This article is mentioned in:

Introduction

Collagen VI is an important component of the extracellular matrix (ECM). It forms a microfibrillar network that is usually found to be in close association with the cell and surrounding basement membrane. Collagen VI is also found in the interstitial space of numerous tissues, including muscles, tendons, skin, cartilage and intervertebral discs (1). Collagen VI is encoded by six different genes, namely collagen α-1 (VI) chain (COL6A1), COL6A2, COL6A3, COL6A4, COL6A5 and COL6A6. Collagen VI-related myopathy (COL6-RD) is caused by pathogenic variants in three collagen VI genes: COL6A1, COL6A2 and COL6A3. These mutations can result in two main types of muscle disorders that can range from severe forms of Ullrich congenital muscular dystrophy (UCMD) to milder forms of Bethlem myopathy (BM) (2).

A child with typical UCMD cannot walk independently, or can only do so for short periods of time. These patients also present with changes to their skin and joints, including variable proximal contractures and distal hyperlaxity, congenital hip dislocations, follicular hyperkeratosis over some of the extensor surfaces of the limbs and soft velvety skin on their palms and soles of their hands and feet. In addition, there is usually abnormal wound healing, which can result in the formation of keloid scars. BM is a relatively mild disorder, and can be characterized by some proximal muscle weakness and contractures, with the onset of these symptoms occurring within the first two decades of life (2). These two major phenotypes are not distinct and an intermediate phenotype, described as mild UCMD with severe BM, has been defined by number of previous studies (2,3). BM is transmitted as an autosomal dominant or autosomal recessive inherited disorder and is usually associated with mutations in the COL6 genes (4,5). The prevalence of BM in the northern region of England in 2007 was 0.77/10,000 (6).

The distribution of the mutations in the COL6 genes is somewhat uniform, and lacks mutation hot spots. The most common mutations in COL6-related myopathy are splicing and missense mutations (3,7). As a consequence of this notable heterogeneity at the clinical and molecular levels in the COL6 genes, it appears to be difficult to establish a correlation between the phenotype and the genotype of this disease. In a previous study, dominant splice-site mutations that cause exon skipping were observed in cases classed as moderate to progressive (8).

Two common splice sites, the canonical splice donor GT and acceptor AG, are present in 98.71% of all mammalian genes (9). Mutations in these canonical dinucleotides always result in splicing errors when pre-mRNA transcripts are transformed into mature mRNAs (10). It was previously reported that sequence variations surrounding certain canonical splice sites may not result in splicing errors because these exon and intron elements are only weakly conserved, and their alterations do not always disrupt processes related to splicing (11). Therefore, it is essential to determine the presence of aberrant mRNAs in order to predict the clinical phenotypes of diseases. RNA-sequencing uses high-throughput sequencing technology to rapidly and comprehensively obtain the transcriptional status of biological samples at a specific time (12). In the present study, RNA-sequencing was successfully used to analyze the influence of splice mutations.

The current study reported three patients from a family who presented with typical clinical features of BM. A novel heterozygous COL6A2 mutation (c.736-1G>C) was identified in this family. The mechanism of this mutation has not been previously reported in BM, to the best of our knowledge. Adiagnosis of BM was made by observation of the clinical characteristics and MRI features of the patients. Bioinformatics, in silico analysis, RNA-sequencing, reverse transcription (RT)-PCR and immunocytochemistry were performed to confirm the pathogenicity and mechanism of the mutation. The present findings may help improve the accurate molecular diagnosis of the disorder, and illustrated a previously unrecognized category of BM.

Materials and methods

Patients and their families

This study was from January 2015 to August 2018. The clinical data of the affected father (proband) and two sons were collected at The First Affiliated Hospital of Guangxi Medical University (Nanning, China). Three male patients age ranged from 40 to 14 years. The median age was 24 years. The family had nine members, the parents, two sisters and a brother of the proband, his wife and two sons. Information such as symptoms at onset and creatine kinase levels were determined by means of a commercially available Creatine Kinase reagent kit (cat. no. 23-666-208; Thermo Fisher Scientific, Inc.) via rate-assay spectrophotometry according to the manufacturer's instructions, as well as ultrasonic cardiogram results, were obtained. Muscle MRI was performed using the thighs of the three patients and a normal control (the grandfather). The MRI results were assessed by two independent and experienced radiologists and a neurologist. The right gastrocnemius was biopsied in the father. Routine periodic acid-Schiff (PAS) staining was performed on muscle cryostat sections to investigate the presence of normal and abnormal glycogen levels.

