Expanding the mutation and phenotype spectrum of MYH3-associated skeletal disorders

Pathogenic variants in MYH3 cause distal arthrogryposis type 2A and type 2B3 as well as contractures, pterygia and spondylocarpotarsal fusion syndromes types 1A and 1B. These disorders are ultra-rare and their natural course and phenotypic variability are not well described. In this study, we summarize the clinical features and genetic findings of 17 patients from 10 unrelated families with vertebral malformations caused by dominant or recessive pathogenic variants in MYH3. Twelve novel pathogenic variants in MYH3 (NM_002470.4) were identified: three of them were de novo or inherited in autosomal dominant way and nine were inherited in autosomal recessive way. The patients had vertebral segmentation anomalies accompanied with variable joint contractures, short stature and dysmorphic facial features. There was a significant phenotypic overlap between dominant and recessive MYH3-associated conditions regarding the degree of short stature as well as the number of vertebral fusions. All monoallelic variants caused significantly decreased SMAD3 phosphorylation, which is consistent with the previously proposed pathogenic mechanism of impaired canonical TGF-β signaling. Most of the biallelic variants were predicted to be protein-truncating, while one missense variant c.4244T>G,p.(Leu1415Arg), which was inherited in an autosomal recessive way, was found to alter the phosphorylation level of p38, suggesting an inhibition of the non-canonical pathway of TGF-β signaling. In conclusion, the identification of 12 novel pathogenic variants and overlapping phenotypes in 17 affected individuals from 10 unrelated families expands the mutation and phenotype spectrum of MYH3-associated skeletal disorders. We show that disturbances of canonical or non-canonical TGF-β signaling pathways are involved in pathogenesis of MYH3-associated skeletal fusion (MASF) syndrome.


INTRODUCTION
Myosin heavy chain 3 (MYH3) encodes the heavy chain of embryonic myosin, a muscle protein composed of a globular motor domain attached to a long coiled-coil rod domain by a short neck and a hinge region. Embryonic myosin exists as a dimer in which the tail domains are intertwined 1 . Hundreds of such dimers assemble with other proteins to form the sarcomere, the subcellular contractile apparatus of skeletal and cardiac muscle cells. MYH3 is highly expressed during embryonic and fetal development, from gestational week 6 to week 24, and continues to be expressed postnatally at both mRNA and protein levels in skeletal and heart muscle and at the mRNA level in several other tissues (www.proteinatlas.org) 2 .
Monoallelic pathogenic variants in MYH3 cause distal arthrogryposis (DA) syndromes, including arthrogryposis, distal, type 2A (DA2A, Freeman-Sheldon syndrome, MIM#193700) and arthrogryposis, distal, type 2B3 (DA2B3, Sheldon-Hall syndrome, MIM#618436) 3 . The phenotypes of DA2A and DA2B3 are variable but both are characterized by contractures of proximal and distal joints 4 . In addition to DA, heterozygous pathogenic variants in MYH3 have been reported in contractures, pterygia and spondylocarpotarsal fusion syndrome 1A (CPSFSIA, MIM#178110), which is characterized not only by joint contractures, but also by multiple vertebral, carpal and tarsal fusions 5,6 . It has been proposed that pathogenic variants in MYH3 inhibit TGF-β signaling, which is essential for the normal muscle function. The dysfunction of small muscles that attach at the distal neural arches of the spine is thought to lead to vertebral fusions 5 .
Recently, biallelic nonsense variants and splicing defects in MYH3 were reported in contractures, pterygia and spondylocarpotarsal fusion syndrome 1B (CPSFSIB, MIM#618469) 7 . In total there are 29 individuals reported with contractures, pterygia and spondylocarpotarsal fusion syndrome 1A and 10 individuals with 1B 5,7-13 . Thus very little is known regarding the phenotype and mutation spectrum of CPSFSIA and CPSFSIB as well as their natural clinical course.
In this study we extend the phenotypic spectrum of MYH3associated disorders reporting additional 17 affected individuals from 10 unrelated families with vertebral fusions, arthrogryposis and multiple pterygia. This report adds 12 novel pathogenic variants in MYH3 and provides experimental evidence supporting pathogenicity of the missense variants in our patients using functional studies in HEK293T cells.

