Article Text

Download PDFPDF

Research paper
Limb girdle muscular dystrophy due to mutations in POMT2
  1. Sofie Thurø Østergaard1,
  2. Katherine Johnson2,
  3. Tanya Stojkovic3,
  4. Thomas Krag1,
  5. Willem De Ridder4,5,6,
  6. Peter De Jonghe4,5,6,
  7. Jonathan Baets4,5,6,
  8. Kristl G Claeys7,8,
  9. Roberto Fernández-Torrón9,
  10. Lauren Phillips2,
  11. Ana Topf2,
  12. Jaume Colomer10,
  13. Shahriar Nafissi11,
  14. Shirin Jamal-Omidi11,
  15. Celine Bouchet-Seraphin12,
  16. France Leturcq13,
  17. Daniel G MacArthur14,15,
  18. Monkol Lek14,15,
  19. Liwen Xu14,15,
  20. Isabelle Nelson16,
  21. Volker Straub2,
  22. John Vissing1
  1. 1 Copenhagen Neuromuscular Center, Rigshospitalet, University of Copenhagen, Kobenhavn, Denmark
  2. 2 John Walton Muscular Dystrophy Research Centre, Institute of Genetic Medicine, Newcastle University, Newcastle upon Tyne, UK
  3. 3 AP-HP, Institute of Myology, Centre de reference des maladies neuromusculaires Paris Est, G-H Pitié-Salpêtrière, France
  4. 4 Neurogenetics Group, Center for Molecular Neurology, Vlaams Instituut voor Biotechnologie, Antwerp, Belgium
  5. 5 Laboratory of Neuromuscular Pathology, Institute Born-Bunge, University of Antwerp, Antwerpen, Belgium
  6. 6 Department of Neurology, Neuromuscular Reference Centre, Antwerp University Hospital, Antwerpen, Belgium
  7. 7 Department of Neurology, Neuromuscular Reference Centre, University Hospitals Leuven, Leuven, Belgium
  8. 8 Department of Neurosciences, Experimental Neurology, Laboratory for Muscle Diseases and Neuropathies, Katholieke Universiteit Leuven, Leuven, Belgium
  9. 9 Department of Neurology, Donostia University Hospital, Biodonostia Health Research Institute, Donostia-San Sebastián, Spain
  10. 10 Servei de Neurologia, Hospital Sant Joan de Déu, Unitatde Patología Neuromuscular, Barcelona, Spain
  11. 11 Department of Neurology, Iranian Center of Neurological Research, Shariati Hospital, Tehran University of Medical Sciences, Tehran, Iran
  12. 12 Département de Biochimie et de Génétique, AP-HP, Hôpital Bichat, Paris, France
  13. 13 Laboratoire de Génétique et Biologie Moleculaires Hopital Cochin, Paris, France
  14. 14 Analytic and Translational Genetics Unit, Massachusetts General Hospital, Boston, Massachusetts, USA
  15. 15 Medical and Population Genetics, Broad Institute of MIT and Harvard, Cambridge, Massachusetts, USA
  16. 16 Center of Research in Myology, Institutede Myologie, Paris, France
  1. Correspondence to Miss Sofie Thurø Østergaard, Copenhagen Neuromuscular Center, Rigshospitalet, 3342, Blegdamsvej 9, DK-2100 Copenhagen, Denmark; sofie.thuroe.oestergaard.02{at}


Background Mutations in the gene coding for protein O-mannosyl-transferase 2 (POMT2) are known to cause severe congenital muscular dystrophy, and recently, mutations in POMT2 have also been linked to a milder limb-girdle muscular dystrophy (LGMD) phenotype, named LGMD type 2N (LGMD2N). Only four cases have been reported so far. ID: NCT02759302

Methods We report 12 new cases of LGMD2N, aged 18–63 years. Muscle involvement was assessed by MRI, muscle strength testing and muscle biopsy analysis. Other clinical features were also recorded.

