ARTICLE https://doi.org/10.1038/s41467-019-09111-2 OPEN Myoglobinopathy is an adult-onset autosomal dominant myopathy with characteristic sarcoplasmic inclusions Montse Olivé et al.# 1234567890():,; Myoglobin, encoded by MB, is a small cytoplasmic globular hemoprotein highly expressed in cardiac myocytes and oxidative skeletal myofibers. Myoglobin binds O2, facilitates its intracellular transport and serves as a controller of nitric oxide and reactive oxygen species. Here, we identify a recurrent c.292C>T (p.His98Tyr) substitution in MB in fourteen members of six European families suffering from an autosomal dominant progressive myopathy with highly characteristic sarcoplasmic inclusions in skeletal and cardiac muscle. Myoglobinopathy manifests in adulthood with proximal and axial weakness that progresses to involve distal muscles and causes respiratory and cardiac failure. Biochemical characterization reveals that the mutant myoglobin has altered O2 binding, exhibits a faster heme dissociation rate and has a lower reduction potential compared to wild-type myoglobin. Preliminary studies show that mutant myoglobin may result in elevated superoxide levels at the cellular level. These data define a recognizable muscle disease associated with MB mutation. Correspondence and requests for materials should be addressed to M.O. (email: 25169mop@comb.cat) or (email: montseolivep@gmail.com) or to N.G.L. (email: nigel.laing@uwa.edu.au). #A full list of authors and their affiliations appears at the end of the paper. NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09111-2 M yoglobin, the pigment that gives muscle its red color, Results is a small cytoplasmic globular hemoprotein highly The clinical phenotype of myoglobinopathy. We studied 14 expressed in cardiac and oxidative skeletal patients from six unrelated European families (F1–F6), (Fig. 1), muscle fibers1. suffering from an autosomal dominant myopathy with variable By reversibly binding O2, myoglobin buffers intracellular O2 cardiac involvement, identified by their highly characteristic concentrations, facilitates intracellular O2 transport and serves as features on muscle biopsies. The clinical presentation was a reservoir of oxygen during hypoxic and anoxic conditions1–3. In homogeneous among all patients and is summarized in Table 1. addition, myoglobin is implicated in vivo in the control of redox Age of onset ranged between 33 and 49 years. Initial symptoms pathways in skeletal and cardiac myocytes, acting as scavenger of were proximal lower limb, pelvic girdle, and axial muscle weak- reactive oxygen species (ROS) and nitric oxide4,5. ness, manifesting with difficulties climbing stairs, rising from Myoglobin was the first protein for which a three-dimensional squatting and standing up from the lying position. Two patients structure was determined by X-ray crystallography6. The back- had, in addition, weakness and atrophy of thenar muscles from bone of myoglobin consists of eight α-helices, assigned the disease onset. Over the following years, the disease slowly pro- letters A to H, that wrap around a central pocket containing gressed to involve distal leg and hand muscles, proximal muscles a heme group, a porphyrin ring that contains a central bound of upper limbs, and neck muscles. Five patients complained of iron atom that is normally in the ferrous oxidation state. dysphagia. Facial and extraocular muscles were not involved. Its heme active site is responsible for reversible binding to Progression was slow; most patients developed respiratory failure various ligands including oxygen, carbon monoxide and nitric requiring nocturnal non-invasive ventilatory support 10 years oxide7. after disease onset and they became wheelchair dependent 15–20 Myoglobin is encoded by the myoglobin gene (MB) on human years after disease onset. Cardiac involvement, revealed by chromosome 22q12.37. ultrasound, magnetic resonance imaging (MRI) or post mortem Here we describe myoglobinopathy, a disease caused by a myocardium examination, was observed in six individuals. Car- recurrent MB mutation found in 14 patients from six unrelated diac MRI in patient F2, II:2 showed a dilated cardiomyopathy European families suffering from an autosomal dominant myo- with extensive and diffuse areas of late gadolinium enhancement, pathy with variable cardiac involvement and characteristic sar- indicative of fibrosis at the epicardium and mesocardium, but no coplasmic inclusions in skeletal and cardiac muscle. We show that involvement of the endocardium, (Supplementary Fig. 1). Six mutation of MB alters the kinetics and thermodynamics of O2 patients died between 18 and 30 years after disease onset, from binding and may result in elevated superoxide levels. respiratory and/or cardiac failure. F1 F2 I I 1 2 1 2 II II 1 2 3 4 5 6 7 1 2 3 III III 1 2 3 4 5 6 7 1 2 3 4 F3 I 1 2 3 II 1 2 3 4 5 6 7 8 9 10 11 12 III 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 IV 1 2 3 4 F4 F5 F6 I I I 1 2 1 2 1 2 II II II 1 2 3 4 5 6 7 8 1 2 3 4 5 1 2 3 4 III III 1 2 3 4 5 6 7 1 2 3 Fig. 1 Pedigrees from the six families affected with myoglobinopathy. F1 and F2 Spanish, F3: Swedish, F4 and F5: French, F6: Dutch. Squares represent males; circles, females; filled black symbols indicate affected individuals; filled gray symbol, clinically affected, but not molecularly tested individuals; and slash, deceased 2 NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09111-2 ARTICLE Table 1 Phenotypic features of myoglobinopathy Family 1 Family 2 Family 3 Family 4 Family 5 Family 6 Inheritance pattern AD AD AD AD unknown unknown Myoglobin mutation His98Tyr His98Tyr His98Tyr His98Tyr His98Tyr His98Tyr Country of origin Spain Spain Sweden France France Netherlands No. of patients 2 1 6 3 1 1 Mean age of onset 37.5 (36–39) 38 44.5 (39–49) 46 (44–48) 40 33 (range) Gender (female/male) 0/2 1/0 2/4 2/1 0/1 0/1 Initial symptoms Proximal LL and Proximal LL and Distal hand weakness Proximal LL and axial Proximal LL Proximal LL and axial weakness axial weakness (2/6) weakness (3/3) weakness (1/1) axial weakness (2/2) (1/1) Proximal LL and axial (1/1) weakness (6/6) Symptoms at advanced disease Distribution of Proximal and Proximal and Proximal and axial > Proximal and distal 4 Proximal and Proximal and weakness axial > distal 4 axial > distal 4 distal 4 EE EE > axial axial > distal 4 axial > distal legs EE EE EE Involvement of hand 2/2 1/1 6/6 2/2** 1/1 0/1 muscles Facial weakness 0/2 0/1 0/6 0/2 0/1 0/1 Muscle atrophy 2/2 1/1 6/6 2/2 0/1 1/1 Dysphagia 2/2 0/1 2/4 0/2 1/1 0/1 Respiratory 2/2 1/1 2/6 1/2 1/1 1/1 insufficiency Cardiac involvement* 2/2 1/1 2/6 0/2 1/1 0/1 Clinical outcome Mean age at wheelchair 54 65 56 (4/6 patients 66 (1/3 patients 56 47 dependency wheelchair wheelchair dependent, 2/6 dependent, 2/3 ambulant) ambulant) Mean age at death 60.5 (54–67) – 64 (58–71) 72 – – (range) Laboratory studies CK fold elevation above 1.5–2 2 1.5–4 2.3–3.7 4 3.5–7 normal levels EMG Myogenic with Myogenic with Myogenic with Myogenic with Myogenic with Myogenic with spontaneous spontaneous spontaneous activity spontaneous activity spontaneous spontaneous activity at rest activity at rest at rest at rest activity at rest activity at rest *As revealed by cardiac ultrasound, cardiac MRI, or by post mortem examination of cardiac muscle. ** Individual III:2 from Fam 4 is in the early stages of the disease and therefore not included in symptoms at advanced disease items. EE: extremities; LL: lower limbs Serum CK levels were normal to mildly increased, except for sarcoplasmic bodies, the majority of muscle biopsies showed non- a single individual who showed CK levels up to sevenfold above specific myopathic changes with increased myofiber size variation, the upper normal limit. EMG was consistent with a myopathy increased numbers of internal nuclei and type 1 fiber predomi- with spontaneous activity at rest. Nerve conduction studies nance (Supplementary Table 1). Most muscle biopsies, particularly were normal. Muscle imaging studies showed a consistent pattern those showing advanced pathological lesions, showed cytoplasmic characterized by fatty degenerative changes in paraspinal and vacuoles filled with granular basophilic material, displaying strong gluteal muscles, especially gluteus maximus and medius. At mid- acid phosphatase activity and lysosome-associated membrane thigh, there was preferential involvement of the adductor magnus, protein 1 (LAMP1) immunoreactivity (Fig. 3), indicating that they semimembranosus, long head of biceps femoris and vastus contained lysosomes. No major architectural changes were intermedius, except in individual F6, II:4 who showed predomi- observed on NADH, SDH, and COX reactions apart from the nant involvement of the anterior compartment. At the mid-leg, lack of oxidative activity at the site of vacuoles. Myoglobin, p62, the soleus was always the first and most affected muscle (Fig. 2). and ubiquitin immunoreactivity was observed in abnormal myofiber regions containing vacuoles and in some but not all sarcoplasmic bodies (Fig. 4), demonstrating abnormal protein Muscle pathology features associated with MB mutation. aggregation (Fig. 4). However, myoglobin immunostaining was Muscle biopsies showed characteristic sarcoplasmic bodies in not particularly useful for the diagnosis. Under electron micro- both Type 1 and Type 2 myofibers in all studied patients, scopy, the inclusions appeared as very dense bodies measuring including pre-symptomatic individuals, irrespective of the site of 0.3–2.5 µm, often located next to myonuclei or dispersed between the biopsy (Fig. 3 and Supplementary Table 1). The sarcoplasmic myofibrils (Fig. 5a–d and Fig. 5e–g). Some were membrane bound, bodies were rounded or oval, appeared brown on hematoxylin whereas others were not. Some were denser than others, likely and eosin and red on the modified Gomori’s trichrome stain. reflecting different stages of the pathological process. Autophagic They had a