Biological materials

All the biological materials, including peripheral blood samples obtained from the cubital vein, skin biopsies, muscle biopsies as well as fibroblasts for cell culture, were collected after the appropriate informed consent was obtained. The use of all appropriate human tissues was approved by The Medical Ethics Committee of the First Affiliated Hospital of Guangxi Medical University (Nanning, China; approval no. 2017-KY-E-154) and the experimental protocols were performed in accordance with the institutional guidelines and regulations. All nine family members donated venous blood samples Genomic DNA was extracted with Qiagen FlexiGene DNA kit (cat. no. 51206; Qiagen China Co., Ltd.) according to the manufacturer's protocol. A muscle biopsy was obtained from the proband and skin biopsies were obtained from the inner side of the thigh in the proband and a son. Normal skin and muscle tissues were collected from two male patients (median age 24 years) receiving skin or muscle resections for non-muscular disease-related reasons, including circumcisions. Fibroblasts were cultured from the skin biopsies and were used as primary cultures.

DNA analysis

To systematically search for the disease-causing genes, a myopathy panel based on next-generation sequencing was performed on the proband. The myopathy panel (Table SI) was a kit designed by the Zhongguancun Huakang Gene Institute (http://www.kangso.net/) and produced using the Agilent SureSelect Target Enrichment technique (13).

High-throughput single-end sequencing was performed on an Illumina NextSeq500 platform (Illumina, Inc.). In order to construct the exome library, 4 µg genomic DNA from peripheral blood samples of each patient was sheared into fragments of 150-250 bp in length by sonication (30 sec on/30 sec off, 0°C for 5 min) and subsequently hybridized for enrichment, using the manufacturer's standard protocol (SureSelect Target Enrichment System Target Sequence Enrichment kit, cat. no. 5190-5931; Agilent). Indexing primers with 8 bp index A01-H12 were used to amplify the captured library, and PCR was performed to amplify the SureSelect-enriched DNA library. The amplified captured library was purified using AMPure XP magnetic beads. To evaluate the DNA quality and quantity, it was necessary to use the 2100 Bioanalyzer instrument (Agilent) and a High Sensitivity DNA kit (Agilent) according to the manufacturer's protocols. The final concentration of the final library was >1.52×10−6 mmol/l. After obtaining the human reference genome from the University of California, Santa Cruz database (build 37.1, version hg19, http://genome.ucsc.edu/), sequence alignment was performed using the Burrows-Wheeler Alignment tool (14). Picard (http://sourceforge.net/projects/picard/) was then employed to mark duplicates resulting from PCR amplification. All candidate mutations were subsequently filtered against several programs, including the single nucleotide polymorphism database (http://www.ncbi.nlm.nih.gov/projects/SNP/snp_summary.cgi), the international HapMap Project (http://hapmap.ncbi.nlm.nih.gov/) and the 1,000 genomes project (2012 April release, http://www.1000genomes.org/) in order to remove loci polymorphisms. Polymorphism phenotyping version 2 (PolyPhen-2) was used for sorting intolerant from tolerant and mutation tasters to predict whether an amino acid substitution could affect the functions of the protein. Next, Sanger sequencing was used to validate whether the identified potential disease-causing variant occurred in the individual family members (the proband's parents, brother, two sisters, two sons and his wife).

In silico analysis

To evaluate the probability of splice site mutations being able to disrupt the normal splicing of COL6A2, two online bioinformatics tools were used for prediction studies: Human Splicer Finder (http://www.umd.be/HSF/) and Alternative Splice Site Predictor (http://wangcomputing.com/assp/index.html). ExPASy (https://www.expasy.org/)was used to predict their impact on the protein that they would produce.

RNA-sequencing analysis

The effect of the splicing mutation on COL6A2 mRNA was determined by RNA-sequencing analysis. Tissue from the muscle of the proband and a non-muscle disease control were selected to extract total RNA for RNA-sequencing analysis. Construction of the sequencing libraries followed the standard protocols of the TruSeq™ RNA Sample Preparation guide (Illumina, Inc.). TruSeq® Stranded mRNA Library Prep kit (Illumina, Inc.) was used to prepare the sequencing libraries. The poly-A containing mRNA molecules are purified from total RNA using poly-T oligo-attached magnetic beads. During the elution of the poly-A RNA, the RNA is also fragmented to 120-200-bp and primed for cDNA synthesis. First-strand cDNA was then synthesized with SuperScript II Reverse Transcriptase (Invitrogen; Thermo Fisher Scientific, Inc.) using random primers. The temperature protocol was follows: 25°C For 10 min, 42°C for 15 min and 70°C for 15 min. AMPureXP beads (Beckman Coulter, Inc.) were then used to isolate double-stranded cDNA, which was synthesized using Second Strand Master mix (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. The adapters were then ligated to the A-tailing fragment. Subsequently, 12 cycles of PCR were performed to enrich those DNA fragments that had adapter molecules at both ends following the program: 98°C For 30 sec; 15 cycles of 98°C for 10 sec, 60°C for 30 sec, 72°C for 30 sec and 72°C for 5 min. This technique also amplified the quantity of DNA in the library. Purified libraries were further quantified using a Qubit2.0 fluorometer (Invitrogen; Thermo Fisher Scientific, Inc.) and then validated using an Agilent 2100 bioanalyzer (Agilent Technologies, Inc.). Clusters were then generated using cBot with the libraries, which were further diluted to 10 pM and were subsequently sequenced using the Illumina HiSeq x Ten system (Illumina, Inc.) for 150 cycles. Shanghai Sangong Biotech Corporation (https://www.sangon.com/) was employed to construct the libraries, and Illumina sequencing was performed for the high-quality reads, which were kept for sequence analysis after passing through Illumina quality filters. The sequencing data sets are stored at http://bioinformatics.jnu.edu.cn/software/sequencing_datasets/ and Gene Expression Omnibus database (accession number, GSE42006) for reference. Integrative Genomics Viewer (version 2.8.0) was used to display the genomic data.

