Splicing Switching of Alternative Last Exons Due to a Deletion Including Canonical Polyadenylation Site in COL6A2 Gene Causes Recessive UCMD

Abstract

Objectives

Collagen VI–related myopathy spans a clinical continuum from severe Ullrich congenital muscular dystrophy to milder Bethlem myopathy, caused by genetic variants in COL6A1, COL6A2, and COL6A3 genes. Our objective was to report a newly identified patient with the pathogenic variants restricted to a polyadenylation signal in the 3′-untranslated region, which have not been reported in hereditary muscle disease.

Methods

We performed clinicopathologic diagnosis and analysis using whole-genome and RNA sequencing.

Results

We report Ullrich congenital muscular dystrophy caused by a homozygous deletion, c.*198_*466del, which includes a polyadenylation signal in the canonical last exon of the COL6A2 gene. The parents were consanguineously married and had the heterozygous variant, but they were completely asymptomatic. In the patient's muscles, collagen VI was deficient in the sarcolemma, but present in the interstitium, showing the pattern of sarcolemma-specific collagen VI deficiency rather than a pattern of complete deficiency despite the lack of a polyadenylation signal. The RNA sequencing of the patient’s muscle showed that alternative last exons were raised in COL6A2 transcript.

Discussion

Our case provides a valuable example of the mechanism of alternative splicing switches for polyadenylation selection.

Introduction

Collagen VI is an important component for maintenance of the interstitium in skeletal muscles and consists of 3 isoforms—alpha 1, 2, and 3, which are encoded by COL6A1, COL6A2, and COL6A3 genes, respectively.¹ Collagen VI–related myopathy shows characteristic clinical phenotypes, which include proximal muscle weakness, skin and joint changes, scoliosis, and respiratory failure.¹ We have previously reported 2 distinct patterns of collagen VI distribution in muscles among patients: completely deficient (CD) or deficient on the sarcolemma but with deposits in the interstitium (sarcolemma-specific collagen VI deficiency: SSCD).^2,3

In this study, we report Ullrich congenital muscular dystrophy (UCMD) caused by a homozygous deletion within the last exon of the COL6A2 gene, which activates alternative splicing of the last exons.

Case Presentation

A 9-year-old girl from Libya born to healthy consanguineous parents was referred to our center for neuromuscular assessment (Figure 1A). She had a history of right hip joint dislocation at birth; took her first steps at 3 years; and showed progressive proximal muscle weakness in both upper and lower extremities, knee, and ankle joint contractures (Figure 1B) with hyperlaxity in the fingers and wrist joints on both sides, and she lost ambulation at 8 years of age. Serum creatinine kinase level was mildly elevated to 694 IU/L (normal 45–195 IU/L). Muscle MRI showed the characteristic target sign and tigroid sign in thigh muscles (Figure 1, C–E).⁴

Figure 1. Proband Family Tree, Contractures, and Lower Limb MRI.

(A) Proband family tree showing positive consanguinity, the proband (black arrow), other affected family members, and family segregation. (B) Photograph of the proband. Bilateral knee and ankle contractures. (C and D) Skeletal muscle imaging on MRI. (C) Axial T1-weighted imaging in thighs shows muscle atrophy involving the periphery of the vastus lateralis bilateral (black arrows) with sparing their central portion giving the characteristic “tigroid sign”. (D) Axial T1-weighted imaging in legs shows 30%–60% fat replacement in both the anterior and posterior compartments of both legs.

This study was approved by the National Center of Neurology and Psychiatry ethics committee. All clinical information and materials used in this study were obtained for diagnostic purposes with written informed consent.

Muscle biopsy was performed from the left biceps brachii. On histochemistry, remarkable diffuse endomysial fibrosis with few necrotic and regenerating fibers was seen (Figure 2A). On collagen VI staining, SSCD was seen (Figure 2B). However, the staining patterns of collagen VI in this case (Figure 2B) were different from other UCMD muscles, a muscle with recessive variants showing the CD pattern (Figure 2C) and a muscle with a dominant variant showing the typical SSCD pattern with remarkable collagen VI deposits in the interstitium (Figure 2D). The present case showed collagen VI deficiency on the sarcolemma, and no deposits in the interstitium (Figure 2B).

Figure 2. Histochemistry and Immunohistochemistry in the Biceps Brachii Muscle.