glassy appearance and were easily visualized with all vacuoles with cellular debris, myelin figures, and filaments stains and even in non-stained sections. measuring 8–12 nm were additional findings. Sarcoplasmic bodies They exhibited autofluorescence emission in a wide range of were present in respiratory and cardiac muscles obtained post visible laser excitation lines (Supplementary Fig. 2). Besides the mortem from individual F1, II:7 and in cardiac muscle from NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09111-2 a d g b e h c f i Fig. 2 Muscle imaging in myoglobinopathy. Muscle CT scan from individuals F1, II: 7 (a–c); F4, II: 4 (d–f); and F5, II: 6 (g–i) at the pelvic (a, d and g), mid- thigh (b, e, and h), and mid-leg (c, f, and i). At the pelvis, there is involvement of the gluteus maximus, medius, and minimus. At mid-thigh, there is preferential involvement of the posterior compartment, specially, of the adductor magnus, biceps femoris, and semimembranosus. At the mid-leg, the soleus is the first and most-affected muscle Fig. 3 Histochemical features of myoglobinopathy. a Anterior tibialis muscle biopsy from individual F3, III: 15, 10 years prior to the onset of symptoms, stained with hematoxylin and eosin, showing several rounded brown inclusions (arrows) (sarcoplasmic bodies) in the majority of myofibers and very small vacuoles in some myofibers (arrowhead in a). b, c Biceps brachii from individual F1, II: 7, 15 years after disease onset. b Note the presence of collections of sarcoplasmic bodies within the rimmed vacuoles. d, e Sarcoplasmic bodies appear red on modified Gomori trichrome stain. e In muscle biopsies with more advanced pathological lesions, large numbers of rimmed vacuoles are observed. f No major architectural changes are seen on NADH reaction, apart from lack of oxidative activity at the site of vacuoles. g Fast myosin immunohistochemistry demonstrate the presence of sarcoplasmic bodies in both type 1 (slow) and 2 (fast) myofibers. h Myofiber regions containing vacuoles display strong phosphatase activity, and LAMP1 i immunoreactivity. Scale bar in a, e, f, g, and h = 50 µm; scale bar in b, c, d, and i = 20 µm individual F3, III:5 (Fig. 5d and Supplementary Fig. 3), thus independent groups used two different strategies. The Australian indicating that cardiomyopathy was a primary consequence of the group used whole-exome sequencing of three affected individuals disease and not secondary to respiratory failure. from two Spanish families (F1, II:5 and II:7 and F2, II:2), and gene prioritization of shared variants using skeletal muscle Identification of p.His98Tyr MB substitution in six families. expression enrichment data from FANTOM58. This identified the To identify the molecular cause of this myopathy two same heterozygous missense variant (c.292C>T, p.His98Tyr) in 4 NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09111-2 ARTICLE Fig. 4 Immunohistochemical features of myoglobinopathy. a, b Small myoglobin aggregates are observed in some myofiber regions and in some sarcoplasmic bodies (double arrow in b), but not in others (arrow in b). c p62 and ubiquitin d immunoreactivity is observed in some myofiber regions, and in some, but not in all sarcoplasmic bodies indicating protein aggregates. Scale bars = 20 µm the myoglobin gene (MB) in all three affected individuals. The multiple compensatory mechanisms10. We therefore hypothe- Swedish group, studying family F3, identified one region on sized a dominant gain of function as the cause of the disease chromosome 22 through genome-wide linkage analysis, with a in the families and used multiple methods to investigate the statistically significant LOD score. Refinement of the peak with pathomechanism. further markers and clarification of the disease state in family members, resulted in a maximum multipoint LOD score of 5.8 at NanoSIMS analysis of the sarcoplasmic bodies. Correlative D22S685 (Supplementary Fig. 4). Targeted capture and sequen- electron microscopy and NanoSIMS (Nanoscale Secondary Ion cing of genes within the linkage region in eight individuals, six Mass Spectrometry) analysis revealed high sulfur content and affected and two unaffected, revealed in all the affected family small iron Fe signals in the sarcoplasmic bodies (Fig. 5h–j). Of members, the presence of the same MB (c.292C>T, p.His98Tyr) note, sulfur is a component of antioxidant systems of cells and variant. Sanger sequencing confirmed segregation of the variant reactive sulfur species are formed under conditions of oxidative with the disease in all available family members and in three stress11,12. Iron signals could be related to degradation of myo- additional families (F4, F5, and F6). The variant involves a highly globin and other metalloproteins inside the lysosomes13. Com- conserved residue in the proximity of the oxygen-binding parative NanoSIMS analysis of myoglobinopathy samples with heme group (Supplementary Fig. 5), it is absent in all unaf- one sample from a patient with Pompe disease containing fected relatives tested, is not present in the 1000genomes, ExAC electron-dense inclusions, two samples showing abundant lipo- (http://exac.broadinstitute.org/) or gnomAD ((http://gnomad. fuscin and two muscle samples from patients affected with distal broadinstitute.org) data sets and was present on different hap- myopathies with rimmed vacuoles, revealed that the amount of Fe lotypes. Moreover, all in silico predictors suggested the observed is higher in the myoglobinopathy sarcoplasmic bodies than in substitution in MB is deleterious: MutationAssessor (medium other similar inclusions. functional impact, FI score 3.175), MutationTaster (disease- causing, p=0.999), PolyPhen-2 (probably damaging, score 1.000), Provean (deleterious, score −4.093) and SIFT (damaging, score µFTIR analysis of muscle biopsy samples. Fourier transform 0.001). Thus, we have compelling evidence across six families that infrared microscopy (µFTIR) analysis of muscle samples the same MB (c.292C>T, p.His98Tyr) variant is responsible for demonstrated increased carbonyl (ester) groups in the sarco- the observed myopathy. plasmic bodies indicating lipid oxidation (Fig. 5k–m). Moreover, Haplotype analysis using microsatellites within 3 Mbp of the some sarcoplasmic bodies showed a band at 1627 cm−1, which is MB gene indicated at least three different haplotypes within the characteristic of intermolecular β-sheet structures14 (Fig. 5n) and six families, suggesting that the c.292C>T variant is a recurrent a feature typical of amyloid formation15,16. However, in vitro, variant originating on different ancestral backgrounds (Supple- only small amounts of non-fibrillar amyloid β-structure were mentary Table 2). Furthermore, the analysis of haplotypes in detected (see Supplementary Note 1 and Supplementary Fig. 6). family F5 showed that the affected and two unaffected siblings Non-fibrillar β-sheet aggregates have been previously described had identical alleles, suggesting that in the affected individual the for other peptides and proteins including β-amyloid peptides variant probably arose de novo (Supplementary Table 2). No related to Alzheimer’s disease16. Both NanoSIMS and µFTIR, samples from the parents in family F5 were available for therefore, suggest lipid oxidation in patients’ muscles contribute confirmation. to sarcoplasmic body formation. Myoglobin knockout mice have a binary phenotype. Two- thirds die in utero at ~ E9.5-E10.59, but those that survive live Biochemical characterization. At the molecular level, the to adulthood with little sign of functional effects, because of His98Tyr variant might impact on different MB functional NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09111-2 k n 0.005 d2A/d(υ)2 -sheet C=O α-helix + unordered 1680 1660 1640 1620 1600 CH3 CH2 Amide Wavenumber (cm–1) 3000 2900 2800 1700 1600 1500 Wavenumber (cm–1) l 0.6 m * 0.5 * 0.5 A2925 cm–1/A1654 cm–1 A1740 cm–1/A2925 cm–1 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0.0 0.0 Sarcoplasmic Surrounding muscle Sarcoplasmic Surrounding bodies tissue bodies muscle tissue properties, crucial to the physiological role: these include the which is eventually reduced back to ferrous MB by cytochrome interaction of MB with dioxygen, the affinity of MB protein b518,19. scaffold to hemin, as well as the ability to keep the heme in the First, we evaluated whether the variant affects the affinity of the reduced state (Fe2+). In vivo, ferrous MB (Fe2+), which is the prosthetic group for the MB polypeptide matrix, as it has been only myoglobin form that can bind and store O2, is prone to demonstrated that mutations in the vicinity of the porphyrin ring autoxidation17, leading to functionally inactive MetMB (Fe3+), can increase heme dissociation rates (k-H) by two-orders of 6 NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09111-2 ARTICLE Fig. 5 Characterization of sarcoplasmic bodies, the morphological hallmark of myoglobinopathy. Electron micrographs a–d showing collections of highly electron-dense bodies with some less dense material at their periphery. The sarcoplasmic bodies are seen under the sarcolemma (a) and often next to the nuclei. Some sarcoplasmic bodies are surrounded by a membrane (b). Sarcoplasmic bodies of different electron densities near several vesicular structures (c). Sarcoplasmic bodies observed in the cardiac muscle obtained post mortem from individuals F1, II:7 (d) and F3, III:5. Electron micrographs (e–g) and the corresponding NanoSIMS images (h–j). Blue indicates sulfur (32S), red phosphorus (31P), and green (56Fe), respectively. Sarcoplasmic bodies interspersed between the myofibrils (e), next to nuclei (f), or inside an autophagic vacuole (g). Note the high-sulfur signal in the sarcoplasmic bodies (h–j), and the iron signal (green dots within the sarcoplasmic bodies in i, j). Scale bar in a = 2 µm, b = 5 µm, c = 0.5 µm, d = 1 µm, h–j = 4 µm. k Typical µFTIR spectra and their second derivative of the