Characterization of severe COL6-related dystrophy due to the recurrent variant COL6A1 c.930+189C>T

Abstract

Collagen VI-related dystrophies manifest with a spectrum of clinical phenotypes, ranging from Ullrich congenital muscular dystrophy (UCMD), presenting with prominent congenital symptoms and characterized by progressive muscle weakness, joint contractures and respiratory insufficiency, to Bethlem muscular dystrophy, with milder symptoms typically recognized later and at times resembling a limb girdle muscular dystrophy, and intermediate phenotypes falling between UCMD and Bethlem muscular dystrophy. Despite clinical and muscle pathology features highly suggestive of collagen VI-related dystrophy, some patients had remained without an identified causative variant in COL6A1, COL6A2 or COL6A3.

With combined muscle RNA sequencing and whole-genome sequencing, we uncovered a recurrent, de novo deep intronic variant in intron 11 of COL6A1 (c.930+189C>T) that leads to a dominantly acting in-frame pseudoexon insertion. We subsequently identified and have characterized an international cohort of 44 patients with this COL6A1 intron 11 causative variant, one of the most common recurrent causative variants in the collagen VI genes.

Patients manifest a consistently severe phenotype characterized by a paucity of early symptoms followed by an accelerated progression to a severe form of UCMD, except for one patient with somatic mosaicism for this COL6A1 intron 11 variant who manifests a milder phenotype consistent with Bethlem muscular dystrophy.

Partial amelioration of the disease phenotype in this individual provides a strong rationale for the development of our pseudoexon skipping therapy to successfully suppress the pseudoexon insertion, resulting in normal COL6A1 transcripts. We have previously shown that splice-modulating antisense oligomers applied in vitro effectively decreased the abundance of the mutant pseudoexon-containing COL6A1 transcripts to levels comparable to the in vivo scenario of the somatic mosaicism shown here, indicating that this therapeutic approach carries significant translational promise for ameliorating the severe form of UCMD caused by this common recurrent COL6A1 variant.

Foley et al. have identified and characterised 44 patients with a recurrent deep intronic variant in COL6A1 (c.930+189C>T), causing a severe form of COL6-related dystrophy. Their study helps lay the foundations for future clinical trials of the COL6A1 c.930+189C>T variant-specific splice-modulating therapeutics currently in development.

Introduction

Ullrich congenital muscular dystrophy [UCMD (MIM 254090)], Bethlem myopathy or rather muscular dystrophy [BM (MIM 158810)] and intermediate phenotypes form a subgroup within the congenital muscular dystrophies (CMDs) known as the collagen VI-related dystrophies (COL6-RDs)^1-3 and result from recessively or dominantly acting causative variants in any of the three collagen VI genes (COL6A1, COL6A2 or COL6A3).^4,5 Ullrich congenital muscular dystrophy was first described in 1930 by Dr Otto Ullrich, who termed the condition ‘Skleratonische Muskeldystrophie’ (scleratonic muscular dystrophy), noting evidence of congenital weakness associated with proximal joint contractures and distal joint laxity.^6,7 The first signs of UCMD can manifest in utero, with decreased fetal movement frequently reported.^1,6-9 At birth, UCMD patients classically demonstrate hypotonia, proximal joint contractures, hip dislocation(s), prominent calcanei and distal joint hyperlaxity, typically resulting in abnormal positioning of the hands and feet (with hands in a position of wrist flexion, resting against the ventral surface of the forearms, and feet in a position of dorsiflexion, resting against the anterior surface of the lower leg). Torticollis and kyphoscoliosis can often be seen at birth also. Although children with UCMD may achieve independent ambulation, this ability is lost in early childhood,^1,9-11 typically by 10 years of age.^3,12 After becoming wheelchair dependent, most UCMD patients demonstrate a relative plateau in progressive muscle weakness (as far as can be assessed within restricted range of movement), whereas joint contractures continue to progress, compounding the motor limitations resulting from the muscle weakness and thus significantly contributing to the overall level of disability.¹³ Although some UCMD patients demonstrate spinal rigidity without an evident spinal curvature, the vast majority develop progressive scoliosis, which can appear as early as the preschool years or even congenitally.^3,12 An early onset of an invariable decline in respiratory function with early hypoventilation is a salient clinical feature of UCMD, necessitating the initiation of nocturnal non-invasive ventilation (NIV) by an average age of 11 years.^2,14

Bethlem myopathy (BM) was first described in 1976 by Drs Jaap Bethlem and George K. van Wijngaarden and is characterized by slowly progressive muscle weakness and distally pronounced joint contractures.¹⁵ Although often described as a slowly progressive ‘myopathy’ of adulthood, muscle histology progresses over time to include dystrophic findings, hence the term ‘Bethlem muscular dystrophy’ more accurately describes this condition.¹⁶ The inclusion of Bethlem muscular dystrophy (MD) within the revised limb-girdle muscular dystrophy (LGMD) nomenclature (with ‘LGMD D5’ for the autosomal dominant and ‘LGMD R22’ for the autosomal recessive forms)¹⁷ provides further evidence of the recognition of the underlying dystrophic process. Bethlem MD is also categorized within the CMDs which probably relates, in part, to the fact that symptoms of Bethlem MD can present as early as birth. In particular, hypotonia, neck flexion weakness, torticollis and joint contractures are among the early symptoms reported. Progressive contractures of the Achilles tendons and elbows usually manifest by the end of the first decade. Patients with Bethlem MD develop proximal muscle weakness but typically maintain the ability to ambulate into adulthood. By 50 years of age, however, more than two-thirds of patients rely on the use of a wheelchair to aid ambulation, classically for outdoor use, while independent ambulation indoors is usually maintained.¹⁸ Although some adults with Bethlem MD develop nocturnal hypoventilation, this does not occur uniformly, as seen in a large natural history study of pulmonary function in the COL6-RDs, which only demonstrated a trend between age and decrease in pulmonary function in patients with Bethlem MD.²

