Homozygous splice variant (c.1741-6G > A) of the COL6A1 gene in three patients with Ullrich congenital muscular dystrophy

The three major collagen VI genes: COL6A1, COL6A2 , and COL6A3 encode microﬁbrillar components of extracellular matrices in multiple tissues including muscles and tendons. Pathogenic variants in the collagen VI genes cause collagen VI-related dystrophies representing a continuum of conditions from Bethlem myopathy at the milder end to Ullrich congenital muscular dystrophy at the more severe end. Here we describe a pathogenic variant in the COL6A1 gene (NM_001848.3; c.1741-6G > A) found in homozygosity in three patients with Ullrich congenital muscular dystrophy. The patients suffered from severe muscle impairment characterised by proximal weakness, distal hyperlaxity, joint contractures, wheelchair-dependency, and use of nocturnal non-invasive ventilation. The pathogenicity was veriﬁed by RNA analyses showing that the variant induced aberrant splicing leading to a frameshift and loss of function. The analyses were in line with immunocytochemistry studies of patient-derived skin ﬁbroblasts and muscle tissue demonstrating impaired secretion of collagen VI into the extracellular matrix. Thereby, we add the variant c.1741-6G > A to the list of pathogenic, recessive, splice variants in COL6A1 causing Ullrich congenital muscular dystrophy. The variant is listed in ClinVar as of “uncertain signiﬁcance” and “likely benign” and may presumably have been overlooked in other patients.

Intracellularly, the alpha chains associate to form triple monomers and assemble into antiparallel dimers, which stag to form tetramers, all stabilized by disulphide bonds [3] . The tetramers are secreted extracellularly where they form larger networks of microfibrils in a tissue specific manner [4] . These networks anchor the cell surface into the ECM components mediating cell adhesion children are often born with hands in a position of wrist flexion and feet in a position of ankle dorsiflexion. Some affected children acquire the ability to walk independently, but the decline in motor function is rapid in the first decade of life [11] . Early and severe respiratory insufficiency occurs by the end of the first decade [12] resulting in the need for nocturnal non-invasive ventilation (NIV). In the first or second decade of life, respiratory failure is a common cause of death unless the affected individual is treated by respiratory support. Other distinctive features observed in patients with COL6A-RDs are as follows: congenital hip dislocation, prominent calcanei, torticollis, kyphotic contracture of the spine, and failure to thrive. Moreover, skin changes, such as mild facial erythema, hypertrophic scars (keloid), and hyperkeratosis follicularis are common [13] . Cardiac involvement is generally absent, and intelligence is normal.

Diagnostics and treatment
Clinical, laboratory, histological, electrophysiological, and imaging findings might all support the suspicion of COL6-RDs [ 12 , 14 , 15 ], but genetic testing is the most accurate diagnostic tool. Currently, there is no cure for COL6A-RD and treatment is focused on managing the symptoms of the disease.

The genetic basis of COL6A-RDs
COL6A-RDs are associated with both autosomal dominant and recessive inheritance and approximately 50-75% of probands have the disorder as a result of a de novo variant [16] . Thus far, 175 COL6A1 -related variants have been reported in patients based on the HGMD database (January 2023) [17] . The variants are almost equally distributed in each of the three genes, including 77 missense/nonsense, 9 gross deletions, 1 indel, and 58 splicing variants. Although all of the splicing variants are predicted to affect RNA splicing and produce abnormal transcripts, most predictions have not been confirmed by published studies. Therefore, the actual effects of mRNA changes in individuals remain unknown. Here we report an intronic variant (NM_001848.3; c.1741-6G > A) of the COL6A1 gene assessed by RNA analysis. The variant was identified homozygous in three patients with congenital muscular dystrophies, which in combination with functional studies that proved pathogenicity of the variant, confirmed the genetic diagnosis UCMD. Thus, the present study extends the allelic spectrum of the COL6A-RDs and shows the value of RNA analyses as a tool to classify unknown variants suspected to affect splicing.

