‘Dusty core disease’ (DuCD): expanding morphological spectrum of RYR1 recessive myopathies

All the muscle biopsies were analysed at the Neuromuscular Morphology Unit of Myology Institute, in Paris. More than 11000 muscle biopsies collected between 1977 and 2015 were screened. Two hundred and thirty belonged to patients with RYR1-related myopathy, of whom 154 had dominant inheritance and 76 were confirmed or suspected to be recessive. In 13 cases the second variant was not found or was not surely pathogenetic (likely pathogenetic, VUS or likely benign). Among the 63 confirmed RYR1-recessive patients, 15 cases have been excluded because of insufficient data (uncomplete clinical and/or morphological data and deteriorated muscle sample for re-analysis). Finally, our study cohort consisted on 48 confirmed RYR1-recessive patients (20 male and 28 female) from 45 unrelated families. Clinical data were obtained from a full revision of all available medical records up to the last clinical examination in all enrolled patients. Disease onset (considered as the first reported clinical sign referred to disease, including perinatal problems or delayed motor milestones), weakness distribution (proximal/distal limb muscles, axial, facial, ocular/extraocular and bulbar muscles), contractures, spinal deformities and dysmorphisms, respiratory and cardiac involvement were considered. Clinical evaluations were performed by different clinicians over 40 years, thus we retrospectively classified patients in three main groups of clinical severity: mild (late-adult onset, walkers, mild muscle weakness, minimal or absent ocular, facial, bulbar or respiratory involvement), moderate (early onset, non-progressive, proximal or diffuse muscle weakness, associate with mild dysmorphism, spine deformities or contractures) and severe (muscle hypotonia at birth, feeding difficulties, severe respiratory involvement requiring ventilation, contractures and/or spinal deformities, diffuse muscle weakness with facial and ocular involvement).

Total RNA was extracted from each skeletal muscle sample lysed in Trizol reagent (Invitrogen, Life Technologies SAS). Complementary DNA was synthesized from 500 to 750 ng of total RNA using 0.5 μl of Transcriptor (Roche) and 0.3 lg of oligo-dT as described [26]. Seven overlapping PCR amplification spanning the entire RYR1 sequence were performed. Each fragment was sequenced as previously described [26, 28]. Each variation was confirmed on DNA sample and on both paternal and maternal DNA sample to establish the transmission. Each variant was analysed by Variant effect Predictors to obtain the different prediction score such as CADD, SIFT, Polyphen and gnomAD exome and gnomAD Genome database frequency. To better assess the functional effect of each missense variation, 3D analysis was performed on Yasara sofware [21]. Due to the large size of the RYR1 gene, we choose not to use the total RyR1 protein structure already described (5gl1/5taz) [3, 8] in the first-round analysis via FoldX prediction. We choose to split the structure in 5 parts spanning the whole human RYR1 structures (amino acid 1 to 627, 628 to 1656, 1657 to 2144, 2145 to 3613 and 3614 to 5038). Then, the sequences were submitted in I-TASSER server [39] to obtain “friendly” usable RyR1 structure. Each structure prediction was matched with the RyR1 global structure (5gl1 and 5taz) [3, 8]. Delta G variations were calculated to estimate protein stability. For delta G variation <0.5 kcal/mol, meaning no destabilization, study of the whole structure was realized (5gl1/5taz) [3, 8]. For ACMG classification, Intervar was used with recessive transmission correction [22].