UCMD and Bethlem MD were initially viewed as two separate phenotypic entities; however, it became apparent subsequently that there is a continuous spectrum of phenotypes, with UCMD at the severe end of the spectrum, Bethlem MD along the mild end, and so-called ‘intermediate’ phenotypes in between the classical UCMD and Bethlem MD phenotypes.^2,3,13,19 A large-scale international natural history study of COL6-RD patients helped in elucidating the parameters of intermediate COL6-RD. Patients in this group uniformly achieved ambulation and walked longer than UCMD patients but never achieved the ability to jump or run, in contrast to patients with Bethlem MD. For this intermediate COL6-RD group, loss of ambulation occurred by ∼19 years of age, and nocturnal NIV was needed by ∼21.5 years of age.²

Anticipatory care of the uniformly progressive decline in respiratory function that characterizes UCMD and intermediate COL6-RD, including the timely initiation of NIV, is essential for decreasing morbidity and mortality, and this care relies on the clinical recognition of COL6-RD. Despite clinical and muscle immunohistochemical features highly suggestive of COL6-RD, some patients have remained without an identified causative/pathogenic variant in the COL6A1–3 genes. By using a combination of muscle RNA sequencing and whole-genome sequencing in four such patients, we previously uncovered a recurrent, de novo deep intronic variant in intron 11 of COL6A1 (c.930+189C>T) that leads to a dominantly acting in-frame pseudoexon insertion.^20,21 Targeted Sanger sequencing for this variant in intron 11 of COL6A1 to complement panel and whole-exome testing has revealed that this deep intronic de novo variant is a surprisingly common cause of UCMD. Here, we describe the consistent phenotype associated with this COL6A1 intron 11 variant, which manifests with a paucity of symptoms congenitally and an accelerated progression to a phenotype of UCMD. The only exception was one patient with somatic mosaicism for this COL6A1 variant, who manifests a significantly milder phenotype that is consistent with Bethlem MD.

Materials and methods

Study subjects

Patients were identified through their local neurology clinics. Written informed consent and age-appropriate assent for research studies, procedures and clinical photographs were obtained by a qualified investigator. Ethical approval was obtained via the Institutional Review Board of the National Institutes of Health (protocol 12-N-0095), the University College London Research Ethics Committee (13/LO/1894) and the Medical Research Council Centre for Neuromuscular Disease Biobank London Research Ethics Committee (06/Q0406/33). Medical history was obtained, and clinical evaluation and muscle imaging were performed as part of the standard neurological evaluation. Muscle MRI was performed using conventional T1-weighted spin echo of the lower extremities. Muscle ultrasound images were obtained using a Siemens/Acuson S2000 with an 18 MHz linear probe. Blood, skin biopsies, muscle biopsies and urine samples were obtained according to standard procedures. Saliva samples were collected using the oragene-discover kit (DNA Genotek).

Biospecimen processing

Fibroblasts were cultured from fresh skin biopsies using a standard enzymatic digestion methodology. DNA was extracted from blood, skin fibroblasts, saliva and the pelleted fraction of the urine specimen using the Gentra Puregene Blood kit (Qiagen). RNA was obtained from blood following the manual purification of total RNA for human whole blood collected in PAXgene blood RNA tubes protocol (PreAnalytiX/Qiagen). RNA was obtained from skin fibroblasts using Trizol (ThermoFisher Scientific). For fresh skin biopsy samples, specimens were frozen in liquid nitrogen, pulverized using a mortar and pestle and homogenized in Trizol. DNA and RNA were both obtained using the interphase/organic and the aqueous phase of Trizol, respectively, following the manufacturer's instructions.

Sanger sequencing

End point PCR was performed on genomic DNA samples using Takara LA Taq (Takara Bio Inc.) according to the manufacturer's specifications. Sequences of the primers were as follows: 5′-TGTTGGGTACCAGGGAATGAAGGT-3′ and 5′-AAACGAAGGCAGGAGTCAGA-3′. PCR products were sent for Sanger sequencing to Genewiz (Azenta Life Sciences).

Pseudoexon amplification and quantification

To quantify the degree of mosaicism, a Taqman assay was designed and synthesized by ThermoFisher Scientific as a Custom Taqman SNP Genotyping Assay, in which the probe hybridizing to the reference (C) allele was labelled with VIC, and the probe hybridizing to the variant (T) allele was labelled with FAM. Genomic DNA samples (between 50 and 600 ng of DNA, depending on the tissue source) were amplified on the Bio-Rad QX200 ddPCR system (Bio-Rad) at the NCI CCR Genomics Core, using the standard protocol provided by the manufacturer and the assay described above. The fractional abundance of the variant (T) allele over the reference (C) allele was calculated from the determined concentrations (in copies per microlitre).

For expression studies, RNA was converted to complementary DNA using the SuperScript IV Reverse Transcriptase (ThermoFisher). For end point PCR and gel electrophoresis, complementary DNA was amplified using the Kapa HiFi HotStart (Roche Sequencing), with the following primers: 5′-ACCTGTTGGGTACCAGGGAATGAA-3′ and 5′-ACCAGGGTCTCCTCTTGGTC-3′. For quantitative PCR, complementary DNA was amplified using the FastStart Universal Master Mix (Roche Life Science), on the QuantStudio 6 Real-Time PCR instrument (ThermoFisher Scientific). Expression levels of the COL6A1 transcripts were determined with quantitative PCR assays detecting either transcripts with the pseudoexon (primer 1: 5′-TACCAGGGAATGAAGGGAG-3′; primer 2: 5′-CCTGGAGCCCTTTGCTG-3′; and probe: 5′-ATCTGGAAGGACAAGGACAGCCAC-3′) or total COL6A1 transcripts [primer 1: 5′-CCGACTGCGCTATCAAGAA-3′; primer 2: 5′-AATCAGGTACTTATTCTCCTTCAGGT-3′; and probe: ROCHE UPL probe #17, Millipore Sigma (product now discontinued)]. Expression of the COL6A1 with pseudoexon transcripts was normalized to the total COL6A1 expression. The patient's samples were calibrated with the control (or average of control) samples, following the 2^−ΔΔCt method.