Patients included
Three patients from Copenhagen Neuromuscular Center and Department of Paediatrics and Adolescent Medicine at Copenhagen University Hospital were included in the study. All analyses were conducted with diagnostic purposes and written informed consent from all patients/parents were obtained for the publication of data.

Biological samples
Peripheral blood samples were obtained from all three patients in EDTA coated tubes to extract DNA for whole exome sequencing (WES). Skin biopsies were performed on all three patients for the RNA analyses and cell studies. Muscle biopsies for evaluation of histopathology and immunohistochemistry were taken from the vastus lateralis of all three patients.

Whole exome sequencing (WES)
The WES analyses were performed on DNA extracted from whole blood using standard methods. The coding regions including flanking intronic sequences were captured using Human Core Exome Kit (Twist Bioscience, South San Francisco, CA) followed by sequencing (100 bp paired-end) on a NovaSeq 60 0 0 (Illumina, San Diego, CA) to a mean depth > 100x (99%, > 20x). Data were processed according to best practice guideline for GATK4 (Broad Institute) with the hg19 human genome as a reference. Variants were filtered for known muscle disease genes, primarily using the Online Mendelian Inheritance in Man (OMIM) database and GeneTable of Neuromuscular Disorders [ 18 , 19 ]. Variant analysis was performed using VarSeq (Golden Helix, Bozeman, MT). All procedures were essentially carried out according to the descriptions of the manufacturers.

Variant evaluation
The ClinVar [20] , HGMD® Professional [17] , and gnomAD [21] databases were used to evaluate the pathogenicity of the variant according to the ACMG guidelines [22] . The effect on splicing was estimated by the splice variant predictor MaxEntScan [23] .

Total RNA extraction and reverse transcription polymerase chain reaction (RT-PCR)
Primary fibroblasts obtained from skin biopsies from the three patients were used for the RNA analyses. The fibroblast culture protocol has previously been described by Witting et al. [24] . The fibroblasts tested negative for mycoplasma contamination (EZ-PCR TM ) immediately before the RNA extraction. For inhibition of nonsense-mediated mRNA decay (NMD), the cells were treated with puromycin (100 μg/ml; Thermo Fisher Scientific, Waltham, MA) and incubated for 4 h. Total RNA was extracted from the patient-derived skin fibroblasts with a Chemagic TM 360-D instrument and used in the production of complementary DNA (cDNA) by RT-PCR according to the recommendations by ENIGMA [25] . Primers for RT-PCR and allele specific analysis were designed from exon 21 and 30 in regions with a frequency of SNPs less than 1% (gnomAD). The primers 21F 5 -GCCCTATCGGACCTAAAGGCTACC-3 and 30R 5 -TCTCGAAGTTCTGCAGGCCAATGC-3 were used.

Visualization and validation of the cDNA variant
RT-PCR products were separated on a 1% agarose gel and the fragments were excised and purified by QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany). The fragments were separated by semiquantitative capillary electrophoresis performed on an ABI3730 (Thermo Fisher Scientific, Waltham, MA) and analysed by GeneMapper TM Software 4.0 (Thermo Fisher Scientific, Waltham, MA). Sanger sequencing was conducted with the use of BigDye TM Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific, Waltham, MA) and the products were analysed on an ABI3730 (Thermo Fisher Scientific, Waltham, MA).

Cell studies of collagen VI and perlecan distribution
Examination of the production of collagen VI in patientderived skin fibroblasts was performed in all three patients, a healthy control, and a control with genetically verified UCMD (heterozygous for c.6309 + 1G > C in the COL6A3 gene, which causes skipping of exon 18 in the COL6A3 transcript). The fibroblast culture and staining protocols have previously been described by Witting et al. and only differ in the time of cultivation (1 week Table 1 Clinical and laboratory findings of three patients with Ullrich congenital muscular dystrophy.