muscle tissue where the lipid region has been highlighted in orange and the protein region in blue; the inset shows the lipid/ protein ratio (calculated from the Infrared spectra) on an optical image of a tissue section with sarcoplasmic bodies. The color bar represents intensity of the ratio: blue and red mean low and high lipid content, respectively. The scale bar is four microns. l Box plot graphic of lipid/protein ratio (2925 cm−1/ 1654 cm−1). m Box plot graphics representing COOH/CH2 ratio indicating lipid oxidation (1739 cm−1/2925 cm−1). Ratios were calculated from the second derivative of the spectra of three different samples from three different patients (at least 10 sarcoplasmic bodies per patient). Boxplots denote the median (center line), interquartile range (box), whiskers that represents the most extreme data that are not >1.5x IQR from the edge of the box and outliers that are the points outside this range. T-tests were used to compare the sarcoplasmic body ratios with the surrounding tissue ratios and determine the p value (*p<0.005). n Infrared second derivative spectrum of the amide region of one sarcoplasmic body (green) showing an increase of β-sheet structures, indicating protein aggregation. Second derivative of the amide region corresponding to the tissue surrounding the sarcoplasmic bodies (black) Table 2 Effect of mutation on biochemical properties of MB KD, O2 (µM) k-H at pH 7.0 (h−1) a k-H at pH 5.0 (h−1) b E°’ by spectroelectr. (V vs SHE) c E°’ by SWV (V vs SHE) c kOX at pH 7.0 (min−1) Heme SASA (nm2) d WT 1.2 ± 1.1 0.22 ± 0.06 1.44 ± 0.09 +0.040 ± 0.005 −0.056 ± 0.015 1.96 ± 0.39 1.45 ± 0.04 His98Tyr 8.4 ± 2.5 1.19 ± 0.05 1.86 ± 0.05 +0.021 ± 0.005 −0.105 ± 0.015 1.97 ± 0.43 1.80 ± 0.05 Conditions: a0.2 M phosphate buffer, pH 7.0, and 0.45 sucrose (37 °C); b0.2 M acetate, pH 5.0, and 0.45 M sucrose (37 °C); c0.2 M phosphate buffer, pH 7.0 (25 °C). dValues obtained from MD simulations magnitude20–22. At pH 7.0, k-H of H98Y was five times higher determines the tendency of electron transfer from the Fe (II) compared with WT (Table 2 and Supplementary Fig. 7). center to the bound dioxygen ligand25, a critical step of the Mutations of the corresponding residue (His97) in sperm whale autoxidation process yielding inactive MetMB. From a purely myoglobin into small and hydrophobic (e.g., Ala, Val) or acidic thermodynamic point of view, lower E°’ values would favor (e.g., Glu, Asp) amino acids enhanced k-H21,22 owing to loss of formation of auto-oxidized MetMB. electrostatic interaction between the side chain of His97 and the Another important difference between WT and mutant MB negatively charged heme propionate-721. His98 of human MB concerns the affinity for molecular oxygen. Oxygen-binding contributes to modulating the solvent accessibility to the heme studies to both WT and mutant deoxymyoglobins showed pocket. Upon exchange of His98 with Tyr, the interaction with formation of oxymyglobin states (Soret band at 418 nm and the propionate is weakened; Tyr98 can still form H-bonds with two distinct bands at 545 and 580 nm) (Fig. 7). Calculation of rate propionate-7, nevertheless, k-H is higher for p.His98Tyr com- constants, kon, of O2 binding from the slope of the linear plots pared with WT MB. Notably, this change in k-H is in line with the of kobs values versus O2 concentration (Fig. 7b) yielded apparent His97Phe mutant of sperm whale myoglobin21,22, reflecting bimolecular binding rates of (1.16 ± 0.7) × 107 M−1 s−1 (WT MB) the similarity between Tyr and Phe in terms of steric hindrance. and (5.6 ± 0.2) × 106 M−1 s−1 (p.His98Tyr), respectively. From Our interpretation is further supported by the fact that at pH the intercept, koff, of these plots, the dissociation constants, KD 5.0 the k-H values for WT and p.His98Tyr are increased (1.44 ± (=koff/kon), were calculated to be (1.2 ± 1.1) µM and (8.4 ± 2.5) µM, 0.09 h−1 and 1.86 ± 0.05 h−1, respectively) and the difference respectively (see Table 2), demonstrating a reduced affinity of the between the two species becomes smaller, because of protonation mutant MB for molecular oxygen. of the heme propionates, resulting in diminished interaction of To shed light onto the structural and dynamic changes the heme group with the protein. caused by the His-to-Tyr substitution underlying the complex Next, we investigated the impact of the mutation on the biochemical picture described above, we performed molecular oxidation state of the heme iron, both from a thermodynamic and dynamics (MD) simulations of both WT and mutant species. kinetic point of view. Oxidized and reduced MB coexist in The analysis of our sampling suggests that the substitution does solution, and the ratio of the two forms is regulated thermo- not lead to major global structural rearrangements. The global dynamically by the reduction potential E°’ and kinetically by the fold of the protein appears to be maintained (Fig. 6), and the protein autoxidation rate17, which in turn are influenced by pH, two proteins feature almost superimposable fluctuation patterns oxygen concentration, and temperature23. The autoxidation (Supplementary Note 2 and Supplementary Figs. 9 and 10). process is kinetically controlled under physiological conditions. Nevertheless, MD simulations indicated that the heme group We observed nearly identical autoxidation rate constants kox for in p.His98Tyr is more solvent exposed than in the WT WT and p.His98Tyr (see Table 2). Moreover, the spectral features protein. This may explain the shift of E°’, which is highly of deoxy-, oxy-, and metmyoglobin forms for both WT and (though not solely) influenced by the extent of exposure of mutant are identical (Supplementary Fig. 8). On the contrary, the heme to solvent26,27; the latter might also contribute WT and p.His98Tyr feature slightly different reduction potential to the increment of KD for O2 binding to p.His98Tyr, owing (E°’) values. Spectroelectrochemical investigations yielded a E°’ of to the more polar environment resulting from enhanced (+0.040 ± 0.005) V vs SHE for WT MB, in excellent agreement accessibility to water. with previous findings24, whereas the reduction potential of the Initial studies using a reporter cell assay, suggest that mutant p.His98Tyr myoglobin is shifted toward lower values (Fig. 6). E°’ myoglobin may result in elevated intracellular superoxide levels. NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09111-2 a Fig. 6 In vitro and in silico studies of WT and mutant MB. a Typical square wave voltammograms recorded for WT (black) and p.His98Tyr mutant 800 His98Tyr (red) human myoglobin immobilized on a Au electrode in 20 mM phosphate buffer, pH 7.0, 25 °C. b Electronic spectra of p.His98Tyr mutant of human WT myoglobin obtained at various applied potentials E in spectroelectrochemical 600 experiments carried out with an optical thin-layer electrochemistry cell at pH 7.0, 25 °C. The corresponding Nernst plot is shown in the inset, where x = [(Amaxλred−Aλred)−(Amaxλox−Aλox). c Overlay of cartoon representations 400 I (nA) of representative structures sampled within the MD simulations for native (red) and p.His98Tyr mutant (black) human myoglobin. His98 is also shown 200 as stick with the same color coding. The heme group and Fe-ligand residue His94 is shown only for WT for sake of clarity 0 on the altered relationship between structure, dynamics, and function in the p.His98Tyr variant. –200 –400 –200 0 200 400 E (V vs SHE) Discussion Myoglobin belongs to the globin superfamily. Mutations in b hemoglobin have been recognized for a long time and result in hemoglobinopathies28. Two other globin proteins (cytoglobin and 0.3 neuroglobin) have not yet been associated with disease29. Our 80 findings allow us to describe the first disorder caused by a 70 mutation in MB. We propose the disorder should be named E (mV vs SHE) 60 Myoglobinopathy. Family 3, first described by Edstrom 50 et al.30,31in 1980, became a classic of the muscle disease literature Abs (a.u.) 0.2 40 but its cause remained unsolved until now. Myoglobinopathy has 30 a recognizable clinical phenotype, characterized by disease onset 20 during the fourth or fifth decade of life, manifesting with prox- –0.2 0.0 0.2 0.4 0.6 0.8 imal weakness in the lower extremities and axial muscles, pro- 0.1 Log x gressing to involve proximal and distal muscles of all limbs with later generalized weakness. Although initially described as a distal myopathy30 our study demonstrates that all patients exhibit proximal weakness as the initial symptom. Respiratory insuffi- ciency is a frequent complication and is the cause of death in the 0.0 majority of patients. Indeed, most advanced patients will require 350 400 450 500 550 600 650 non-invasive respiratory support. Cardiomyopathy was demon- λ (nm) strated in less than half of the patients. It is uncertain whether c myocardial involvement is an inconstant feature of the disease, or it remains subclinical for a long time. In support of this last hypothesis, post mortem examination of cardiac tissue in two individuals who did not manifest symptoms of cardiac failure showed large numbers of sarcoplasmic bodies. The combination of proximal and axial weakness and respiratory insufficiency share some similarities with the pheno- type of patients suffering from late-onset Pompe disease. How- ever, cardiomyopathy is usually not a feature of late onset Pompe disease; moreover, distal muscle involvement as revealed by clinical examination and muscle imaging studies does not occur in late onset Pompe diseases until very late in the course of the illness32. The sarcoplasmic bodies are the pathological hallmark of the disease and are the principal aid to diagnosis. They have some similarities with lipofuscin33 and also with the globular dense inclusions found in Pompe disease33,34 but they are much denser and homogeneous, and usually lack the vacuolar lipid droplets seen in these two conditions. They are also highly reminiscent of the inclusions