Recruitment of patients with pathogenic MYH3 variants
We recruited ten families with congenital vertebral malformations and pathogenic variants in MYH3. Families A, B, D and E are of Chinese origin while families C, F, G, H, I and J are of European descent (Fig. 1). The phenotypic features, radiographic findings and the identified pathogenic variants are summarized in Figs. 1 and 2 and listed in Table 1. Information on the variant frequency, evolutionary conservation and in silico analyses is summarized in Table 2.
Autosomal dominantly inherited or de novo MYH3 variants and phenotype of the patients In families A, B and C, de novo or dominantly inherited heterozygous variants in MYH3 were identified. All these variants are novel.
Proband A-II:1 from family A is a 7-year-old Chinese boy born at GA 37 + 3. Skeletal survey showed multiple vertebral fusions, short neck with pterygium, increased distance between nipples, pectus excavatum and moderate short stature (Table 1) Fig. 3).
The proband from family B, B-II:1, is a 10-year-old male born at GA 38 + 4. He presented with torticollis at birth. Skeletal examinations showed congenital scoliosis, multiple vertebral fusions and moderate short stature (Table 1 and Fig. 2). The gross motor development of the proband is mildly delayed and cognitive development is normal. This patient has a de novo heterozygous variant in MYH3: c.1400A>C, p.(Glu467Ala), also located in the head domain of MYH3 (Table 1, Fig. 3).
The proband (IV-2) in family C, C-IV:2, is a 9-year-old Caucasian male, born at gestational week 38 + 3 to nonconsanguineous Swedish parents. He presented at birth with joint contractures and enlargement, popliteal pterygium, short neck, low set posteriorly rotated ears, which are different in size, pterygium colli, bilateral cryptorchidism and unilateral inguinal hernia. He developed scoliosis and spine radiographs revealed multiple vertebral fusion of the cervical, thoracic and lumbar spine (Fig. 2). Postnatally his height z-score decreased from −1.5 at birth to −2.9 at 6 years of age ( Supplementary Fig. 1). He has late speech development and is undergoing developmental evaluation for mild learning disabilities. Family C includes seven affected individuals, all with short stature and vertebral anomalies, but otherwise variable features, summarized in Table 1. All affected members in this family are heterozygous for an in-frame deletion in MYH3: c.3661_3663delGAG, p.(Glu1221del) in exon 27, which is not present in their unaffected relatives (n = 7). Exon 27 encodes the neck domain of MYH3 (Table 1, Fig. 3).

