The neurobiology of the dystrophin-associated glycoprotein complex

Abstract

While the function of dystrophin in muscle disease has been thoroughly investigated, dystrophin and associated proteins also have important roles in the central nervous system. Many patients with Duchenne and Becker muscular dystrophies (D/BMD) have cognitive impairment, learning disability, and an increased incidence of some neuropsychiatric disorders. Accordingly, dystrophin and members of the dystrophin-associated glycoprotein complex (DGC) are found in the brain where they participate in macromolecular assemblies that anchor receptors to specialized sites within the membrane. In neurons, dystrophin and the DGC participate in the postsynaptic clustering and stabilization of some inhibitory GABAergic synapses. During development, α-dystroglycan functions as an extracellular matrix receptor controlling, amongst other things, neuronal migration in the developing cortex and cerebellum. Several types of congenital muscular dystrophy caused by impaired α-dystroglycan glycosylation cause neuronal migration abnormalities and mental retardation. In glial cells, the DGC is involved in the organization of protein complexes that target water-channels to the plasma membrane. Finally, mutations in the gene encoding ε-sarcoglycan cause the neurogenic movement disorder, myoclonus-dystonia syndrome implicating ε–sarcoglycan in dopaminergic neurotransmission. In this review we describe the recent progress in defining the role of the DGC and associated proteins in the brain.

• Dystonia
• glycosylation
• mental retardation
• muscular dystrophy
• neuronal migration

Introduction

In the past two decades staggering progress has been made in elucidating the genetic basis for the muscular dystrophies. The genes for the most common forms of muscular dystrophy have been identified and their functions well defined [1]. The DMD gene, mutated in Duchenne and Becker muscular dystrophies (D/BMD), is the largest gene in the human genome, spanning approximately 2.5 Mbp of the X chromosome and encodes several different proteins, including the giant actin-binding protein dystrophin. Dystrophin is expressed at high levels in striated muscle and to a lesser extent in neurons. In both cell types, dystrophin is found at the cytoplasmic face of the plasma membrane, in multiprotein complexes that can interact with both the cytoskeleton and extracellular matrix (ECM).

In common with many important proteins, dystrophin is part of multiprotein complex known as the dystrophin-associated glycoprotein complex (DGC). The DGC is a large, membrane-spanning protein complex that is thought to protect the muscle plasma membrane (sarcolemma) from mechanical damage (Figure 1A). Biochemically, the core DGC can be divided into three subcomplexes: the sarcoglycan subcomplex, the dystroglycan subcomplex, and the cytoplasmic subcomplex composed of dystrophin, the dystrobrevins, and the syntrophins [2]. Several forms of muscular dystrophy are caused by defects in the function and assembly of the DGC [3]. Dystrophin can act as a macromolecular scaffold and is critically required for the assembly and stability of the DGC at the sarcolemma. In patients with D/BMD, the absence of dystrophin or a mutation that affects its function results in disassembly of the DGC at the membrane, effectively severing one of the links between the subsarcolemmal cytoskeleton and the ECM. Thus, the absence of dystrophin results in a secondary deficiency of other components of the DGC at the sarcolemma. In addition to the core DGC, many accessory proteins have been found to interact with this complex in muscle and non-muscle tissues. In this review, we have used a manually annotated protein interaction network to depict the conceptual organization of the DGC and associated proteins (Figure 2).

Key messages

• Duchenne and Becker muscular dystrophies are complex disorders that are often associated with cognitive impairment, behavioural, and psychiatric abnormalities.

• Dystrophin and components of the dystrophin-associated glycoprotein complex (DGC) are associated with inhibitory GABAergic synapses in the hippocampus and cerebellum.

• Congenital muscular dystrophies caused by defects in α-dystroglycan glycosylation are frequently associated with neuronal migration abnormalities and mental retardation.

• DGC-like complexes anchor membrane-associated proteins, such as the water-channel aquaporin-4, to specialized sites in neurons and glia.

• Mutations in the SGCE gene that encodes ε-sarcoglycan cause the movement disorder myoclonus-dystonia syndrome.