Histoenzymological analysis was conducted on 54 muscle biopsies (4 patients had two muscle biopsies and 1 patient three muscle biopsies available in the Myology Institute Lab). Age at muscle biopsy ranged from 1 day of life (30 weeks of adjusted gestational age) to 76 years (median 16 years, IQR 3-34). Open muscle biopsies were obtained from deltoid or quadriceps muscles in most of patients. Histological and histochemical slides were systematically re-analysed by two authors (MG and NBR) with experience in skeletal muscle morphology, blinded to clinical and molecular data. For the oldest, deteriorated or not interpretable slides, new slides were obtained from the best muscle specimen available in the lab. Conventional histological and histochemical techniques, 8-10 μm thick cryostat sections were stained with haematoxylin and eosin (HE), modified Gomori trichrome (GT), Periodic acid Schiff technique (PAS), Oil red O, reduced nicotinamide adenine dinucleotide dehydrogenase-tetrazolium reductase (NADH-TR), succinic dehydrogenase (SDH), cytochrome c oxidase (COX), and adenosine triphosphatase (ATPase) preincubated at pH 9.4, 4.63, 4.35. Digital photographs of biopsies were obtained with a Zeiss AxioCam HRc linked to a Zeiss Axioplan Bright Field Microscope and processed with the Axio Vision 4.4 software (Zeiss, Germany). Fibre type pattern was determined in ATPases reactions, and by calculating the percentage of type 1 and type 2 fibres. We considered type 1 fibres predominance when there were more than of 60% type 1 fibre in deltoid muscles, and more than 40% in quadriceps muscle. CFTD was considered when all the type1 fibres were consistently (at least 35-40%) smaller than type2 fibres in absence of other pathological findings. Centronuclear pattern was considered only when myonuclei were centrally placed in almost 50% of muscle fibres showing nuclear internalization. Rods were considered in the presence of numerous, multiple, small nemaline bodies both in cytoplasmic or subsarcolemmal areas. Central cores were considered when single or multiple sharply demarked ovoidal areas devoid of oxidative stains were observed in transversal sections of type1 muscle fibres, centrally or peripherally placed. Multiminicore were considered in the presence of boundless, small areas of decreased enzymatic activity at oxidative stains [30]. “Dusty cores” were defined as irregular areas of reddish-purple granular material deposition at GT stain corresponding to decreased or/and increased enzymatic activity at oxidative stains and devoid of ATPase activity. Patients with available ultrastructural study were finally classified considering both histological and ultrastructural features. In the five patients with two or three muscle biopsies, final morphological classification was reached considering both muscle biopsies and most relevant findings.

IHC was performed in new sections from available frozen muscle samples of enrolled patients, rejecting oldest and/or deteriorated specimens. Finally, IHC analysis was available for 23 muscle biopsies. Antibodies against Desmin (Anti-Human Desmin, Clone D33, Dako Laboratories, Denmark A/S ), Myotilin (NCL-Myotilin, Novocastra Laboratories, Newcastle Upon Tyne, United Kingdom) and αB-crystallin (CRYAB, GeneTex International Corporation, Irvine, USA) were visualized using immunoperoxidase techniques. Immunofluorescence study was performed for RyR (anti-Ryanodine Receptor, clone 34C, Sigma Laboratories, Saint Louis, Missouri, USA), DHPR (anti-CACNA1S, ab2862, abcam Laboratories, Cambridge, UK) and alpha-actinin (anti-α actinin sarcomeric, clone EA-53, Sigma Aldrich Laboratories, Saint Louis, USA) antibodies on 10-μm-thick cryosections over night at 4°C. Subsequently, sections were incubated with appropriate conjugated secondary antibodies (Alexa Fluor-488 goat anti-rabbit antibody and Jackson IR goat anti-mouse antibody) for one hour. A set of control slides was prepared with omission of the primary antibodies.

Ultrastructural study was newly performed in all available muscle biopsies. EM images were obtained for 39 muscle biopsies. Small muscle specimens were fixed with glutaraldehyde (2.5%, pH 7.4), post fixed with osmium tetroxide (2%), dehydrated and embedded in resin. Longitudinally oriented ultra-thin sections were obtained at different level of deepness from 1 to 3 small blocks and stained with uranyl acetate and lead citrate. Ultra-thin sections of transversally oriented blocks were obtained only for the most significant findings. The grids were observed using a Philips CM120 electron microscope (80 kV; Philips Electronics NV, Eindhoven, The Netherlands) and were photo documented using a Morada camera (Soft Imaging System).