Fig. 2
Phenotypic and radiographic features of the patients with pathogenic MYH3 variants. Please note the following features: A-II:1 has small mouth with dowslanting corners, low-set ears, prominent nasal bridge and camptodactyly. C-IV:2 has small mouth, broad chest, increased distance between the nipples, enlarged knee joints with pterygia, flat feet and camptodactyly. D-II:1 has small mouth with downslanting corners, small low set ears, prominent philtrum and increased distance between the nipples. D-II:2 has small mouth with downslanting corners, small low set ears, prominent philtrum and ptosis. F-II:2 has broad chest, scoliosis, enlarged knee joints, elbow and knee contractures and bilateral clinodactyly of the 5th finger. G-II:1 has scoliosis and pterygia. H-II:2 has broad chest, scoliosis and mild pterygia of the knee. Radiographic features: A-II:1 has multiple cervical and thoracic vertebral fusions, rudimentary disc spaces and scoliosis. B-II:1 has multiple, mainly thoracic, vertebral fusions with rudimentary disc spaces, rib crowding and scoliosis. C-III:2 and C-III:4 both have moderate scoliosis and a few thoracic and lumbar/lumbosacral fusions. C-IV:2 has multiple posterior fusions of the cervical, thoracic and lumbar spine and scoliosis. D-II:1 has several thoracic and lumbosacral fusions with a distinct curve due to asymmetric T9 vertebral body. D-II:2 has some posterior vertebral fusions, mostly thoracic, without scoliosis. E-II:1 has multiple thoracic vertebral fusions, mild rib crowding and scoliosis. F-II:2 has multiple thoracic and lumbar vertebral fusions with rudimentary disc spaces, rib crowding and marked lumbosacral lordosis without scoliosis. G-II:1 has multiple thoracic vertebral fusions, rib crowding and scoliosis. H-II:2 has multiple thoracic and lumbar/lumbosacral posterior fusions, rib crowding and scoliosis. I-II:1 has multiple thoracic and lumbar vertebral fusions, rib crowding and scoliosis. Hand radiograms of I-II:1 at 5 and 8 years of age show development of lunotriquetral fusion. J-II:1 has multiple thoracic and lumbar vertebral fusions, rib crowding and scoliosis at 4 years 4 months of age. Hand radiogram shows lunotriquetral fusion.  T3-6, T7-8, T9 11, L4-S1 Self-reported fusion no radiograms available L4-S1 Self-reported fusion no radiograms available T4-5, T9-10, L4-S1    The proband from family E, E-II:1, presents with multiple thoracic vertebral fusions, joint contractures, downslanting palpebral fissures, ptosis, long nose, low-set ears and wide philtrum (Table  1 and Fig. 2, for this individual only spinal CT is available for publication). He is compound heterozygous for two variants in MYH3: c.3062delT, p.(Leu1021*); c.4244T>G, p.(Leu1415Arg) (Fig. 5).
The proband in family F, F-II:2 is a 7 years 5 months old Caucasian girl born at GA 41 + 1. At birth, she presented with short extremities, broad chest with wide intermammillary distance, downslanting palpebral fissures, short neck with mild pterygium and reduced extension of the knees and fingers. Her skeletal survey shows multiple segmentation anomalies, reduced vertebral body and disc height, abnormally angulated and crowded ribs, anteriorly angled sacrum and subtle mesomelia of the upper extremities (Fig. 2). She suffers from severe progressive postnatal growth failure ( Supplementary Fig. 1). She received continuous physical therapy from birth with improvement in mobility and has not needed any orthopedic surgical corrections. Cognitive development is normal and gross motor development mildly delayed. During her first years of life she experienced mild feeding problems, recurrent middle ear infections as well as recurrent episodes of obstructive bronchitis. She is compound heterozygous for two splicing variants in MYH3: c.-9+1G>A; c.5658+3_5658+6del (Fig. 5).
The proband in family G, G-II:1, was born to nonconsanguineous parents of Caucasian origin from a second IVF pregnancy. IVF was due to ovarian cysts in the mother. She was born at GA 39 + 2 with Apgar 8, 9 at 1 and 5 min. Newborn physical examination revealed bilateral single transverse palmar creases, sacral skin fold with slight hypertrichosis and left sided mild hip dysplasia. MRI at 2 weeks of age showed scoliosis at the thoracic portion of the spine, fused arches in vertebral blocks T2-T4, T6-T8 and T10-T12 and butterfly vertebrae at T10 and T12. Her psychomotor development is normal. She has short neck, 10°c ontractures of elbows and knees, micrognathia, left palpebral ptosis and hypoplastic alae nasi. At 28 months she had hypermetropia +3.5D. She is compound heterozygous for two splice variants: MYH3: c.-9+1G>A; c.1002+1G>A (Fig. 5).