Muscle immunofluorescence

Muscle tissues frozen in optimal cutting temperature embedding medium were cross-sectioned (10 µm thickness) from control and patient muscle biopsies, fixed in precooled 100% methanol at −20°C for 5 min and washed in phosphate buffered saline (PBS). Sections were blocked in PBS with 10% fetal bovine serum and 10% goat serum with 0.1% Triton X-100 for 30 min. Primary antibodies were diluted in the blocking buffer and incubated overnight at 4°C [anti-collagen VI mouse monoclonal antibody MAB3303 (1:2500) and anti-laminin rabbit antibody L9393 (1:800)]. Alexa-568- and Alexa-488-conjugated secondary antibodies (1:500; Invitrogen) were used, and immunofluorescence images were obtained using Zeiss Airy confocal microscope.

Statistical analysis

Summary statistics for onset of ambulation, loss of ambulation and onset of NIV were described using the mean ± standard deviation (SD). All statistical tests were conducted with a significance level of 0.05. A Mann–Whitney U-test was performed (with Holm–Šídák multiple comparison) to compare the onset of ambulation, loss of ambulation and onset of NIV for UCMD patients heterozygous for the COL6A1 c.930+189C>T variant with UCMD patients having causative variants in the COL6 genes (other than the COL6A1 c.930+189C>T variant). Log-rank tests were performed for the Kaplan–Meier curves.

Results

COL6A1 variant

We recently reported the initial identification of the COL6A1 c.930+189C>T variant by RNA sequencing of muscle biopsy samples from patients with a clinical diagnosis of COL6-RD without an identified pathogenic variant in the COL6 genes (n = 4).²⁰ We subsequently used targeted Sanger sequencing of this variant on DNA samples of additional patients with a clinical phenotype of COL6-RD who remained without an identified causative variant in the COL6 genes (COL6A1, COL6A2 and COL6A3) on panel or whole-exome sequencing, or so-called ‘causative variant negative’ COL6-RD, and identified the COL6A1 c.930+189C>T variant (n = 27).²¹ Following the inclusion of this variant on diagnostic next-generation sequencing panels, it was identified in additional patients (n = 13), including one patient (US16) with evidence of somatic mosaicism for the COL6A1 c.930+189C>T variant. In all patients in whom parental segregation testing was performed, the variant was confirmed to be de novo.

Clinical presentation

Demographic and clinical features of the 44 patients, including the patient with somatic mosaicism, are listed in Table 1, with additional details provided in Supplementary Table 1. Sixteen patients were identified in the USA (U1–U16), five in the UK (UK1–UK5), five in Italy (I1–I5), three in Canada (CA1–CA3), two patients each in France (F1–F2), Spain (S1–S2) and Latvia (L1–L2), and one patient each was identified in Australia (A1), Chile (CH1), Finland (FI1), Germany (G1), Hong Kong (HK1), Ireland (I1), Israel (IS1), Romania (R1) and Turkey (T1). Twenty-seven of the patients were female (61%) and 17 male (39%). Three patients (IT1, IT2 and R1) were included in a report of a cohort of patients with COL6-RD in Italy.²² Phenotypic data on the clinical presentation of patients with a phenotype of UCMD attributable to causative variants in the COL6 genes other than the COL6A1 c.930+189C>T variant were evaluated to serve as a comparison cohort to the COL6A1 c.930+189C>T-specific cohort. Seventeen patients with UCMD, twelve females (71%) and five males (29%) evaluated at the National Institutes of Health (NIH), were included in this comparison cohort (Supplementary Table 2).

Table 1.

Core phenotypic features of patients with Ullrich congenital muscular dystrophy attributable to COL6A1 c.930+189C>T

Patient identifier	Sex	Race/ethnicity	Age at last clinical assessment (years)	Age started walking (months)	Age started using wheelchair full-time (years)	Age started BiPAP (years)
US1	F	Hispanic/Mexican	15	18	6	8
US2	M	Caucasian/Ashkenazi Jewish	29	14	11	18
US3	M	Hispanic	15	22	3.5	14
US4	F	Caucasian/American	30	24	12	21
US5	F	Caucasian/American	26	13	10	21
US6	F	Caucasian/American	23	15	13	14
US7	F	Caucasian/American	22	15	5	13
US8	F	Hispanic	17	13	10	Not yet
US9a	M	Black/American	18	15	10	Refused
US10	M	Hispanic/Guatemalan and Costa Rican	16	13	12	8
US11	M	Hispanic	10	14	8	11
US12	M	Asian/Indian	7	15	Still ambulant	5
US13	F	Caucasian/American	9	24	4	7
US14	F	Caucasian/American	7	Never	N/A	5
US15	F	Hispanic/Colombian	17	24	7	11
UK1	M	Caucasian/British	12	17	8	13
UK2	M	Caucasian/Irish	16	23	11.5	12
UK3	F	Caucasian/British	38	21	13	15
UK4	F	Caucasian/Greek	16	Unknown	8	8
UK5	F	Caucasian/Spanish	6	12	Still ambulant	Not yet
IT1	F	Caucasian/Italian	10	18	6.5	11
IT2	F	Caucasian/Italian	11	18	4.5	11
IT3	M	Caucasian/Italian	9	14	5	9
IT4	F	Caucasian/Italian	7	18	Still ambulant	7
IT5	M	Caucasian/Italian	17	14	12	14
CA1	F	Asian/Indian	12	Never	N/A	12
CA2	M	Asian/Bengali	13	18	9	13
CA3b	M	Caucasian/Canadian	28	24	9	12
F1	M	Caucasian/French	18	15	6	12
F2	F	Caucasian/French and African/Algerian	31	13	14	Initially refused 26
S1	M	Caucasian/Spanish	7	16	Still ambulant	Not yet
S2	F	Caucasian/Spanish	8	11	Still ambulant	Not yet
L1	M	Caucasian/Latvian	16	12	6	16
L2	F	Caucasian/Latvian	7	16	Still ambulant	Not yet
A1	F	Caucasian/Australian	12	24	8	11
CH1	F	Hispanic/Chilean	24	14	10	21
FI1	F	Caucasian/Finnish	11	18	6	Not yet
G1	M	Caucasian/German	19	14	10	Not yet
HK1	F	Asian/Chinese	12	18	8	Not yet
IR1	M	Caucasian/Irish	16	13	9	11
IS1	F	Caucasian/Arab	9	20	Still ambulant	Not yet
R1	F	Caucasian/Romanian	12	18	9	12
T1	F	Caucasian/Turkish	8	14	6	Not yet
US16c	F	Caucasian/American	9	12	Still ambulant	Not yet