Skin fibroblast cultures

To extract total RNA for transcript analysis, skin biopsies from two patients (proband and his elder son) and normal controls were immersed into Dulbecco's modified Eagle's medium (cat. no. 22390; Gibco; Thermo Fisher Scientific, Inc.) supplemented with 20% heat-inactivated fetal bovine serum (cat. no. A11-043; PAA Laboratories GmbH; GE Healthcare) and penicillin-streptomycin (cat. no. 15070-063; Gibco; Thermo Fisher Scientific, Inc.). Fibroblasts from skin biopsies were cultured in Dulbecco's modified Eagle's medium containing 20% fetal bovine serum (Sigma-Aldrich; Merck KGaA) and penicillin-streptomycin (Sigma-Aldrich; Merck KGaA) at 37°C in an atmosphere of 95% air and 5% CO2 (15).

Transcript analysis

In order to validate the results of RNA-sequencing analysis and the effect of the splice site mutation, transcript analysis was performed using RT-PCR with RNA from fibroblasts derived from two patients (proband and his elder son) and two healthy individuals as controls. TRIzol® (0.5 ml; Invitrogen; Thermo Fisher Scientific, Inc.) was used to extract total RNA using the standard manufacturer's instructions.

In order to reveal the COL6A2 cryptic splice acceptor, RT-PCR was performed with a forward primer within the 3′ of exon 3 and a reverse primer within exon 5. The primer pairs were as follows: Forward, 5′-CTC TAC CGC AAC GAC TAC GC-3′ and reverse, 5′-TTC CAG GCA GCT CAC CTT-3′. RT-PCR, revealing COL6A2 exon skipping, was performed using a forward primer within exon 3 and a reverse primer within exon 6. The primer pairs were as follows: Forward, 5′-CCC AAC CAG AAC CTG AAG GA-3′ and reverse, 5′-TCG GCT CCA AAT TCA CCC T-3′. PCR was conducted with Premix Taq™ DNA Polymerase (Takara Biotechnology Co., Ltd.) under the following conditions: Initial denaturation at 95°C for 4 min, followed by 36 cycles at 95°C for 30 sec, 56°C for 30 sec and 72°C for 40 sec. PCR products were separated on 2% agarose gels and extracted using a gel purification kit (Favorgen Biotech Corp.) according to the manufacturer's protocols. The eluted products were sequenced using an ABI 3730 XL Genetic Analyser with BigDye™ Terminator Cycle Sequencing kit (Applied Biosystems; Thermo Fisher Scientific, Inc.). The COL6A2 sequence of the patients was compared with that observed in the control and reference sequences of COL6A2.

Immunohistochemistry of muscle biopsies and immunocytochemistry of skin fibroblasts

The morphology was observed under a microscope via hematoxylin and eosin (H&E) staining. Fresh specimens were fixed in 10% neutral buffered formalin and further processed into paraffin-embedded blocks, sectioned at 6 -µm thickness, dehydrated in graded ethanol (70, 80, 90, 95 and 100%) solutions and processed according to the H&E staining protocol (16). The stained tissues were observed using a biomicroscopy imaging system microscope (cat. no. DP73; Olympus Corporation). Samples derived from skin fibroblast cultures and muscle biopsies were subsequently prepared for immunohistochemistry to evaluate collagen VI protein expression. Briefly, sections were incubated at 4°C overnight with the primary antibody which was a monoclonal antibody raised in rabbits (clone #EPR17072) that targeted collagen VI (cat. no. ab182744; 1:100; Abcam), and then incubated at room temperature for 30 min with Supervision™ Universal (Anti-Mouse/Rabbit) Detection reagent (HRP) (cat. no. D-3004; Thermo Fisher Scientific, Inc.) conjugated to peroxidase in Tris-HCI buffer containing carrier protein and anti-microbial agent. The experiment was performed according to the manufacturer's protocol.