Figure 1.  The molecular organization of the DGC and ‘DGC-like’ complexes in muscle and brain. Conceptual organization of the core DGC, ‘DGC-like’ complexes and associated proteins in skeletal muscle (A), neurons (B) and glia (C). The composition of the core DGC in muscle was determined by the direct biochemical purification of the complex from tissue. By contrast, the complexes shown in panels B and C have been inferred from different studies (yeast two-hybrid interactions, co-localization and co-immunoprecipitation) and do not represent a biochemically defined entity. For the purposes of clarity, only a subset of syntrophin-associated proteins, such as the Na⁺ channels, nNOS and aquaporin-4 have been included in each complex. A comprehensive list of syntrophin-associated proteins is presented in Table III. In each panel, the dystrophin/β-dystroglycan binding region is reduced in scale in comparison with the similar region in dystrobrevin. Question marks denote imputed interactions that have not been demonstrated in vivo. (Abbreviations: DGC = dystrophin-associated glycoprotein complex; SBD = syntrophin-binding domain; nNOS = neuronal nitric oxide synthase; SAST = syntrophin-associated serine/threonine kinase; SSPN = sarcospan; ABD = actin-binding domain; PH = plecstrin homology domain; PDZ = PSD-95, (postsynaptic density protein 95), discs–large and ZO-1 (zonula occludens-1); SU = syntrophin-unique region; Kir 4.1 = inwardly rectifying K⁺ channel 4.1; WW = WW domain ; EFI/EFII = EF hand domains ; ZZ = zinc finger domain; H1/H2 = helical domains.)

Figure 2.  Protein interaction network for the dystrophin-associated glycoprotein complex (DGC). A systems view of the molecular interactions between the core DGC and associated proteins. The map was created using the Human GRID (General Repository of Interaction Datasets, http://www.thebiogrid.org/) database and was assembled using the Osprey network visualization tool [106]. Gene names: LAMA2 = laminin-α2; AGRN = agrin; HSPG2 = perlecan (heparan sulphate proteoglycan-2); NRXN = neurexin; DAG1 = dystroglycan; CAV3 = caveolin-3; SGCA = α-sarcoglycan; SGCB = β-sarcoglycan; SGCG = γ-sarcoglycan; SGCD = δ-sarcoglycan; SSPN = sarcospan; FLNC = filamin C; DMD = dystrophin; ACTG1 = γ-(cytoplasmic) actin; PGM5 = phosphoglucomutase 5; SNTA1 = α-syntrophin; SNTB1 = β1-syntrophin; SNTB2 = β2-syntrophin; DNTA = α-dystrobrevin; SYNC1 = syncoilin; DMN = desmuslin; DTNBP1 = dystrobrevin-binding protein-1; SCN5A = Nav1.5; SCN4A = Nav1.4; NOS = neuronal nitric oxide synthase; DGKZ = diacylglycerol kinase, zeta; CALM1 = calmodulin 1; GRB2 = growth factor receptor bound protein 2; MAST = microtubule associated serine/threonine kinase 1; KCNJ4 = potassium inwardly-rectifying channel J4 (Kir2.3); KCNJ12 = potassium inwardly-rectifying channel, subfamily J12 (Kir2.2). Colour coding: red = core DGC components; green = cytoskeletal proteins; yellow = signalling proteins; dark blue = transmembrane proteins; light blue = extracellular matrix proteins; grey = trafficking proteins.

In addition to its mechanical role in muscle, dystrophin and the DGC are also involved in intracellular Ca^{2 +} homeostasis. In dystrophin-deficient muscle, raised intracellular [Ca^{2 +}] has been shown to affect the function of some Ca^{2 +}-permeable channels and activate Ca^{2 +}-dependent proteases [4]. It has also been suggested that the DGC has a regulatory function, participating in signal transduction between the ECM and the cytoskeleton [5]. Furthermore, central nervous system (CNS) involvement in some D/BMD patients supports the notion that dystrophin and the DGC have pleiotropic roles in excitable cells. Several recent reviews have focused on the role of dystrophin and DGC in muscle [5–7]. In this review, we will discuss the role of the DGC in the CNS.

Abbreviations

Table

ADHD	attention deficit hyperactivity disorder
ATP	adenosine triphosphate
BLOC-1	biogenesis of lysosome-related organelles complex-1
BMD	Becker muscular dystrophy
CMD	congenital muscular dystrophy
DGC	dystrophin-associated glycoprotein complex
dko	double knockout
DMD	Duchenne muscular dystrophy
DTNBP1	dystrobrevin-binding protein-1
ER	endoplasmic reticulum
FCMD	Fukuyama congenital muscular dystrophy
FKRP	fukutin-related protein
FKTN	fukutin
GABA	γ-aminobutyric acid
GTP	guanosine triphosphate
IQ	intelligence quotient
LGMD	limb-girdle muscular dystrophy
LTP	long-term potentiation
MEB	muscle-eye-brain disease
MDS	myoclonus-dystonia syndrome
POMGnT1	protein O-mannose β-1,2-N-acetyl glucosaminyltransferase
POMT1/2	protein O-mannosyltransferase-1 or -2
PSD	postsynaptic density
PDZ	PSD-95 (postsynaptic density protein 95) discs–large and ZO-1 (zonula occludens-1)
PH	plecstrin homology
SEA	sea urchin, enterokinase, agrin
SU	syntrophin-unique
WWS	Walker-Warburg syndrome