Results Presenting symptoms were difficulties in walking, pain during exercise, delayed motor milestones and learning disabilities at school. All had some degree of cognitive impairment. Brain MRIs were abnormal in 3 of 10 patients, showing ventricular enlargement in one, periventricular hyperintensities in another and frontal atrophy of the left hemisphere in a third patient. Most affected muscle groups were hip and knee flexors and extensors on strength testing. On MRI, most affected muscles were hamstrings followed by paraspinal and gluteal muscles. The 12 patients in our cohort carried 11 alleles with known mutations, whereas 11 novel mutations accounted for the remaining 13 alleles.

Conclusion We describe the first cohort of patients with LGMD2N and show that unlike other LGMD types, all patients had cognitive impairment. Primary muscle involvement was found in hamstring, paraspinal and gluteal muscles on MRI, which correlated well with reduced muscle strength in hip and knee flexors and extensors. The study expands the mutational spectrum for LGMD2N, with the description of 11 novel POMT2 mutations in the association with LGMD2N.

Clinical trial registration NCT02759302.

View Full Text

Statistics from


Recessive mutations in the gene coding for protein O-mannosyl-transferase 2 (POMT2) are known to cause severe congenital muscular dystrophies (CMD), such as Walker-Warburg syndrome and muscle–eye–brain disease. These conditions are characterised by structural brain and muscle involvement at birth and a low chance of survival past childhood. In 2007, mutations in POMT2 were linked to a milder limb-girdle muscular dystrophy (LGMD) phenotype, named LGMD type 2N (LGMD2N).1 LGMD is a group of heterogeneous diseases characterised by wasting and weakness of the muscles of the shoulder and hip region.

The POMT2 protein forms a complex with protein O-mannosyl-transferase 1 (POMT1), and catalyses the first step in the synthesis of O-mannosyl glycan, located on the extracellular protein, α-dystroglycan. Alpha-dystroglycan (α-DG) is part of the dystrophin-associated glycoprotein complex at the sarcolemma. Here, it forms an essential link between the subsarcolemmal cytoskeleton and the extracellular matrix of the muscle cells, and plays an important role in membrane integrity and force transmission.3

Only four young patients (age 4–18 years) with a LGMD2N phenotype have been reported as separate cases so far.1 4–6 Thus, detailed knowledge about the disease characteristics is lacking for LGMD2N. Through an international collaboration between seven clinical centres, and facilitated by the use of next-generation sequencing (NGS) in unclassified myopathies with limb girdle weakness, we have identified a group of 12 patients affected by LGMD2N. We report on the specific clinical features of this rare form of LGMD, extend the mutational spectrum and describe findings on MRI of brain and muscle as well as immunohistochemical analyses on muscle biopsies.


All patients consented to participate.


Patients affected by LGMD2N were identified primarily by NGS in cases with undiagnosed limb girdle and in a few cases by direct Sanger sequencing. Twelve patients with genetically verified LGMD2N were included in the study (table 1). Two of the patients (cases 1 and 2) were siblings of consanguineous parents, and cases 9 and 10 were siblings of non-consanguineous parents. All other patients were unrelated.

Table 1

Baseline characteristics of 12 patients with LGMD2N

Clinical evaluation

A thorough medical history was taken and focused on presenting symptom, age at disease onset, muscle cramps and myalgia. Sometimes disease onset and presenting symptom had to be obtained from parents or medical records. Walking capability was examined by the 10 m walk test and by recording the use of walking aids. Limb muscle strength was evaluated at the current ages indicated in table 1 by manual muscle testing (Medical Research Council scale). Respiratory function was assessed by a spirometry (forced vital capacity (FVC)) and cardiac function by ECG and echocardiography. Cognitive function was screened by using the Minimal Mental Status Examination (MMSE).


Whole-body muscle MRI was performed in all patients and brain MRI in 10 of the 12 patients. The MRI scanners differed among the participating clinical centres, but only axial T1-weighted images were assessed. Four cross-sectional slices at the level of calves, thighs, L4 and pelvis were chosen for the evaluation of muscle involvement (figure 1). Replacement of muscle by fat was graded according to the Mercuri scale.7

Figure 1

T1-weighted, cross-sectional MR images of muscles in six cases. Images were acquired at L4 at spine level, at pelvic level, the middle of the thighs and at the thickest part of the calves.

