Background

Dystroglycanopathies are a group of muscular dystrophies resulting from abnormal glycosylation of α-dystroglycan (α-DG) that leads to reduced binding affinity for extracellular matrix proteins [1]. The clinical phenotypes span a broad range from the congenital muscular dystrophies (CMDs) with brain and eye malformations to adult-onset limb-girdle muscular dystrophy (LGMD) [2]. Dystroglycan is encoded by DAG1 and is cleaved into α-DG and β-DG after translation [3]. DAG1 is widely expressed in different human tissues, consistent with the multi-organ phenotypes of many individuals with the most severe forms of dystroglycanopathy [4].

Mutations in DAG1 itself as well as 17 other genes have been reported in patients with dystroglycanopathy. These include glycosyltransferases (POMT1, POMT2, POMGNT1, POMGNT2, B3GALNT2, B3GNT1, LARGE, TMEM5), a kinase (POMK), five genes encoding enzymes necessary for dolichol-P-mannose (dol-P-man) synthesis (DOLK, DPM1, DPM2, DPM3, and GMPPB), and three genes encoding proteins necessary for joining the α-DG-linked core glycan structure with the distal ligand-binding region of the structure via a ribitol phosphate disaccharide (FKTN, FKRP, ISPD) [5]. To date, no membrane trafficking proteins have been implicated in dystroglycanopathies.

In this study, we report clinical, histopathological, biochemical, and molecular genetic data on two families with CMD and hypoglycosylation of α-DG. Two genes, TRAPPC11 and GOSR2, that each have a role in membrane trafficking in the biosynthetic pathway have been implicated as candidate dystroglycanopathy genes. They represent the first membrane trafficking proteins implicated in α-DG hypoglycosylation. Since TRAPPC11 mutations have been reported in a number of individuals suffering from a muscular dystrophy, and these individuals also display membrane trafficking defects in cultured fibroblasts, this gene should be considered in the diagnostic evaluation of patients with CMD.

Case presentation

Family 1

Subject 1 presented with status epilepticus in the setting of a vomiting illness at 6 months of age. Magnetic resonance imaging (MRI) of the brain showed bilateral multifocal restricted diffusion of the cortex, the cerebral white matter, and the pons (Fig. 1a). Her serum aminotransferases were elevated with alanine aminotransferase (ALT) of ~ 1600 U/L and aspartate aminotransferase (AST) ~ 400 U/L as well as a prolonged prothrombin time of 20.7 s (normal range is 12–15 s), consistent with synthetic liver dysfunction. The approximately 4:1 ALT to AST ratio was consistent over multiple measurements. Creatine kinase (CK) at initial presentation was 3500 U/L. She had significant regression of development with loss of rolling and sitting, loss of fine motor and verbal skills, and inability to feed orally after this illness.

Fig. 1
figure 1

Subjects 1 and 3 display brain, liver, and muscle abnormalities. a Diffusion-weighted (B1000) MRI showing restricted diffusion of the medial occipital cortex and underlying white matter at 6 months of age in subject 1 at the time of initial presentation. b Fluid-attenuated inversion recovery (FLAIR) MRI for subject 1 at 15 months notable for marked cerebral volume loss. c Short tau inversion recovery (STIR) shows symmetric high signal in the posterior compartments of the legs of subject 1 at 12 months of age. Subject 1 has microvesicular steatosis of the liver; light microscopy hematoxylin and eosin (d) and electron microscopy (e). Note the lipid accumulations marked by the arrows in e. f–h Muscle biopsies from control (f), subject 1 (g), and subject 3 (h) were stained with hematoxylin and eosin. Dystrophic features are present in subjects 1 and 3. The size bar denotes 50 μm in d and f–h. The size bar denotes 5 μm in e

Full size image

MRI of the lower extremities showed high signal on short tau inversion recovery (STIR) sequences of the deep and superficial posterior compartments bilaterally (Fig. 1c). Skeletal muscle and liver biopsies were obtained at 9 months of age. Liver biopsy showed microvesicular steatosis (Fig. 1d, e). Skeletal muscle showed an active dystrophic process (Fig. 1g) and hypoglycosylation of α-DG by both immunofluorescence and western blotting (Fig. 2). In contrast, α-DG in cultured fibroblasts was indistinguishable from control fibroblasts in on-cell and in WGA glycoprotein western blots (data not shown). This is not uncommon and has been reported in the case of other genes involved in dystroglycanopathy [6, 7].