The amount of RyR1 in muscle samples was determined by quantitative Western Blot (WB) analysis using antibodies directed against RyR1 [24] and normalized to the amount of myosin heavy chain as described previously [26]. Briefly, the muscle sample (20-40mg) was homogenized in 200 mM sucrose, 20mM HEPES (pH 7.4), 0.4mMCaCl2, 200mM phenylmethylsulfonyl fluoride, 1 mM diisopropyl Fluorophosphate using a Minilys homogenizer (Bertin, France). After electrophoretic separation on a 4–20% gradient acrylamide gel (Biorad, France) and electrotransfer to Immobilon P (Biorad, France) during 4h at 0.8A to ensure a complete transfer of the loaded proteins, the membrane was incubated with anti-RyR1 antibodies and then HRP-labelled secondary antibodies (Jackson ImmunoResearch Laboratories). Signal quantification was performed using a ChemiDoc Touch apparatus (Biorad, France) and the Image Lab software (Biorad). The total amount of RyR1 in each experiment was corrected from the amount of myosin and normalized to the amount of RyR1 present in the control referred as 100% as described previously [6]. Controls (muscle biopsy from individuals non-affected by neuromuscular disease) of different age have been used: 10 days, 3 years, 23 years and 46 years, and the amount of RyR1 in the muscle biopsy of the patient was compared to the age-related control. For each muscle biopsy, at least 3 western blots have been performed, and the value for each patient is presented as mean ± SEM of the different western blots.

Data have been expressed as range, median and interquartile range (IQR) for continuous variables (age and % of RyR expression) and as absolute values and frequencies for categorical variables (gender, disease severity, age at onset, morphology and occurrence of ocular involvement). Age at disease onset has been categorized as <1 year or ≥1 year, based on the median value in the sample. Chi-Square test has been used to compare categorical variables, while Kruskal-Wallis or Mann-Whitney tests have been used to evaluate the distribution of continuous variables with respect to demographic, clinical and morphological characteristics. P values lower than 0.05 were considered to be statistically significant. Statistical Package for Social Science (SPSS®) version 20.0 (IBM Corp. Released 2011. IBM SPSS Statistics for Windows, Version 20.0. Armonk, NY: IBM Corp.) was used for statistical analysis.

Age at onset ranged from antenatal period to 66 years (median 1 year, IQR 0-7). The first symptom/sign occurred within the first year of life in 23 patients (48%), of whom five had also antenatal manifestation. Disease onset occurred after the first year of life in 25 patients (52%) of whom 4 had a relative late onset (>20 years). Motor milestones were delayed in 25 patients (52%) and 4 out of 8 severe patients never acquired ambulation. All the patients with latest onset (>45 years) had not delayed motor milestones, as well as 9 (19%) with early-infantile onset. Ocular involvement was assessed in 38 of 48 patients and showed isolated eyelid ptosis in 4 patients, isolated ophthalmoplegia (sometimes isolated upper gaze movements) in 7 and both in 9. Overall, ocular involvement occurred in 20/38 patients and prevailed in subjects with moderate/severe disease (88%) compared to those with mild disease (24%) (p<0.0001). Overall clinical severity was mild in 25 (52%), moderate in 15 (31%) and severe in 8 patients (17%). All the clinical data are summarised in Table 2.