Muscle biopsies

Muscle biopsies were obtained from the tibialis anterior, gastrocnemius or deltoid muscles and stained with H&E for general histopathological evaluation. For immunohistochemical evaluation of α-DG glycosylation, muscle sections were stained with the antibodies VIA4-1 and IIH6C (Merck-Millipore, Temecula, California, USA) and goat anti-mouse Alexa Fluor 594 antibodies (ThermoFisher, Waltham, Maryland, USA) using standard protocols.8 Biopsies for immunohistochemistry were only available for cases 1, 5, 9 and 10.

Molecular findings

Blood samples were drawn for determination of plasma creatine kinase (CK) concentration and for extraction of DNA from leucocytes, according to standard procedures. Mutations in POMT2 were identified by whole exome sequencing of leucocyte DNA at the Broad Institute’s Genomics Platform, using Illumina exome capture, 38 Mb baited target and the Broad’s in-solution hybrid selection process. The mutations were confirmed by Sanger sequencing. In two cases (3 and 4), mutations were found directly by Sanger sequencing. Mutation frequencies were estimated using Exome Aggregation Consortium database with 60 706 unrelated individuals as control population (table 2).

Table 2

Mutations identified with estimated frequency in control population


Clinical evaluation

Disease onset varied from birth to 55 years (table 1). Three participants had disease onset at birth and were classified as CMD/LGMD, because they had a typical LGMD phenotype as adults. Two with onset at birth still walked unassisted at age 18 years and one was still ambulatory with assistance at age 54 years. Similar reasoning to classify patients as LGMD2N, despite disease onset at birth, was also applied in two of the four previously reported young cases of LGMD2N.1 5 Presenting symptoms were in most cases related to ambulatory function, showing either as a delay in the ability to walk, or troubles in walking, climbing stairs or running. Ten patients were ambulatory while one was wheelchair users, and one could only walk when assisted. Some patients presented with learning difficulties in school, especially case 3, who attended a special school. Cases 3 and 4 also experienced pain during mild exercise which preceded muscle weakness in childhood. Delay in cognitive function in case 5 was followed by delayed motor milestones due to muscle weakness. The MMSE score was decreased relative to normal in most patients (table 1), and all were described by their physicians as having some degree of cognitive impairment, although no formal neuropsychological tests were performed.

Two cases had reduced left ventricular ejection fraction (LVEF) (table 1). Case 7 was diagnosed with dilated cardiomyopathy at age 18 years and treated with an angiotensin-converting enzyme inhibitor. The echocardiography of case 5 revealed a mild reduction of LVEF without any cardiac symptoms. Case 3 had hypertension and was treated with an antihypertensive drug. FVC was measured in eight patients and was reduced in all with an average FVC of 66% of the predicted value. Cases 1, 2 and 5 could not cooperate during pulmonary function tests, because of poor intellectual capacity, and case 10 was unavailable for FVC measurements, because he was lost to follow-up.

Muscle strength examination showed reduced force in all patients (figure 2), especially in hip and knee flexors and extensors. Knee flexor muscles were weaker than the extensors and the opposite pattern was seen across the hip. As apparent from figure 2, muscle weakness was mostly symmetric, except in case 6, who had a much weaker left relative to right leg. Muscle strength of forearm and finger muscles showed normal force in all patients.

Figure 2

Muscle strength evaluation by using the MRC scale. Values range from 0 to 5, including plus and minus for 4 and 5 (4+ equals 4.33 and 5− equals 4.66). Boxplots show the distribution of MRC scores for each motion, including a median line. Dots representing minimal and maximal values. ext, extension; flx, flexion, L, left; MRC, Medical Research Council; R, right.

Proximal muscle atrophy was observed in most patients, and atrophy of the shoulder girdle was found in cases 9, 10 and 12, who had significant scapular winging (figure 3). CK levels were highly elevated in all participants, except in two cases (cases 4 and 5) in whom it was slightly elevated (table 1).