BiPAP = bilevel positive airway pressure; F = female; M = male; N/A = not applicable. Additional details are provided in Supplementary Table 1.

^aDied (age 21 years, owing to probable cor pulmonale related to respiratory insufficiency).

^bDied (age 33 years owing to respiratory failure and failure to thrive).

^cThis patient has somatic mosaicism for COL6A1 c.930+189C>T.

Phenotype of patients heterozygous for the COL6A1 c.930+189C>T variant

A history of decreased fetal movements during pregnancy was reported by the respective mothers of 43% (17/40) of the patients. Of those patients with findings documented at birth, 34% (14/41) had hip dislocation and/or hip dysplasia, 27% (11/41) had evidence of hypotonia, 15% (6/39) had abnormal positioning of the hands and feet, and 10% (4/41) had torticollis (Fig. 1A). The mean onset of independent ambulation was 1.4 ± 0.3 years, with two patients reported never to have attained independent ambulation (Fig. 1B).

Figure 1.

Clinical phenotype of patients heterozygous for the COL6A1 c.930+189C>T variant versus patients with classical UCMD (UCMD patients who do not have the COL6A1 c.930+189C>T variant). (A) Bar graph demonstrating symptoms at birth in patients with the classical UCMD phenotype (blue) versus the COL6A1 intron 11 phenotype (green). (B) Box plot demonstrating distribution of ages at the time of major motor function and pulmonary function milestones. Onset of independent ambulation: (classical UCMD phenotype, n = 17; blue): box range 1.3–2.0 years; whiskers: 1.0–3.0 years. Onset of independent ambulation (COL6A1 intron 11 phenotype, n = 40; green): box range: 1.2–1.5 years; whiskers: 0.9–2.0 years. Loss of ambulation (classical UCMD phenotype, n = 15; blue): box range: 5.3–10.0 years; whiskers: 4.0–11.0 years. Loss of ambulation (COL6A1 intron 11 phenotype, n = 34; green): box range: 6–10.3 years; whiskers: 3.5–14 years. Onset of non-invasive ventilation (classical UCMD phenotype, n = 12; blue): box range: 7.3–11.0 years; whiskers: 5.0–13.0 years. Onset of non-invasive ventilation (COL6A1 intron 11 phenotype, n = 31; green): box range: 9.0–14.0 years; whiskers: 5–21 years. Asterisks indicate significance at the 0.05 level for Mann–Whitney U-tests. (C) Kaplan–Meier curve depicting independent ambulation in patients with classical UCMD (blue) and patients with COL6A1 intron 11 (green) (P = 0.04). (D) Kaplan–Meier curve depicting ventilation-free status in patients with classical UCMD (blue) and patients with COL6A1 intron 11 (green) (P = 0.003). NIV = non-invasive ventilation; UCMD = Ullrich congenital muscular dystrophy.

The age at the time of last clinical evaluation ranged 7 to 38 years. Thirty-four patients had lost independent ambulation (as defined by full-time wheelchair dependence) by the time of the last clinical evaluation, with a mean age at loss of ambulation of 8.5 ± 2.8 years (Fig. 1B and C). Thirty-two patients had started NIV, in the form of bilevel positive airway pressure (BiPAP), by the time of the last clinical evaluation, at a mean age of 12.0 ± 4.2 years (Fig. 1B and D). Two patients in this cohort died: patient US9 at 21 years of age from probable cor pulmonale, in the setting of choosing not to use NIV, and patient CA3 at 33 years of age, in the setting of respiratory failure and failure to thrive.

Joint contractures of the elbows, knees and wrists were assessed on clinical examination and categorized as ‘mild’ (<45°), ‘moderate’ (45°–90°) and ‘severe’ (>90°). At the time of detailed clinical evaluation of contractures, 16% (7/43) of patients had ‘mild’ joint contractures (ages 4–10 years), 37% (16/43) of patients had ‘moderate’ joint contractures (ages 3–16 years), and 47% (20/43) of patients had ‘severe’ joint contractures (ages 7–38 years) (Fig. 2A–E).

Figure 2.

Joint contractures in heterozygous patients versus a patient with somatic mosaicism for COL6A1 c.930+189C>T. (A) Mild elbow contractures already noticeable at 3.5 years of age in Patient US1. (B) Severe long finger flexor contractures at age 10 years in Patient IR1. (C) Severe wrist flexion and long finger flexor contractures at age 15 years in Patient US9. (D) Severe elbow, wrist flexion and long finger flexor contractures at age 24 years in Patient US5. (E) Severe elbow contracture (>90° in elbow flexion) at age 29 years in Patient US2. (F) Joint hyperlaxity of the elbow at age 9 years in Patient US16, who has somatic mosaicism for COL6A1 c.930+189C>T.