found in the skeletal muscles from patients with chronic vitamin E deficiency, which are known to be the result of lipid oxidation as a consequence of oxidative stress35. HEK cells transfected with MBp.His98Tyr-EGFP had significantly MB accomplishes its physiological tasks thanks to an extremely elevated levels (1.45-fold, p = 0.007) of intracellular superoxide delicate balance of efficient oxygen binding, resistance to auto- compared with those transfected with wild-type MB-EGFP xidation, and high heme affinity. The full interplay between the (Fig. 7c). Further experimental investigations of structural and molecular properties affected by the pathogenic mutation is functional changes caused by the substitution will shed more light obviously complex. However, it is apparent from our biochemical 8 NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09111-2 ARTICLE a 0.14 1.0 Relative absorbance at 418 nm 418 434 0.12 0.8 0.6 0.10 Experimental time trace/wild-type 0.4 Single exponential fit/wild-type 0.08 Absorbance Experimental time trace/H98Y 0.2 Single exponential fit/H98Y 0.06 0.0 0.04 0 20 40 60 Time (ms) 0.02 545 555 580 0.00 10 μM O2 400 500 600 700 Wavelength (nm) b 1400 1200 Human myoglobin wild-type kon = 11.6 (± 0.7) × 106 M–1 s–1 koff = 13.6 (± 11.1) s–1 1000 KD = 1.2 (± 1.1) μM 800 kobs (s–1) 600 Human myoglobin H98Y kon = 5.6 (± 0.2) × 106 M–1 s–1 koff = 47.5 (± 12.5) s–1 400 KD = 8.4 (± 2.5) μM 200 0 0 50 100 150 200 O2 (μM) c * 1.0 (au, normalised to t = 0 min) 0.8 DHE fluorescence 0.6 0.4 0.2 0.0 WT (26) His98Tyr (23) Fig. 7 Kinetics of dioxygen binding to wild-type human myoglobin and the variant His98Tyr. a Spectral changes upon reaction of 1 µM ferrous wild-type hMb (black line) with 10 µM O2. The final spectrum represents oxymyoglobin (red line, 68 ms after mixing). Gray lines represent spectra obtained at 0.68, 2.72, 4.08, 6.12, 8.84, 12.24, 34.00, and 51.00 ms after mixing. The inset depicts experimental time traces at 418 nm of wild-type hMb (solid black line) and p.His98Tyr MB (dashed black line) mixed with 10 µM O2 and corresponding single-exponential fits (solid red line, wild-type MB; dashed red line, His98Tyr MB). b Linear dependence of kobs values from the O2 concentration for wild-type MM (gray circles, solid line) and p.His98Tyr MB (white squares, dashed line). c Basal intracellular superoxide levels in HEK293FT cells expressing WT or mutant MB-EGFP. Data presented as individual data points and the mean ± SEM, numbers in parenthesis represent n. *indicates p = 0.007 (Mann–Whitney test, two-tailed) NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09111-2 characterization that some key properties are altered upon from families F1 and F2. Exome enrichment was performed on DNA from the mutation, each with potential repercussions at the physiological proband (using an Ampliseq Whole Exome kit, Thermofisher Scientific). In brief, a total of 100 ng of DNA was amplified in 12 separate PCR pools, each containing level. The increased k-H would make the mutant more prone to ~ 25,000 primer pairs under the following conditions: 99 °C for 2 min followed by heme loss, as previously seen with variant sperm whale 10 cycles of 99 °C for 15 s and 60 °C for 16 min. After amplification, the individual myoglobin21,22 and this could contribute to the formation of the reactions were pooled and digested using the supplied FuPa enzyme that degrades lipid and/or protein aggregates found in patients’ muscle owing the PCR primers. Next, barcoded sequencing adaptors were ligated and the library to the generation of ROS by free heme36. In support of this was purified using AMPure beads (Beckman Coulter) and amplified for five cycles using Platinum high-fidelity polymerase. The final amplified library was purified hypothesis, we have shown experimentally, via a reporter cell again using AMPure beads and analyzed on a 2100 Bioanalyser (Agilent Tech- assay, that mutant myoglobin appears to be associated with ele- nologies). Libraries were diluted to 18–26 pM. Attachment to Ion Sphere Particles vated levels of intracellular superoxide. As a consequence, was done with the Ion Proton Template 200 V3 kit. Sequencing was performed on although myoglobin is expressed in oxidative myofibers, the P1 sequencing chips (520 flows) using an Ion Proton sequencer and Ion Sequen- cing 200 kit V3. After pooling, two samples were sequenced on one chip. After impact of the mutation may be reflected in both myofiber types. sequencing, to remove low-quality bases from the 3’ end, reads were trimmed. The E°’ shift may impact on myoglobin’s role as a regulator of Reads were mapped to the genome reference sequence (Hg19) using tmap (Torrent redox pathways in muscle, for example unbalancing the redox- Suite 4.2). Torrent Variant caller was used to call variants with exome-optimized dependent mechanisms of oxygen-controlled NO formation37–39, custom settings. Ion Reporter 4.0 was used to annotate the data. Filtering of var- iants was performed with ANNOVAR against three databases: ENCODE GENE- whereas the higher KD of the mutant MB for O2 might affect the CODE v.19, 1000genomes (threshold > 0.5%), dbSNP138 and also against a list of capability of MB to act as an oxygen storage species. common in-house variants42. The filtered exome data were interrogated for var- To date, all myoglobinopathy cases have been caused by the iants in a set of genes that were most highly enriched (>1000-fold) in skeletal same c.292C>T, (p.His98Tyr) missense variant in MB, thus muscle samples within the FANTOM5 (functional annotation of the mammalian genome) data set8. Using the FANTOM5 promoter level expression atlas, we suggesting that this position may be a mutational hotspot. We generated a ranked list of skeletal muscle enriched genes. Source file with pre- demonstrate that the c.292C>T, (p.His98Tyr) missense variant computed relative expression was downloaded from http://fantom.gsc.riken.jp/5/ is associated with altered oxygen binding and increased tendency tet/data/hg19.cage_peak_phase1and2combined_rel_expr.txt.gz. We identified the to heme loss. It remains to be seen whether other variants in promoter for every gene with the most enriched expression in 76 skeletal muscle MB result in the same disease. samples and compared this with the median expression across the entire FAN- TOM5 collection [log10(max expression in any skeletal muscle sample + 1)—log10 (median expression in the FANTOM5 collection + 1)]. A promoter for 82 anno- Methods tated protein-coding genes was identified showing > 1000-fold enriched expression Patients. The 14 patients described in this study constitute a clinically and in skeletal muscle. Thirty-one of these genes had previously been associated with a pathologically homogeneous cohort of adult patients, belonging to six unrelated genetic muscle disease in humans. The 51 genes that had not previously been families (F1–F6), (Fig. 1), originating from Spain (2), Sweden (1), France (2), and associated with human muscle disease we prioritized as candidate myopathy genes. The Netherlands (1). The families were principally identified by their highly characteristic features on muscle biopsies. The clinical and pathological features of Genome-wide linkage analysis of the Swedish family—Family 3. Genome-wide the Swedish family have been described34,35 and the Dutch patient was briefly linkage analysis was performed using genotypes of 550 microsatellite markers. described in a workshop report40. Genotype errors were checked and removed prior to analyses by using PedCheck43 The study was approved by the ethics committees of the participating GENEHUNTER v. 2.144 as used for both parametric and non-parametric linkage institutions: IDIBELL-Hospital de Bellvitge, Karolinska Institutet, University of analysis. Parametric linkage analysis was performed assuming an autosomal Western Australia, Institut de Myologie, GHU Pitié-Salpêtrière, Hospices Civils de dominant model with a penetrance of 0.999 for heterozygotes, disease-allele fre- Lyon, University Hospitals Leuven and Henri Mondor University Hospital Sample quency of 0.001, and phenocopy rate of 0.001. LOD scores were calculated under collection was performed with written informed consent from the patients the same dominant model. Both multipoint and single-point analyses were per- according to the declaration of Helsinki. formed. The positions of marker loci were based on published human genetic maps (National Center for Biotechnology Information). Muscle biopsy. A muscle biopsy was performed for at least one affected patient per family after written informed consent (Supplementary Table 1). In addition, post Target sequencing in the Swedish Family. Targeted capture and sequencing mortem cardiac, skeletal, and diaphragm muscle samples were obtained from included > 400 genes surrounding the peak multipoint LOD score. Enrichment for individuals F1, II: 7 and post mortem cardiac and skeletal muscle was obtained the target-sequencing region was obtained by using a custom-designed NimbleGen from individual F3: III: 5. Samples were processed for routine histochemical SeqCap EZ Developer library (Roche NimbleGen, Inc. Madison, WI 53719 USA). analysis and for acid phosphatase activity, and Perls’ Prussian blue stain and Eight samples, two healthy controls, and six patients, from the Swedish family, F3, Alizarin Red for iron and calcium detection, respectively. Immunohistochemical were selected for targeted sequencing. The sequencing was done at SciLifeLab analysis was performed for myoglobin, ubiquitin, p62, and LAMP1. Cryostat sec- Stockholm, Sweden. All samples yielded the expected number of sequencing reads. tions, 8 μm thick, were incubated with 1% hydrogen peroxide in 1 × PBS (Phosphate Each DNA library was prepared from 3 μg of the pooled genomic DNA. DNA was Buffer Saline) for 5 min in order to block endogenous peroxidase. Then, unspecific sheared to 300 bp using a Covaris S2 instrument and enriched by using the