Figure 3

Picture of case 10, showing prominent scapular winging.


Brain MRIs were abnormal in 3 of the 10 patients in whom brain imaging was carried out. Mild ventricular enlargement due to central and cortical atrophy was found in case 1, periventricular hyperintensities in case 4 and frontal atrophy of the left hemisphere in case 5 (figure 4).

Figure 4

T1-weighted brain MRI with sagittal and transverse slices showing central and cortical atrophy and mild ventricular enlargement in case 1 and frontal atrophy of left hemisphere in case 5.

Muscle MRI revealed a pattern of selective muscle involvement, most strikingly affecting the hamstring, paraspinal and gluteal muscles (figure 1). Consistent with the evaluation of muscle strength, the hamstring muscles were more severely affected (average 3.8 on the Mercuri scale) than the anterior thigh muscle group (average 3.3 for the quadriceps).

In the muscles of the lower leg, both degree of fatty infiltration and muscle involvement differed among the participants (figure 1), but there was a predilection for selective involvement of the muscles of the posterior compartment; on average, the gastrocnemius muscles (average 3.2 on Mercuri) were more severely affected compared with the tibialis anterior muscle (average 1.7 on Mercuri).

Muscle biopsy

A dystrophic pattern with fibre size variation and central nuclei was present in muscle biopsies from six cases. Immunohistochemical staining of α-DG glycosylation demonstrated a clear signal reduction (figure 5). Glycosylation was also absent or severely reduced in muscle biopsies of cases 1, 9 and 10 (data not shown). In the last two patients (cases 3 and 4) for whom a muscle biopsy was available, immunohistochemistry was performed using only the VIA-4 antibody, because IIHC6 antibodies were not routinely used in that lab. The stains showed reduced glycosylation of α-DG in cases 3 and 4 with the VIA-4 antibody. The four other cases showed highly reduced glycosylation using both antibodies, except case 1, in whom glycosylation was significantly reduced using the IIHC6 antibody, but only mildly reduced using the VIA-4 antibody.

Figure 5

Muscle biopsy from a healthy control and case 5, displaying myopathic features (increased internalised nuclei and increased variation of muscle fibre diameter). Staining, using α-dystroglycan glycosylation-specific antibodies IIH6C and VIA4-1, demonstrates loss of glycosylation. Bar is 50 µm. α-DG, alpha-dystroglycan.

Molecular findings

The patients carried 11 novel mutations in POMT2 (table 2). Case 3 carried the known mutation, c. 1997A>G, which previously has been linked to a CMD phenotype.6 9


We describe the first cohort of patients affected by LGMD2N due to mutations in the POMT2 gene, which so far has only been described in a few single-case reports. The major new findings of the study are: (1) patients with LGMD2N, unlike other recessively inherited LGMDs, are cognitively impaired; (2) the disease primarily affects hamstring, paraspinal and gluteal muscles; (3) the mutational spectrum of LGMD2N is expanded by the addition of 11 new mutations in POMT2, adding to the 11 known mutations causing LGMD2N and (4) patients with LGMD2N seem to have a wide range of disease onset.

The muscle involvement and clinical presentation of patients with autosomal-recessive LGMD differ highly among the various subtypes. In this study, the degree of muscle involvement also varied among the 12 patients. One patient had an asymmetric appearance of muscle involvement, although generally the pattern on strength testing and MRI showed consistent involvement, preferentially of the hamstring, paraspinal and gluteal muscles in a symmetrical pattern. This pattern of muscle involvement was similar to that found in the worldwide most prevalent form of recessive LGMD, LGMD2A, which is caused by mutations in the CAPN3 gene. Besides the pattern of leg involvement, prominent scapular winging, as seen in three of our patients with LGMD2N, and asymmetry are also occasionally encountered in LGMD2A.10 The same pattern of muscle involvement is also seen in another glycosylation defect of α-DG, LGMD2I, caused by mutations in FKRP (fukutin-related protein gene). Patients with LGMD2I present with a similar, preferential involvement of the posterior thigh. When calf muscles are involved in LGMD2I, the medial gastrocnemius and soleus muscles are affected first, which was also seen in our patients with LGMD2N together with a rather hypertrophic appearance of the calf muscles.11 Assessment of calf muscle hypertrophy was based on the clinical examination performed by an experienced myologist at each centre; cross-sectional area measurements of different muscle groups could aid in reliably judging muscle bulk. The combination of different MRI features can, however, yield valuable clues, as for example, LGMD2I is typically associated with muscle hypertrophy, contrasting with the atrophic phenotype of LGMD2A. If imaging is performed in relatively early disease stages, a similar pattern seems to be observed in different dystroglycanopathies.

