TRAPPC11-Related Muscular Dystrophy with Hypoglycosylation of Alpha-Dystroglycan in Skeletal Muscle and Brain

Background: TRAPPC11, a subunit of the transport protein particle (TRAPP) complex is important for complex integrity and anterograde membrane transport from the endoplasmic reticulum (ER) to the ER-Golgi intermediate compartment. Several individuals with TRAPPC11 mutations have been reported with muscle weakness and other features including brain, liver, skeletal and eye involvement. A detailed analysis of brain and muscle biopsies will further our understanding of the presentation and etiology of TRAPPC11-disease. Methods: We describe ve cases of early-onset TRAPPC11–related muscular dystrophy with a systematic review of muscle pathology and post-mortem brain pathology ndings in one individual, and membrane tracking assays in another. Results: All affected individuals presented in infancy with muscle weakness, motor delay and elevated serum creatine kinase (CK). Additional features included cataracts, liver disease, intellectual disability, cardiomyopathy, movement disorder, and structural brain abnormalities. Muscle pathology in all ve revealed dystrophic changes, universal hypoglycosylation of alpha-dystroglycan and variably reduced dystrophin-associated complex proteins. Membrane tracking assays showed defective Golgi tracking in one individual. Neuropathological examination of one individual revealed cerebellar atrophy, granule cell hypoplasia, Purkinje cell (PC) loss and dendritic neurodegeneration, reduced alpha-dystroglycan (IIH6) expression in PC and dentate neurons, and absence of neuronal migration defects. Conclusions: This report suggests that recessive mutations in TRAPPC11 are linked to muscular dystrophies with hypoglycosylation of alpha-dystroglycan. The structural brain involvement that we document for the rst time resembles the pathology previously reported in N-linked congenital disorders of glycosylation (CDG) such as PMM2-CDG, suggesting defects in multiple glycosylation pathways in this condition.


Background
Muscular dystrophies are a heterogeneous group of disorders characterised by progressive muscle weakness with a dystrophic muscle biopsy (1). Their severity ranges from congenital onset within the rst year of life (CMDs) to later onset variant, within the spectrum of limb girdle muscular dystrophies (LGMDs) (2). Multi-system involvement is a feature of several CMDs and LGMDs, particularly with structural brain abnormalities and mental retardation occurring especially in some CMDs, and cardiac involvement coexisting in several CMD and LGMD subtypes. Extracellular matrix proteins and proteins involved in glycosylation of alpha-dystroglycan are frequently involved in CMDs, while proteins located in the sarcolemma, members of the dystrophin associated protein complex (DAPC), and nuclear envelope proteins are the most common proteins involved in LGMDs. Despite signi cant advances, some individuals remain undiagnosed. Despite increasing discovery of novel disease-associated genes by application of whole exome and genome sequencing (WES and WGS), understanding the molecular mechanisms underpinning these novel genes and proteins can be challenging.
In 2013, Bogershausen et al (3) described 3 families with an unusual phenotype including muscle weakness and elevated creatine kinase (CK) resembling a LGMD, but with associated dyskinesia, ataxia, and intellectual disability. Recessive mutations in the TRAPPC11 gene were identi ed, leading to the description of LGMD2S. Since then, a few isolated cases affected by either LGMD2S or a more severe congenital form have been reported (4,5). In this study, we describe ve cases of muscular dystrophy associated with mutations in TRAPPC11 and discuss a possible role of aberrant glycosylation of alpha-dystroglycan in the disease process. We also present post mortem pathology ndings in one individual and membrane tra cking assays in another. The structural defects identi ed are nevertheless different from those typically associated in dystroglycanopathies and more closely resemble the ndings in defects of N-glycosylation disorders. Our ndings broaden the phenotype and improve our understanding of the pathology of TRAPPC11-associated disorders.