Scoliosis requiring surgical repair was reported in 39% (15/38) of patients, with surgery performed in 10 patients, ranging in age at the time of surgery between 6 and 15 years. Three families declined scoliosis surgery, and two families opted to postpone surgical intervention at the time of the last assessment.

Phenotype of a patient with somatic mosaicism for the COL6A1 c.930+189C>T variant

Patient US16 has a distinctly milder clinical phenotype. Her mother denied noting evidence of decreased fetal movement during the pregnancy. At the time of birth, hypotonia was noted, but there was no evidence of hip dislocation and/or hip dysplasia, abnormal positioning of hands and feet or torticollis. Patient US16 walked independently at 12 months of age. At the time of her last clinical evaluation (age 9 years), she maintained the ability to ascend and descend stairs without holding onto the railing. She demonstrated the ability to ambulate independently with a Trendelenburg gait and was able to run with an exaggerated arm swing. Pulmonary function testing performed at age 9 years demonstrated a forced vital capacity of 91% of predicted upright and 79% of predicted supine. There was evidence of only mild contractures of the long finger flexors and Achilles tendons and hyperlaxity of the elbow joints (Fig. 2F). There was no evidence of scoliosis.

UCMD patient phenotype attributable to causative COL6 variants (apart from the COL6A1 c.930+189C>T variant)

Of those patients with findings documented at birth, 88% (15/17) had hip dislocation and/or hip dysplasia, 63% (10/16) had evidence of hypotonia, 38% (6/16) had abnormal positioning of hands and feet, and 25% (4/16) had torticollis (Fig. 1A). All patients had attained independent ambulation at a mean age of 1.6 ± 0.5 years (Fig. 1B). The age at the time of last clinical evaluation ranged from 5 to 29 years. Fifteen patients had lost independent ambulation (as defined by full-time wheelchair dependence) by the time of the last clinical evaluation, with a mean age at loss of ambulation of 7.1 ± 2.6 years (Fig. 1B and C). Twelve patients had started NIV, in the form of bilevel positive airway pressure, by the time of last clinical evaluation, at a mean age of 8.9 ± 2.3 years (Fig. 1B and D). One patient in this cohort died at age 17 years of respiratory failure and failure to thrive in the setting of choosing not to use NIV.

Comparison of the phenotype of ‘COL6A1 intron 11’ and the phenotype of ‘classical UCMD’

Based on this study, there was no statistically significant difference for the mean age of onset of independent ambulation for the ‘COL6A1 intron 11’ patients (owing to heterozygosity for the COL6A1 c.930+189C>T variant) and the ‘classical UCMD’ patients (owing to causative variants in the COL6 genes other than the COL6A1 c.930+189C>T variant) (P = 0.08) or the mean age at loss of ambulation (P = 0.31) (Fig. 1B). The difference between the mean age at onset of NIV between patients with the COL6A1 intron 11 phenotype and patients with the classical UCMD phenotype was statistically significant (P = 0.04) (Fig. 1B). Kaplan–Meier curves depicting the probability of independent ambulation (Fig. 1C) and the ventilation-free probability (Fig. 1D) for the COL6A1 intron 11 patients and the classical UCMD patients demonstrated a statistically significant difference (P = 0.04 and P = 0.003, respectively).

Muscle imaging

Muscle MRI (not available at all centres, and challenging to perform in this cohort owing to the need for NIV while in a supine position for the MRI, in addition to joint contractures that complicate positioning in the MRI scanner) was performed in 30% (13/43) of patients heterozygous for the COL6A1 c.930+189C>T variant and demonstrated abnormal signal on T1-weighted images, with a ‘central cloud’ pattern of abnormal signal along the central fascia of the rectus femoris muscle and an ‘outside-in’ pattern of abnormal T1 signal along the periphery of the vastus lateralis, as appreciated in Patient US10 at age 8 years (Fig. 3A) and as is classically seen in patients with COL6-RD.^23-25 Muscle MRI performed in Patient US16 (who has somatic mosaicism for the COL6A1 c.930+189C>T variant) at age 9 years (Fig. 3D) demonstrated a mildly abnormal T1 signal in the vastus lateralis muscle.

Figure 3.

Muscle MRI and ultrasound in a heterozygous patient versus a patient with somatic mosaicism for COL6A1 c.930+189C>T. (A) Axial TI-weighted MRI of the upper leg in Patient US10 (heterozygous for COL6A1 c.930+189C>T) at age 8 years demonstrating abnormal T1-weighted signal, consistent with a ‘central cloud’ pattern in the rectus femoris muscle (black arrows) and an ‘outside-in’ pattern in the vastus lateralis muscle (white arrowheads). (B) Ultrasound of the rectus femoris muscle in Patient US10 at age 8 years, demonstrating increased echogenicity, consistent with a ‘central cloud’ pattern (white arrow) with a loss of bone echogenicity. (C) Ultrasound of the vastus lateralis muscle in Patient US10 at age 8 years, demonstrating increased echogenicity with a loss of bone echogenicity. (D) Axial TI-weighted MRI of the upper leg in Patient US16 (with somatic mosaicism for COL6A1 c.930+189C>T) at age 9 years, demonstrating mildly abnormal signal in the vastus lateralis muscle suggestive of a subtle ‘outside-in’ pattern (black arrowheads). (E) Ultrasound of the rectus femoris muscle in Patient US16 at 9 years of age, demonstrating an increase in echogenicity of the rectus femoris muscle consistent with a ‘central cloud’ pattern (white arrow) with bone echogenicity (asterisk) preserved. (F) Ultrasound of the vastus lateralis muscle in Patient US16 at 9 years of age, demonstrating an increase in echogenicity of the vastus lateralis with bone echogenicity (asterisk) preserved. RF = rectus femoris; VL = vastus lateralis.