The cognitive and neuropsychiatric spectrum in D/BMD

Duchenne muscular dystrophy (DMD) is a complex, multisystemic disorder that can affect the CNS. It is now well established that a significant proportion of boys with DMD have mild cognitive impairment [8]. Although normally distributed, full-scale intelligence quotient (IQ) scores are shifted downwards by one standard deviation from the population mean of 100, such that the mean IQ for DMD boys is 80.2 (see [9] for meta-analysis). Furthermore, approximately 35% of DMD boys, compared to only 2% within the general population, have full IQ scores of 70 or below [9]. DMD boys also have an increased incidence of three types of learning disabilities: dyslexia, dyscalculia, and dysgraphia (disorder of written communication) [10]. Although there is little correlation between the site of the mutation in the DMD gene and cognitive impairment, mutations in the 3’-end of the gene are frequently associated with lower IQ scores [11]. Although generally rarer, truncating mutations in the 3′-end of the DMD gene are often associated with moderate to severe mental retardation and have the propensity to disrupt the assembly of the DGC by deleting the β-dystroglycan binding site at the C-terminus of dystrophin [12], [13].

In addition to cognitive impairment, D/BMD patients have been reported to have an increased incidence of neuropsychiatric and behavioural disorders including autism spectrum disorder and attention deficit hyperactivity disorder (ADHD) (Table I). A recent questionnaire-based study by Hendriksen and Vles found that 11.7% of males with DMD had a co-morbid diagnosis of ADHD [14]. Similarly, Hinton and colleagues found that 1 in 4 boys with DMD had significant problems with attention [15]. These studies point to a complex pattern of neuropsychiatric and neurocognitive deficits in a significant proportion of D/BMD patients. Educational provision is clearly an important aspect for the treatment and management of D/BMD [16].

Table I.  Neurocognitive and neuropsychiatric deficits associated with Duchenne and Becker muscular dystrophies (D/BMD). The Table summarizes some of the findings in research articles reporting cognitive, behavioural, and psychiatric abnormalities described in some boys with DMD and BMD. Key references are given.

Deficit	Reference
Cognitive impairment	[9], [107]
Learning disability	[10], [108]
Attention deficit hyperactivity disorder (ADHD)	[14], [109]
Obsessive-compulsive disorder (OCD)	[14]
Autism spectrum disorder	[14], [109], [110]
Sensory processing deficit	[16]

There have been surprisingly few neuroimaging studies examining brain function in DMD patients. Positron emission tomography (PET) imaging has been used to study glucose metabolism in the brains of ten boys with DMD. This study found that region-specific glucose hypometabolism might be linked to cognitive and behavioural abnormalities in this cohort [17]. Further metabolic defects have been reported in the brains of DMD patients using phosphorus-31 magnetic resonance spectroscopy. DMD patients were found to have higher ratios of inorganic phosphate to adenosine triphosphate (ATP), phosphocreatine and phosphomonoesters compared to normal controls [18]. Thus, altered brain metabolism may contribute to the cognitive and behavioural deficits in DMD patients.

Dystrophin and its isoforms in the brain

There are a variety of reasons why the DGC in brain differs in composition and function from the core DGC in muscle. First, there are several distinct dystrophin isoforms that are present in the brain. Second, the DGC has not been purified from the brain using the biochemical techniques initially used to isolate the complex from striated muscle. Several distinct complexes have been shown to exist in neurons and glia, suggesting that there is considerable heterogeneity in the composition of the DGC in brain (Figure 1B and 1C) [19], [20]. For the purposes of this review we have called these complexes ‘DGC-like’, differentiating them from the core DGC in muscle. Finally, with the exception of ε- and ζ-sarcoglycan, most of the sarcoglycans are expressed predominantly in striated and smooth muscle and are only weakly expressed in the CNS. Thus, if a sarcoglycan complex exists in the brain it is unlikely to resemble the cognate complex in muscle.

As mentioned previously, in addition to dystrophin, the DMD gene encodes several other proteins, most of which are found in the CNS and peripheral nervous system (for review see [3]). Three forms of full-length dystrophin are encoded by transcripts originating from independent promoters in the 5′-end of the DMD gene. Specifically, the P-promoter produces dystrophin that is specifically expressed in cerebellar Purkinje cells whereas the C-promoter transcribes dystrophin in the cortex and subcortical regions including the hippocampus [21]. The most abundant product of the DMD gene that is expressed in the CNS is Dp71 [22]. The Dp71 mRNA is transcribed from a promoter located between exons 62 and 63. The Dp71 protein contains the binding sites for dystroglycan as well as the dystrobrevins and syntrophins but does not have an N-terminal actin-binding domain [23]. In addition to these variants, Dp140 and Dp260 are also found in the CNS [24]. Dp140 appears to be restricted to the microvasculature, whereas Dp260 is found almost exclusively in the retina. It is noteworthy that the dystrophin isoform Dp116 is found in Schwann cells of peripheral nerves but is not associated with myelinated structures in the adult CNS.