Fig. 2
figure 2

Subjects 1 and 2 display abnormalities in both α-dystroglycan staining and glycosylation. Control muscle or muscle taken from subject 1 (S1) and subject 3 (S3) were stained for alpha dystroglycan using VIA4-1 antibody (a) or β-DG (b). Note the reduced staining for α-DG but not β-DG in subjects 1 and 3. The size bar denotes 50 μm for all panels in a and b. c Western blot analysis of muscle tissue from control and subjects 1 and 3. Samples were probed with peptide-specific antibody AF6868 and the glycoepitope-specific antibody IIH6 as indicated. The location of α-DG and β-DG is indicated. Note that control shows a higher molecular size immunoreactive species for α-DG with both antibodies while S1 and S3 show a more heterogeneous species of much smaller molecular size, suggesting hypoglycosylation of the protein

Full size image

The patient had several subsequent acute infectious illnesses with seizures and neurological regression. Follow-up brain MRI at 15 months of age showed marked progressive volume loss (Fig. 1b). Glycosylation analysis of transferrin and ApoCIII proteins in serum by affinity chromatography-mass spectrometry (Mayo Medical Laboratory) as well as by MALDI-TOF mass spectrometry (Emory Genetics Laboratory) was normal. In her last evaluation, at 3 years and 6 months of age, she was areflexic with limited antigravity strength and severe hypotonia. She fed exclusively via gastrostomy tube with no verbal communication. Seizures were well controlled on levetiracetam monotherapy. CK remained elevated with values as high as 19,000 U/L. She had mild hepatomegaly and aminotransferases were still significantly elevated with 4:1 ALT to AST ratio, but there was no coagulopathy or hyperbilirubinemia. She chronically required noninvasive positive pressure ventilation with sleep. After multiple hospital admissions for viral respiratory infections, she underwent immunological evaluation and was found to have impaired natural killer cell function on multiple repeated analyses. She did not have peripheral neuropathy, cataracts, alacrima, achalasia, renal disease, hearing loss, or cholestasis.

Exome trio sequencing showed compound heterozygous rare variants in trans in TRAPPC11 (NM_021942): c.851A>C (p.Q284P) and c.965+5G>T (intron 9 splice site disruptor). The p.Q284P missense mutation was absent from the Exome Aggregation Consortium (ExAC) database, and c.965+5G>T was present in 2/119,770 alleles [8]. The latter mutation resulted in a transcript that lacks exon 9 and the first 88 bases of exon 10 (Fig. 3a) and is predicted to result in an in-frame deletion of amino acids 278–351 (p.I278_Q351del). Cultured fibroblasts had greatly reduced levels of TRAPPC11 (Fig. 3b) suggesting the p.Q284P protein and the predicted p.I278_Q351del protein are unstable. These fibroblasts showed a delay in the maturation of the marker protein VSVG-GFP ts045 (Fig. 3c, d). Analysis of live-cell trafficking revealed a delay in the release of VSVG-GFP ts045 from the Golgi (Fig. 3e, f) as well as a delay in arrival of a Golgi marker (sialyl transferase-SBP-GFP) from the endoplasmic reticulum (Fig. 3g, h). The delayed release of protein from the Golgi is consistent with the initial findings reported by Bögershausen et al. in LGMD2S patients with TRAPPC11 mutations [9], and the delayed arrival of protein to the Golgi is consistent with the findings of Scrivens et al. [10].