Muscle ultrasound was performed in 28% (12/43) of patients heterozygous for the COL6A1 c.930+189C>T variant and demonstrated significantly increased echogenicity in a granular quality and a ‘central cloud’ pattern of increased echogenicity along the central fascia of the rectus femoris muscle and an ‘outside-in’ pattern of increased echogenicity along the outer region of the vastus lateralis, a classic muscle ultrasound pattern in patients with COL6-RD,²⁴ as appreciated in muscle ultrasound images performed in Patient US10 at age 8 years (Fig. 3B and C). Muscle ultrasound performed in the mosaic Patient US16 at age 9 years demonstrated increased echogenicity in a ‘central cloud’ pattern in the rectus femoris muscle, with generally increased echogenicity in the vastus lateralis and maintenance of bone echogenicity (Fig. 3E and F), in contrast to patients heterozygous for COL6A1 c.930+189C>T, in whom bone echogenicity was lost owing to the degree of increased echogenicity seen in the muscles.

Muscle immunofluorescence

Collagen VI immunofluorescence was performed on available muscle biopsy tissue in 53% (23/43) of patients heterozygous for the COL6A1 c.930+189C>T variant and demonstrated mislocalization of collagen VI immunoreactivity, which was found to be accumulated in the interstitial space instead of co-localizing with laminin at the basement membrane, as demonstrated in confocal microscopy images of the muscle biopsy of patient US1 when compared with control muscle (Fig. 4).

Figure 4.

Muscle immunofluorescence. Confocal imaging of muscle co-stained with collagen VI (red) and basement membrane marker laminin (green) along with the nuclear stain DAPI (blue). (A) In control muscle, collagen VI is co-localized with laminin at the basement membrane. (B) In the muscle of Patient US1 from a biopsy performed at age 2 years, collagen VI signal is observed in the interstitial space, indicative of collagen VI mislocalization relative to the basement membrane. (Magnification = ×63; scale bars = 75 µm).

Somatic mosaicism for COL6A1 c.930+189C>T

Patient US16 was found to have somatic mosaicism for COL6A1 c.930+189C>T based on Sanger sequencing of genomic DNA performed in various tissues (Fig. 5A). The degree of mosaicism, calculated from the fractional abundance (or percentage) of the variant (T) allele as measured by digital droplet PCR, was determined to be ∼20%, meaning that ∼40% of cells in this individual harbour the c.930+189C>T variant (Fig. 5B). Tissues tested were derived from the endoderm (bladder epithelium in the urine sample), the mesoderm (blood cells and dermis of the skin biopsy sample) or the ectoderm (epidermis of the skin biopsy sample and buccal epithelium in the saliva sample) (Fig. 5B). Consistent with this finding, expression levels of COL6A1 transcripts including the pseudoexon as assessed in a fresh skin biopsy sample and in cultured dermal fibroblasts, were lower in patient US16 compared with samples obtained in patients heterozygous for COL6A1 c.930+189C>T (Fig. 5C). By quantitative PCR performed in fresh skin biopsies, we determined that the COL6A1 transcripts with inclusion of the pseudoexon were 7.2-fold lower in the patient with somatic mosaicism for COL6A1 c.930+189C>T (US16) compared with a patient heterozygous for COL6A1 c.930+189C>T (US5) (Fig. 5D), although additional samples would be needed to determine statistical significance.

Figure 5.

Somatic mosaicism for COL6A1 c.930+189C>T. (A) Genomic DNA sequencing chromatograms at the COL6A1 c.930+189C>T locus, showing comparable peak heights for the cytosine and thymine alleles in a patient heterozygous for COL6A1 c.930+189C>T (Patient CA1), but a higher cytosine peak height compared with thymine in the patient mosaic for COL6A1 c.930+189C>T (Patient US16), in various tissue samples (skin fibroblasts, skin biopsy and blood). (B) Determination of the degree of mosaicism was achieved by droplet digital PCR quantification using a genotyping probe assay and genomic DNA as input. The graph shows the fractional abundance of the thymine (‘T’) allele, calculated as the ratio of ‘T’ concentration (in copies per microlitre) over total (‘T’ + ‘C’) concentration (in copies per microlitre). Error bars represent the Poisson confidence interval. [The heterozygous patients were Patient HK1 (blood) and Patient US12 (skin fibroblasts).] (C) RNA isolated from skin biopsies or from skin-derived primary fibroblasts was reverse transcribed and amplified with primers spanning COL6A1 exons 10–20. In Patient US16 (mosaic for COL6A1 c.930+189C>T), the upper band (transcripts with pseudoexon) appears fainter than in Patient US12 and Patient US5 (heterozygous for COL6A1 c.930+189C>T). (D) Relative expression of COL6A1 with pseudoexon transcripts normalized to total COL6A1 levels in skin fibroblasts and skin biopsy of the patient mosaic for COL6A1 c.930+189C>T (Patient US16), compared with two patients heterozygous for COL6A1 c.930+189C>T (Patient US5 and Patient US12). A fresh skin biopsy was not available for Patient US12.

Discussion

The COL6-related dystrophies typically manifest symptoms at birth and thus are at their core congenital muscular dystrophies. The COL6-RD subtype UCMD is characterized by prominent symptoms at the time of birth, including significant hypotonia, proximal joint contractures, distal hyperlaxity, abnormal positioning of the hands and feet, prominent calcanei, hip dislocation(s), torticollis and kyphoscoliosis.^1,6-9 Here, we report an international cohort of 44 patients habouring a de novo COL6A1 c.930+189C>T deep intronic pseudoexon-inducing variant, and we delineate their presentation and natural history in comparison to patients with classical UCMD.