custom- binding reactions were blocked with 10% (goat) or 3% (horse) normal serum in 1 × designed NimbleGen SeqCap kit. The clustering was performed on a cBot cluster PBS for 2h, followed by incubation of the corresponding primary antibody diluted generation system using a HiSeq paired-end read cluster generation kit according in 1 × PBS with 3% (goat) or 1% (horse) normal serum at 4 °C overnight. Mouse to the manufacturer’s instructions. The samples were sequenced on an Illumina monoclonal anti-myoglobin (Novus biologicals), anti-LAMP1 (Santa Cruz Bio- HiSeq 2000 as paired-end reads to 100 bp. The sequencing runs were performed technology, H-228:sc-5570), rabbit polyclonal anti-ubiquitin (Dako, z-458), and according to the manufacturer’s instructions. Base conversion was done using guinea pig polyclonal antibody against the C-terminus of p62 (Progen, GP62-C) Illumina’s OLB v1.9. were used at a dilution of 1:50, 1:100, 1:100, and 1:100, respectively. After washing, the sections were processed with Super Sensitive Link-Label IHC Detection System (BioGenex) following the instructions of the manufacturer. As sarcoplasmic Sanger sequencing. The MB c.292C>T (p.His98Tyr) variant was identified in bodies are brown and diaminobenzidine (DAB) immunoreactions appear brown, Families F1 to F3 through NGS. The presence of the variant in these three families the immunoreactions were visualized as a dark blue precipitate using NH4NiSO4 was Sanger confirmed. Families F4, F5 and F6 were screened for mutations in MB diluted in phosphate buffer with 0.03% DAB, 0.04% NH4Cl, and 0.001% using Sanger sequencing. Segregation studies were performed in all families hydrogen peroxide41. Sections were viewed with an Olympus BX43 microscope. including samples from all available affected and unaffected relatives using Sanger Non-stained sections were also examined with a Leica TCS-SL confocal microscope sequencing. MB primers are included in Supplementary Methods 1. under a wide range of visible laser excitation lines. A small sample of biopsy tissue was fixed in 2% glutaraldehyde, postfixed with 1% osmium tetroxide, and In silico predictions. The in silico pathogenicity predictors used were: Mutatio- embedded in araldite. Ultrathin sections were stained with uranyl acetate and lead nAssessor45, MutationTaster46, PolyPhen-247, Provean48, and SIFT49. citrate, viewed with a JEOL 1011 electron microscope and photographs taken with a gatan 782 camera. Haplotype analysis. To test for possible founder effects, we studied eight infor- mative microsatellite markers (D22S685, D22S691, D22S1152, D22S1265, (MB), Molecular genetics—the two Spanish families. Targeted capture and next D22S277, D22S683, D22S692, IL2RB) spanning 3 Mb around MB using multiplex generation sequencing (ProtonTM, Life Technologies) of 336 genes including 254 polymerase chain reaction with labeled primers and fragment analysis. Gene- neuromuscular disease genes listed in the Neuromuscular Disorders Gene Table at Mapper software was used for the fragment analysis. The primers used in haplo- the end of December 2012 (www.musclegenetable.fr) was performed in probands type analysis are included in the Supplementary Methods 2. 10 NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09111-2 ARTICLE Correlative electron microscopy and NanoSIMS imaging. For NanoSIMS minigrid working electrode (Buckbee-Mears, Chicago, IL), an Ag/AgCl/3M KCl (Nanoscale Secondary Ion Mass Spectrometry) imaging, 500 nm sections of resin microreference electrode (373/SSG/6, Amel electrochemistry, Milano, Italy) and a embedded muscle samples from individuals F1 II: 5, and II: 7 and F2 II:2 were platinum wire as the counter electrode. Potentials were applied across the OTTLE mounted on platinum coated coverslips, and then transferred into a FEI Verios cell with an Amel model 2053 potentiostat/galvanostat. A constant temperature scanning electron microscope (SEM) for backscattered electron (BSE) imaging50. A was maintained by a circulating water bath, and the OTTLE cell temperature was 1–2 kV electron beam with current of ~ 100pA was used for BSE imaging. Sections monitored with a micro thermo-couple. UV−vis spectra were recorded using a were then coated with 5 nm platinum by a SC 7640 Polaron sputter coater, and Varian Cary C50 spectrophotometer. The OTTLE cell was flushed with argon gas transferred into the NanoSIMS 50L (CAMECA, France) for chemical analysis. The to establish an oxygen-free environment in the cell. Experiments were conducted at same areas imaged by SEM were analyzed by the NanoSIMS for chemical dis- a constant temperature of 25 °C, using 600 μL samples containing 15 μM of protein tribution. A 16 keV Cs+ primary beam is used to detect 31P–, 32S–, and 56Fe16O. in 50 mM phosphate buffer (pH 7.0) 20 mM NaCl and 30 μM phenazine metho- For comparative purposes, in addition to myoglobinopathy samples, we sulfate. Nernst plots consisted of at least five points and were invariably linear with performed NanoSIMS analysis in one sample from a patient with Pompe disease a slope consistent with a one-electron reduction process. Square wave voltam- containing electron-dense inclusions, two samples showing abundant lipofuscin mograms were recorded using a Potentiostat/Galvanostat mod. 273A (EG&G PAR, and two muscle samples from patients affected with distal myopathies with Oak Ridge, USA). A graphite disk electrode, a Pt wire and a saturated calomel rimmed vacuoles. electrode (SCE) were used as working, counter and reference electrode, respec- tively. The electric contact between the SCE and the working solution was achieved Fourier transform infrared microscopy. 10 μm-thick cryostat muscle sections with a Vycor® (PAR) set. The working electrode was cleaned before each use with were obtained from individuals F1, II: 5 and II: 7, F2, II:2, F3, III:5, II:15 and mechanical polishing with 0.5 μm aluminum oxide followed by 5 min in an IV:6 and deposited on CaF2 windows and allowed to dry. Samples were kept at ultrasound bath. Myoglobin was deposited on the electrode by drop casting 10 μL vacuum conditions before the infrared spectroscopy analysis. μFTIR experiments of a solution containing 100 μM protein and 20 mM phosphate buffer (pH 7.0). The were performed at the MIRAS beamline in the ALBA synchrotron, Spain using electrode was then incubated overnight at 4 °C. The following step was to drop cast a Hyperion 3000 Microscope coupled to a Vertex 80 spectrometer (Brucker) 20 μL of a solution containing 10% m/m of gelatin type A (Sigma-Aldrich, G1890) equipped with 36x magnification objective. The measuring range was 900 and 20 mM phosphate buffer (pH 7.0), followed by incubation at room temperature −4000 cm−1 and the spectra collection was done in transmission mode at 4 cm−1 for 2 hours. The electrode was then ready for electrochemical measurements. The spectral resolution, 10 μm × 10 μm aperture dimensions using an mercury cad- working solution was 20 mM phosphate buffer (pH 7.0). Experiments were carried mium telluride (MCT) detector. For each spectrum 128 scans were co-added. out under argon atmosphere at 25 °C. The errors associated to the E°’ values are the Background spectra were collected from a clean area of each CaF2 window. FTIR standard deviations calculated over at least three independent sets of measurements data were analyzed using OPUS 7.5 (Brucker) and Origin 9.1. The spectra exhi- for each species. biting strong Mie scattering were eliminated and second derivation of the spectra was applied in order to eliminate the baseline and improve the resolution of the Measurement of hemin loss rates for WT and p.His98Tyr. Hemin different bands using a Savitzky−Golay algorithm with a 15-point filter and a dissociation kinetics was investigated as follows: we monitored the decrease with polynomial order of two. The assignments of Infrared bands are well-established time (0–350 min) of the absorbance at 410 nm for the transfer of the heme from for proteins and lipids and are shown in Fig. 5c. The derivative of each FTIR holo WT (or p. His98Tyr) MB to the apo MB double mutant p.His65Tyr/ spectrum was computed and the ratios for lipid oxidation (1741 cm−1/2925 cm−1, Val69Phe20,21 in a 1:10 concentration ratio. Time courses were fitted to a single- CO/−CH2) and lipid protein ratio (2925 cm−1/1654 cm−1 Amide I/−CH2) exponential expression in GraphPad Prism 5.0 (GraphPad Software, Inc., San calculated51. Diego, CA) to obtain the first order dissociation rate constant for hemin loss k-H. The experiments were conducted at 37 °C in the presence of 0.45 M sucrose, in 0.2 M Expression and purification of recombinant WT and mutant MB. pET-19b phosphate buffer at pH 7 or 0.2 M acetate buffer pH 5. UV−vis spectra were plasmids with human WT, p.His98Tyr or p.His65Tyr/Val69Phe myoglobin recorded using a Varian Cary C50 spectrophotometer. cDNA (the latter to be used for heme loss kinetic studies, see below) cloned into them were purchased (GenScript USA Inc.) and then transformed in Escherichia Coli BL21(DE3) for the production of N-terminally His-tagged fusion myoglo- Molecular modeling. For each protein, three 350 ns-long MD simulations were bins. Luria-Bertani medium supplemented with ampicillin (100 μg/ml) was produced using the 5.0.4 version of the GROMACS software. The starting inoculated with a freshly prepared overnight culture (at a dilution ratio of 1:100). structure of human myoglobin was taken from the PDB file 3RGK52. The pro- The culture was grown at 37 °C under agitation (220 rpm) until OD600 = 0.6 was gram Modeller53–56 was used to edit the 3RKG.pdb file. The structure of the C- reached. The temperature and the shaker speed were reduced to 20 °C and 100 terminus was taken from the PDB file 1MWD57. The structure of the p.His98Tyr rpm and hemin was added to a final concentration of 15 μg/mL. After 30 min, mutant was obtained by homology modeling using Modeller and taking the isopropyl-β-d-thiogalactopyranoside was added to a final concentration