Fig. 3
figure 3

TRAPPC11 compound heterozygous mutations affect membrane trafficking in patient fibroblasts. a mRNA was collected from control and subject 1 (S1), converted to cDNA and amplified by PCR using oligonucleotides annealing to exons 8 and 11. The amplicons were sequenced and found to represent exons 8-9-10-11 (higher molecular size amplicon) and exons 8-part of 10-11 (lower molecular size amplicon). b Lysates from control and subject 1 (S1) fibroblasts were probed for TRAPPC11 and tubulin as a loading control. c Fibroblasts were infected with VSVG-GFP ts045, and the protein was arrested in the endoplasmic reticulum (ER) by shifting the cells to 40 °C. The protein was synchronously released from the ER upon downshifting the temperature to 32 °C, and the acquisition of Endoglycosidase H (EndoH) resistance was assayed at the times indicated. A representative western blot is displayed, and quantification of a minimum of three such blots is shown in d. e The same assay as in b was performed on live cells and the arrival and release of the GFP signal was quantified over time. Representative images from the movies are displayed in e, and quantification of the signal in the Golgi region is shown in f. To more accurately measure ER-to-Golgi trafficking, the RUSH assay [36] was performed using ST-SBP-GFP with the Ii hook (g). Images were acquired over time in live cells upon addition of biotin to initiate release of the protein from the ER. Quantification of the signal in the Golgi is displayed in h. Size bars in e and g denote 25 μm. Error bars represent SEM from a minimum of three replicates in d. N values for f and h are indicated in the figure

Full size image

Family 2

Subject 2 presented for medical evaluation at age 6 months for hypotonia. She was found to have CK values of up to ~ 5000 U/L. She developed absence seizures at age 2 years. She had steadily progressive muscle weakness. On examination at age 4 years and 9 months, she was areflexic. She had low muscle bulk and myopathic facial appearance and did not have antigravity strength. She had fine nystagmus but otherwise intact extraocular movements. MRI of the brain was normal at ages 2 and 4 years. She died due to respiratory failure at age 5 years.

Subject 3 is the younger sister of subject 2. She was noted to have muscle weakness and hypotonia at 3 months of age. At 9 months, she had antigravity strength only. Serum CK value was 1760 U/L. At 19 months, skeletal muscle biopsy was obtained showing an active dystrophic process (Fig. 1h) and hypoglycosylation of α-DG by both immunofluorescence and western blotting (Fig. 2). In contrast, α-DG in cultured fibroblasts was indistinguishable from control fibroblasts in on-cell and in WGA glycoprotein western blots (data not shown). Furthermore, the VSVG-GFP membrane trafficking assay kinetics as well as arrival of the Golgi marker was indistinguishable from control fibroblasts (Fig. 3c–f).

At age 2.5 years, she developed a seizure disorder characterized as focal seizures but later as both focal and generalized, which often became intractable and required hospitalization. Evaluation showed no evidence of nystagmus and ocular range of motion was full. There were no focal deficits and her cranial nerves were normal. She demonstrated severe weakness and muscular hypotonia. MRI of the brain showed diffuse volume loss resulting in ex vacuo ventriculomegaly. EEG at 2 years and 7 months of age showed runs of spike and wave discharges originating in the occipital lobe which were exacerbated by photic stimuli. Head circumference was at the 30th centile, height at the 10th centile, and weight below the 1st centile.

At 3.5 years of age, she developed episodes of vomiting and apparent abdominal pain. This led to the detection of elevated ALT of up to 700 U/L. An extensive evaluation for infectious, anatomical, autoimmune, and metabolic etiologies of liver disease was nondiagnostic.