Methods
All human tissues in this study were acquired and processed under appropriate consent and institutional research ethics cover (REC reference: 13/LO/1894 (NRES Committee London). Whole exome sequence was performed on Patient 5 as part of the Deciphering Developmental Disorders (DDD) Study (Cambridge South REC ref: 10/H0305/83 and Republic of Ireland REC ref: GEN/284/12). Muscle biopsies were processed in accordance with standard protocols as previously described (6).
Quantitative ow cytometric assay for alpha-dystroglycan was performed on cultured dermal broblasts from individual 1, in accordance with a procedure previously described (8).
The tra cking of VSVG-GFP ts045 and quanti cation of the data was performed in cultured dermal broblasts from individual 1 as described by Koehler et al (9). The retention using selective hooks (RUSH) assay was performed and quanti ed as described by Milev et al (10).

Molecular genetics
Whole exome sequencing on our index case (individual 1) revealed compound heterozygous mutations in TRAPPC11: c.100C > T p. (Arg34*) and c.2938G > A p. (Gly980Arg). We further identi ed four unrelated families with four affected members who all had compound heterozygous mutations as indicated in Table 1. previously described (5), but here we provide immunoanalysis of alpha-dystroglycan glycosylation not previously performed.
All individuals acquired independent ambulation with the exception of individual 5. Cognitive function was impaired in one individual (severely affected infant with CMD, individual 5). The extra muscular features described in previous TRAPPC11 reports were sought, but were not consistently present. Two individuals had bilateral cataracts (individuals 1 and 2), two had liver disease (individuals 2 and 5), and one had a choreiform movement disorder with cerebral and cerebellar atrophy on magnetic resonance imaging (MRI) (individual 5; Fig. 1). This child additionally had cardiomyopathy. Clinical, imaging and laboratory features for all individuals are summarised in Table 1.

Muscle biopsies revealed dystrophic changes in all ve individuals
All 5 individuals underwent a muscle biopsy taken from either biceps brachialis or the quadriceps (Fig. 2). The age at biopsy ranged from 14 months to 4 years. The biopsies revealed dystrophic changes including abnormal size variation, increased internal nuclei, necrosis, regeneration (fetal/developmental myosin positive bres), bre splitting, whorling, and bro -fatty in ltration (individual 1; Fig. 2: a-e).

Muscle biopsies revealed abnormalities in dystrophy-associated proteins in all ve individuals
Immunoanalysis for a broad panel of dystrophy-associated proteins was performed (Fig. 2). Individuals 1-4 showed moderate, patchy or mosaic reduction in sarcolemmal labelling with an antibody to alpha-dystroglycan (IIH6, recognising an epitope within the O-glycosylated domain crucial in binding to laminin alpha-2) ( Fig. 2: g, p, r, t) and subtle reduction in individual 5 ( Fig. 2: w). There was variable reduction of several other DAPC proteins including dystrophin and sarcoglycans in two individuals (only individual 1 illustrated; Fig. 2i, j). Unusually, in individual 1, the sarcolemmal depletion of the DAPC proteins was more pronounced in the larger bres ( Fig. 2: i-k). These bres also showed reduced caveolin-3 labelling ( Fig. 2: n). In all cases, a variable but small number of bres showed cytoplasmic retention of caveolin-3 and/or dysferlin ( Fig. 2: o). Labelling for laminin-alpha 2 (300 and 80 kDa forms) and laminin alpha 5 (excluding regenerating bres) was normal, except few larger bres with an intact basal lamina (normal labelling for laminin gamma 1, a marker of basal lamina integrity, Fig. 2: m) showing patchy loss of laminin alpha-2 ( Fig. 2: l).

Altered alpha-dystroglycan expression and glycosylation in muscle from all ve individuals
Quantitative western blotting on muscle lysates using an antibody against alpha-dystroglycan (IIH6) was performed in all cases (data shown only for individual: Fig. 3) and showed marked reduction in alpha-dystroglycan expression (82% average intensity reduction in all 5 individuals compared to control). This was accompanied by an aberrant expression pattern comprising 2-3 bands of lower molecular weight, strikingly observed in individual 1 ( Fig. 3: a). Furthermore, laminin overlay assay revealed a reduction in the glycosylated smear, both in intensity (89% decrease) and in molecular weight in this individual compared to control ( Fig. 3: b). A similar pattern was observed in the remaining individuals (69% decrease average intensity compared to control).