(A) Hematoxylin and eosin staining indicated a moderate to marked variation in fiber size. Some fibers with internalized nuclei are seen. Marked and diffuse endomysial fibrosis was seen. Necrotic and regenerating fibers are not prominent. Bar = 50 μm. (B) On immunohistochemistry of collagen VI (VI-26, 1:1,000; MP Biomedicals, LLC, Irvine CA) in this case, collagen VI deficiency in the sarcolemma and presence in the interstitium showing a sarcolemma-specific collagen VI deficiency (SSCD) pattern. Bar = 50 μm. (C) In an UCMD patient with recessive variants in the COL6A2 gene, collagen VI is absent in both the sarcolemma and the interstitium, showing a complete deficiency pattern. Bar = 50 μm. (D) In an UCMD patient with a dominant variant in the COL6A2 gene, collagen VI is reduced in the sarcolemma and increased staining in the interstitium, showing the typical SSCD pattern. Bar = 50 μm. (E) As a normal control, the normal collagen VI staining pattern in a control individual. Bar = 50 μm.

In whole-genome sequencing (WGS) analysis, no apparent pathogenic variants in coding regions of known causative genes were reported in hereditary muscle disease.⁵ Furthermore, analysis of structural variants on WGS detected a homozygous 269-bp deletion, c.*198_*466del, in canonical last exon 28 in the COL6A2 gene, which was included in the 3′UTR of transcript isoform 2C2 (NM_001849.4) (Figure 3A). This deletion has not been reported and unlisted in gnomAD. We confirmed that both asymptomatic parents possessed this deletion in heterozygous state by Sanger sequencing (data not shown).

Figure 3. RNA and Whole-Genome Sequencing Analyses.

(A) In RNA sequencing using poly A capture, alternative exon 28 in the COL6A2 gene is expressed in the skeletal muscle with transcripts of reference sequences NM_001849.4, isoform 2C2, and NM_058174.3, isoform 2C2a, the major one being NM_001849.4 in the control. In this patient, RNA expression in exon 28 of isoform 2C2 was lower in the patient than in the control. RNA expression in exon 28 of isoform 2C2a′, NM_058175.2, and exon 28 of isoform 2C2a were compensatory and upregulated in the patient. In whole-genome sequencing, a homozygous deletion, c.*198_*466del, in the 3′UTR lesion including the PAS in the COL6A2 gene. Alternative polyadenylation and polyadenylation signal sites are, respectively, indicated by white and black arrowheads. (B) Schematic diagram of the collagen VI α2 chain. Each chain contains a triple helical domain (TH) flanked by N- and C- globular domains, designated N1 and C1–C2. The minor 2C2a and 2C2a′ chains have a shorter 2C2a and 2C2a′ domain at the C-terminal end compared with the major 2C2 chain, as a result of alternative splicing of exon28 at the 3′ end of the gene.

We further performed RNA sequencing of constructed library using the poly A capture technique. There are 3 isoforms 2C2a, 2C2a′, and 2C2, using different last exons; isoform 2C2 is mostly used in skeletal muscles. In control, COL6A2 transcripts predominantly use canonical exon 28 of isoform 2C2 and minorly use alternative exon 28 of isoform 2C2a (NM_058174.3) in skeletal muscles as previously reported.⁶ The COL6A2 transcripts shifted to use alternative exon 28 of isoform 2C2a and alternative exon 28 of isoform 2C2a’ (NM_058175.2), although canonical exon 28 of isoform 2C2 was used to a small extent, and of interest, this transcript extended in length probably because of using the downstream polyadenylation signal (PAS) (Figure 3B). Considering this variant which causes aberrant splicing in RNA, it meets the standard for being classified as pathogenic in the ACMG guideline.⁷ The major protein product in the patient was predicted to be isoform 2C2a and has a different C2 subdomain in the C-globular domain of the COL6A2 gene.

Discussion

In this case, a novel homozygous deletion, c.*198_*466del, was detected in the COL6A2 gene. This is a report of muscle disease resulting from a variant restricted to the 3′-UTR region, which includes PAS, which has not been reported previously. In general, loss of the PAS causes diseases due to reduced mRNA stability.⁸ There have been some reports related to loss of the PAS as a cause of hereditary diseases, and among them, thalassemia is the most famous.⁹ In thalassemia, single-nucleotide substitutions have been observed within the PAS of the HBA2 and HBB genes, resulting in reduced mRNA amounts and the detection of long transcripts that extend beyond the normal PAS to downstream PAS.⁹ From that indication, it was assumed that immunohistochemistry would result in a CD pattern because of the decay of mRNA, but in this case, the SSCD pattern was observed, suggesting a different mechanism from that discussed previously.