The other way around, MRI imaging can aid in the interpretation of candidate sequence variants obtained by NGS techniques.

Dystroglycanopathies comprising >18 separate disorders and more remain to be discovered.12 The disorders affect glycosylation of α-DG, and in the majority of cases result in CMD associated with brain abnormalities. POMT2 deficiency was first reported to cause CMD,13 and only later was the defect associated with a rare phenotype compatible with LGMD.1 4–6 Only in rare cases have patients with a LGMD phenotype been systematically associated with brain involvement. First, LGMD2I has been proposed to have occasional involvement of the frontal and posterior parietal lobes of the brain, as suggested by brain MRI scans from 10 patients with LGMD2I.14 Patients were reported to have mild problems in graphic element integration, but this cognitive impairment was not related to the MRI findings.14 Thus, although the general clinical experience is that patients with LGMD2I have normal cognitive function, there is evidence to suggest a mild impairment.14 15 Patients with LGMD2M, caused by mutations in the FKTN gene, have been reported to have normal cognitive function and brain structure.16 17 A recent study of two siblings with POMT1 deficiency and a LGMD phenotype (LGMD2K) were reported to have intellectual disability and focal cortical dysplasia on brain MRI.18 The same pattern of cognitive and structural brain involvement has been found in a cohort of five patients with LGMD2K.19 POMT1 and POMT2 dimerize to attach the initial O-linked mannose onto α-DG. It can be argued that the mutations in POMT1 and POMT2 generally cause more severe phenotypes due to the requirement of this initial mannose for extracellular matrix anchoring. However, most of the enzymes responsible for the transfer of sugar moieties to the initial mannose, and subsequent expanding glycan, fukutin, FKRP, POMTGNT1 and LARGE, may result in CMD, muscle–eye–brain disease or Walker-Warburg syndrome when the genes are mutated.16 20 21 The absence of phenotypes without some brain involvement in patients with POMT2 deficiency suggests that minor changes to the structure of POMT2, which has nine transmembrane helices and requires N-glycosylation at multiple sites, are likely to affect function or binding to POMT1 significantly.22 23 Cognitive function was not quantitatively assessed in our patients, due to practical difficulties in doing so across seven centres in six countries. However, the cognitive impairment was evident from MMSE and the patients’ inability to succeed in school. Cognitive function in our patients was not quantitatively assessed using a neuropsychological examination, due to practical difficulties in coordinating this effort across seven centres in six countries. The use of the MMSE score is not validated for this cohort, but was used as an instrument to gain access to an index of cognitive function, which together with information on performance at school allowed a general judgement of the intellectual capabilities. Three of our patients showed abnormalities on brain MRI, but no uniform brain changes across their scans. The patients with MRI abnormalities were the ones with the lowest MMSE scores, suggesting a link with cognitive impairment. However, cognitive impairment was also present in patients with normal brain MRI scans (table 1). Abnormal brain MRI findings could also have other causes than LGMD2N. As shown in figure 4, case 5 had frontal atrophy of the left hemisphere, which could relate to the patient’s hypoxia at birth. T2-weighted periventricular hyperintensities as found in case 4 are often linked to cerebrovascular diseases, and this patient was also treated for hypertension. On the other hand, learning difficulties had been present since age 13 in school, which suggests that the periventricular lesions likely played no role in the cognitive function. The finding of diffuse central and cortical atrophy in case 1, a 25-year-old woman with no history of other organic diseases, is likely related directly to the POMT2 deficiency. Although our study is the first to suggest consistent cognitive dysfunction in a LGMD subtype, case reports of patients with LGMD affected by glycosylation defects of α-DG suggest that brain abnormalities may be present in some LGMD subtypes, especially caused by POMT1 and POMT2 mutations. Attention to cognitive aspects should therefore be exercised when diagnosing patients with dystroglycanopathies and an LGMD phenotype.