Membrane tra cking into and out of the Golgi is delayed in broblasts from individual 1 Since TRAPPC11 has been implicated in membrane tra cking in the biosynthetic pathway, we examined broblasts from control and the affected individual (individual 1) for the ability to tra c cargo from the endoplasmic reticulum (ER) to and through the Golgi (Fig. 4). We rst used the RUSH assay (24) to examine the tra cking of two different cargo proteins from the endoplasmic reticulum (ER) to the Golgi. As shown in Fig. 4 (a, b), both cargo proteins showed a delay in arrival at the Golgi, consistent with a defect in the early secretory pathway. We also examined the marker VSVG-GFP ts045, a protein that can be conditionally retained in the ER at elevated temperature and released upon downshifting the temperature (25). Similar to the RUSH assay, we noted a delay in the arrival of the uorescent signal to the Golgi (Fig. 4: c). In addition, there was a delay in release of the uorescent signal from the Golgi, suggesting that import into and export from the Golgi are both affected in the presence of the bi-allelic TRAPPC11 variant.
Post-mortem ndings in individual 5 Individual 5 died at the age of 32 months following an unexpected cardiovascular collapse. A full post-mortem examination was performed. External examination showed right frontal plagiocephaly with a high forehead, downturned mouth, and slightly down-slanting eyes with epicanthic folds. The ears were asymmetric. The skin over the neck was loose, and the nipples were widely spaced. There was peripheral oedema. Internal examination revealed moderate ascites, and large bilateral pleural effusions. There was moderate cardiomegaly (111.7 g; expected weight for age 74 g) and marked left ventricular dilatation, with brosis of the left anterior subepicardial myocardium, extending to the interventricular septum. There were no structural cardiac anomalies. The liver was of normal size and weight but appeared pale and fatty. The muscles appeared pale and abby, with reduced bulk in the chest and calves. The brain was signi cantly small for his age (876 g; expected weight for age is 1120 g). The gyral pattern in the forebrain was normal. The cerebellum was atrophic.
Histology revealed mild myocardial hypertrophy with a distinct wide band of myocyte loss and brosis in the left ventricular myocardium with septal extension (Fig. 5: a, b). Minor chronic interstitial in ammation, and fatty in ltration in the right ventricle was also noted (data not illustrated). A marked perivenular hepatocyte fatty change and patchy perivenular necrosis associated with bridging portal brosis and bile ductular proliferation was seen (Fig. 5: l). Dystrophic changes in muscles (psoas, quadriceps, calf pectoralis, diaphragm) including myopathic size variation, internal nucleation, brosis and fatty in ltration were noted (Fig. 5: c-k). In the quadriceps there was a dramatic progression of the dystrophic changes from the time of the rst biopsy at 14 months, with large areas of complete bro-fatty replacement adjacent to dystrophic areas in the post-mortem samples (Fig. 5: c, d).
Neuropathological examination showed intact cortical organisation in the forebrain. There was diffuse neuronal loss, more severe in the outer and middle cortical layers, with neuropil vacuolation. The deep white matter was rare ed, with diffuse cortical and white matter gliosis. The deep grey nuclei were well-formed. There was no evidence of brainstem hypoplasia or dysplasia. The ventricles were dilated. The spinal cord appeared normal.
The cerebellum was hypoplastic, with more severe involvement of the vermis. The histological changes were similar in the vermis and hemispheres. There was diffuse atrophy of the cerebellar folia with widened inter-folial spaces ( Fig. 6: b, d).
