient III-2 (PIII-2) (right) and control cells with primers amplifying exons 3 to 11 (ZAK_a) or 9 to 11 (ZAK_b). GAPDH and 18 S were used as housekeeping gene controls. (B) Gene reads counts from RNA-seq data on muscle biopsies from Patient II-1 and two control samples. (C) Western blot analysis of muscle extract from Patient III-2 and two controls with an anti-ZAK 4-23 antibody. (D) Comparative RT-PCR analysis of different mouse tissues. GAPDH was used as a housekeeping gene control to normalize the values, which are displayed as a ratio compare to quadriceps. (E) Comparative RT-PCR analysis of mouse and human myoblasts and myotubes from 0 to 13 days of in vitro differentiation. GAPDH was used as a housekeeping gene control to normalize the values, which are displayed as a ratio compare to mouse C2C12 muscle cells at 3 days. Error bars represent standard error of the mean.

Physiopathological insight

ZAK was previously reported to be expressed in different tissues with a higher expression in skeletal muscle and heart where the shorter isoform 2 is predominant (Liu et al., 2000; Bloem et al., 2001; Gotoh et al., 2001; Gross et al., 2002). Mining the GTEx database confirmed highest expression in skeletal muscle and bladder among 43 human tissues and the predominant presence of isoform 2 in skeletal muscle (Supplementary Table 3) (Consortium, 2015). We used RT-PCR from different mouse tissues and differentiating mouse and human muscle cells in culture and found ZAK is expressed in several tissues, especially in type I skeletal muscles such as diaphragm and soleus, while it appeared more expressed in dividing myoblasts than differentiating myotubes (Fig. 3D, E and Supplementary Fig. 3). Together, these data support the importance of ZAK isoform 2 in skeletal muscle.

We compared the transcriptome of Patient II-1 muscle with two controls and identified a total 518 differentially expressed genes (124 downregulated and 394 upregulated) (Supplementary Table 4). GO enrichment analysis considering biological process, molecular function as well as cellular compartment was performed using DAVID and was consistent with analysis obtained for KEGG pathways (Table 2). Among the upregulated GO terms and pathways and the most relevant to a muscle disease were cell adhesion, extracellular matrix (ECM), glycosaminoglycan/carbohydrate binding, muscle development, hypertrophic cardiomyopathy and dilated cardiomyopathy, enzyme receptor protein signalling, and TGF-beta signalling. The main downregulated categories were gluconeogenesis, glucose metabolism and insulin pathway, sarcomere protein, muscle development and differentiation, calcium and MAPK signalling pathway. A similar pattern of up- and downregulated genes and pathways are observed in other muscle diseases, especially in Ullrich muscular dystrophy but also in X-linked myotubular myopathy and others (Supplementary Table 4) (Noguchi et al., 2005; Saenz et al., 2008; Kotelnikova et al., 2012; Paco et al., 2013). Genes such as MYH8, TNN2 and MYH3, involved in muscle contraction, regeneration and differentiation, are upregulated in disease muscles (Paco et al., 2013). These genes are usually predominantly expressed in foetal skeletal muscles and could be associated with attempt to regenerate the damage muscle fibres. Collagens and extracellular matrix genes, including biglycan (BGN) and lumican (LUM), are upregulated (Saenz et al., 2008; Paco et al., 2013) and are known to be an indication of muscle fibrosis (Zanotti et al., 2005; Zanotti and Mora, 2006). In contrast to the increased expression of extracellular matrix genes, sarcomeric components of the actin cytoskeleton and Z-disc (ACTA1, ALDOA, CSRP3, FLNC, FHL3, MYOZ3, TTN and TCAP) show a lower expression. Genes involved in the glycolytic pathway (PGM1, PGAM2 and FBP2) have a lower expression (Noguchi et al., 2005). It is also interesting to observe a decrease expression of the MAPK signalling pathway, as ZAK is a member of the MAPKKK family. The downregulation of this pathway did not stand out in the transcriptome analysis performed on other muscle diseases and might thus be a more specific signature for ZAK myopathy.

Table 2.

