Introduction

Desmin is an intermediate filament protein expressed in cardiac, skeletal and smooth muscle. Desminopathies, are typically caused by autosomal dominant mutations in DES [1] and are clinically heterogeneous [2]. Desminopathy may manifest as myopathy, typically with adult onset (OMIM #601419), cardiomyopathy (OMIM #604765), cardiac conduction defects, arrhythmias; or a combination of these features. Respiratory problems are also common [2]. DES variants are a common cause of myofibrillar myopathy [2]; characterised by myofibrillar disarray, granulofilamentous inclusions on muscle pathology and desmin-containing protein aggregates by immunohistochemistry.

DES variants are most commonly autosomal dominant or de novo dominant missense mutations [1,2,3]. Among DES missense variants, the most common amino acid substitution is to a proline, as this disrupts helix formation in the rod domain of desmin [3]. Data-mining of all DES variants classified in ClinVar as pathogenic or likely pathogenic, or classified as disease-causing in the Human Gene Mutation Database (HGMD; see Supplemental Table 1) identified only seven splice site variants (and see also [3]), with a synonymous and missense variant affecting the last nucleotide of exon 3 predicted to disrupt splicing. No deep intronic variants in DES have yet been identified in an individual affected with suspected desminopathy, which disrupt the production or function of desmin protein due to splicing aberrations; in this case, inclusion of an artificial exon/intronic fragment encoding a premature stop codon and subject to mRNA silencing by nonsense-mediated decay.

A few rare cases of autosomal recessive desminopathy have been described [4,5,6,7,8,9,10,11], ranging from severe infantile to adult-onset, with variable phenotypic presentations affecting skeletal, cardiac and smooth muscle (Table 1; extended descriptions of clinical and histopathological phenotypes are provided within Supplemental Table 2).

Table 1 Clinical features of autosomal recessive DES cases
Full size table

Methods

This study was approved by the SCHN human ethics committee 10/CHW/45 with informed, written consent. Genomic sequencing was performed on gDNA extracted from blood as described [12, 13]. PCR-free whole genome sequencing was performed at the Broad Institute of Harvard and MIT. Samples were prepared utilizing custom next generation sequencing index adaptors (Integrated DNA Technologies, Iowa, 52241, USA) and Kapa Biosciences HyperPrep library construction kit and sequenced on an Illumina HiSeqX v1 (2 × 150 bp reads). Data processing included read de-multiplexing, aggregation, alignment (Burrows-Wheeler Aligner: Bwa-Mem) with 30 × mean coverage. Segregation analysis of DES variants was performed by Sanger sequencing. RNA isolated from 30 × 8 µm muscle cryosections (10 mm2 surface area) was used for RNA sequencing and RT-PCR [13]. cDNA was synthesised from 1 µg of total skeletal muscle RNA [14] and PCR performed using Immolase DNA polymerase (Bioline, NSW, Australia). Western blotting [15] and immunohistochemistry [16] were performed as described. Antibodies: anti-desmin NCL-L-Des-DerII, anti-myotilin NCL-myotilin (Leica Biosystems, VIC, Australia); anti-mouse IgG light chain and anti-rabbit HRP-conjugated secondary antibodies (GE healthcare, NSW, Australia); Cy3-conjugated goat anti-mouse IgG secondary antibody (Jackson Immunoresearch Laboratories, PA, USA).

Case report

Clinical history

The proband (II:3) and her affected older sister (II:2), born to non-consanguineous Italian parents (Fig. 1ai), presented in their mid to late 20’s with limb-girdle pattern of muscle weakness and elevated serum creatine kinase (up to 1070 U/L; reference range 0–175 U/L). Both sisters had normal developmental milestones and academic performance. The weakness was slowly progressive, and associated with cardiomyopathy and respiratory weakness. Echocardiogram in II:3 showed severe systolic dysfunction requiring a defibrillator insertion, with an ejection fraction of 15%, dilated left ventricle (LV) and prolonged QRS trace (Fig. 1aii). A skeletal muscle biopsy from II:3 was performed at the age of 27 years that showed dystrophic changes, with variation in fibre size, scattered necrotic fibers and an increase in central nuclei (Fig. 1aiii); oxidative stains were normal. Electron microscopy showed extensive myofibrillar disarray, reduplication of basal lamina, no inclusions or granulofilamentous aggregates (Fig. 1b). By the age of 34 years, II:3 developed respiratory weakness requiring continuous positive airway pressure (CPAP), and by the age of 43 years, became dependent on use of a wheelchair. Her older sister II:2 showed a somewhat less severe clinical course; presenting with progressive limb-girdle weakness with cardiac involvement at the age of 29 years, requiring use of a wheelchair at 45 years of age. II:2 has not required insertion of a defibrillator, or any respiratory support to date.