The proband in family H, H-II:2, is a 4-year-old female, born to unrelated parents of Caucasian origin at GA of 40 + 6. Her family history is unremarkable except for mild scoliosis in her mother. Her parents noticed spine deformity at 3 months of age and she was diagnosed with multiple vertebral fusions at age of 3 years. She has scoliosis, a relatively wide thorax, contractures of the knees and hips, normal psychological development and presents with mild short stature at 4 years. The patient is compound heterozygous for variants MYH3: c.-9+1G>A and c.4000del, p.(Leu1334Cysfs*75) (data not shown).
The proband from family I, I-II:1, is an 18-year-old female, born to unrelated parents of Swedish origin at GA 41 + 5 after induction of delivery and vacuum extraction. Prenatal ultrasound examinations were normal. Apgar was 9, 10, 10 at 1, 5 and 10 min respectively. Delivery was complicated by a clavicular fracture. Short stature and scoliosis came to medical attention at 15 months. From 3 years of age, she suffered from a few episodes of painful, swollen knee joints. Juvenile idiopathic rheumatoid arthritis was suspected and the patient received various treatments, including methotrexate from 3 years and 8 months to 6 years of age. A mild 5-degree extension defect of one knee joint, some limitations/ stiffness in the flexion of the wrist joints and an inability to elevate the right shoulder beyond 160 degrees were noted. She was treated with growth hormone from 4 years and 8 months until 15 years of age. Her psychomotor development is normal. The patient has a short neck with pterygium, broad and relatively short chest, bilateral single palmar creases, scoliosis and extensive vertebral anomalies in the thoracic and lumbar region. The patient has a bicuspid aortic valve, which is also present in her father. Her father does not show any signs of skeletal dysplasia or joint contractures and we suspect that his heart malformation is due to another yet unidentified genetic reason. The patient is compound heterozygous for variants MYH3: c.-9+1G>A and c.5220_5223del, p.(Ser1740Argfs*2) (data not shown). The proband in family J, J-II:1, is a 21-year-old female born at term to non-consanguineous Swedish parents. During the first year a congenital scoliosis became evident and spine radiographs revealed multiple vertebral fusions (Fig. 2). Further radiologic examinations showed a tethered cord and a combined surgery for tethered cord release and vertical expandible prosthetic titanium rib was performed at 4 years of age. She has a broad thorax and mild dysmorphic features including slight facial asymmetry (Table 1). Limited extension of the knees, hips, elbows and fingers became more evident with age. Length at birth was normal (z-score +0.2) but she showed successive growth failure and her final height is at z-score of −2.9 ( Supplementary Fig. 1). During the first year she developed obstructive bronchitis that is still treated. Her motor and cognitive development is normal. She is compound heterozygous for two splicing variants in MYH3: c.-9+1G>A; c.4357-10G>A (Fig. 5).
Significant phenotype variability and overlap among the patients with autosomal dominant and autosomal recessive MYH3-associated diseases The affected individuals with dominant and recessive MYH3associated conditions show variable degrees of vertebral fusions, short stature and dysmorphic features. We have collected available growth data from our patients' medical records (n = 17) and calculated the height z-scores ( Supplementary  Fig. 1) using the WHO growth standards. For nine individuals in our study some retrospective growth data was available, for seven individuals we had only single height measurements and in one case no growth data was available. We depicted the height data from our patients comparing it with the height data from the patients reported in the literature 8,10-13 (n = 17, Supplementary  Fig. 1 and Supplementary Table 1). We find an overlap regarding the degrees of short stature in patients with autosomal dominant and recessive MYH3-associatedd diseases ( Supplementary Fig. 1). In addition, the number of vertebral fusions was overlapping in our patients (Table 1). Height z-scores in our cohorts ranged from 0 to −4.4 indicating that pathogenic variants in MYH3 lead to mild to moderate growth failure ( Supplementary Fig. 1).