Results

Three patients in a family display typical features of BM

The three patients in this study were from three generations of a Chinese family (Fig. 1). The proband (II-4) was 40 years old. From 4 years of age, he experienced difficulty getting up from the floor and climbing stairs; at 8 years of age, he began toe walking and needed to use a rail to climb stairs. He underwent Achilles tendon release aged 8 years. He had mild weakness in the shoulder girdle and upper limb muscles, and moderate weakness in the axial and hip girdle muscles. At the age of 40, he was able to walk unassisted but with a marked Trendelen burg gait. Physical examination revealed proximal muscle weakness and amyotrophy. Joint contractures of the shoulders, hips, knees and feet were observed (Fig. 2A). Skin examination revealed that he had keratosis pilaris along the extensor surfaces of his arms and legs.

Figure 1 Pedigree of the family with Bethlem myopathy.

Figure 2 Typical clinical features of the family with BM. Typical finger flexor contractures in (A) II-4 and (B) III-1 with BM made it almost impossible to put the fingers close together with the elbows extended. (C) The younger son could still put his fingers close together. Three patients shared similar clinical symptoms. (D) Skin examination revealed scar and (E) and keratosis pilaris in proband. BM, Bethlem myopathy.

Two sons (III-1 and III-2) of the proband shared similar clinical symptoms. The father was the most severe case, the younger brother the mildest. Examination in our hospital occurred when the older boy was 18 years old. He could not get up from the floor. The younger 14-year-old brother experienced difficulty when getting up from the floor and climbing stairs, and he walked on his toes (Table I). Both underwent Achilles tendon release 10 years of age.

Table I Clinical data and biochemical results of three patients in the family.

Table I

Clinical data and biochemical results of three patients in the family.

Variable	Patient
	II:4	III:1	III:2
Age, years	40	18	14
Approximate age at which muscle weakness was diagnosed, years	4	4	5
Capable of climbing stairs	No	No	No
Capable of walking 100 m	Yes	Yes	Yes
Capable of jumping up and down for 1 min	No	No	No
Capable of standing for 10 min	No	No	No
Skin appearance	Some scarring	Keratosis pilaris	Normal
Rigidity of spine	No	No	No
Plasma creatine kinase levels, U/l	576	747	759
Ultrasonic cardiogram	Normal	Normal	Normal

To further demonstrate the patients' pattern of muscle involvement and to determine the potential relevance of these unknown genetic variants, muscle MRI for the lower extremities was recommended. This revealed a distinct pattern of fatty infiltration in both lower extremities, and the anterior group muscles were more impaired than the rear groups in the thigh (Fig. 3A-C). This is consistent with a pattern of symptoms termed 'central shadow' in the rectus femoris and a pattern further described as 'outside-in' in the vastus lateralis muscle. Fatty infiltration of the thigh muscles varied greatly among the three patients. The father (Fig. 3A) was the most severe case, while the youngest son (Fig. 3C) exhibited the mildest symptoms.

Figure 3 Muscle MRI with axial cuts performed through the mid-thigh region of (A) II:4, (B) III:1 and (C) III:2 and (D) the grandfather, I:1, with normal muscle with a low-signal intensity and high-signal intensity fat intensity. Arrow 1 indicates vastus lateralis muscle; it reveals abnormal signaling along the periphery of the vastus lateralis muscle ('outside-in' pattern). Arrow 2 indicates rectus femoris; it reveals abnormal signaling in the central region of the rectus femoris muscle ('central shadow' pattern). Arrow 3 indicates sartorius. Arrow 4 indicates gracilis muscles. L, left; R, right.

H&E staining of the gastrocnemius muscle of the proband showed mild dystrophic features, such as muscle degeneration and necrotic fibers. In addition, connective tissue hyperplasia was observed in the muscle fibers and these were unequal in size and had certain structural distortion (Fig. 4A). Abnormalities of glycogen deposition were not observed when tissue sections of muscle were stained with PAS staining (Fig. 4B).

Figure 4 H&E and PAS staining of skew muscle in patient. (A) Patient muscle biopsy H&E staining reveals variations in the size of fibers (green arrow), evidence of degeneration and connective tissue hyperplasia (red arrow). (B) PAS staining showed no abnormalities of glycogen deposition. Glycogen was expressed normally in the cytoplasm. Magnification, ×10. PAS, periodic acid-Schiff; H&E, hematoxylin and eosin.