A hallmark of this COL6A1 c.930+189C>T-specific or ‘COL6A1 intron 11’ cohort is that all patients who are heterozygous for this causative variant (n = 43) demonstrate a paucity of congenital symptoms, followed by an apparent accelerated progression of symptoms, ultimately demonstrating a phenotype consistent with the well-defined phenotype of UCMD (as distinguished from other COL6-RD phenotypes).^2,12,14,26 In fact, only approximately one-third or less of the patients heterozygous for COL6A1 c.930+189C>T had evidence of any symptoms at birth. Furthermore, with a mean age of 1.4 years at the onset of independent ambulation, most patients in this COL6A1 c.930+189C>T-specific cohort did not present to a neurologist until the time of delayed independent ambulation or afterwards, when evidence of difficulty arising from the floor or frequent falls were noted. In contrast, in a comparison cohort of patients with ‘classical UCMD’ phenotype who do not harbour the COL6A1 c.930+189C>T variant (n = 17), symptoms at birth were present in up to 88% of patients, thus typically prompting a neurological evaluation in the early neonatal period.

The mean age at loss of independent ambulation in this COL6A1 intron 11 cohort is 8.5 ± 2.8 years, which is not statistically different from our comparison cohort of patients with classical UCMD (non-COL6A1 c.930+189C>T UCMD; n = 17) of 7.1 ± 2.6 years (Fig. 6). The mean age at the time of initiation of NIV in patients heterozygous for COL6A1 c.930+189C>T is 12.0 ± 4.2 years. Of note, although this mean age is statistically different from the mean age at the time of NIV initiation of 8.9 ± 2.3 in our comparison cohort of classical UCMD, it is similar to the data in the largest international natural history study of pulmonary function in patients with UCMD (n = 75), in which the mean age at the time of NIV initiation was 11.3 ± 4.0 years² (Fig. 6). Given differences in practice among centres internationally relating to frequency of pulmonary function testing and polysomnogram assessments and to thresholds for starting NIV, it would probably be more accurate to compare patients heterozygous for the COL6A1 c.930+189C>T variant (n = 43) with the international classical UCMD cohort (n = 75) than with the comparison cohort of classical UCMD patients evaluated at one centre (n = 17) where the initiation of NIV tends to be more proactive.

Figure 6.

Comparison of the COL6A1 intron 11 phenotype and the classical UCMD phenotype. (A) Schematic diagram of the natural history of the motor and pulmonary function in patients with the classical UCMD phenotype in the comparison cohort (n = 17). Natural history of pulmonary function data from the largest published international natural history study of patients with UCMD (n = 75)² is included (dotted line). (B) Schematic diagram of the natural history of motor and pulmonary function in this cohort of patients who are heterozygous for COL6A1 c.930+189C>T (n = 43). UCMD = Ullrich congenital muscular dystrophy. Created in BioRender. Or Bach, R. (2025) https://BioRender.com/r9a5g60.

Overall, what distinguishes this COL6A1 c.930+189C>T-specific phenotype is a delayed onset of symptoms followed by an accelerated rate of progression of symptoms, in contrast to the classical UCMD phenotype, in which striking symptoms are evident at the time of birth (Fig. 6). All patients who are heterozygous for this causative COL6A1 intron 11 variant (not with somatic mosaicism) ultimately arrive at a highly consistent phenotype of clinical severity of motor and pulmonary function, characterized by loss of ambulation and respiratory insufficiency close in age to the patients with the classical UCMD phenotype (depending on the comparator cohort), despite the COL6A1 intron 11 patients manifesting first symptoms at a mean age of 1.4 ± 0.3 years (Fig. 6). Given this convergence of phenotypic features in patients in this COL6A1 intron 11 cohort and patients with classical UCMD, outcome measures validated in patients with the classical UCMD phenotype could be used for patients with the COL6A1 intron 11 phenotype.^27,28

In COL6-RDs, respiratory insufficiency is largely attributable to disproportionate weakness of the diaphragm.^2,29 In this particular cohort of patients with de novo COL6A1 c.930+189C>T, it is possible that the severity of joint contractures noted in this cohort might contribute to additionally decreasing the compliance of the chest wall, thus probably further exacerbating patients’ respiratory insufficiency. Studying the prospective natural history of patients with the COL6A1 intron 11 phenotype longer term in comparison with long-term natural history data collected in larger cohorts of patients with the classical UCMD phenotype internationally will further strengthen our understanding of the longer-term trajectories of respiratory insufficiency and of skeletal muscle weakness and joint contractures.

Patients heterozygous for COL6A1 c.930+189C>T have muscle imaging findings by muscle ultrasound and muscle MRI that are consistent with those findings described in association with COL6-RDs.^23-25 Thus, muscle imaging remains a very helpful tool in supporting efforts to identify a causative variant in the COL6 genes in the setting of a clinical suspicion of COL6-RD. Furthermore, muscle immunohistochemistry studies in patients harbouring COL6A1 c.930+189C>T demonstrate mislocalized collagen VI expression, as classically seen in COL6-RD attributable to dominant mutational mechanisms.⁵ Thus, in the setting of clinical examination, muscle imaging and muscle pathology findings (i.e. evidence of mislocalization of collagen VI on muscle immunohistochemistry) suggestive of COL6-RD, it is essential to consider the diagnostic possibility of COL6-RD attributable to COL6A1 c.930+189C>T and to ensure that genetic testing laboratories adequately assess for the presence of this deep intronic variant.

The exact pathomechanisms by which the COL6A1 c.930+189C>T causative variant results in a severe phenotype of UCMD characterized by an accelerated progression of symptoms in comparison to classical UCMD have been only partially elucidated. One hypothesis to explain the initial delay in the presentation of symptoms is that the splicing events leading to inclusion of the pseudoexon might be differentially regulated pre- and postnatally. In muscle biopsy tissue samples from patients, we found that ∼25% of total COL6A1 transcripts include the 72-nucleotide pseudoexon (as opposed to the expected 50%), which is probably attributable to the ‘leakiness’ of the c.930+189C>T variant that allows for some normal splicing to occur.^20,21 It is conceivable, however, that during early development the splicing events leading to inclusion of the pseudoexon might occur even less frequently, and occur more frequently as patients age, then resulting in higher expression levels of the pseudoexon over time. Additional hypotheses relate to the unique properties of the mutant protein produced, which might be more stable and less prone to normal turnover and which might lead to a cumulative effect over time, amplifying the dominant-negative effect. Whether this dominant-negative effect is based on interference with collagen VI assembly and function or abnormal protein aggregation or a combination thereof remains to be elucidated.