Physical exam revealed findings similar to her sister. Apart from her sister, there is no family history of neuromuscular disease. The subject is now 6 years of age with medically refractory epilepsy and progressive severe muscle weakness. Clinical exome trio sequencing was performed, and no relevant sequence variants were initially reported. In a targeted sequencing panel, subject 3 was found to have compound heterozygous rare variants in GOSR2 (NM_001012511): c.430G>T (p.G144W) and c.2T>G. Retrospective evaluation of GOSR2 in the whole exome sequencing (WES) data confirmed that both variants were present in subject 3 and were in trans. Extensive re-evaluation of seizure and dystroglycanopathy loci in the WES failed to identify any other pathologic variants. The GOSR2 p.G144W missense variant is a previously reported disease-causing mutation and is present in 5/121,408 alleles in the ExAC database with no homozygous individuals. The second variant (c.2T>G) is present in 1/18,808 alleles in the ExAC database [8]. The mutation is likely to result in the use of an alternate start codon with elimination of 18 amino acids from the amino-terminus of the protein according to MutationTaster2 and is presumed to be pathogenic as a result [11].

Discussion and conclusions

In this report, we show that mutations in two genes encoding proteins involved in membrane trafficking, TRAPPC11 and GOSR2, are associated with CMD and dystroglycanopathy. Biallelic mutations in TRAPPC11 were initially reported as the etiology of LGMD2S in 2013 [9] and have since been associated with a variety of multisystemic phenotypic findings including intellectual disability, seizures, microcephaly, cerebral atrophy, cataracts, alacrima, achalasia, hepatic steatosis, and cholestatic liver disease, in addition to muscular dystrophy [9, 12,13,14,15]. Comparisons between subject 1 and all published mutations in TRAPPC11 and associated phenotypes are summarized in Table 1. Our study now adds two new mutations with functional validation and categorizes TRAPPC11-related disease as a dystroglycanopathy.

Table 1 Comparison of all known TRAPPC11 and GOSR2 mutations
Full size table

TRAPPC11 dysfunction may contribute to disease pathophysiology in several ways. Extensive functional studies of cultured fibroblasts were conducted by Bögershausen et al. [9]. They demonstrated that cells had abnormally fragmented and diffuse Golgi; delayed traffic out of the Golgi and the proteins LAMP1 and LAMP2 were found to be abnormally glycosylated. TRAPP (transport protein particle) forms several related multisubunit trafficking complexes (MTCs) that participate in the tethering of vesicles to target membranes, including vesicles associated with the Golgi [10]. Since the Golgi is the major site of protein glycosylation in the cell [16], defects in Golgi morphology and traffic can result in protein glycosylation defects. Recently, abnormal glycosylation of serum transferrin was described in a patient with compound heterozygous mutations in TRAPPC11, consistent with a type 2 disorder of glycosylation [12]. We were unable to detect abnormalities in glycosylation of serum transferrin using two different commonly employed methods. Thus, while TRAPPC11-related disease is a disorder of glycosylation, analysis of glycoepitopes of secreted proteins may not be a sensitive test for diagnostic purposes.

The zebrafish model of TRAPPC11-related disease shows generalized impairment of N-linked glycosylation as well as depletion of lipid-linked oligosaccharides (LLOs) [17]. The inability to synthesize dolichol-P-mannose (dol-P-man), a lipid-linked saccharide, is a known cause of dystroglycanopathy [7]. Expression of multiple glycosylation-related genes (including the known etiologies of dystroglycanopathy gmppb, dpm1, dpm2, and dpm3) showed significant compensatory upregulation in the trappc11 fish [17]. Interestingly, TRAPPC11 siRNA knockdown in HeLa cells had a specific inhibitory effect on glycosylation that was not present with knockdown of other components of the TRAPP complex. This led to the conclusion that TRAPPC11 may have another function that is independent of its role in vesicle transport and led to speculation that impaired LLO synthesis may be the most relevant function of TRAPPC11 in the process of protein glycosylation [17]. Finally, trappc11 zebrafish mutations were shown to lead to fatty liver via a pathological activation of the unfolded protein response. This may be relevant to subject 1 as well as the other reported individuals with hepatopathy and TRAPPC11-related disease [13]. Taken together, several mechanisms for the role of TRAPPC11 in muscular and hepatic phenotypes are known and can explain many clinical features of subject 1.