Missense variants result in abnormal TGF-β signaling
As protein-truncating variants in MYH3 probably cause LoF of the protein, missense variants are more likely to have variable effects. Therefore, we examined the missense variants of unknown clinical significance found in this study using an over-expressing cell model. Using a MYH3-EGFP fusion plasmid we transfected constructs harboring the variants associated with the dominant disorder (c.841G>A, c.1400A>C, c.3661_3663del-GAG) and with the recessive disorder (c.4244T>G) into HEK 293T cells. We found that none of the variants affected the expression of MYH3, SMAD3, or the phosphorylation levels of Erk1/2 (Fig. 4a, b, c, f). Phosphorylation levels of SMAD3 were significantly decreased in cells transfected with the dominantly inherited or de novo variants as compared with cells transfected with control wild-type plasmid (Fig. 4e, P = 0.0013, P = 0.0051 and P = 0.0075, respectively), suggesting an inhibition of the canonical TGF-β signaling pathway. In contrast, we found a decrease in p38 phosphorylation in HEK 293T cells transfected with the recessively inherited variant (c.4244T>G), indicating an inhibition of the non-canonical TGF-β signaling (Fig. 4d, P = 0.0051). These results indicate that different variants in MYH3associated disorders may act through different mechanisms, i.e. interfere with the canonical or non-canonical TGF-β signaling, respectively. Abnormal splicing due to variants located outside the canonical splice sites To confirm the predicted abnormal splicing caused by the variants located outside canonical splice sites, we performed a minigene assay for c.1582-6A>G and c.3249-9_3249-5delTCTTC. The c.1582−6A>G variant was shown to insert a 5-bp intronic sequence before exon 16, causing a frameshift and a premature stop codon after 55 amino acids (Fig. 5a, b, c). In contrast, c.3249-9_3249-5delTCTTC resulted in partial retention of intron 25 (Fig. 5d, e, f).
Sanger sequencing of cDNA from the proband in family F and her healthy parents showed that her variant c.5658+3_5658 +6del led to a 9 bp intron retention between exon 39 and exon 40 (Fig. 5g, h). When applied to the in silico translation tool in SIB ExPASy Bioinformatics Resources Portal, this 9 bp inclusion is predicted to incorporate two additional amino acids followed by a termination codon as shown in Fig. 5h. Sanger sequencing of cDNA from the proband in family J, and her healthy parents showed that her variant c.4357-10G>A led to exon 32 skipping (Fig. 5i, j).
Notably, The c.-9+1G>A variant was detected in all Caucasian patients in trans with other splice or frameshift variants. This variant has been reported as pathogenic by previous functional studies 7 .