Fig. 1
figure 1

a (i) Family pedigree showing inheritance of recessive DES variants. (ii) Echocardiogram for Patient II:3 showing severe end diastolic (normal LVIDd < 5.3 cm) dilatation and intraventricular conduction delay on ECG. (iii) Histology haemotoxylin and eosin stain of frozen skeletal deltoid muscle from II:3 demonstrating increased internal nuclei, variation in fibre size, scattered necrotic fibres and nuclear clumps. b Electron microscopy of Patient II:3. (i) Myofibrillar loss and disarray, and an internalised nucleus. Scale bar 5 µm. (ii) Longitudinal orientation showing misalignment of sarcomeres and variation in myofibril thickness. Scale bar 5 µm. (iii) Higher magnification images showing disorganisation of myofibrils lying at divergent angles. Scale bar 2 µm. (iv) Multiple layers of reduplicated basal lamina. Scale bar 2 µm. c (i) Western blot loading decreasing amounts of Control or Patient II:3 skeletal muscle lysate showing a reduction in levels of desmin in II:3, with no change in levels of myotilin, actinin or actin (myofibrillar protein controls). (ii) Immunohistochemistry of frozen muscle sections stained with NCL-spectrin as a marker of sarcolemmal integrity and NCL-desmin. Variable intensity cytoplasmic staining was observed for desmin compared to uniform cytoplasmic staining in control

Full size image

Targeted gene sequencing of CNBP, LMNA, FHL1, FKRP and CAPN3 was normal, and exome sequencing failed to identify the genetic cause (N.B. the DES c.569 T > C, p.(Leu190Pro) variant was not identified due to poor coverage in the region).

Genomic sequencing results

Whole genome sequencing identified compound heterozygous variants in DES in the affected sisters (Figs. 1a and 2b): a paternal missense variant Chr2(GRCh37):g.220283753 T > C, NM_001927.3, c.569 T > C, p.(Leu190Pro), also inherited by the two unaffected brothers; and maternal deep intronic variant NG_008043.1, g.220289644 G > A, NM_001927.3 c.1289-741 G > A. Neither DES variant is present in the gnomAD database. Variants have been submitted to the gene variant database at www.LOVD.nl/DES (variant IDs 0000446054, 0000446055).

Fig. 2
figure 2

a Disparate patterns of mRNA expression levels from DES alleles. (i) Schematic representation of allelic percentages derived from RNA-seq read depth. The paternal g.220283753 T > C, c.569 T > C variant shows allele bias, accounting for 92% of RNA-seq reads. Three other common SNPs in the coding region also show allele bias, with the reference sequence present in 92% of reads. Inheritance of these SNPs could not be determined; although allele bias data infers maternal inheritance. No splicing aberrations were observed to result from these common SNPs. (ii) Screenshot showing the RNA sequencing reads over the paternal missense c.569 T > C variant and substantial allele bias of variant C nucleotide. b Schematic representation of DES exon structure and consequence of the DES intronic variant c.1289-741 G > A, r.1288_1289ins1289-739_1289-622 (118 bp artificial exon/intronic fragment and abnormal termination codon). Enlarged representation of the artificial exon/intronic fragment and flanking essential splice sites. Predicted strengths of the splice donor and accepter using Alamut Visual® Interactive Biosoftware are annotated beneath. C: cDNA sequencing electropherogram of a RT-PCR amplicon showing inclusion of the 118 nt artificial exon/intronic fragment into the DES mRNA, flanked by exon 7 and exon 8 sequences

Full size image

The missense p.(Leu190Pro) lies in the rod domain of desmin, and is predicted to be damaging (Polyphen2: 0.997, probably damaging, Mutationtaster: disease causing, FATHMM: damaging, CADD-PHRED: 24.9, SIFT: 0, deleterious, and Align GVGD: Class C65). Sanger-sequencing confirmed segregation in the father and two unaffected brothers (Fig. 1ai). Thus, the missense variant p.(Leu190Pro) was viewed as unlikely to be causative of an autosomal dominant desminopathy. However, suspicion of a DES-related cardio/myopathy was alerted. Despite absence of biopsy findings consistent with myofibrillar myopathy, closer scrutiny was applied to all variants affecting the DES locus. A maternal deep intronic variant c.1289-741 G > A was identified in the two affected siblings; absent in the two unaffected brothers. c.1289-741 G > A was predicted to create a cryptic acceptor site by all five splicing prediction programs in Alamut Visual (Interactive Biosoftware, Rouen, France) (Fig. 2b).