Human mutations in GOSR2 were first reported in 2011 in six individuals with the same homozygous missense mutation (c.430G>T) who had progressive myoclonus epilepsy (PME), ataxia, scoliosis, and mildly elevated serum CK (see Table 1 for a comparison between subjects 2 and 3 with all reported GOSR2 mutations) [18]. All individuals were areflexic in early childhood and were non-ambulatory by adolescence or early adulthood. Muscle histology and EMG were normal. An additional eleven individuals with similar clinical presentations and the same homozygous mutation were reported in 2013 and 2014 [19, 20]. The maximum CK value reported in any of the patients was 2467 U/L. There was no specific assessment of α-DG glycosylation in their muscle biopsies. Subjects 2 and 3 in our study have a much more severe phenotype. Since CMD represents the severe end of the clinical spectrum of GOSR2-related disease and PME represents the milder end of the spectrum, the new c.2T>G mutation resulting in CMD reported in our study likely cause more severe perturbation of Golgi function than the common c.430G>T mutation. It remains unclear which aspect of Golgi function is affected since a membrane trafficking defect in neither the VSVG-GFP marker protein nor a resident Golgi enzyme was detected. Future studies should examine the trafficking of Golgi-localized glycosyl transferases that are responsible for α-DG processing.

GOSR2 encodes a Golgi Qb-SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) protein. In the cell, GOSR2 localizes to the cis-Golgi and mediates docking and fusion of vesicles originating from the ER. There is precedent for Golgi dysfunction leading to diseases manifesting with abnormal glycosylation and multisystemic disease. Examples include the disease caused by mutations in genes that encode the COG (conserved oligomeric Golgi) complex, an MTC that localizes to the Golgi [21]. Additionally, an individual has been described with CMD due to homozygous mutations in GOLGA2, a golgin protein that also impacts Golgi function [22]. The potential for a link between aberrant Golgi trafficking and dystroglycanopathy stems from an experiment employing a modified virus that required normally glycosylated α-DG for cell entry. Knockouts of known dystroglycanopathy genes in cultured fibroblasts resulted in impaired viral cell entry. Among the other knockouts shown to impair viral cell entry were those cells with mutations in several of the COG complex genes [23].

Dystroglycanopathies result in muscular dystrophy due to dysfunctional linkage of the sarcolemma to the extracellular matrix. This linkage occurs via α-DG and relies on the synthesis of a complex LARGE-glycan for normal function [5]. Since the initial descriptions of dystroglycanopathy [1, 24,25,26], a variety of molecular mechanisms of the diseases have been discovered. Specific glycosyltransferases such as POMT1/POMT2 are required to construct the core glycan structure that is linked to α-DG [25, 26]. Mutations in DOLK, DPM1, DPM2, DPM3, and GMPPB likely lead to a deficiency of dol-P-man (a lipid-linked monosaccharide) resulting in abnormal N-linked glycosylation as well as the O-linked mannosylation defect that results in dystroglycanopathy [6, 7, 27,28,29,30,31]. LARGE synthesizes the extracellular matrix binding region of the glycan structure (matriglycan) that is distal to the core region [32]. Most recently, FKTN, FKRP, and ISPD have been implicated in the addition of ribitol phosphate molecules to link the core and ligand-binding regions of the α-DG glycan structure [33,34,35]. Our study suggests that TRAPPC11 and GOSR2 are also involved in the trafficking and glycosylation of dystroglycan in the Golgi. This represents the first report of an association between these genes and α-DG hypoglycosylation. It remains to be seen if other GOSR2 mutations associate with similar cellular and clinical phenotypes. Given the number of individuals with TRAPPC11 mutation-associated muscular dystrophy, it may be prudent for this gene to now be considered in the diagnostic evaluation of patients with dystroglycanopathy.