Identification of a heterozygous splicing mutation in the COL6A2 gene

To make a genetic diagnosis, 2.84 million 75-bp single-end sequence reads were generated from the proband. Of these, 2.83 million reads (99.93%) aligned to the human reference genome. Of the 169 selected genes (Table SI), a heterozygous splicing mutation (c.736-1G>C) in the COL6A2 gene was identified in the proband and Sanger sequencing validated that the same mutation existed in the two sons (Fig. 5A), while the spouse (II-5), proband's siblings (II-1, 2 and 3), father (I-1) and mother (I-2) were normal (Fig. 5B). The members who had the mutation in the COL6A2 gene had a similar phenotype, while those without this mutation did not have muscle weakness or motor disability. This splice acceptor site mutation observed in intron 4 was co-segregated within the family.

Figure 5 Sequence analysis of the Bethlem myopathy family. (A) Sequence analysis of heterozygous mutation c.736-1G>C was found in proband (II:4), two sons (III:1 and III:2). (B) Sequence analysis of the wife (II:5), proband's siblings (II-1, 2 and 3) and his father (I-1) and mother (I-2). (C and D) Prediction of the change of protein assessed by ExPASy. (C) A new cryptic splice acceptor site activation resulted in a frameshift in 246 amino acid and led to a premature stop codon after 264 amino acids (green arrow). (D) Exon 5 skipping resulted in the deletion of amino acids between positions 246 to 267 (blue arrow).

Splicing mutations can result in aberrant splicing

In silico analysis

In order to determine the effects of the mutation on the splicing of COL6A2 mRNA, in silico and in vitro analyses, which involved RNA-sequencing analysis and RT-PCR, was performed.

Human Splicer Finder and Alternative Splice Site Predictor, which are two freely available online bioinformatics tools, were used to predict the effect of the c.736-1G>C variation on COL6A2 mRNA splicing. Human Splicer Finder shown that wild-type sequencing existed a normal splice acceptor sites of 3′ end; while the normal acceptor sites disappeared and replaced by a new one in the mutant-type. The same result was found by Alternative Splice Site Predictor. Wild-type sequencing existed a splice acceptor sites of 3′ end; while the acceptor sites disappeared in the mutant-type. These tools suggested that this mutation would probably result in the elimination of any wild-type splice acceptor sites that would create a new cryptic splice acceptor site. Thus, a 129 bp sequence at the 3′ end of intron 4 was retained (data not shown).

Prediction of influence on the protein

To understand the influence of the two abnormal splicing variants, ExPASy was used to predict their impact on the protein that they would produce. The new cryptic splice acceptor site obtained suggested that the loss of the wild-type splice site could result in a frame-shift in amino acid 246, which would eventually lead to a premature stop codon in amino acid 264 (p.Cys246fs264*) of the protein. The skipping of exon 5 appeared to result in the deletion of 21 amino acids, namely from 246 to 267, in the protein (p.Cys246_Lys267del) (Fig. 5C and D). The truncated protein lacked the triple-helical domain (THD) region.

RNA-sequencing analysis

Integrative Genomics Viewer was used to display the genomic data. In addition to the normal base sequence, two abnormal splicing variants adjacent to the mutation site were identified in the COL6A2 gene. One variant had a 141-bp sequence retained at the 3′ end of intron 4, causing a new cryptic splice acceptor site. Another variant was due to exon 5 skipping. The percentage of presence of the normal, 3′ end of intron 4 retention and exon 5 skipping variants was 50, 36 and 14%, respectively (Table II). The coverage (reads) of the patient was lower than compared with that of the control. Differential gene expression analysis revealed that the expression of COL6A2 in the patient was ~5-fold higher compared with that of the normal control.

Table II Coverage of splicing variants in proband and normal control.

Table II

Coverage of splicing variants in proband and normal control.

Splicing variant	Patient		Normal control
	Reads	Percentage, %	Reads	Percentage, %
Wild-type transcripts sequence	18	50	4,846	99.5
Intron 4 retention	13	36	18	0.4
Exon 5 skipping	5	14	7	0.1

RT-PCR

To validate the results demonstrated by RNA-sequencing analysis, RNA derived from cultured skin fibroblasts from the proband and his elder son, and two healthy controls were used to perform RT-PCR experiments. The difference between the wild-type transcripts and the mutant transcripts in patients was that the wild-type transcripts were longer (which was caused by a cryptic splice acceptor site activation) or shorter (which was caused by exon skipping) compared with the mutant transcripts when identical primers were used to amplify the PCR products that flanked the region of mutation. Therefore, this method can serve as a practical tool to distinguish the two transcripts by when the samples are subjected to gel electrophoresis. The amplification of exon 3 to exon 5 in COL6A2 generated a 130-bp product for all the cDNA samples from the controls and patients. Notably, an additional larger product of 260 bp was only detected in the cDNA samples obtained from the patients (Fig. 6A). The RT-PCR products were subjected to gel extraction and Sanger sequencing, which showed that a 129-bp sequence from the 3′ end of intron 4 was retained between exon 4 and exon 5. This result was consistent with the in silico analysis. RT-PCR amplification of exon 3 to exon 6 also produced a 371-bp product when using cDNA samples from either the control individuals or the patients as templates. However, a smaller product of 305 bp was only detected in the patients' samples (Fig. 6B). Gel extraction and Sanger sequencing also revealed that exon 4 directly spliced the in-frame to exon 6 and hence skiped exon 5.