Granule cells were virtually absent or very severely depleted ( Fig. 6: f, g), with depletion of parallel bres in the molecular layer. The Purkinje cells (PC) were depleted, and most of the surviving neurones showed striking dendritic dystrophy in the form of 'asteroid bodies' which appeared as globular eosinophilic dendritic swellings within the molecular layer ( Fig. 6: f, g).
Immunostaining with a high and low molecular weight neuro lament cocktail antibody (NFC) showed a range of abnormalities, including a chaotic and/or dendritic arbor with misdirected, often swollen proximal primary dendrites, and dystrophic swellings on primary dendrites seen more often distally, showing anomalous dendritic spines radiating in all directions, corresponding to the 'asteroid bodies' (Fig. 6: i-m). A proportion of these swellings showed an 'empty core' devoid of any immunoreactivities ( Fig. 6: m). The dendritic swellings were noted to preferentially accumulate SMI32 + nonphosphorylated neuro laments ( Fig. 6: o). The swellings were immunonegative for ubiquitin, p62 and LC3B (markers of autophagy and ubiquitin-proteosomal stress). There were patches of the molecular layer containing a 'disconnected' Purkinje cell arbor devoid of cell bodies within the depleted Purkinje cell layer underneath the molecular layer ( Fig. 6: l). There was patchily reduced labelling for alpha-dystroglycan (IIH6) in a proportion of surviving Purkinje cells (Fig. 2: z) compared to an age-matched control (Fig. 6: x), although, in many other cells, the staining appeared indistinguishable from the control. In contrast, there was virtually no labelling for IIH6-reactive glycans in the dentate nucleus ( Fig. 6: za) compared to the control ( Fig. 6: y). TRAPPC11 expression was retained in the residual PCs ( Fig. 6: zb) and dentate neurones and was comparable to an age-matched control. There was atrophy of the dentate grey ribbon, with neuronal shrinkage (Fig. 6: v) and marked activation of CD68 + microglia (Fig. 6: w). The cerebellar white matter was rare ed but there was no evidence of axonal dystrophy or dysmyelination, and there was no cerebellar dysplasia.

Discussion
TRAPPC11, a subunit of the TRAPP complex, is important for complex integrity and anterograde membrane transport from the ER to the ER-Golgi intermediate compartment. Recently a role in autophagy was also suggested (11). At least 18 individuals with TRAPPC11 mutations have been reported, ranging from the classical LGMD phenotype at the milder end of the spectrum (12), with the clinical course complicated in some instances by scoliosis, or mild intellectual disability and cataracts (3). At the severe end of the spectrum, individuals with CMD with variable degrees of extra muscular involvement are described (4,5), including the description of individuals with achalasia, alacrimia, intellectual disability, and cerebral or cerebellar atrophy on brain MRI (9). A phenotype overlapping congenital disorder of glycosylation (CDG) type II was described in a child with neonatal hypotonia, cholestasis, thrombocytopenia, nephropathy, cerebral atrophy and a combined defect of N-and O-linked glycosylation (13).
Previous studies in mammalian cellular models elucidated the crucial role of TRAPPC11 in maintaining the integrity of the TRAPPIII complex (14). This work also established TRAPPC11 as a This study identi ed a potential link between TRAPPC11 and the dystroglycanopathies via TRAPPC11's role in the tra cking and glycosylation of dystroglycan in the Golgi. More recently, hypoglycosylation of alpha-dystroglycan was described (22) in siblings with TRAPPC11-associated LGMD and hepatopathy.