Muscle histopathology and distribution of collagen VI and perlecan
Muscle biopsies were sectioned on a cryostat and stained for hematoxylin and eosin (H&E) for evaluation of general histopathology and sections from patient 3 were stained for collagen VI 1:100 and perlecan (an ECM protein) 1:200 in 3% fetal calf serum in PBS and detected using Alexa Fluor goat-anti-rabbit and goat anti-mouse antibodies. All stained sections were imaged on an inverted microscope with a 20x objective using Neo (Andor, Belfast, UK) and Nikon DS Fi-3 (Tokyo, Japan) cameras.

Clinical report of three patients with UCMD
The patients were not related to each other but originated from the same region near Ankara, Turkey ( Table 1 ).

Patient 1:
The patient is a woman of 31 years of age, born to consanguineous parents from Turkey. She has no family history of neuromuscular diseases and has an unaffected brother and sister. She presented at birth/as a young child (exact age is unclear) with walking difficulties and relied on a wheelchair for mobility by the age of 8 years. Muscle impairment is characterised by proximal weakness, distal hyperlaxity and joint contractures of the elbows, hips, knees, and Achilles tendons as well of scoliosis. Since 15 years of age, she has used nocturnal noninvasive ventilation (NIV). She has a forced vital capacity (FVC) of 10% and unaffected heart function. She depended on personal assistance 14 h a day. For the last years, the disease has been non-progressive.

Patient 2:
The patient is a man of 48 years of age, born to unaffected, consanguineous parents from Turkey. He has 3 unaffected brothers and one brother with a similar clinical picture, who died as a young child. He presented at birth/as a young child (exact age is unclear) with myopathic symptoms and relied on a wheelchair for mobility by the age of 8-9 years. He has joint contractures of the elbows, knees, and Achilles tendons as well as scoliosis. Since 28 years of age, he has used nocturnal non-invasive ventilation (NIV). He has a forced vital capacity (FVC) of < 15% and unaffected heart function. He depends on personal assistance 24 h a day.
In 20 01, 20 02, 20 04, and 2014 the patient was examined for pathogenic variants in the SMN1, DMD , and FKRP genes with normal results.

Patient 3:
The patient is a girl of 10 years of age, born to non-consanguineous parents originating from the same village in Turkey. She has no family history of neuromuscular diseases. Prenatal screening was normal despite fetal growth restriction at −16% and maternal perception of reduced fetal movements. She was born by cesarian section with need for cardiopulmonary resuscitation with achievement of spontaneous breathing after 6 min. She presented at birth with severe hypotonia, and muscle impairment was characterised by proximal weakness, distal hyperlaxity and joint contractures of the elbows, shoulders, hips, and knees as well of scoliosis. She never achieved independent ambulation. Since few months of age, she has used nocturnal NIV and supplemental at daytime. Her heart function is unaffected. Furthermore, the patient is suffering from epileptic encephalopathy and severe intellectual disability due to another unrelated genetically confirmed disease (a novel autosomal recessive disease, manuscript in preparation).
In 2012-2016 the patient was examined for pathogenic variants in the SMN1 gene and genes associated with dystrophia myotonica and limb-girdle muscular dystrophy all with normal results. In 2016, respiratory chain enzymes and deletions in the mtDNA (mitochondrial DNA) were examined also with normal results.  ( Fig. 1 A). If the new acceptor site was used, it was predicted that 4 intronic nucleotides (ACAG) upstream of exon 27 would be included causing a frameshift at p.581 involving 29 amino acids leading to the formation of a premature stop codon (PTC) in exon 29, described on protein level as p.(Gly581Thrfs * 29) ( Fig. 1 D).