DISCUSSION
In this study, we report 12 novel pathogenic variants in MYH3, examine the functional consequences of disease associated variant alleles and summarize the clinical phenotypes of 17 affected individuals from 10 unrelated families. The affected individuals show variable manifestations of vertebral fusions, short stature and dysmorphic features. The clinical data indicate that there is a significant overlap in the degree of short stature as well as the number of fused vertebrae between our patients with autosomal dominant and autosomal recessive MYH3-associated disorders. These diagnoses are currently annotated as contractures, pterygia and spondylocarpotarsal fusion syndromes IA and IB (Supplementary Table 1, OMIM MIM#178110 and #618469). By noting this considerable clinical overlap, we suggest that these two Mendelian conditions represent variable expression of the same clinical entity and consequently should be termed MYH3associated skeletal fusion (MASF) syndrome, with annotation of two different inheritance patterns. Even if the number of patients in this report is limited, our results showing the broad phenotypic variability between and within the families indicate that future studies are unlikely to find consistent and clear-cut phenotypic differences between autosomal and recessive MASF syndromes.
Until the recent identification of MYH3 variants associated with contractures, pterygia and spondylocarpotarsal fusion syndromes 1A and 1B, most pathogenic variants in MYH3 were associated with DA2A (Freeman-Sheldon syndrome, MIM #193700) and DA2B3 (Sheldon-Hall syndrome, MIM #601680), which are both characterized by limb contractures without vertebral anomalies 2,3,14,15 as summarized in Supplementary Table 1. In some patients with DA2A, scoliosis and short stature have been described 2 , indicating presence of a phenotypic continuum between MYH3-associated syndromes with arthrogryposis with or without signs of skeletal fusions. DA2A and DA2B3 are associated with missense variants which are mostly located in the head and the neck domain of the myosin protein (Fig. 3). Consistent with the Cameron-Christie et al. study 7 of contractures, pterygia and spondylocarpotarsal fusion syndrome IB, our results show that biallelic variants in MYH3 causing recessive vertebral fusions are more likely LoF and might be protein-truncating or leading to nonsense mediated decay (NMD).
Cameron-Christie et al. 7 and our study identified the c.-9+1G>A allele as a recurrent variant in European patients with contractures, pterygia and spondylocarpotarsal fusion syndromes 1B (AR MASF). This variant has a minor allele frequency of 0.01 in the Finnish population, 0.002 in the non-Finnish European population and is observed once in the African population (frequency according to gnomAD: https://gnomad.broadinstitute.org/variant/17-10559406-C-T). However, we did not observe this variant in any patient from the Chinese cohort or the East Asian population from gnomAD. It is possible that the position of the MYH3:c.−9+1G>A variant was not covered by the exome kits used in ES of the Chinese cohort, but the absence of it in East Asian population in gnomAD indicates that it might be very rare. This variant was present in all our Caucasian patients with AR MASF in compound heterozygous state with other splice site or frameshift variants. The high MAF of the c.-9+1G>A variant in the Finnish population seem to be contradicting with the rare prevalence of MYH3-associated diseases. A similar phenomenon has been observed in TBX6associated congenital scoliosis, where a rare null TBX6 variant in compound heterozygosity with a common haplotype causes congenital vertebral malformation 16,17 . It is possible that the c.-9+1G>A in homozygous state does not lead to phenotypic abnormalities, since having that high population carrier frequency the condition would have been reported as a genetic skeletal disorder commonly occurring in Finland. Taken together, we suggest that the hypomorphic variants should be explored when a highly pathogenic variant exists in trans in an affected individuals with features of MASF syndrome, particularly in populations that have a relatively high carrier rate of this variant, e.g. Finnish population.
TGF-β signaling is considered to be a crucial pathway for development and function of skeletal muscle and bone, including osteogenic or myogenic differentiation and regeneration 18,19 . Specifically, it has been shown to control proliferation, differentiation, apoptosis, migration, extracellular matrix (ECM) remodeling, immune functions and tumor invasion/metastasis 20 . In this study, we show that variants causing AD or AR forms of MASF act through different mechanisms. While variants causing the AD form of disease perturb canonical TGF-β signaling, pathogenic variants involved in AR form of disease either lead to truncated transcripts or interfere with the non-canonical TGF-β signaling. The canonical TGF-β signaling directly phosphorylates Smad3 to initiate signal transduction through the canonical cascades via activation of type I and type II serine/threonine kinase receptors 21,22 . In contrast, the non-canonical pathway mediates the activation of the mitogen-activated protein kinases (MAPKs) extracellular signal-related kinase (ERK), c-Jun N-terminal kinase (JNK) and p38 23,24 . Both canonical and non-canonical TGF-β signaling pathways are associated with musculoskeletal differentiation, migration and ECM remodeling 20 . The perturbation of canonical TGF-β signaling may confer more deleterious defects than non-canonical TGF-β signaling since the heterozygous variant c.4244T>G, p.(Leu1415Arg), which was inherited autosomal recessively in this study, did not lead to congenital vertebral malformations in the carrier parent. Similarly, functional studies in the Cameron-Christie et al. study 7 demonstrated that the c.-9+1G>A variant resulted in a partial LoF of MYH3. Therefore, we hypothesize that the variants associated with the recessive forms of MASF are null (e.g., protein-truncating variants or variants leading to NMD) or hypomorphic (e.g., the c.-9+1G>A variant) in heterozygous state and therefore two affected alleles are necessary to lead to phenotypic abnormalities 25 . Further studies are needed to prove or discard this hypothesis, since the number of affected individuals in this report is relatively small.
In summary, we present both monoallelic and biallelic pathogenic variants in MYH3 associated with heterogeneous and clinically overlapping features in patients with congenital vertebral malformations, providing further evidence for impaired TGF-β signaling in pathogenesis of MYH3-associated skeletal fusion syndrome.