In the present study, we report 5 individuals with congenital/early onset muscular dystrophy related to TRAPPC11 mutations. All 5 underwent a muscle biopsy providing an opportunity for detailed pathology studies. The biopsies revealed dystrophic changes and consistent but variable reduction of alpha-dystroglycan hypoglycosylation. Interestingly in one individual there was a rapid progression of dystrophic changes between the biopsy taken at 14 months and following post-mortem examination of the same muscle at 32 months. Immunohistochemical analysis revealed patchy mosaic mild-to-moderate reduction in sarcolemmal labelling for IIH6-reactive glycans. Quantitative immunoblotting for alpha-dystroglycan showed severe hypoglycosylation in all cases, accompanied by signi cant reduction of laminin binding in overlay studies, thereby unequivocally demonstrating glycosylation defects. Conversely, quantitative analysis of alpha-dystroglycan from cultured dermal broblasts in individual 1 did not demonstrate a reduction, but instead, an increase compared to the control. A similar lack of abnormal labelling in broblasts was observed by Larson et al (21) in both their individuals. This may suggest that there are tissue-speci c differences in TRAPPC11-mediated Golgi processing and tra cking of alpha-dystroglycan. A similar discrepancy has been reported in other genes implicated in dystroglycanopathy (23,24). The defective tra c into and out of the Golgi seen in our membrane tra cking assays is consistent with what has been noted for other TRAPPC11 variants (refs 3, 9, 21). In all our cases, we also documented variable, mildly reduced uneven labelling for dystrophin and/or dystrophinassociated proteins (DAPC). This could be secondary to the reduced labelling for alpha-dystroglycan., In individual 1, there was also patchy reduction in sarcolemmal labelling for caveolin-3, and sarcoplasmic retention of caveolin-3 and dysferlin in a number of non-regenerating, non-necrotic bres. Caveolin-3 and dysferlin are not part of the DAPC, and labelling is normally retained in cases with primary or secondary reduction of one or more DAPC proteins.
While brain involvement has been reported previously in individuals with TRAPPC11 mutations, there is no information on its neuropathological basis. Brain involvement with cobblestone lissencephaly including brainstem and cerebellar dysplasia and cysts are well documented in individuals with dystroglycanopathies, indicating a major role of dystroglycan in multiple developmental processes including maintenance of the basement membrane integrity, normal radial glia morphology, organisation of neocortical proliferation, and cortical plate lamination (25)(26)(27).
Neuropathological examination of the brain in individual 5 with severe CMD revealed interesting differences to the structural brain abnormalities associated with the 'classic' dystroglycanopathies. The brain was globally small, but with no evidence of neuronal migration defects in the forebrain. There was marked cerebellar hypoplasia/cerebellar atrophy (CA/CH). An absolute distinction between CA/CH is di cult in absence of serial imaging (28). Histologically, there was no evidence of the neuronal migration defects described in the 'classic' dystroglycanopathies. Instead, we noted a virtually complete loss of the granule cells with accompanying moderate, patchy Purkinje cell loss and a remarkable dendritic dystrophy affecting surviving Purkinje cells. Morphologically, these changes are strikingly similar to the descriptions of primary degeneration of the granular layer of the cerebellum (Norman type) from the pre-molecular era (29). Subsequently, it was suggested that the diseases described by Norman and Jaeken are the same pathological entity, with cerebellar pathology identical to that observed in PMM2-related CDG (30). Aronica et al (31) reported post-mortem ndings from another case of PMM2-related CDG1a, with isolated CA/CH, granule cell depletion, Purkinje cell loss, and morphological changes in the surviving Purkinje cell dendritic arbor. PMM2-CDG is a frequent cause of cerebellar ataxia (32) and pathological ndings suggest an early-onset atrophic process rather than primary hypoplasia (33). Taken together, these neuropathological ndings in our case closely resemble brain involvement in CDG, in particular PMM2-CDG, rather than classic dystroglycanopathies. Immunostaining revealed the presence of IIH6-reactive glycans in Purkinje cells, albeit reduced compared to age-matched control, and a virtually total loss of staining in the dentate neurones. Glycosylation analysis of serum transferrin was normal in this individual (not shown), but analysis of glycoepitopes of secreted proteins may not be sensitive in detection of tissue-speci c defects (21).