Full-length dystrophin is expressed almost exclusively in neurons [25]. By contrast, many of the shorter isoforms, notably Dp71 and Dp140, are expressed in glial cells, notably perivascular astrocytes and Bergmann glia [19], [24]. This diversity is mirrored by the distribution of the brain-expressed components of the DGC and suggests that there are several distinct DGC-like complexes in the brain (see below). Dystrophin is located within membrane-associated puncta in cerebellar Purkinje cells, pyramidal neurons in the neocortex, and pyramidal neurons of the hippocampal formation [25]. In these brain regions, dystrophin co-localizes with inhibitory GABA_A receptor clusters [26]. In the mdx mouse there is a significant reduction in the number of clusters that are immunoreactive for the α1 and α2 GABA_A subunits [26].

A role for dystrophin and the DGC at inhibitory synapses has gained increasing weight and may explain some of the behavioural and electrophysiological findings described in the preceding sections. In cultured hippocampal neurons, Brunig and colleagues found that dystrophin is extensively co-localized with the α2 subunit of the GABA_A receptor [27]. Dystrophin is also able to recruit other components of the DGC, including the syntrophins and β-dystroglycan, to nascent receptor clusters. This study also made the intriguing observation that selective signalling from GABAergic terminals contributes to postsynaptic clustering of dystrophin [27]. The clustering of dystrophin at inhibitory GABAergic synapses is independent of postsynaptic GABA_A receptors and gephyrin, a scaffolding protein that clusters inhibitory neurotransmitter receptors [27]. As alluded to above, dystroglycan is also clustered at GABAergic synapses. GABAergic receptor clusters from mice lacking dystroglycan do not contain dystrophin, suggesting that dystroglycan is required for the assembly of DGC-like complexes at inhibitory synapses in the hippocampus [28]. Conversely, dystroglycan remains clustered in neurons from both mdx and gephyrin-deficient mice [28]. Finally, in cerebellar Purkinje cells from mdx mice, the frequency and amplitude of spontaneous inhibitory postsynaptic currents were reduced compared with littermate controls [29]. This observation is consistent with the reported reduction in the number and size of GABA_A receptor clusters described above. Although these molecules are not essential for GABAergic synaptogenesis, these findings suggest that dystrophin and the DGC are involved in modulating synapse function at a subset of GABAergic synapses in the hippocampus and cerebellum [28], [29].

Dystrophin has also been reported to be a component of the postsynaptic density (PSD) [30]. The PSD is composed of a detergent-resistant matrix containing excitatory neurotransmitter receptors and their subsynaptic scaffolds. Whilst this finding appears to contradict the association of dystrophin with inhibitory synapses, differences in the methodologies for PSD preparation may result in the apparent association of dystrophin with the PSD. GABA_B receptors have been detected in PSD preparations although GABA_A receptors were not similarly identified [19], [31]. Furthermore, β-dystroglycan (see below) is not PSD-enriched and is effectively removed from synaptosomes with the non-ionic detergent Triton X-100 [32]. Finally, in cultured hippocampal neurons, core DGC components are selectively associated with inhibitory synapses but were not found at excitatory glutamatergic synapses [28]. Taken together, these data suggest that dystrophin and the DGC are found at GABAergic rather than glutamatergic synapses.

Behaviour and synaptic plasticity in dystrophin-deficient animal models

The vast majority of studies addressing the function of dystrophin and DGC in brain have utilized the dystrophin-deficient mdx mouse. The mdx mouse is a commonly used model of muscular dystrophy that also has biochemical, physiological, and behavioural abnormalities attributable to the absence of dystrophin in the brain. Mdx mice have impaired retention in a T-maze but no deficit in locomotor activity [33]. The T-maze tests spatial working memory and is often used to assess hippocampal and forebrain function. Retention impairments at long delays in mdx mice suggest a role for dystrophin in long-term consolidation processes. By contrast, mdx^3cv mice lacking dystrophin and its C-terminal isoforms (see below) showed no retention deficit in the T-maze but displayed increased anxiety and reduced locomotion compared to wild-type mice [34].