of 0.5 3RGK file as template structure. All the simulations were performed in the mM. Please note that p.His65Tyr/Val69Phe for the hemin loss rate experiments reduced ensemble. The starting structures were first relaxed in vacuum using the was grown using the same procedure but no hemin was added. To make sure steepest descent algorithm, 3000 steps, and then put in a rhombic dodecahedral that the double mutant was in the apo form, 3.5 mg of protein at pH 2.5 were box, filled with TIP3P water molecules. Na+ and Cl− were added to neutralize extracted using an equal volume of 2-butanone, shaking three times for 5 min the protein total charge and to have a final concentration of 100 mM. The sys- and then incubating the mixture for 10 min at 4 °C20. The aqueous phase was tems were then relaxed with the conjugate gradients algorithm in two steps. dialyzed at 4 °C against 1 l of water and then against 1 l of 0.2 M phosphate buffer Then the systems were annealed to 300 K, starting from 50 K, in a 100 ps run. pH 7 or 0.2 M acetate buffer pH 5. After 12 h, the culture was centrifuged, The systems were then equilibrated in three successive runs, each 500 ps long: (i) resuspended in 50 mM phosphate buffer, 150 mM NaCl, and 20 mM imidazole an NVT at 300 K58, (ii) an NPT at 300 K, 1 bar (iii) an NPT at 300 K and 1 bar59. (pH 7.5), lysed by sonication and clarified by centrifugation. The proteins were The last frame was used as a starting structure for the MD production run in the purified by affinity chromatography (HisTrap FF crude columns (GE Health- NPT ensemble, at a temperature of 300 K and a pressure of 1 bar. All bond care) connected to an AKTA Prime chromatographic system (Amersham lengths were constrained using the LINCS algorithm60 and an integration step of Pharmacia Biotech) and subsequently by gel filtration chromatography (HiLoad 2.0 fs was used. Long-range electrostatics was computed by the Particle Mesh 16/60 Superdex 75 prep grade (Amersham Biosciences). The purity of the Ewald method61, with a grid spacing of 0.12 nm combined with a fourth-order fractions was checked by sodium dodecyl sulfate polyacrylamide gel electro- cubic interpolation and a real space cutoff of 0.9 nm. During all the runs, the phoresis and UV-Vis spectroscopy. CHARMM27 force field62 with CMAP correction63 was used. The three repli- cates for each system were made using the same starting structure, but changing the seed for the velocities random generator. The determination of the Solvent WT and His98Tyr variant myoglobin aggregation test. WT MB and the variant Accessible Surface Area (SASA), the analysis of the distances, the analysis of His98Tyr MB were incubated in 10 mM phosphate buffer at pH 7.4 at 37 °C and 200 secondary structure, the Essential Dynamics (ED), the root mean square fluc- rpm agitated for 1 week. At different times, aliquots were taken for ThT Fluores- tuations calculations and the clustering analysis, were performed using the cence and Infrared Spectroscopy measurements. For fibril fluorescence detection, corresponding Gromacs packages. The ED analyses were performed using the Thioflavin T (ThT) (SIGMA-Aldrich), a fluorescence dye specific to fibrils, was backbone atoms of residues 1–150 both for the fitting and the calculation of the added to the solution. In total, 8 mM ThT stock solutions were prepared in 10 mM covariance matrix. The clustering procedure was made with a cutoff of 0.11 nm phosphate buffer at pH 7.4 buffer and filtered through a 0.45 μm syringe driven and the Gromos method64. The centroids of the first clusters were used for the filter unit (Millipore). ThT final concentration in the fluorescence cuvette was 35 μM. structure comparison. ED analysis was used to investigate the conformational Excitation and emission wavelengths were set at 450 and 490 nm, respectively, using space spanned by WT and mutant p.His98Tyr myoglobin within the molecular a PTI fluorimeter. Temperature was 37 °C. For Infrared spectra, 3 μL of the different dynamics (MD) simulation and visualize the principal motions associated with samples were deposited on CaF2 windows and dried under vacuum. Infrared spectra each species. To assess the differences between the fluctuation patterns char- were measured at MIRAS beamline in ALBA synchrotron. acteristic of WT myoglobin and the p.His98Tyr mutant, the equilibrated tra- jectories were linked for the two species into a single molecular dynamics Electrochemical measurements. All spectroelectrochemical experiments were ensemble (concatenated trajectory) and essential dynamics analysis was per- conducted in a homemade OTTLE (optical transparent thin-layer spectro- formed. The concatenated trajectory was projected onto a plane defined by the electrochemical) cell, in a three-electrode configuration that consisted of a gold first two eigenvectors of motions, which are usually referred to as the essential NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09111-2 plane. This allowed a two-dimensional representation of the conformational 7. Blanchetot, A., Price, M. & Jeffreys, A. J. The mouse myoglobin gene. landscape explored during the MD simulations, and for the calculation of the Characterisation and sequence comparison with other mammalian myoglobin corresponding Free Energy Landscape. genes. Eur. J. Biochem. 159, 469–474 (1986). 8. Forrest, A. R. et al. A promoter-level mammalian expression atlas. Nature 507, Oxygen binding to wild-type human myoglobin and the variant His98Tyr. 462–470 (2014). Time-resolved binding of oxygen (O2) to wild-type human myoglobin and the 9. Meeson, A. P. et al. Adaptive mechanisms that preserve cardiac function in variant His98Tyr was monitored by stopped-flow spectroscopy in the conventional mice without myoglobin. Circ. Res. 88, 713–720 (2001). mixing mode. The stopped-flow apparatus (model SX 18MV, Applied Photo- 10. Godecke, A. et al. Disruption of myoglobin in mice induces multiple physics) was equipped with a diode-array detector and an optical quartz cell with a compensatory mechanisms. Proc. Natl Acad. Sci. USA 96, 10495–10500 (1999). path length of 10 mm (volume: 20 µL). The fastest time of mixing was 0.68 ms. All 11. Giles, G. I. & Jacob, C. Reactive sulfur species: an emerging concept in measurements were performed at 25 °C. In order to establish O2-free starting oxidative stress. Biol. Chem. 383, 375–388 (2002). conditions, all solutions were degassed with Argon. The stopped-flow apparatus 12. Kasamatsu, S. et al. Redox signaling regulated by cysteine persulfide and (all syringes and tubings) was extensively flushed with O2-free buffer (20 mM protein polysulfidation. Molecules 21, 1721 (2016). phosphate buffer, pH 7.0; 20 mM NaCl). The concentration of wild-type MB and of 13. Terman, A. & Kurz, T. Lysosomal iron, iron chelation, and cell death. the variant in the cell was 1 µM. Prior to measurements, the degassed O2-free Antioxid. Redox Signal. 18, 888–898 (2013). protein solutions were reduced with sodium dithionite (final concentration 25 µg 14. Yang, H., Yang, S., Kong, J., Dong, A. & Yu, S. Obtaining information about mL−1). To test oxygen binding, O2-free buffer was flushed with O2 for different protein secondary structures in aqueous solution using Fourier transform time spans and the O2 concentration was determined using a thermostatted (25 °C) IR spectroscopy. Nat. Protoc. 10, 382–396 (2015). Clark-type electrode (Oxygraph Plus; Hansatech Instruments, Norfolk, United 15. Benseny-Cases, N., Klementieva, O., Cotte, M., Ferrer, I. & Cladera, J. Kingdom). The electrode was equilibrated to 100% O2 saturation by bubbling O2 Microspectroscopy (μFTIR) reveals co-localization of lipid oxidation and through the buffer and for 0% saturation by bubbling with N2 until the respective amyloid plaques in human Alzheimer disease brains. Anal. Chem. 86, plateaus were reached to derive an offset and a calibration factor. Experiments were 12047–12054 (2014). conducted in 20 mM phosphate buffer, pH 7.0, supplemented with 20 mM NaCl. 16. Benseny-Cases, N., Klementieva, O., Malý, J. & Cladera, J. Granular non- Dioxygen binding was determined by monitoring the absorbance increase at 418 fibrillar aggregates and toxicity in Alzheimer’s disease. Curr. Alzheimer Res. 9, nm. The apparent second-order rate constants, kon, were obtained from the slope of 962–971 (2012). a plot of kobs values (derived from single-exponential fit of the respective time 17. Brantley, R. E., Smerdon, S. J., Wilkinson, A. J., Singleton, E. W. & Olson, J. S. traces at 418 nm) versus the dioxygen concentration; koff was derived from the The mechanism of autooxidation of myoglobin. J. Biol. Chem. 268, 6995–7010 intercept of this plot. (1993). 