mRNA studies confirm splice-altering effects of the deep intronic variant

RNA sequencing using muscle biopsy specimens from the proband II:3 showed significant allele bias among DES transcripts for 3 heterozygous single-nucleotide variants spanning the DES gene (Fig. 2a). With the c.569 T > C, p.(Leu190Pro) variant specific for paternal transcript, collective data infer paternal transcripts account for 92% of the DES mRNA pool, with 8% maternal transcripts (Fig. 2a). RNA sequencing also showed ~ 100 reads immediately following the c.1289-741 G > A variant creating a cryptic splice acceptor, which may correspond to inclusion of a 118 nucleotide (nt) artificial exon/intronic fragment into the DES mRNA, inducing a frameshift, p.Thr431Serfs*43 (data not shown). DES allele bias of this nature is consistent with active nonsense-mediated decay (NMD) of the maternal allele due to the frameshift induced by inclusion of the 118 nt artificial exon/intronic fragment. We were unable to confirm precisely inclusion of the pseudo-exon by RNA sequencing, due to the combination of effective nonsense-mediated decay silencing this allele, and, technical limitations of short-read RNA sequencing that do not effectively bridge multiple exons.

Therefore, RT-PCR was performed on cDNA derived from a skeletal muscle biopsy from II:3, using forward and reverse primers within the predicted intron 7 118 nt fragment, paired with primers in exon 5 or exon 9. PCR amplicons corresponding to an aberrantly spliced product incorporating the artificial exon/intronic fragment into DES transcripts were observed for the proband; absent from two control muscle cDNA specimens. Sanger sequencing of amplified products confirmed the presence of a 118 nt artificial exon/intronic fragment in DES mRNA transcripts from II:3, r.1288_1289ins1289-739_1289-622 (Fig. 2b, c).

Western blot reveals 60–75% reduction in desmin levels, likely comprised only of missense p.Leu190Pro desmin

Western blot was performed on skeletal muscle lysate from the proband and an aged matched control. A standard curve demonstrates that desmin protein levels in the proband are reduced, approximately one-quarter to one-third the levels of the control, while myotilin levels are unchanged. α-actinin and actin show appropriate loading for the standard curve (Fig. 1c).

Discussion

In summary, we show compound heterozygosity for a missense p.(Leu190Pro) DES variant and deep intron-7 cryptic splice variant are the likely cause of recessive DES myopathy with severe cardiomyopathy in two siblings. RNA-seq and RT-PCR studies of muscle-derived mRNA confirms the c.1289-741 G > A variant induces abnormal inclusion of a 118 nt artificial exon/intronic fragment, r.1288_1289ins1289-739_1289-622; effecting a frameshift, abnormal stop codon, and nonsense-mediated decay of maternal transcripts. Therefore, the pool of DES mRNA in skeletal muscle is comprised largely of paternal transcripts encoding the p.(Leu190Pro) missense variant; predicted by in silico tools to be deleterious or damaging. The p.(Leu190Pro) substitution occurs in the coil 1B domain of desmin involved in homodimerisation. Many pathogenic missense mutations in DES introduce prolines into coil domains [3]. Proline residues induce a kink in polypeptide chains, which, in the setting of an α-helix are disruptive; distorting secondary structure, with proline residues unable to form hydrogen bonding required for α-helix stability. Likely pathogenicity of identified recessive DES variants is supported by western analyses, showing desmin protein levels are reduced to around one-quarter to one-third levels in control muscle. As desmin levels are less than one-half that observed in controls, data also suggests inherent instability and/or elevated turnover of the missense p.(Leu190Pro) desmin. Desmin deficiency in this case of autosomal recessive desminopathy is a feature distinct from autosomal dominant desminopathy, where levels of desmin can often appear elevated due to desmin aggregates [17].

Our results extend the phenotypic spectrum of recessive desminopathy. In contrast to severe infantile presentation associated with homozygous loss-of-function DES variants [18], our two affected siblings presented in their late teens to early twenties with a limb girdle pattern of progressive weakness and cardiomyopathy, expressing diminished levels of variant p.(Leu190Pro) desmin. Muscle biopsy showed extensive myofibrillar disarray, but did not show aggregates typical of DES myofibrillar myopathy (aggregates were also not present in AR desminopathy described in [5, 11]). Thus, recessive desminopathy may take different forms depending whether characterised by desmin deficiency, or (reduced) expression levels of different missense forms of desmin protein, that each may elicit different impacts on intermediate filament structure and function. Our results show that many tricky diagnoses in neuromuscular conditions depend upon analyses of available muscle biopsy specimens, combined with interdisciplinary clinical, pathological and functional genomics expertise. Importantly, deep intronic variants inducing inclusion of a frameshifting artificial exon (intronic sequences) make very attractive candidates for morpholino-based genetic exon-skipping therapies, currently advancing through therapeutic pipelines.