Electrophysiological studies of synaptic function in the mdx mouse have generated important but conflicting insights into the role of dystrophin and its isoforms in synaptic plasticity. In the first experiments designed to examine synaptic plasticity in mdx mice, Sesay and colleagues found no deficit in hippocampal long-term potentiation (LTP) in mdx mice that lack only full-length dystrophin [35]. This study also found no difference in hippocampal-dependent spatial learning tasks between mdx mice and wild-type controls [35]. Using different training paradigms, Vaillend et al. found abnormal enhancement of hippocampal LTP in mdx mice and deficits in long-term spatial memory and object recognition [36]. The same study found that short-term potentiation (STP) and depression (STD) are also enhanced in the mdx mouse [37]. Furthermore, the GABA_A antagonist bicuculline increased basal neurotransmission in wild-type mice but had no effect upon the mdx mouse. Interestingly, bicuculline was able to prevent enhanced STP and STD in the mdx mouse [37]. Defects in long-term depression have been found in the cerebellum of mdx mice [38]. Bicuculline did not alter the LTD deficit, suggesting that another class of receptor may be affected by the absence of dystrophin [38].

Dystroglycan

Dystroglycan is a matrix receptor that spans the plasma membrane linking dystrophin and the cytoplasmic components of the DGC to the ECM. α-Dystroglycan and β-dystroglycan are produced by proteolytic cleavage and glycosylation of a single precursor propeptide. Dystroglycan proteolysis appears to be functionally important and occurs via the autocatalytic SEA (sea urchin, enterokinase, agrin)-module that contains the processing site for the precursor [39], [40]. α-Dystroglycan binds to a number of laminin G-domain-containing ECM proteins, including some laminins, perlecan, agrin, and the neurexins. Moreover, laminin-binding to α-dystroglycan is mediated by O-linked glycans, specifically O-linked mannose structures, in a Ca^{2 +}-dependent manner (see below) [41]. β-Dystroglycan associates with α-dystroglycan at the plasma membrane following proteolytic processing and binds to dystrophin and its isoforms on the cytoplasmic face of the membrane. Whilst it is beyond the scope of this review to describe in detail the function of dystroglycan, the reader is referred to the following reviews that cover this subject [41], [42].

No mutations have been described in the DAG1 gene that encodes dystroglycan in humans. It has been suggested that DAG1 is an essential gene in mammals. Indeed, mice lacking Dag1 (the murine dystroglycan gene) die in utero due to a defect in the formation of Reichert's membrane during early embryogenesis [43]. Conditional knockout of dystroglycan specifically in transgenic mice have provided important insights into the function of dystroglycan during brain development (see below) [44]. In addition to the neuronal migration abnormality in these mice, loss of dystroglycan in the hippocampus results in a severe deficit in LTP but has no effect upon basal neurotransmission and paired pulse facilitation, a measure of presynaptic neurotransmitter release [44]. Whether this defect in LTP is caused by cortical disorganization or reflects a direct role of dystroglycan in synaptic function is at present unclear. Nevertheless, these studies explain how defects in dystroglycan can cause severe CNS abnormalities in congenital muscular dystrophy (CMD) patients (see below).

α-Dystroglycan glycosylation defects, CMD, and brain development

While DMD is the most common inherited muscle disease in humans, in the last few years the genetic defects underlying several rare CMDs have been identified. Perhaps the most surprising finding to emerge from these studies is that that many of the CMDs are associated with defects in the posttranslational processing of α-dystroglycan. Mutations in the genes encoding several confirmed or putative glycosyltransferases have been discovered in patients with different forms of CMD (Table II). These mutations are associated with incomplete glycosylation of α-dystroglycan that abrogates its function as a matrix receptor. Many of the CMDs have a broad phenotypic overlap with mutations in several functionally related genes producing similar clinical diagnoses (Table II). For example, FKTN mutations in the gene that encodes fukutin were originally found in Japanese families with Fukuyama congenital muscular dystrophy (FCMD) [45]. Subsequently, FKTN mutations were found to cause a disease phenotypically resembling Walker-Warburg syndrome (WWS), steroid-responsive limb-girdle muscular dystrophy, and more recently dilated cardiomyopathy with mild muscle weakness [46–48].

Table II.  Muscular dystrophies associated with hypoglycosylation of α-dystroglycan. The Table lists the salient features of the diseases associated with a defect in α-dystroglycan glycosylation. Underlining indicates the original disease in which mutations were found. Severe structural changes include neuronal migration defects (causing lissencephaly), cerebellar hypoplasia, and cerebellar dysplasia [48].

Gene	Disease	OMIM	Locus	CNS involvement
FKTN	FCMD	253800	9q31	SSC, MR
	WWS	236670		SSC
	LGMD2M ^a	611588		None
	CMD1X ^b	611615		None
FKRP	LGMD2I	607155	19q13.3	Brain atrophy in some
	MDC1C	606612		CC, MR in some
	MEB	253280		SSC
	WWS	236670		SSC
POMT1	WWS	236670	9q34.1	SSC
	LGMD2K	609308		Mild MR, microcephaly
	CMD + MR	unassigned		MR
POMT2	WWS	236670	14q24.3	SSC
POMGNT1	MEB	253280	1p34-p33	SSC
	LGMD2M ^a			None
LARGE	WWS	236670	22q12.3-	SSC
	MDC1D	608840	q13.1	MR, WMC

^a The designation LGMD2M has been attributed to diseases caused by mutations in POMGnT1 and FKTN.