Figure 6 Gel electrophoresis of RT-PCR product of RNA isolated from fibroblasts of patients and normal control. (A) Products amplified with primers intron retention, with a larger PCR product (260 bp) with part of intron 4 found in patients. (B) Products were amplified with primers to analyze exon skipping. It demonstrates a smaller PCR product (305 bp) lacking exon 5 found in patients. M, marker; P1, patient (II:4) fibroblasts cDNA; P2, patient (III:1) fibroblasts cDNA; N1, normal control 1 fibroblasts cDNA; N2, normal control 1 fibroblasts cDNA; negative, negative control; blue indicates exon 4; green square indicates exon 5; red oval indicates intron 4.

Collagen VI is absent in fibroblasts but is expressed normally in muscle

To further investigate the consequences of the c.736-1G>C mutation on the protein product, immunohistochemical analysis was performed using a rabbit monoclonal antibody against collagen VI (cat. no. EPR17072). Collagen VI immunohistochemistry indicated that this protein was absent in the skin fibroblasts derived from patients but showed a normal pattern in the fibroblasts derived from the controls. In contrast, a normal pattern was found in the patient muscle fibers, attesting to the integrity of the basement membrane (Fig. 7C and D). Fig. 7C and D do not have the same background with stain. The difference between the two figures may be due to non-specific factors, for example the degree of pathological staining.

Figure 7 COL6 immunohistochemistry of patient and normal control. (A) Immunohistochemistry revealed COL6 was normally expressed in patient's muscle and (B) in normal control. Magnification, ×100. (C) COL6 was absent in patient's skin fibroblast, but (D) normally expressed in normal control. Magnification, ×40. Red arrow indicates COL6 protein expressed in the intercellular substance. Green arrow indicates no COL6 protein expressed. COL6, collagen VI.

Discussion

BM is a well-defined collagen VI disorder. BM has been associated with heterozygous, dominant and recessive mutations in one of the COL6A1, COL6A2 or COL6A3 genes, and is either inherited from an affected parent or occurs de novo (5). The present study was able to identify a heterozygous mutation (c.736-1G>C) in a splice-site of the COL6A2 gene within a Chinese family. The pathogenicity of this mutation has not been reported previously, to the best of our knowledge. Three patients in the family shared similar classical BM presentation and disease progression, presenting with an early onset, slow progression and a relatively mild disease. The severity of the clinical symptoms had a tendency to worsen with advancing age.

The majority of patients with COL6-RD exhibit a pathogenomic pattern in muscle MRI scans, and this is particularly noticed in patients with certain degree of muscle weakness (17). Patients with BM can exhibit patterns that were termed 'central shadow' in the rectus femoris and 'outside-in' in the vastus lateralis muscle in previous reports (2). In the present study, the presentation of muscle MRI scans were similar to the previous study; fatty infiltration of the thigh muscles varied greatly among the three patients studied. The degree of fatty infiltration observed in the MRI scans was associated with the severity of the clinical symptoms. Thigh muscle MRI scans can also be used to assess the progression of the disease in patients with BM (18).

To confirm the clinical findings, a myopathy panel was performed on the proband. A heterozygous mutation (c.736-1G>C) was identified in a splice-site of the COL6A2 gene. This mutation has not been previously reported in previous literature, to the best of our knowledge. The splice acceptor site mutations at positions 1 and 2 are considered pathogenic In silico analysis predicted that c.736-1G>C, which is a splice site mutation, likely damages the splice acceptor site. This evidence supports the likely pathogenicity of the novel splice-site mutation.

To identify the aberrant splice variant caused by the splice-site mutation, RNA-sequencing was used to analyze the muscle obtained from the proband. The results revealed a 129-bp retention of the 3′ end of intron 4 and skipping of exon 5. Finally, RT-PCR was performed to validate the results of RNA-sequencing. The splice site mutation (c.736-1G>C) destroyed the normal splice acceptor site, which led to two main consequences. Firstly, the activation of a new cryptic splice acceptor site in intron 4 caused a frameshift and ultimately resulted in premature termination of the protein. Secondly, there was a skipped sequence of 66 bp on exon 5, which introduced a deletion comprising of 22 amino acids. However, this resulted in no alteration of the open reading frame of COL6A2. As reported in a previous study, the mutant mRNA, exon skipping or the retention of the intron in the mRNA was detected, which was similar to the mutant mRNA in our study. It formed was degraded by the nonsense mediated decay (NMD) pathway (19). Even if the present protein escaped the NMD pathway, the truncated protein lacked the triple-helical domain (THD) region. Hence, these variants may result in aberrant splicing and affect the structure of collagen VI.