Aberrant splicing was detected by RNA analyses
To investigate the effects of the variant on splicing of COL6A1 , RNA was extracted from skin fibroblasts from all three patients and two control samples that were cultivated with or without puromycin to inhibit NMD, and RT-PCR was performed with primers targeting exons 21 and 30. The amplicons were subjected to semiquantitative capillary electrophoresis analyses, which for all samples resulted in a peak of 487 bp corresponding to the canonical transcript (Fig. S1A). Another peak of 491 bp was, however, observed exclusively in the patient samples, suggesting that the in silico predicted acceptor site does cause aberrant splicing of exon 27, and this fraction was estimated to 68.7 -74.3% and 75.0 -81.7% wo/w puromycin ( Fig. 1 B and Fig. S1B). The results of the semiquantitative capillary electrophoresis analyses were in line with the Sanger sequencing result that confirmed the inclusion of 4 nucleotides upstream of exon 27 and that both the new splice site and the wild type splice site were used ( Fig. 1 C). Thereby, the c.1741-6G > A variant was confirmed to lead to aberrant splicing resulting in frameshift and a PTC (p. (Gly581Thrfs  *  29)). The variant was submitted to ClinVar [26] , classified as a class 5/pathogenic variant in accordance with ACMG guidelines [22] .

Muscles are myopathic with reduced level of collagen VI and perlecan
H&E stained sections of muscle demonstrated myopathic rounded and sometimes centrally nucleated fibers of variable size ( Fig. 2 A). Sections stained for collagen VI and perlecan showed substantially reduced levels of both proteins compared to a healthy control ( Fig. 2 B).

The COL6A1 variant leads to intracellular retention of collagen VI
Examinations of the production of collagen VI in patientderived skin fibroblasts were performed in all three patients. Fibroblasts from all patients demonstrated intracellular stain for retained collagen VI and nearly no extracellular stain for collagen VI. Perlecan, a marker for ECM, appeared at a normal level in the fibroblasts ( Fig. 2 C).