^b CMD1X is a form a cardiomyopathy with minimal muscle involvement caused by mutations in the FKTN gene [47].

Abbreviations: CC = cerebellar cysts; CMD = congenital muscular dystrophy; CMD1C = congenital muscular dystrophy type 1C; CMD1D = congenital muscular dystrophy type 1D; CMD1X = congenital muscular dystrophy type 1X; CNS = central nervous system; FCMD = Fukuyama congenital muscular dystrophy; LGMD2I = limb-girdle muscular dystrophy type 2I; LGMD2K = limb-girdle muscular dystrophy type 2K; LGMD2M = limb-girdle muscular dystrophy type 2M; MEB = muscle-eye-brain disease; MR = mental retardation; OMIM = Online Inheritance In Man (http://www.ncbi.nlm.nih.gov/omim/); SSC = severe structural changes; WMC = white matter changes; WWS = Walker-Warburg syndrome.

Mutations that cause FCMD, muscle-eye-brain disease (MEB), and WWS are commonly associated with defects in cortical migration, cerebellar cysts, and mental retardation [49], [50]. A subset of mutations in the gene encoding fukutin-related protein (FKRP) is also associated with CMD and similar brain pathologies [51]. Mutations in fukutin and its relative FKRP can cause a broad range of phenotypes; however, the function of each of these proteins has still to be determined. Each protein is predicted to be a glycosyltransferase having weak homology to bacterial phosphoryl ligand transferases, some of which are involved in synthesis of capsular polysaccharides [52], [53].

The enzymes, POMT1 (protein O-mannosyltransferase-1), POMT2 (protein O-mannosyltransferase-2), and POMGnT1 (protein O-mannose β-1,2-N-acetyl glucosaminyltransferase), catalyse the initial steps in protein O-linked mannosylation within the endoplasmic reticulum (ER). O-linked mannose is further modified by as yet unknown glycosyltransferases to produce mucin-like modifications of the α-dystroglycan peptide. One of the enzymes that are thought to generate terminal modifications on α-dystroglycan is LARGE. LARGE has two distinct glycosyltransferase domains. The N-terminal domain is similar to the glycosyltransferase 8 and 24, (uridine diphosphate, UDP-glucose glycoprotein glucosyltransferase) families whereas the C-terminal domain has similarities to the glycosyltransferase 31 (β-1,3-N-acetylglucosaminyltransferase 6) family [54], [55]. Interestingly, ectopic expression of LARGE can restore α-dystroglycan functionality in cellular models of CMD caused by defects in other CMD-associated genes [56].

Significant progress has been made in deciphering the role played by α-dystroglycan during brain development and neurogenesis. In addition to muscle, dystroglycan is also a laminin receptor in the brain. Laminin-α2 chain deficiency also causes CMD (merosin-negative CMD), and there is some evidence that this form of CMD is associated with mental retardation, structural abnormalities of the cortex and cerebellum, and white matter changes [57]. By contrast, dy/dy (dystrophia muscularis) mice harbouring mutations in the Lama2 gene that encodes the α2 chain of laminin have muscle disease, white matter changes, but no evidence of a neuronal migration defect [58]. Thus, laminin-2 may not be directly involved in the neuronal migration defects seen in the CMDs associated with hypoglycosylation of α-dystroglycan.

Myd (myodystrophy) mice homozygous for a deletion in the LARGE gene also have a neuronal migration defect that affects cortical lamination, neuronal placement in the hippocampal formation, and the organization of the laminated folia in the cerebellum [44], [59]. Conditional deletion of dystroglycan in the brain causes a similar neuronal migration defect suggesting that the deficiency of glycosyltransferases results in loss of dystroglycan function as a matrix receptor [44]. Clearly the identity of the brain-expressed ECM ligands for α-dystroglycan would contribute to our understanding of the neuropathology of CMD. These ligands may include other forms of laminin, perlecan, or agrin. Dystroglycan is also a receptor for the synaptic adhesion molecule neurexin [60]. Neurexins function as transsynaptic adhesion molecules that play important roles in synapse specification and postsynaptic differentiation [61]. Dystroglycan has also been found to influence the differentiation of neuroepithelial cells during CNS development [62]. The mechanism for this effect is as yet unknown; however, these studies raise the possibility that a defect in neuroepithelial proliferation may contribute to the severe brain abnormalities in some forms of CMD.