Differential gene expression analysis revealed that the expression of COL6A2 in the patient was ~5-fold higher compared with that of the normal control. The low level of COL6A2 mRNA most likely resulted from NMD in one allele. In a previous report, nonsense-mediated decay was found in patients with splicing mutations. Lucarini et al (20) reported a patient with UCMD and a homozygous A to G transition at position -10 of intron 12. Northern blot hybridization reported that the COL6A2 mRNA level in the patient was substantially reduced compared with that in the normal control. The data suggested that the majority of the COL6A2 mRNA in the patient's fibroblasts was degraded, most likely through nonsense-mediated mRNA decay. This possibility was further investigated. The patient's fibroblasts were treated with cycloheximide to inhibit RNA degradation. Next, RT-PCR analysis of total RNA prepared from the patient's fibroblasts indicated that aberrant splicing had occurred. Zhang et al (21) reported the existence of three patients from two families with UCMD resulting from recessive mutations of the COL6A2 gene. A splice acceptor site mutation in intron 17 was found in two of the patients, causing NMD in patients and in their unaffected parents. Fibroblasts from the patients expressed low levels of COL6A2 mRNA. In the parents, the COL6A2 mRNA levels were reduced to 57-73% of those of the control. In the present study, the splicing mutation led to a truncated protein. To avoid the accumulation of the truncated protein, NMD may decrease the level of aberrant mRNA. This might be the reason why the patients in the present study had a mild phenotype.

In order to identify the consequence of the mutation at the protein level in the current study, immunohistochemical analyses of the patients' muscles and fibroblasts were performed. Collagen VI was expressed differently between these two samples. This finding is consistent with those of previous studies. In patients with BM, muscle biopsies immunostained with various anti-collagen VI antibodies tended to be positive and therefore could not be used for diagnostic purposes (22). This phenomenon also occurs in other muscular dystrophies. For example, in patients with limb-girdle muscular dystrophy1B, lamin A/C labeling of muscle tissue is a common event (23). The cause of this phenomenon is unclear. It can be hypothesized that the protein expression is normal but structure and function are impaired. Further research is needed to confirm this hypothesis. However, further tests, such as western blotting, on the samples obtained from the patients, were not possible in our study, due to the limited quantity of sample available.

COL6A2 encodes the α2 chain of type VI collagen and comprises a signaling peptide domain and three different von Willebrand factor type A domains. These are typically protein binding domains that have been shown to bind ECM proteins as well as several collagen units (1). In the patients with BM in the present study, the splice-site mutation (c.736-1G>C) was located within the α2 THD region. The majority of mutations in the THD region reported previously were mostly associated with UCMD (24), and their phenotypes were different from those of the current patients. It is conceivable that non-penetrance and incomplete expression could be the causes for this difference. Therefore, the correlation between phenotype and genotype is not clear in patients with COL6-RD. Zhang et al (25) reported a mutation (c.736-1G>A) at the same position in a patient with UCMD, who acquired ambulation after a delay of 16 months. Subsequently, he lost ambulation at the age of 8 years old. Hip dislocation, torticollis, extended talipes and skin changes were also present in this patient.

In contrast, the patients in the current study presented only with mild symptoms. The proband patient was 40 years old and still walking independently. It was hypothesized that the fact that certain transcripts 'escape' exon 5 skipping (by retaining intron 4 instead) contributes to reducing the load of the dominant-negative protein isoform, whereas the c.736-1G>A mutation primarily produces the exon 5-skipped variant, which could explain the reportedly severe phenotype. Finally, it should be noted that similar types of mutations or mutations within the same domains of the protein do not necessarily result in comparable phenotypic severity of the disease, since these have been identified in patients with differing clinical severity by the present group and other researchers (26).

The interpretation of numerous splicing variants of unknown biological and clinical significance found in general genetic screenings represents a major challenge when attempting to assess the molecular diagnosis of genetic diseases (12). A fraction of splicing mutations may be deleterious because they may eventually affect mRNA splicing (27). The majority of online splice prediction programs tend to offer tools that predict putative 5′-splice donor and 3′-splice acceptor sites (including several cryptic splice sites). Additionally, these will give probability scores or potential alterations of the splicing, including Splice-Port, ESEfinder, BDGP Spice Site Prediction, Human Splicing Finder and Alternative Splice Site Predictor (28). However, these programs do not offer direct predictions of functional consequences of splice-site mutations with regards to exon skipping, exon deletion or intron retention (10). Researchers often have to make their own predictions based on the calculated probability scores. Some splice-site mutations can result in both exon skipping and partial exon deletion, which is occasionally due to activation of certain cryptic splice sites within the exon (29). Therefore, it appears to be unwise to predict the eventual functional consequences of these alterations in the genomic sequence. Therefore, assays for detecting mutant mRNA splicing isoforms are needed and can determine the mechanism of mutation at a splice site.