GO terms and KEGG pathways enrichment

Upregulated
GO terms	Enrichment scores
Cell adhesion	14.95
Extracellular matrix	13.16
Glycosaminoglycan binding	6.3
Plasma membrane	5.23
Skeletal development	4.62
Enzyme receptor signalling pathway	3.91
Collagen - skin development	3.81
Ion binding	3.58
Cell-cell signalling	3.35
Muscle development	2.59
Muscle contraction process	2.45
Proteoglycan binding	2.39
Response to wounding	2.25
KEGG pathways	Number of genes	P-value	FDR
ECM-receptor interaction	7	1.29 × 10−8	2.00 × 10−7
Focal adhesion	8	1.66 × 10−7	5.15 × 10−6
Cell adhesion molecules	6	9.98 × 10−7	4.64 × 10−5
Complement and coagulation cascades	5	4.40 × 10−6	2.73 × 10−4
Hypertrophic cardiomyopathy	5	2.92 × 10−5	0.0022
Dilated cardiomyopathy	5	2.99 × 10−5	0.0027
Tight junction	5	1.41 × 10−4	0.015
Pathways in cancer	6	2.03 × 10−94	0.025
p53 signalling pathway	4	3.16 × 10−4	0.044
TGF-beta signalling pathway	4	8.29 × 10−4	0.128
Downregulated
GO terms	Enrichment scores
Sarcomere	4.33
Muscle development	4.26
Glucose metabolic	3.33
Calcium homeostasis	2.17
KEGG pathways	Number of genes	P-value	FDR
Glycolysis	6	4.72 × 10−8	7.59 × 10−7
Insulin signalling pathway	6	4.13 × 10−6	1.33 × 10−4
Calcium signalling pathway	5	4.60 × 10−5	0.0022
MAPK signalling pathway	5	6.76 × 10−94	0.0434

ECM = extracellular matrix.

Discussion

We have identified recessive mutations in the kinase ZAK as a novel cause of congenital myopathy. Different mutations were found in three unrelated families with overlapping phenotypes. The mutations were homozygous as suggested by the known consanguinity of the families, and segregated as a recessive trait with the disease. All mutations caused a premature stop codon within the kinase domain at the N-terminus of the protein, and lead to nonsense mRNA decay.

Clinical and histological spectra

Patients with ZAK mutations share a generalized slowly progressive muscle weakness of neonatal or childhood onset with a mild to moderate decrease in vital capacity (63–98% of predicated vital capacity). Developmental delay and scoliosis are also important clinical features of ZAK myopathy. Unlike other congenital myopathies such as centronuclear, core or nemaline myopathies, no prominent facial weakness is present. The presence of mild contractures with distal hyperlaxity is a major feature in Family 2, but is not observed in the other families with the exception of Patient III-2 for whom hyperlaxity was described. Inherited muscle diseases are known to have a high clinical heterogeneity, and genetic and environmental heterogeneity may link a single gene with variable clinical features, severity and age of onset (Cirak et al., 2013). We believe the patients in this study share sufficient clinical and pathological features to support a common disease mechanism. Noteworthy, although patients of Family 3 presented with a congenital/early childhood onset, they displayed a LGMD phenotype in adulthood, suggesting ZAK mutations as a cause of LGMD phenotype as well. At the histopathological level, all patient biopsies share fibre size variation and predominance of type 1, as in other congenital myopathy. The presence of rimmed vacuoles and especially the spicular subsarcolemmal accumulation of mitochondria could be a pathological feature useful to identify additional patients to be tested for ZAK mutations. Targeted or exome sequencing of a larger cohort of patients with myopathies will be needed to characterize the full clinical and pathological spectrum linked to ZAK mutations. Thus, ZAK represents an excellent novel candidate for patients with undefined congenital myopathies or LGMD and no genetic cause identified so far. Moreover, preliminary data suggested clinical and neuromuscular improvement of Patient I-1 following treatment with an anticholinesterase inhibitor. Alteration of the neuromuscular junction and a myasthenic component are thus possible in this patient and a tight correlation with ZAK mutations will have to be confirmed by future studies.