Every patient has at least one variation on each allele of RYR1 gene (Additional file 1). Each variant was identified based on its low frequency (less than 0.5% in database gnomAD). Thirteen variants were not previously reported. Twenty-six hypomorphic variants were found (non-sense, out of frame deletion, splice variant). The others were in frame deletion (p.M4000del), in frame duplication (p.E1651_L1656dup), missense variants and a combination of 3 or 4 variants resulting in the mutant haplotype p.[I1571V;R3366H;Y3933C] or p.[I1571V; R3119H; R3366H; Y3933C] already described [10]. Each variant was found in a conserved region (in the vertebrate subphylum). Only L4650P, R2140W, L883P, P4501L and R4179H variations were classified as moderate, a similar type of amino acid change being observed in other species of vertebrate phylum. All the bioinformatic scores lead to consider all these variations as impacting the protein. Each variant was localized on the RyR1 3D structures and 26 variants were predicted to induce destabilisation by foldX software. Other 10 variants, localized on the RyR1 3D structures (in the so-called domains 5gl1 and 5taz), were predicted to induce the loss of nearby interaction. Five variants (N2342S, R4179H, V4842M, E4911K) were predicted to induce no particular change in RyR1 structure at the interaction or steric hindrance level [3, 8]. The variant V4842M was associated with an additional splice variant. The N2342S has already been associated with a functional defect, the E4911K is located in the triadin interaction motif, the A4946V is located in a transmembrane domain and the R4179H is located in a calcium affinity domain [38]. Overall, 17 different missenses variants (40.5%) are located in the bridge solenoid (B-sol) domain (which contains the FKBP binding domain), 2 in the S2S3 domain, 4 in the Csol (5%), 3 in the J sol (7.5%), 5 in the NTD domain (12.5%), 6 in the pore domain (15%), 3 in the pVSD domain (7.5%), 1 in the RYR1&2 domain (2.5%), 2 in the SPRY1 (5%), 2 in the SPRY2 (5%), 1 in the SPRY3 (2.5%), 3 in the TaF domain (7.5%) and 1 in undefined regions (2.5%). For the haplotype variant [I1571V; R3366H; Y3933C], the variants were respectively localized in SPRY3, BSol and CSol domains. Among all mutations only 2 were previously associated to MHS (T2206M, R2458C). To note 5 were located in MH1 domain (p.E39K, p.S71Y, p.Q155P, p.G215E, p.R614C), 6 in the MH2 domain (p.T2206M, p.N2283H, p.R2355W, p.M2423K, p.R2435L, p.R2458C) and 17 in the MH3 domain (p.Y3933C, p.M4000del, p.R4179H, p.E4181K, p.A4287_A4289dup, p.P4501L, p.R4564Q, p.L4650P, p.T4709M, p.K4724Q, p.Y4796H, p.L4828R, p.V4842M, p.A4846V, p.M4875T, p.A4906V, p.E4911K). Among all variants only 2 were previously associated to MHS (T2206M, R2458C). According to the ACMG classification, 19 variants were considered as class 5 (pathogenic), 31 variants were classified as class 4 and 19 variants of unknow signification (class3).

RyR1 expression was obtained from 31/54 muscle biopsies: 17 from DuCD group (68%), 7 CCD (70%), 5 C&R (62%) and 2 T1P+ (40%). All the patients had reduced expression of RyR1 in muscle tissue (median 40%, range 7-79%, IQR 25-53). Patients with a very large reduction in the amount of RyR1 (<15% of total) were severely affected. An amount as low as 20% resulted only in a moderate phenotype in at least 5 patients. The patients with latest disease onset (>50 years) had more than 50% protein. Patients with MH mutations have a lower reduction in RyR1 amount, estimated at 7 to 9% for the R2458C variant and 11,5% for the T2206M variant. The reduction related to haplotypes ranged between 17-24%. The E4911K variant leads to a RyR1 loss estimation ranging between 22 to 37.5%.

Percentage of RyR1 expression was significantly lower in patients with disease onset <1 year (range 7-79; median 25.00; IQR 19-38) compared to those with later onset (≥1 year) (range 18-76; median 51.50; IQR 38.25-57.75) (p=0.011) (Fig. 4a). An interesting correlation (Kruskal-Wallis test) was also found between clinical severity and RYR1 expression, with lower RyR1 expression in more severe patients (p=0.016) (Fig. 4b). Concerning morphology, the lowest RYR1 expression was observed in DuCD group (range: 7-63%; median: 32%, IQR 24.0-40.5), compared to CCD (range: 14-76%; median: 47%, IQR 40-52), C&R (range: 20-60%; median: 52%, IQR 36.0-58.5) and T1P+ (range: 54-79%) (p=0.059). Comparing DuCD group with non-DuCD groups, RYR1 expression was significantly lower in the former (p=0.015) (Fig. 4c,d).