Cytosolic components of DGC in the brain

In addition to dystrophin and dystroglycan, the dystrobrevins and syntrophins are also components of the core DGC that are expressed in the brain (Figure 1B and 1C). The dystrobrevins are a family of dystrophin-related proteins encoded by two separate genes: DTNA encoding the α-dystrobrevins and DTNB encoding β-dystrobrevin. The dystrobrevins bind directly to dystrophin through reciprocal coiled-coil interactions. In addition, the dystrobrevins and dystrophin bind to the syntrophin protein family [63]. Whilst the α-dystrobrevins are part of the DGC in striated muscle, the related protein β-dystrobrevin is found in many non-muscle DGC-like complexes, including those found in the brain [64]. Mice lacking α-dystrobrevin have a mild form of muscular dystrophy, which does not impair the assembly of the core DGC [65]. By contrast mice lacking β-dystrobrevin appear to be phenotypically normal [66]. However, mice lacking both α- and β-dystrobrevin have impaired motor function, performing poorly in tests assessing sensorimotor function such as the ability to balance on a rotating rotorod [67]. Double knockout (dko) mice also have a reduction in cerebellar GABA_A receptors that is similar in magnitude to the deficit in the mdx mouse described above [67]. Dystrophin was also absent from GABA_A clusters in dko mice, demonstrating the interdependence of each protein for physiological synaptic targeting [67].

In addition to the DGC, several dystrobrevin-binding proteins have been described. These proteins include syncoilin, dysbindin (dystrobrevin-binding protein-1 (DTNBP1)), and neuronal kinesin heavy chain 5A [68]. Interestingly, dysbindin has been shown to be an important schizophrenia susceptibility gene [69]. Variants in the DTNBP1 gene that are associated with an increased risk of schizophrenia are also associated with poor educational performance and the so-called negative symptoms of cognitive decline in patients with schizophrenia [70]. Although the function of dysbindin in the context of the DGC has not been determined, dysbindin is part of another protein complex, the biogenesis of lysosome-related organelles complex-1 (BLOC-1) that is involved in protein trafficking between the endosomal and lysosomal compartments [71]. Thus BLOC-1 may play a role in the trafficking and assembly of DGC-like complexes in the brain.

The syntrophin family of adaptor proteins is composed of five members, α-, β1-, β2-, γ1-, and γ2-syntrophin [72–74]. Each syntrophin contains a PDZ (PSD-95 (postsynaptic density protein 95) discs–large and ZO-1 (zonula occludens-1)) domain and two PH (plecstrin homology) domains (the first PH domain of each syntrophin is split by the insertion of the PDZ domain). PDZ domains bind to the C-terminal tails of many different classes of transmembrane protein but can also participate in homotypic interactions with other PDZ domain-containing proteins including neuronal nitric oxide synthase (nNOS) [75]. The different syntrophins have been found to interact with a wide variety of proteins in the brain (see Table III for details). Many of these proteins have been identified in yeast two-hybrid screens for proteins interacting with the tails of transmembrane receptors. In most cases, the PDZ domain of the syntrophins is responsible for binding to specific conserved sequences at the tails of these proteins (Table III). Many of the syntrophin-associated proteins, such as nNOS, aquaporin-4 (AQP4), and the neuroligins, have important functions in the brain (Table III). However, in the majority of studies reporting these syntrophin-dependent interactions, the effect on the DGC has not been explored.

Table III.  Syntrophin-binding proteins. The table collates the repertoire of proteins that bind to the syntrophin protein family in different tissues. The majority of these interactions occur between the PDZ domain in syntrophin and the C-terminal tails of membrane associated receptors. For clarity, components of the core DGC and DGC-related complexes, such as Dp71 and the dystrophin paralogue utrophin, have been omitted.

Syntrophin	Binding protein	Domain	Reference
α	heterotrimeric G protein α subunits	PH1a and SU	[111]
γ2	neuroligins 3 and 4x	-HSTTRV*, -HSTTRV*	[112]
β2	erbB-4	-HRNTVV*	[113]
α	transient receptor potential-1 (TRPC1)	PDZ	[114]
α, β1, β2	α1D-adrenergic receptor	-LRETDI*	[115]
α, β1, β2	ABCA1	-VKESYV*	[116]
α, β2, γ1	TAPP1	-LPVSDV*	[117]
γ1	diacylglycerol kinase-ζ	-DQETAV*	[118]
α, β1, β2	Kir2x channels	-RRESEI*	[119]
α	Kir4.1	-VRISNV*	[120]
α	aquaporin-4^a	-EVLSSV*	[121]
β2	microtubule-associated serine/threonine kinase (MAST205) syntrophin-associated serine/threonine kinase (SAST)	PDZ:PDZ	[122]
α	neuronal nitric oxide synthase (nNOS)	PDZ:PDZ	[75]
α	stress-activated protein kinase 3 (SAPK3)	-SKETPL*	[123]
α, β1, β2	voltage-gated Na^+ channels, SCN4A and SCN5A	-VKESLV*, -DRESIV*	[124]
α	plasma membrane Ca^2 +-transporting ATPase 1 (PMCA1); plasma membrane Ca^2 +-transporting ATPase 4 (PMCA4)	-SLETSL*; -SLETSV*	[125]
α	α-actin-2	PH1a and SU	[126]
α	calmodulin	PH1a/PDZ/SU^b	[126], [127]
α	glutaminase	-NLESMV*	[128]

^a A direct interaction has not been shown biochemically. The asterisk (*) denotes the end of the peptide sequence such that most interactions between PDZ domains and their ligands occur within the extreme C-terminus of the protein.