Conclusions
Our case studies further consolidate recessive TRAPPC11 mutations as being causally associated with muscular dystrophies of different clinical severity and hypoglycosylation of alpha-dystroglycan. The CNS shows a lack of the classic neuronal migration defects of alpha-dystroglycanopathies, but instead a distinctive type of cerebellar neurodegeneration with remarkable similarities to the cerebellar pathology seen in PMM2-CDG. The multisystem involvement with cardiomyopathy and steatohepatitis shows further phenotypic overlap with CDGs. Future work with a focus on animal models to unravel tissue-speci c differences in TRAPPC11-mediated processing of N-and O-linked glycans should be helpful in discerning the molecular mechanisms of TRAPPC11-associated disease.   u). The biopsy from individual 1 taken at 2 years showed moderate-to-marked dystrophic changes (a). There was overall slow bre predominance (b). Fast myosin showed considerable co-expression (c) with preferential expression in smaller bres clustering around larger preferentially slow myosin expressing bres. There were many bres of all sizes and intensities expressing fetal (d) and developmental (e) myosins, consistent with a dystrophic pattern. Most bres retained robust spectrin labelling indicating an intact sarcolemma (f). There was moderate mosaic/uneven depletion of sarcolemmal alpha-dystroglycan (g), with normal expression of beta-dystroglycan (h). C-terminal dystrophin (i) and alpha-sarcoglycan (j)

List Of Abbreviations
were patchily reduced, and NNOS (k) was markedly depleted. Laminin alpha 2 (300 kDa) was reduced in a few larger bres (l) that retained laminin gamma a1 labelling (m) suggesting that the basal lamina was intact in these bres. There was mild uneven labelling for caveolin-3 (n) and several bres of varying size showed sarcoplasmic retention of dysferlin (o).
Sarcolemmal alpha-dystroglycan labelling was moderate-to-severely reduced in individuals 2-4 (p, r, t). Alpha-sarcoglycan Membrane tra cking is delayed in individual 1 Cells from the TRAPPC11 bi-allelic variant display a defect in membrane tra cking. The RUSH assay was performed in broblasts from a control and the affected individual (Individual 1) using sialyl transferase (ST)-eGFP (a) and mannosidase II (ManII)-mCherry (b). The signal in the Golgi was quanti ed as described previously  and is plotted beneath the representative images. (c) An assay for the tra cking of VSVG-GFP ts045 into and out of the Golgi of the broblasts was performed and quanti ed as previously described (Koehler et al, 2017). Error bars represent SEM. Bars in the representative images are 20 µm. Post mortem pathological ndings from the brain in individual 5 Compared to sections from the vermis (a) and hemisphere (c) from an age-matched control, the individual showed marked vermian (b) and hemispheric (d) atrophy of the folia and widened inter-folial spaces. Compared to the control (e) there was sub-total loss of granule cells and frequent, patchy loss of Purkinje cells. Surviving Purkinje cells showed remarkable dendritic dystrophy, with eosinophilic swellings emanating from proximal or distal dendrites within the molecular layer (f, g). Neuro lament cocktail (NFC) staining showed an orderly Purkinje cell dendritic arbor in the control (box, h) with robust network of parallel bres. In the individual, the surviving Purkinje cells showed abnormally oriented dendrites with thickened primary branches and frequent NFC+ swellings on terminal, and less frequently, proximal primary dendrites with anomalous spines radiating in all directions, so called asteroid bodies. The dendritic arbor was reduced, and there were patches of the molecular layer with 'disconnected' dendritic arbor with absent Purkinje cell bodies (i-m). A proportion of these swellings displayed immunonegative 'empty cores' (m). Staining with antibodies to phosphorylated (SMI31) (n) and non-phosphorylated (SMI32) (o) neuro laments showed preferential aggregation of non-phosphorylated neuro laments within the dendritic swellings. There were scattered Purkinje cell axonal swellings or 'torpedoes' aggregating SMI31+ phosphorylated neuro laments (n). There was commensurate depletion of NFC+ axons and rarefaction of SMI94+ myelinated bres in the subcortical white matter (p, q) and in the hilum of the dentate nucleus (r). There was prominent GFAP+ Bergmann/radial gliosis within the molecular layer (s). The inferior olivary