Discussion
With the present study, we add a new splice variant (NM_001848.3; c.1741-6G > A) to the list of pathogenic variants in the COL6A1 gene causing Ullrich congenital muscular dystrophy. To date (Jan. 23), 175 disease-causing variants have been reported in HGMD [17] , including 58 splice variants resulting in myopathic phenotypes mostly described as BM or UCMD [27][28][29][30] . The c.1741-6G > A variant is recorded heterozygous in 34 out of 140,545 people worldwide (gnomAD v2.2.1) primarily in the ethnic Finnish population (29/12,445) followed by non-Finnish Europeans (3/64,073). No record of homozygosity is noted in gnomAD (v2.2.1 of v3.12) (Jan. 23) [21] . As our patients are of Turkish/Kurdish descent, the variant exists across ethnic groups. Functional characterization in our study provided evidence that the splice variant (c.1741-6G > A) outperformed the normal acceptor splice site upstream of exon 27 and produced an abnormal transcript with a PTC in exons 29/30. Until now, the variant has been recorded in ClinVar [31] as "likely benign" and of "uncertain significance", which might have led geneticists to discard the variant as a potential cause of disease. Even if an in silico prediction tool had been used in the assessment of the variant, the effect on splicing would still have been unclarified as the prediction tools tend to report many pseudo splice sites, especially for 3' splice sites, and therefore in silico prediction should never stand-alone [32] . Generally, the role of aberrant splicing as a cause of genetic diseases is severely underestimated as RNA analysis is rarely implemented in clinical practice [ 33 , 34 ].
The aberrant transcript was shown by RNA analysis to cause a frameshift from p.581 and produce a PTC in the junction between exons 29 and 30, which might cause loss-of-function, consistent with the typical result of pathogenic splice site variants [34] . Semiquantitative capillary electrophoresis showed that the mutant splice site was used in 71.6% of the transcripts, on average, in the three patients. Variation in the ratio between wild type and mutant transcripts in the three patients could possibly relate to genetic modifiers, culture conditions, single nucleotide polymorphisms (SNPs) located near the splice site, and/or tissuedependent splicing. Moreover, patient 3 was also suffering from severe epileptic encephalopathy due to another recessive disease, which might play a role in her physical functional ability. All three patients were homozygous for the variant, and no COL6-RDs have been noted among people with the variant in the heterozygous state, consistent with recessive inheritance. No parental phenotypes were noted.
Loss-of-function variants in COL6A1 are known to be diseasecausing for autosomal recessive COL6A1 conditions [ 29 , 35 ], whereas certain in-frame variants affecting donor or acceptor splice sites in the triple-helical domain of COL6A1 have been reported to cause autosomal dominant COL6-RDs [36] . Furthermore, homozygous loss-of-function variants in COL6A1 due to PTCs in the triple-helical domains are known to cause the most severe phenotypes (ambulation never achieved) [ 29 , 35 ]. Under in vivo conditions, the c.1741-6G > A variant may either produce a truncated α1 chain, be degraded by NMD, or both. However, the large percentage of aberrant transcripts in the untreated samples together with the fact that addition of NMD-inhibitor increased the ratio of aberrant mRNA by only 10.7% on average, indicate, that most aberrant transcripts might not be degraded by NMD. Furthermore, the collagen VI staining of fibroblasts and muscle tissue showed that collagen VI was retained inside the fibroblasts, thereby indicating that (some of) the variant transcripts were translated into truncated α1 chains ( Fig. 2 C). In general, around half of all possible PTCs in the human genome do not trigger NMD efficiently, but in most of these cases, the PTC is located in either: the last exon, in the last 50 bp of the penultimate exon, less than 150 bp away from the start codon, or in a long exon ( > 407 bp); none of these scenarios being the case with the variant identified in this study [ 37 , 38 ], as the frameshift was located in the triple-helical domain and the PTC in the junction between the triple-helical domain and the C-terminal domain. If the aberrant mRNA is translated into a truncated α1 chain with no C-terminal, this might lead to incorrect triple helix formation intracellularly because the collagenous Gly-X-Y motifs at the C-terminal ends of the α chains are crucial for the initiation of the triple helix formation [39] . The RNA analyses indicated that the wild type splice site was used in around 28% of the transcripts and as each collagen VI tetramer, that is secreted to the ECM, should contain four wild type α1 chains to assemble correctly, the chance of having a tetramer entirely consisting of wild type α chains would be very small (0.28 4 % = 0.65%). Furthermore, assembly and/or export of wild type tetramers might be impeded due to steric hindrance. According to the manufacturer of the antibody used for staining of collagen VI (ab6588, Abcam, Cambridge, UK), the antibody binds the α1 chains, although the exact epitope has not been mapped, but the specificity of the antibody is reduced when the strands are reduced and separated; thus the intracellular staining does not indicate whether or not the precursors are assembled correctly [40] . The immunofluorescent labeling of collagen VI in cultured fibroblasts and muscle tissue is consistent with the typical pattern of patients affected by recessive COL6-RDs where collagen VI is reduced in the ECM [ 39 , 41 , 42 ]. Collagen VI-stained sections demonstrate near absence of collagen VI in patient 3 compared to a healthy control. Perlecan stained sections demonstrate that the loss of wild-type collagen VI affects perlecan levels, which is severely reduced in patient 3. C. Collagen VI-stained fibroblasts from patient 1, 2 and 3 resembles negative control fibroblasts from a patient with Ullrichs congenital muscular dystrophy (UCMD) with all collagen VI positive stain retained within the cells as opposed to the extensive extracellular staining in a healthy control. Unlike in muscle, all patients had the same level of extracellular perlecan as the healthy control. Bar is 50 μm.
The collagen VI staining of fibroblasts as a diagnostic tool in patients with symptoms of COL6-RDs has a positive predictive value of 75%, a sensitivity and negative predictive value of 100%, and a specificity of 63% [42] . Perlecan levels, as a marker of extracellular matrix proteins, are reduced in the patient muscle section compared to the healthy control but not in fibroblasts where the extracellular level appears normal, suggesting that perlecan at least in fibroblasts can be found in the extracellular space independently of the level of collagen VI. This also means that a perlecan stain of fibroblasts has little predictive value for the status of the extracellular matrix in muscle and should not be used for this.
Although further research in the molecular pathogenesis is still needed, our study provides strong evidence that the splice site variant detected in the three patients indeed causes their phenotypical traits of UCMD, and therefore, we hereby add the variant c.1741-6G > A to the list of pathogenic splice variants in COL6A1 causing autosomal recessive UCMD [31] .

Declaration of Competing Interest
The authors have no conflicts of interest to disclose.