The transcript isoforms 2C2, 2C2a, and 2C2a′ of the COL6A2 gene use different final exons, resulting in alternative polyadenylation (APA) selection. Considering data from scAPAdb,¹⁰ in addition to the transcriptome analysis results of the control case in Figure 3A, it is indicated that exon 28 in 2C2 is predominantly selected as an APA over exon 28 in 2C2a; in addition to the PAS, U or GU-rich downstream and less well-defined upstream sequence elements are known to enhance cleavage efficiency.¹¹ APA selection occurs when 2 or more PAS sites are present in the gene. At least 22% of mRNA undergoes APA selection, often in a tissue-specific manner.¹² The 3′-UTR is an important regulatory element for the transcript that, for example, affects its stability, localization, translation because APA strongly affects mRNA, and thus gene function. The gene responsible for thalassemia does not have APA, so when the PAS is disrupted, mRNA elongates to downstream PAS sequences. While in the COL6A2 gene in the patient, the PAS of isoform 2C2 was deleted, which may have enhanced the use of APA. In addition, in the synthesis of mature RNAs, variants in the PAS site inhibit removal of the last intron but not the more upstream introns, and alternative last exons of 2C2a with APA may be selected for effective intron splicing.¹³ That is consistent with the result that this variant inhibited splicing of upstream intron of exon 28 in 2C2, which resulted in the use of exon 28 of 2C2a in matured transcripts.

The isoform 2C2a was mainly produced in the patient's muscles. A previous study demonstrated that the formation and secretion of collagen VI was confirmed after introducing the expression cassette of COL6A2-C2a together with A1 and A3 into HEK cells.¹⁴ From that finding and our observation, collagen VI with A2-C2a is formed up to the tetramer and secreted, but probably failed to be incorporated into the microfibrils because of the lack of a normal C-terminal globular domain, C2. On the tetramer structure of collagen VI, N-terminal globular domains are believed to be an attaching site for microfibril formation,¹⁴ but one hypothesis is that the C-terminal globular domain also works for it. With Col6A1-3 molecules with a variant in triple helical regions, as the collagen VI tetramers have the intact globular domains and are incorporated into microfibrils, they would show a dominant-negative effect, but as the collagen VI tetramers with C2a would fail to be incorporated into microfibrils, they do not show a dominant-negative effect, but loss of function (LOF). The fact that the heterozygous parents for this variant are asymptomatic supports this idea. As this indication, collagen VI is detected in the patient's muscles, although it is in the SSCD pattern because the sarcolemma staining was reduced and interstitial staining was observed because of enhanced fibrosis. However, on fine observation in the comparison between the dominant-inherited monoallelic cases and this case, interstitial staining patterns of collagen VI are a little bit different; the staining in this patient is rather weaker and deposition is inconspicuous. The typical SSCD pattern in dominant cases shows reduced staining in the sarcolemma and extensively increased staining in the interstitium to make deposits, reflecting dominant-negative effects of mutant molecules on oligomeric filament formation of collagen VI. The distribution of collagen VI observed in the patient is different from the SSCD pattern that is observed in patients with the monoallelic variant in the triple helical region and well compatible with the collagen VI distribution in recessive cases, indicating that the 2C2a chain would be a LOF molecule in the skeletal muscles.

In conclusion, we reported a novel homozygous deletion in the 3′ UTR of the COL6A2 gene containing the PAS. The transcriptome results in this case also provide a valuable example of the mechanism of alternative splicing switches for polyadenylation selection.

Ethics Approval

This study was approved by the ethics committee of the National Center of Neurology and Psychiatry (NCNP) in Japan (A2019-123). All clinical information and materials used in this study were obtained for diagnostic purposes, with written informed consent.

Data Availability

The data sets used during the current study are available from the corresponding author on reasonable request.

Appendix. Authors

Name	Location	Contribution
Rasha El Sherif, MD	Myo-Care Neuromuscular Center, Myo-Care National Foundation; School of Medicine, New Giza University, Cairo, Egypt	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data
Yoshihiko Saito, MD, PhD	Department of Neuromuscular Research, National Institute of Neuroscience, National Center of Neurology and Psychiatry, Tokyo, Japan	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data
Tomonari Awaya, MD, PhD	Department of Anatomy and Developmental Biology, Graduate School of Medicine and Faculty of Medicine, The University of Kyoto, Japan	Drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data
Satoru Noguchi, PhD	Department of Neuromuscular Research, National Institute of Neuroscience, National Center of Neurology and Psychiatry, Tokyo, Japan	Drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data
Ichizo Nishino, MD, PhD	Department of Neuromuscular Research, National Institute of Neuroscience, National Center of Neurology and Psychiatry, Tokyo, Japan	Drafting/revision of the manuscript for content, including medical writing for content

Study Funding

This study was supported partly by JSPS KAKENHI Grant Number (20H03592, 21K06961, 22K11562, 22K07497, 23K14766, 23K14553, 23K06973, 23H03065); Intramural Research Grant (2-5, 3-9, 4-6, 5-4, 5-5, 5-6, 5-7) for Neurological and Psychiatric Disorders of NCNP; AMED under Grant Numbers, JP23bm1223001s0102, JP23ek0109617s0402, JP23ak0101195s0101; and Health, Labour and Welfare Sciences Research Grants (JPMH23FC1014, JPMH21FC1006, JPMH23FC1017, JPMH23FC0201).

Disclosure

The authors report no relevant disclosures. Go to Neurology.org/NG for full disclosures.