Pathological mechanisms

ZAK is a serine-threonine kinase that was implicated in the activation of the ERK, JNK and p38 pathways (Liu et al., 2000; Bloem et al., 2001; Gotoh et al., 2001; Christe et al., 2004). Overexpression in cultured cells induces apoptosis (Liu et al., 2000) and preliminary experiments in zebrafish confirmed that exogenous expression of ZAK is toxic (unpublished data) suggesting a tightly regulated level of ZAK is needed, consistent with its role in signal transduction. ZAK was also implicated in actin fibre modulation and in the TGF-beta-induced cardiomyocyte hypertrophy in vitro while ZAK overexpression in vivo induced cardiac hypertrophy (Christe et al., 2004; Huang et al., 2004; Hsieh et al., 2015). The link between ZAK and these different pathways is interesting as most are involved in myogenesis (Burks and Cohn, 2011). Signalization through the p38 pathway regulates muscle regeneration and causes the irreversible withdrawal from the cell cycle, which is necessary for the myoblast differentiation. The JNK pathway is known to act as a negative regulator of myogenesis, whereas ERK has many roles such as both enhancing myoblast proliferation and repressing their differentiation. A possible role of ZAK in muscle regeneration is in concordance with the variable muscle fibre size and the presence of central nuclei observed on the muscle biopsies.

Our pathological data support that loss of ZAK in human leads to fibre size variation and predominance of type 1 fibres as well as accumulation of mitochondria at the periphery of the fibres in some cases. In vitro assays have shown that ZAK isoform 2 (also known as MRK-beta) preferentially activates the p38γ/ERK6 pathway via MKK3/MKK6 (Gross et al., 2002). Moreover, p38γ is constitutively activated in type I skeletal muscle but not in type II fibres muscle and a p38γ knockout shows specific reduction of slow muscle size over time, indicating a role for p38γ in slow muscle growth (Foster et al., 2012). Interestingly, the number of type I fibres was significantly increased in soleus from p38γ knockout mice (Foster et al., 2012). Thus, it is plausible that disruption of this MAPKKK cascade underlies the predominance of type I fibres observed in patients.

Recently, mutations in ZAK were identified in patients affected by development limb defects. Spielmann et al. (2016) reported a missense variant (Phe368Cys) and an in-frame deletion both in the SAM domain, which is only present in the longer ZAK isoform 1 that is barely expressed in muscle (Fig. 2D and Supplementary Table 3). ZAK isoform 2 appears to be more expressed in muscle than in bones in mice (Supplementary Fig. 3). The six patients reported in our study carry truncating mutations in the kinase domain and have a distinct phenotype with no split-foot abnormality. This suggests that limb defects are specifically due to variants present in the SAM domain without loss of ZAK. The mutated long isoform of ZAK could potentially act as a dominant negative resulting in a downregulation of Trp63 in the limb bud (Spielmann et al., 2016). The complete inactivation of ZAK seems to have a more severe effect in mice, as the knockout mouse was reported to be embryonic lethal due to severe cardiac oedema and growth retardation (Spielmann et al., 2016). In contrast, we are reporting six patients with null mutations in ZAK and presenting a myopathy phenotype. Although we have no evidence for this apparent species difference, it maybe that ZAK is more essential for normal developmental in mice than in humans.

Transcriptome analysis from muscle of one ZAK patient showed similarities with other muscle diseases, in particular congenital myopathies such as Ullrich muscular dystrophy with upregulation of extracellular matrix components as well as a downregulation of sarcomeric genes and downregulation of gluconeogenesis as in X-linked myotubular myopathy. Upregulation of extracellular matrix genes seems to be common to many dystrophic processes and is potentially an indication of signal transduction defects. While alteration in the regulation of signal transduction can have a wide cellular impact, future studies will be needed to identify the precise role of ZAK and its specific substrates in skeletal muscle and to decipher the molecular cascade of events leading to the associated myopathy.

Web resources

Gene table of neuromuscular disorders: www.musclegenetable.fr

1000 genomes: www.1000genomes.org

NHLBI exome variant server: evs.gs.washington.edu/EVS

dbSNP: www.ncbi.nlm.nih.gov/SNP

Affymetrix: www.affymetrix.com

Illumina: www.illumina.com

AutoSNPa: http://dna.leeds.ac.uk/autosnpa/

ExAC Browser: http://exac.broadinstitute.org

GTEx: http://www.gtexportal.org/home/

Glossary

Abbreviations

MAPK

mitogen-activated protein kinase

RT-PCR

reverse transcription-polymerase chain reaction

ZAK

mitogen-activated protein triple kinase/sterile alpha motif and leucine zipper containing kinase