18. Livingston, D. J., McLachlan, S. J., La Mar, G. N. & Brown, W. D. Myoglobin: Assessment of cellular superoxide. HEK293FT cells (p15-20) (Invitrogen) cytochrome b5 interactions and the kinetic mechanism of metmyoglobin were seeded onto coverglass pre-coated with Matrigel (Corning) in 12-well reductase. J. Biol. Chem. 260, 15699–15707 (1985). tissue culture plates. Cells were transfected with pcDNA3.1 expression plasmids 19. Hagler, L., Coppes, R. I. Jr & Herman, R. H. Metmyoglobin reductase. containing full length human wild-type (MBWT) or c.292C>T MB (MBp.His98Tyr) Identification and purification of a reduced nicotinamide adenine cDNA fused with an EGFP tag (Genscript). Transfections were performed using dinucleotide-dependent enzyme from bovine heart which reduces 1 μg of DNA and Lipofectamine® 3000 (Thermofisher) as per manufacturer’s metmyoglobin. J. Biol. Chem. 254, 6505–6514 (1979). instructions. Experiments were performed 48–72 h post transfection and were 20. Hargrove, M. S. et al. His64(E7) → Tyr apomyoglobin as a reagent for conducted over three independent experiments. Plasmids are available upon measuring rates of hemin dissociation. J. Biol. Chem. 269, 4207–4214 (1994). request. Basal superoxide generation was assessed65 in individual cells expressing 21. Hargrove, M. S., Wilkinson, A. J. & Olson, J. S. Structural factors governing MBWT and MBp.His98Tyr (as indicated by EGFP positivity) using the hemin dissociation from metmyoglobin. Biochemistry 35, 11300–11309 fluorescent indicator dihydroethidium (DHE, 5 μM, 515–560 nm ex filter, 590 (1996). long pass em) at 37 °C. Fluorescent DHE signal was measured on a Hamamatsu 22. Liong, E. C., Dou, Y., Scott, E. E., Olson, J. S. & Phillips, G. N. Jr. Orca ER digital camera attached to an inverted Nikon TE2000-U microscope. Waterproofing the heme pocket. Role of proximal amino acid side chains Fluorescent images were taken at 1 min intervals with 200 ms exposure. Meta- in preventing hemin loss from myoglobin. J. Biol. Chem. 276, 9093–9100 morph 6.3 was used to quantify the signal by manually tracing individual cells. (2001). An equivalent region not containing cells was used as background and was 23. Shikama, K. The molecular mechanism of autoxidation for myoglobin and subtracted. DHE fluorescence for each cell was recorded as the slope of the hemoglobin: a venerable puzzle. Chem. Rev. 98, 1357–1374 (1998). signal measured over the course of 20 min, normalized to fluorescent signal 24. Fuller Taylor, J. & Morgan, V. E. Oxidation-reduction potentials of the recorded at 1 min. metmyoglobin-myoglobin. J. Biol. Chem. 144, 15–20 (1942). 25. Arcon, J. P., Rosi, P., Petruk, A. A., Marti, M. A. & Estrin, D. A. Molecular Data availability mechanism of myoglobin autoxidation: insights from computer simulations. The source data underlying Figs. 5, 6, and 7 are provided in the Source Data file. The novel J. Phys. Chem. B 119, 1802–1813 (2015). variant here reported has been deposited in ClinVar, accession number: SCV000882438. 26. Bortolotti, C. A. et al. The reversible opening of water channels in cytochrome Patients’ consents were not obtained for public deposition of sequencing data, but these are c modulates the heme iron reduction potential. J. Am. Chem. Soc. 134, available upon reasonable request. Any additional data that support the findings of this 13670–13678 (2012). study are available from the corresponding author upon reasonable request. 27. Akif Tezcan, F., Winkler, J. R. & Gray, H. B. Effects of ligation and folding on reduction potentials of heme proteins. J. Am. Chem. Soc. 120, 13383–13388 (1998). Received: 14 September 2017 Accepted: 14 February 2019 28. Piel, F. B. The present and future global burden of the inherited disorders of hemoglobin. Hematol. Oncol. Clin. North. Am. 30, 327–4 (2016). 29. Burmester, T. & Hankeln, T. Function and evolution of vertebrate globins. Acta Physiol. (Oxf.). 211, 501–514 (2014). 30. Edström, L., Thornell, L. E. & Eriksson, A. A new type of hereditary distal myopathy with characteristic sarcoplasmic bodies and intermediate (skeletin) filaments. J. Neurol. Sci. 47, 171–190 (1980). References 31. Engvall, M., Ahlberg, G., Hedberg, B., Edström, L. & Ansved, T. Sarcoplasmic 1. Ordway, G. A. & Garry, D. J. Myoglobin: an essential hemoprotein in body myopathy—a rare hereditary myopathy with characteristic inclusions. striated muscle. J. Exp. Biol. 207, 3441–3446 (2004). Acta Neurol. Scand. 112, 223–227 (2005). 2. Wittenberg, J. B. & Wittenberg, B. A. Myoglobin function reassessed. J. Exp. 32. Schoser, B. G. et al. Adult-onset glycogen storage disease type 2: clinico- Biol. 206, 2011–2020 (2003). pathological phenotype revisited. Neuropathol. Appl. Neurobiol. 33, 544–559 3. Kanatous, S. B. & Garry, D. J. Gene deletional strategies reveal novel (2007). physiological roles for myoglobin in striated muscle. Respir. Physiol. 33. Sohal, R. S. & Brunk, U. T. Lipofuscin as an indicator of oxidative stress and Neurobiol. 51, 151–158 (2006). aging. Adv. Exp. Med. Biol. 266, 17–26 (1989). 4. Flögel, U., Gödecke, A., Klotz, L. O. & Schrader, J. Role of myoglobin in the 34. Tsuburaya, R. S. et al. Acid phosphatase-positive globular inclusions is a good antioxidant defense of the heart. FASEB J. 18, 1156–1158 (2004). diagnostic marker for two patients with adult-onset Pompe disease lacking 5. Richards, M. P. Redox reactions of myoglobin. Antioxid. Redox Signal. 18, disease specific pathology. Neuromuscul. Disord. 22, 389–393 (2012). 2342–2351 (2013). 35. Neville, H. E., Ringel, S. P., Guggenheim, M. A., Wehling, C. A. & Starcevich, 6. Kendrew, J. C. et al. Structure of myoglobin: a three-dimensional Fourier J. M. Ultrastructural and histochemical abnormalities of skeletal muscle in synthesis at 2 Å resolution. Nature 185, 422–427 (1960). patients with chronic vitamin E deficiency. Neurology 33, 48348 (1983). 12 NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09111-2 ARTICLE 36. Kumar, S. & Bandyopadhyay, U. Free heme toxicity and its detoxification 65. Viola, H. M., Arthur, P. G. & Hool, L. C. Transient exposure to hydrogen systems in human. Toxicol. Lett. 157, 175–188 (2005). peroxide causes an increase in mitochondria-derived superoxide as a result of 37. Maia, L. B. & Moura, J. J. How biology handles nitrite. Chem. Rev. 114, sustained alteration in l-type Ca2+ channel function in the absence of 5273–5357 (2014). apoptosis in ventricular myocytes. Circ. Res. 100, 1036–1044 (2007). 38. Rassaf, T. et al. Nitrite reductase function of deoxymyoglobin: oxygen sensor and regulator of cardiac energetics and function. Circ. Res. 100, 1749–1754 (2007). Acknowledgements 39. Shiva, S. et al. Deoxymyoglobin is a nitrite reductase that generates nitric We thank the patients and their families for their collaboration. This study has been oxide and regulates mitochondrial respiration. Circ. Res. 100, 654–661 (2007). funded by Instituto de Salud Carlos III through the project PI14/00738 (co-funded by 40. Goebel, H. H. & Bönneman, C. G. Rare structural myopathy consortium. European Regional Development Fund. ERDF, a way to build Europe) and from a grant 169th ENMC International Workshop rare structural myopathies. Neuromusc. from Genzyme Foundation to MO. We thank CERCA Programme/Generalitat de Cat- Disord 21, 363–374 (2011). alunya for institutional support. GR, NGL, and ARRF are supported by Australian 41. Janué, A., Olivé, M. & Ferrer, I. Oxidative stress in desminopathies and National Health and Medical Research Council Fellowships APP1002147, APP1035955, myotilinopathies: a link between oxidative damage and abnormal protein APP1154524, and Project Grants APP1080587 and APP1146321. ARRF was also sup- aggregation. Brain Pathol. 17, 377–388 (2007). ported by a Senior Research Fellowship from the Cancer Research Trust, the MACA Ride 42. Ravenscroft, G. et al. Mutations of GPR126 are responsible for severe to Conquer Cancer and an Australian Research Council Discovery Project Grant arthrogryposis multiplex congenita. Am. J. Hum. Genet. 96, 955–961 (2015). (DP160101960). MC is supported by ISCIII (JR15/00042). AWr is a Ragnar Söderberg 43. O’Connell, J. R. & Weeks, D. E. PedCheck: a program for identification of Fellow in Medicine (M77/13) and is supported by the Swedish Research Council (AW genotype incompatibilities in linkage analysis. Am. J. Hum. Genet. 63, (VR2016-02179). AWe is supported by the Swedish Research Council (2016-01082), 259–266 (1998). Stockholm County Council (20170022), the Swedish Brain Foundation (FO2015-0146), 44. Kruglyak, L., Daly, M. J., Reeve-Daly, M. P. & Lander, E. S. Parametric and and the Knut & Alice Wallenberg Foundation (KAW 2013.0026 and KAW 2014.0293). nonparametric linkage analysis: a unified multipoint approach. Am. J. Hum. KGC is supported by the KOOR, UZ Leuven, Belgium. AG is supported by the Research Genet. 58, 1347–1363 (1996). Council KU Leuven (C24/16/045). JK is a recipient of the Royal Society Wolfson 45. Grimm, D. G. et al. The evaluation of tools used to predict the impact of Research Merit Award. TS is supported by Stockholm County Council (20170831), missense variants is hindered by two types of circularity. Hum. Mutat. 36, Stiftelsen Frimurare Foundation, and Promobilia. LH is supported by an Australian 513–523 (2015). National Health and Medical Research Council Fellowship (APP1117366). HV is sup- 46. Schwarz, J. M., Cooper, D. N., Schuelke, M. & Seelow, D. MutationTaster2: ported by a National Heart Foundation of Australia Fellowship (101930). µFTIR mutation prediction for the deep-sequencing age. Nat. Methods 11, 361–362 microscope analysis was performed at MIRAS beamline at ALBA Synchrotron with the (2014). collaboration of ALBA staff. We thank professor Josep García Valero from the University 47. Adzhubei, I., Jordan, D. M. & Sunyaev, S. R. Predicting functional effect of of Barcelona for helpful comments on the ultrastructural features. We thank Dolores human missense mutations using PolyPhen-2. Curr. Protoc. Hum. Genet. Moreno, Susana Arnedo, Guy Brochier, and Benjamín Torrejón-Escribano for excellent Chapter 7, Unit 7. 20 (2013). technical assistance and to the Audiovisual Department from Hospital Universitari de 48. Choi, Y. & Chan, A. P. PROVEAN web server: a tool to predict the functional Bellvitge for their help with the preparation of the figures. KCG wants to thank Klara effect of amino acid substitutions and indels. Bioinformatics 31, 2745–2747 Mallants for technical support and to Dr. Sjoerd G. van Duinen (Leiden), and professsor (2015). Joachim Weis (Aachen). Muscle sample and associated data from Family 5 were obtained 49. Flanagan, S. E., Patch, A. M. & Ellard, S. Using SIFT and PolyPhen to predict from Cardiobiotec Biobank (CRB-HCL Hospices Civils de Lyon BB-0033-00046). loss-of-function and gain-of-function mutations. Genet. Test. Mol. Biomark. 14, 533–537 (2010). Erratum in: Genet. Test Mol. Biomarkers. 14, 730 (2010). Author contributions 50. Jiang, H., Kilburn, M. R., Decelle, J. & Musat, N. NanoSIMS chemical M.O., M.E., A.W. and N.G.L. designed and oversaw the study. M.O., G.R., M.C.-S., imaging combined with correlative microscopy for biological sample analysis. HoJ, C.A.B., M.P., M.L., HaJ, N.B.-C., S.H., C.O., G.B., MBe, MBo, G.D.R., H.V., L.H., Curr. Opin. Biotechnol. 