METHODS Patients
Patients with disease-causing variants in MYH3 were selected from a Chinese cohort of vertebral malformations under the framework of Deciphering disorders Involving Scoliosis and COmorbidities (DISCO) study (http://discostudy.org/) at Peking Union Medical College Hospital (PUMCH) (total cohort size n = 583) and from an European cohort of patients with congenital skeletal abnormalities studied at the Karolinska University Hospital (total cohort size n = 277). The study was approved by the Ethics Committee of Peking Union Medical College Hospital (JS-908) and by Regional Ethical Review Board for Karolinska University Hospital (protocol numbers are 2014/983-31/1 and 2012/2106-31/4). Informed consent was obtained from all affected patients or their parents/legal guardians. Clinical data were collected from the patients' records by their referring physicians. The z-scores of patients length/height were calculated using WHO Child growth standards (https://www.who.int/childgrowth/standards/en/) 26 .

Sequencing and in-house genomic databases
For the Chinese cohort, exome sequencing (ES) was performed as previously described 27 . For the Swedish cohort, clinical whole-genome sequencing (WGS) was performed as previously described including calling for SNVs, CNVs, and SVs 28 . As a part of a standard clinical genome analysis we look at CNVs and SVs in all our patients with suspected genetic syndromes 29 . For the Swedish cohort, we have also looked for clinically significant CNVs and SVs in the patients who underwent WGS, but no significant abnormalities were detected. DNA of the proband in family J, J-II:1, who underwent clinical ES in 2013, also was examined using aCGH as previously described 30 with normal result. MYH3 (NM_002470.4) variants were queried from the ES or WGS data and segregation analysis was performed in the families using Sanger sequencing. The variants were prioritized according to expected disease-causing potential, their presence in coding exons ±20 base pairs intronic sequence and a max minor allele frequency of 0.005 according to public (1000 G, dbSNP, ExAC/gnomAD, and ClinVar) and two country-specific in-house databases. An in-house database consisting of ES data from 4246 unrelated Chinese individuals without apparent scoliosis was utilized to determine the frequency of candidate variants in the Chinese Han population. The Swedish population reference database consists of whole genomes of 5015 individuals. The sample of patient J in the Swedish cohort was analysed using ES and conventional Sanger sequencing for the European hot spot variant MYH3: c.-9+1G>A, since it was not covered by ES (primers available at request).

MYH3 plasmid site-directed mutagenesis
The wild-type full-length cDNA of MYH3 amplified from the peripheral blood RNA using primer 5'TGGGAGGTCTATATAAGCAGAG3' (forward) and primer 5' CGTCGCCGTCCAGCTCGACCAG3' (reverse) and cloned into the XhoI and HindIII sites of the pEGFP expression vector (pcDNA3.1-EGFP). Inverse PCR-based site-directed mutagenesis (TOYOBO, Japan) was used to construct the MYH3 missense variants.

RT-PCR and cDNA sequencing
RT-PCR and RNA sequencing were performed in Swedish families F and J to explore whether their intronic variants (MYH3: c.5658+3_5658+6del and c.4357-10G>A) lead to abnormal mRNA splicing. RNA was extracted from white blood cells using the RNeasy Mini Kit (QIAGEN, Hilden, Germany). First-strand cDNA was synthesized with M-MLV Reverse Transcriptase (Life Technologies, Carlsbad, CA, USA). cDNA was amplified, examined by gel electrophoresis and Sanger sequenced according to standard protocols. The sequences of the novel transcripts were analyzed using the in silico translation tool in SIB ExPASy Bioinformatics Resources Portal 31 .

Minigene assay
The splicing variants in MYH3 were experimentally evaluated for potential functional impact on splicing using minigene plasmid pEGFP. The wildtype and mutant plasmids were transfected into HEK293T cells. After incubation for 48 h, total RNA was extracted using the RNeasy Mini Kit (Qiagen, Germany), and cDNA was obtained with the PrimeScript RT reagent kit (Takara, Japan) according to the manufacturer's instructions. RT-PCR products were examined by agarose gel electrophoresis and Sanger sequencing.

Patient consent
Written informed consent was obtained in accordance with protocols approved by ethics committee of Peking Union Medical College Hospital (IRB number: JS-908) and from the Regional Ethical Review Board (protocol numbers are 2014/983-31/1 and 2012/2106-31/4). Written consents to use the photographs and radiograms/CT images in this report were obtained from the adult patients and from parents/guardians of all children included in this study. The authors affirm that human research participants provided informed consent for publication of the images in Fig. 2.
Statistics SPSS Statistics V22.0 software was used for statistical analyses. Data were analyzed by Student's t-test; the results are shown as the mean ± standard error of n = 3 biological replicates. P ≤ 0.05 was considered statistically significant.

Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.