Alpha-DG is also glycosylated in cardiac muscle cells, which might account for the frequent occurrence of cardiac affection in CMD and LGMD caused by glycosylation defects of α-DG. Dilated cardiomyopathy has been reported in patients with mutations in the FKTN gene,24 25 and a third of patients with LGMD2I develop cardiomyopathy.26–28 Two patients with POMT1 mutations and a LGMD phenotype have also been reported with ventricular dilatation of the heart.29 In the present study, two cases had reduced LVEF (cases 5 and 7), suggesting that patients with LGMD2N are at risk of developing pump failure. In accordance with this, Martinez et al recently reported three siblings with a CMD phenotype and a homozygous c.1997A>G mutation in POMT2, which was also present in case 3 in our study. The three siblings had reduced LVEF, dilatation of the aortic root and/or left ventricular wall motion abnormalities.30 These and our findings suggest that regular cardiac investigations should be carried out in patients with LGMD2N.

Our study disclosed 11 new mutations in POMT2, which were either inherently pathogenic, because they were frameshift mutations, or predicted by various in-silico prediction tools (PolyPhen, Mutation Taster and FATHMM) to be pathogenic. No prediction was given for the mutation c.1654–5T>G in case 12, as it was predicted to be located at an extended splice site. These new mutations were all absent or extremely rare in the background population. As mutations affect wide range of sites in the POMT2 gene, including the transmembrane helices, N-glycosylation sites and variable effects on hydrophobic/hydrophilic changes, there appears to be little resilience in the POMT2 structure before mutations become pathogenic (table 2).

In conclusion, we demonstrate the clinical features in the first cohort of patients with LGMD2N, showing a pattern of muscle affection similar to other LGMDs, most notably LGMD2A and 2I, and a consistent cognitive affection, which has not been described as a signature feature of other LGMD types. Our study quadruples the number of cases of LGMD2N reported in the world, and therefore suggests that the condition may be more prevalent than hitherto considered.


The authors would like to thank the patients and their relatives for their willingness to cooperate in this study. They would like to thank Mojgan Reza and Dan Cox for technical support. Also, they would like to thank the Myocapture project, France Génomique National infrastructure. They also thank Dr Monica Ensini for her support in managing the MYO-SEQ study.


View Abstract


  • Contributors STØ: design of study, analysis, acquisition and interpretation of data, and drafting the manuscript. KJ, TS, PDeJ, JB, KGC, RF-T, LP, AT, JC, WDeR, SN, SJ-O, CB-S, FL, DGMacA, ML, LX, IN and VS: acquisition of data and revision of manuscript. TK: acquisition and interpretation of data, and revision of manuscript. JV: design of study, acquisition and interpretation of data, and revision of manuscript.

  • Funding MYO-SEQ is funded by Sanofi Genzyme, Ultragenyx Pharmaceutical, the LGMD2I Research Fund, the Kurt + Peter Foundation, the LGMD2D Foundation and the Samantha J Brazzo Foundation. Also the Myocapture project, France Génomique National infrastructure, was funded as part of the ‘Investissements d’Avenir’ for performing the whole exome sequencing of our French patients. The study was also supported by the Medical Research Council UK (reference G1002274, grant ID 98482). This work was supported by the Association Belge contre les Maladies Neuromusculaire (ABMM) – Aide à la Recherche ASBL and the EUFP7/2007-2013 under grant agreement number 2012-305121 (NEUROMICS). JB is supported by a Senior Clinical Researcher mandate of the Research Fund – Flanders (FWO).

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval Danish National Committee on Health Research Ethics (H-3-2012-163 withamendment #41665, #43449 and #50556) and the local Ethical Review Boards of the participating centers.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.