41, 130–135 (2016). J.C., K.L.-R., F.X., C.F., P.C. and T.S. performed experiments. M.O., M.E., G.R., M.C.-S., 51. Vileno, B. et al. Evidence of lipid peroxidation and protein phosphorylation C.A.B., N.B.-C., S.H., C.O., H.V., L.H. F.X., HoJ, HaJ, A.R.R.F., N.V., I.F., A.W. and in cells upon oxidative stress photo-generated by fullerols. Biophys. Chem. N.G.L. analyzed data. M.O., M.E., M.C.-S., F.M., J.J.B., E.M., N.B.R., N.S., C.V., K.G.C., 152, 164–169 (2010). C.S.M.S., A.G., M.L., G.B., J.K., B.U., L.E., A.W. provided samples and clinical data. M.O., 52. Hubbard, S. R., Hendrickson, W. A., Lambright, D. G. & Boxer, S. G. X-ray M.E., G.R., M.C.-S., H.J., C.A.B., C.O., N.B.-C. and N.G.L. wrote the manuscript. All crystal structure of a recombinant human myoglobin mutant at 2.8 A authors contributed to and approved the final manuscript. resolution. J. Mol. Biol. 213, 215–218 (1999). 53. Webb, B. & Sali, A. Comparative protein structure modeling using Modeller. Curr. Protoc. Bioinform. 47, 5.6.1–5.6.32 (2014). Additional information 54. Martí-Renom, M. A. et al. Comparative protein structure modeling of genes Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- and genomes. Annu. Rev. Biophys. Biomol. Struct. 29, 291–325 (2000). 019-09111-2. 55. Sali, A. & Blundell, T. L. Comparative protein modelling by satisfaction of spatial restraints. J. Mol. Biol. 234, 779–815 (1993). Competing interests: The authors declare no competing interests. 56. Fiser, A., Do, R. K. & Sali, A. Modeling of loops in protein structures. Protein Sci. 9, 1753–1773 (2000). Reprints and permission information is available online at http://npg.nature.com/ 57. Krzywda, S. et al. Stabilizing bound O2 in myoglobin byvaline68 (E11) to reprintsandpermissions/ asparagine substitution. Biochemistry 37, 15896–15907 (1998). 58. Berendsen, H. J. C., Postma, J. P. M., van Gunsteren, W. F., DiNola, A. & Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in Haak, J. R. Molecular dynamics with coupling to an external bath. J. Chem. published maps and institutional affiliations. Phys. 81, 3684–3690 (1984). 59. Hoover, W. G. Canonical dynamics: equilibrium phase-space distributions. Phys. Rev. A. Gen. Phys. 31, 1695–1697 (1985). Open Access This article is licensed under a Creative Commons 60. Hess, B., Bekker, H., Berendsen, H. J. C. & Fraaije, J. G. E. M. LINCS: a linear Attribution 4.0 International License, which permits use, sharing, constraint solver for molecular simulations. J. Comput. Chem. 18, 1463–1472 adaptation, distribution and reproduction in any medium or format, as long as you give (1997). appropriate credit to the original author(s) and the source, provide a link to the Creative 61. Darden, T., York, D. & Pedersen, L. Particle mesh Ewald: an N•log(N) method Commons license, and indicate if changes were made. The images or other third party for Ewald sums in large systems. J. Chem. Phys. 98, 10089–10092 (1993). material in this article are included in the article’s Creative Commons license, unless 62. Best, R. B. et al. Optimization of the additive CHARMM all-atom protein indicated otherwise in a credit line to the material. If material is not included in the force field targeting improved sampling of the backbone φ, ȥ and side-chain article’s Creative Commons license and your intended use is not permitted by statutory Ȥ(1) and Ȥ (2) dihedral angles. J. Chem. Theory Comput. 8, 3257–3273 (2012). regulation or exceeds the permitted use, you will need to obtain permission directly from 63. Mackerell, A. D. Jr, Feig, M. & Brooks, C. L. 3rd Extending the treatment the copyright holder. To view a copy of this license, visit http://creativecommons.org/ of backbone energetics in protein force fields: limitations of gas-phase licenses/by/4.0/. quantum mechanics in reproducing protein conformational distributions in molecular dynamics simulations. J. Comput. Chem. 25, 1400–1415 (2004). 64. Daura, X. et al. Peptide folding: when simulation meets experiment. Angew. © The Author(s) 2019 Chem. Int. Ed. 38, 236–240 (1999). NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications 13 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09111-2 Montse Olivé 1, Martin Engvall2,3, Gianina Ravenscroft4, Macarena Cabrera-Serrano5, Hong Jiao6,7, Carlo Augusto Bortolotti 8, Marcello Pignataro 9, Matteo Lambrughi8, Haibo Jiang10, Alistair R.R. Forrest4, Núria Benseny-Cases11, Stefan Hofbauer 12, Christian Obinger 12, Gianantonio Battistuzzi9, Marzia Bellei8, Marco Borsari 9, Giulia Di Rocco8, Helena M. Viola13, Livia C. Hool13,14, Josep Cladera15, Kristina Lagerstedt-Robinson16, Fengqing Xiang17, Anna Wredenberg3,18, Francesc Miralles19, Juan José Baiges20, Edoardo Malfatti21,22, Norma B. Romero21,22, Nathalie Streichenberger23, Christophe Vial24, Kristl G. Claeys25,26, Chiara S.M. Straathof27, An Goris28, Christoph Freyer 3,18, Martin Lammens29,30,31, Guillaume Bassez32, Juha Kere 6,7,33,34, Paula Clemente18, Thomas Sejersen 17, Bjarne Udd35,36,37, Noemí Vidal1, Isidre Ferrer1,38, Lars Edström39, Anna Wedell2,3 & Nigel G. Laing4 1 Neuropathology Unit, Department of Pathology and Neuromuscular Unit, Department of Neurology, IDIBELL-Hospital de Bellvitge, Hospitalet de Llobregat, Barcelona 08907, Spain. 2Department of Molecular Medicine and Surgery, Science for Life Laboratory, Karolinska Institutet, Stockholm SE-17176, Sweden. 3Centre for Inherited Metabolic Diseases, Karolinska University Hospital, Stockholm SE-17177, Sweden. 4Centre for Medical Research, University of Western Australia, Harry Perkins Institute of Medical Research, Perth 6000 WA, Australia. 5Neurology Department and Instituto de Biomedicina de Sevilla, Hospital Universitario Virgen del Rocío, Seville 41013, Spain. 6Department of Biosciences and Nutrition, Science for Life Laboratory, Karolinska Institutet, Stockholm SE-14157, Sweden. 7Clinical Research Centre, Karolinska University Hospital, Huddinge SE- 17177, Sweden. 8Department of Life Sciences, University of Modena and Reggio Emilia, Modena 41121, Italy. 9Department of Chemical and Geological Sciences, University of Modena and Reggio Emilia, Modena 41121, Italy. 10School of Molecular Sciences, The University of Western Australia, 35 Stirling Highway, Crawley 6009 WA, Australia. 11ALBA Synchrotron Light Source, Cerdanyola del Vallès, Barcelona 08290, Spain. 12 Division of Biochemistry, Department of Chemistry, Vienna Institute of BioTechnology, BOKU—University of Natural Resources and Life Sciences, Vienna A-1180, Austria. 13School of Human Sciences, The University of Western Australia, Perth 6000 Western Australia, Australia. 14Victor Chang Cardiac Research Institute, Darlinghurst 2010 NSW, Australia. 15Unitat de Biofísica, Departament de Bioquímica i de Biologia Molecular, Facultat de Medicina, Universitat Autònoma de Barcelona, Bellaterra 08193, Spain. 16Department of Molecular Medicine and Surgery, Karolinska Institutet, and Department of Clinical Genetics, Karolinska University Hospital, Solna, Stockholm SE-17176, Sweden. 17Department of Women’s and Children’s Health, Karolinska Institutet, Stockholm SE-17177, Sweden. 18Department of Medical Biochemistry and Biophysics, Karolinska Institutet, Stockholm SE-17177, Sweden. 19Neurology Department, Hospital Son Espases, Palma de Mallorca 07120, Spain. 20Neurology Department, Hospital Verge de la Cinta, Tortosa 43500, Spain. 21Université Sorbonne, UPMC Univ Paris 06, INSERM; UMRS974, CNRS FRE3617, Center for Research in Myology, GH Pitié-Salpêtrière, 47 Boulevard de l’hôpital, 75013 Paris, France. 22Centre de Référence de Pathologie Neuromusculaire Paris-Est, Institut de Myologie, GHU Pitié-Salpêtrière, Assistance Publique-Hôpitaux de Paris, Paris 75013, France. 23Centre de Pathologie et Neuropathologie Est, Hospices Civils de Lyon; Université Claude Bernard Lyon1, Institut NeuroMyogène CNRS UMR 5310—INSERM U1217; Institut NeuroMyogène, Villeurbanne 69677, France. 24Electromyographie—Groupement Hospitalier Est, Hospices Civils de Lyon 69677, France. 25Department of Neurology, University Hospitals Leuven, Leuven 2333, Belgium. 26KU Leuven—University of Leuven, Laboratory for Muscle diseases and Neuropathies, Department of Neurosciences, Experimental Neurology, Leuven 2333, Belgium. 27Department of Neurology, Leiden University Medical Center, Leiden 2333, The Netherlands. 28KU Leuven—University of Leuven, Laboratory for Neuroimmunology, Department of Neurosciences, Experimental Neurology, Leuven 2333, Belgium. 29Department of Pathology, Antwerp University Hospital, Edegem 2650, Belgium. 30 Laboratory of Neuromuscular Pathology, Institute Born-Bunge, University of Antwerp, Wilrijk 2610, Belgium. 31Department of Pathology, Radboud University Medical Center, Nijmegen 6525, The Netherlands. 32Neuromuscular Reference Center, Henri Mondor University Hospital AP-HP, INSERM U955, Team 10, Biology of the Neuromuscular System, East-Paris University (UPEC), Paris 94010, France. 33Molecular Neurology Research Program, University of Helsinki and Folkhälsan Institute of Genetics, Helsinki 00014, Finland. 34School of Basic and Medical Biosciences, King’s College London, London WC2R2LS, UK. 35Neuromuscular Research Center, Tampere University Hospital, University of Tampere, Tampere 33521, Finland. 36Folkhälsan Genetic Institute, University of Helsinki, Helsinki 00250, Finland. 37Neurology Department, Vasa Central Hospital, Vasa, Finland Neuromuscular Research Center, Tampere University Hospital, University of Tampere, Tampere 65100, Finland. 38Department of Pathology and Experimental Therapeutics, University of Barcelona, CIBERNEDHospitalet de LLobregat, Barcelona 08907, Spain. 39Center for Molecular Medicine, Karolinska Institutet, Stockolm 17177, Sweden. These authors contributed equally: Montse Olivé, Martin Engvall, Gianina Ravenscroft. These authors jointly supervised this work: Anna Wedell, Nigel G. Laing. 14 NATURE COMMUNICATIONS | (2019)10:1396 | https://doi.org/10.1038/s41467-019-09111-2 | www.nature.com/naturecommunications