The mutant collagen α1(VI) protein produced includes a stretch of 24 amino acid residues that disrupts the Gly-X-Y repeat in its amino-terminus,²¹ prior to the cysteine residue involved in dimerization, a ‘hot-spot’ where dominant-negative COL6 variants are known to allow assembly of mutant chains into tetramers. It can thus be assumed that the mutant collagen α1(VI) protein containing the pseudoexon-encoding sequence exerts a dominant-negative effect by assembling into tetramers and consequently hampering polymerization of collagen VI tetramers. In fact, co-staining with a mutation-specific antibody and a collagen α3(VI) N-terminus-specific antibody showed a broad overlap in immunofluorescence microscopy in patient muscle.³⁰ However, when patient fibroblast cell culture supernatants were studied by composite agarose/polyacrylamide gel electrophoresis and immunoblotting, it was found that the mutant collagen α1(VI) was secreted as single chains.³⁰ Interestingly, as in the supernatant from healthy control fibroblasts, wild-type collagen VI tetramers were found in the supernatant of patient fibroblast cultures. Indeed, in negative stain electron micrographs of immunogold-labelled supernatants of patient fibroblasts, collagen VI microfibrils with the typical spacing of the globular beads were detected with a gold-labelled collagen α3(VI) N-terminus antibody. In contrast, the gold-labelled mutation-specific antibody bound to protein aggregates and sometimes decorated collagen VI microfibrils.³⁰ The two effects of aggregated mutant collagen α1(VI) chains and the decoration of collagen VI microfibrils are not mutually exclusive and could be cumulative with mutant collagen VI accumulating in the matrix over time, which might contribute to the acceleration of clinical symptoms once the disease starts.

The strikingly milder clinical phenotype observed in Patient US16, who has somatic mosaicism for the COL6A1 c.930+189C>T variant, for whom expression of pseudoexon transcripts was 7.2-fold lower compared with the heterozygous patient US5 (as assessed in respective skin biopsies), highlights how, in principle, a reduction in the abundance of the pseudoexon would translate to an amelioration of clinical symptoms. In particular, the observation that Patient US16 continues to ascend and descend stairs without use of the railing at age 9 years suggests a phenotype consistent with Bethlem MD.²⁶ In keeping with the milder motor phenotype of this patient is her mildly affected pulmonary function at age 9 years, with a forced vital capacity of 91% upright and 79% supine. Thus, this patient might represent an in vivo scenario of our previously published in vitro rescue of the pseudoexon insertion with splice-modulating antisense oligomers, which effectively decreased the levels of pseudoexon transcripts by ∼7-fold (for PMO-PEX1, at the highest concentration tested, in cultured fibroblasts) and, consequently, levels of the aberrant protein product.²¹

Conclusion

Taken together, given the clinical severity of UCMD described in this large international cohort of patients harbouring COL6A1 c.930+189C>T, the recognition that this variant is one of the most common recurrent causative variants in COL6-RDs and the promise of therapeutic rescue as demonstrated by our pseudoexon skipping/splice-modulating in vitro work,^21,31,32 it is imperative that patients harbouring this deep intronic variant are clinically recognized and diagnosed. To this end, it is necessary for next-generation sequencing panels to include an algorithm to ensure that intron 11 of COL6A1 is captured, including libraries built for high throughput that would capture this variant. Moreover, we suggest designating the COL6A1 c.930+189C>T-specific phenotype described here as a ‘COL6A1 intron 11’ phenotype, given the need to distinguish this phenotype, characterized by a delayed onset of clinical symptoms followed by an accelerated progression of symptoms, from other patients with COL6-RD for the purpose of natural history studies in preparation for future clinical trials. Distinguishing this so-called ‘COL6A1 intron 11’ phenotype is essential from the perspective of inclusion criteria for clinical trial stratification, including for non-variant-specific therapeutic approaches for COL6-RDs, such as therapeutic approaches targeting transforming growth factor beta (TGF-β), fibrosis and apoptosis. Our characterization of this COL6A1 c.930+189C>T-specific cohort, including the comparative retrospective natural history of patients with the COL6A1 intron 11 phenotype to patients with the classical UCMD phenotype, contributes to the clinical trial readiness of the COL6A1 intron 11 patient population, thus helping to enable the realization of the promise of the COL6A1 c.930+189C>T variant-specific splice-modulating therapeutic approaches currently in development.^21,31,32

Data availability

All de-identified data are available upon request from the corresponding author.

Funding

This work was supported by intramural funds from the NIH National Institute of Neurological Disorders and Stroke (grant to C.G.B.) and Muscular Dystrophy UK funds (to C.G.B., F.M. and H.Z.). F.G. is supported by the NIH Medical Research Scholars Program, a public–private partnership supported jointly by the NIH and contributions to the Foundation for the NIH from the Doris Duke Charitable Foundation, Genentech, the American Association for Dental Research, the Colgate-Palmolive Company and other private donors. G.H. is supported by GREGoR Consortium (Genomics Research to Elucidate the Genetics of Rare Diseases), and research in this publication was supported by the National Human Genome Research Institute of the National Institutes of Health under Award Number U24HG011746. Al.Sc. is supported by Deutsche Forschungsgemeinschaft through project ID 384170921: SCHI 1627/2-2. C.J.M. was supported by Plan Nacional de I + D + I and Instituto de Salud Carlos III (ISCIII), Subdirección General de Evaluación y Fomento de la Investigación Sanitaria (PI22/01382).

Competing interests

The authors report no competing interests.

Supplementary material

Supplementary material is available at Brain online.