According to the literature data, RT-PCR analysis (29) of patient mRNA and transfection of mini-gene constructs (30) expressing the mutated exons in cell lines are two of the most widely used methods for validating mutant mRNA splicing isoforms. RT-PCR and mini-gene assays can probe the specific variant in question but do not reveal the relative abundance of each of the different isoforms. In addition, mini-gene construct expression only reveals the molecular consequence of a splice-site mutation in vitro, while alterations in vivo cannot be detected by mini-gene assays (31). RT-PCR can identify any mutant mRNAs directly, but is sophisticated for certain genes, such as the dystrophin gene, which contain several alternative promoters, markedly large introns and cryptic exons (32).

In the current study, RNA-sequencing was successfully used to analyze the influence of splicing mutations. This approach can directly identify the sequences of aberrant mRNAs and the relative abundance of each isoform of the protein (12). This appears to be much simpler and faster than RT-PCR and mini-gene assays. However, a limitation of high-throughput sequencing is the existence of sequencing errors (33). In the present study, discrepancies between the RNA-sequencing and RT-PCR results were observed. Specific RT-PCR was needed to validate the results of RNA-sequencing. RNA-sequencing holds expanded promise for the splicing mutation pathogenesis in various diseases, including myopathy, developmental disorders and neurodegenerative disorders. RNA-sequencing can be beneficial due to its diagnostic, prognostic and therapeutic applicability (34). Our future studies will focus on processing additional cases of BM in order to evaluate this method, so that the specific impact of sequence variants on splicing can be assessed.

Muscle biopsies are invasive operations (35). In our study, limited muscle sample was obtained from the proband. The total RNA extracted from the muscle was consumed by the RNA-sequencing and the rest of the sample was used hematoxylin and eosin staining and immunochemistry. Unfortunately, there was not enough RNA for RT-PCR to validate the results of RNA-sequencing. BM is a disease involving both muscle and skin. COL6 is abundantly expressed by skin fibroblasts, and human skin biopsies and in vitro primary skin fibroblast cultures have been widely used to characterize COL6 defects in BM (5,15). Notably, fibroblast are relatively easy to culture from the skin biopsies (36), therefore the present study performed reverse transcription PCR using RNA from fibroblasts derived from two patients and two healthy individuals as controls. Our future studies will extract more total RNA for the RT-PCR validation, aiming to compare the expression of the mRNA in muscle and skin fibroblast.

In conclusion, the present study described a BM-affected family resulting from a novel splice-site mutation within intron 4 of the COL6A2 gene. Using a number of bioinformatics and experimental methods, the hypothesis that the c.736-1G>C mutation caused the BM phenotype in this family was supported. These findings extend our present knowledge of the possible mutations of the COL6A2 gene associated with BM. The splicing mutation c.736-1G>C in the COL6A2 gene can produce two different mRNA splice variants: One retaining intron 4 and one skipping exon 5 (in-frame), which possibly causes aberrant splicing and leads to premature termination and truncation of translation. Ultimately, this would affect the quaternary structure of collagen VI. RNA-sequencing can be a powerful tool to determine the underlying mechanism of a disease-causing mutation at a splice site.

Supplementary Data

Availability of data and materials

The datasets generated and/or analyzed during the current study are available in the Genome Sequence Archive repository (https://bigd.big.ac.cn/gsa-human/). The results of Human Splicer Finder and Alternative Splice Site Predictor. Human Splicer Finder additional data are available upon request.

Authors' contributions

DL and JZ summarized the clinical information, analyzed the genetic test results and drafted the manuscript. YX helped to collect the clinical information and analyze the genetic test results. YD assessed the pathological biopsy results and performed the pathological biopsies. JZ collected the additional clinical information. YS performed the MRI. All authors read and approved the final manuscript.

Ethics approval and consent to participate

All procedures performed in this study involving human participants were conducted in accordance with the ethical standards of the institutional and national research committees and was approved by The Medical Ethics Committee of the First Affiliated Hospital of Guangxi Medical University (Nanning, China; approval no. 2017-KY-E-154). Written informed consent was obtained from each participant included in the study or a parent.

Patient consent for publication

Consent for publication was also obtained from all participants included in this study or a parent.

Competing interests

The authors declare that they have no competing interests.

Acknowledgments

The authors would like to thank Dr Dev Sooranna, Imperial College London, for editing the manuscript.

Abbreviations:

References