^b The location of the calmodulin binding sites on α-syntrophin differs between reports.

Abbreviations: ABCA1 = adenosine triphosphate (ATP)-binding cassette, subfamily A member 1; DGC = dystrophin-associated glycoprotein complex; PDZ = PSD-95 (postsynaptic density protein 95), discs–large and ZO-1 (zonula occludens-1); Kir = inwardly rectifying K⁺ channel; PH = plecstrin homology; SU = syntrophin unique; TAPP1 = tandem PH-domain-containing protein 1; MDC1C = congenital muscular dystrophy type 1C; MDC1D = congenital muscular dystrophy type 1D.

Mice lacking α-syntrophin have muscle hypertrophy and abnormal neuromuscular junctions but do not appear to have any gross brain pathology or behavioural abnormalities [76], [77]. In common with muscle, the assembly of syntrophin-based complexes in the brain is likely to be dependent upon dystrophin, dystrophin isoforms, and the DGC. Mdx-βgeo mice that are devoid of dystrophin and all of its isoforms have a compound deficiency of syntrophins in the brain (Esapa and Blake, unpublished results) [78]. It is therefore possible that some of CNS deficits in DMD are related to the mistargeting of syntrophin-interacting proteins in the brain (see below). The lack of overt CNS phenotypes in mice lacking individual syntrophins may be due to genetic redundancy such that other syntrophin family members may have a compensatory role.

Involvement of the DGC in H₂O channel clustering

As mentioned above, the syntrophins can interact with the C-terminal tails of several different membrane-associated channels (Table III). Components of the DGC, notably Dp71, the syntrophins, and dystroglycan are highly expressed in perivascular astrocytes [19], [79], [80]. In perivascular astrocytes the DGC has been shown to be part of a protein complex that regulates intracellular water transport (Figure 1C). Aquaporin-4 (AQP4) is the major water-channel in the central nervous system and is ostensibly found at the end-feet of perivascular astrocytes and in ependymal cells that line the brain ventricles. Mice lacking AQP4 are resistant to brain oedema induced following acute water intoxication or ischaemic stroke [81]. The C-terminal tail of AQP4 contains the consensus sequence for PDZ binding (Table III). Although a direct interaction between the PDZ domain of α-syntrophin and the tail of AQP4 has not been demonstrated, AQP4 co-purifies with syntrophin and Dp71 from mouse in a cross-linked protein complex. Furthermore, a pool of AQP4 is mislocalized in α-syntrophin-deficient mice suggesting that α-syntrophin is required to anchor AQP4 to the membrane in perivascular astrocytes but not the endothelium [82]. AQP4 localization is also dependent upon the major brain isoform of dystrophin, Dp71. Dp71 null mice fail to concentrate AQP4 at glial end-feet, possibly due to impaired assembly of a DGC-like complex containing α-syntrophin. Furthermore, mdx-βgeo mice have an attenuated response to brain oedema compared to controls following water intoxication and fail to target AQP4 to astroglial end-feet surrounding capillaries [83]. Taken together these results suggest that DGC-like complexes in perivascular astrocytes are required for membrane targeting of AQP4. It should be noted that there is also a DGC-independent pool of AQP4 in the brain found in the granule cell layer of the cerebellum, in subpial glial cells and in ependymal cells of the ventricles [84].

The sarcoglycan complex in the brain

The sarcoglycans are a family of plasma membrane proteins that are expressed predominantly in striated and smooth muscle and peripheral nerve. The sarcoglycan family is composed of six members (α-, β-, γ-, δ-, ε-, and ζ-) that form a heterotetrameric complex that associates with dystroglycan at the plasma membrane. Although the precise function of the sarcoglycan complex is unknown, mutations in α-, β-, γ-, and δ-sarcoglycan cause different forms of autosomal recessive limb-girdle muscular dystrophies (LGMD), demonstrating its importance for normal muscle function [85]. Initial studies on the sarcoglycans suggested that their expression was ostensibly limited to striated and smooth muscle and to a lesser extent peripheral nerve. However, recent research has shown that ε-sarcoglycan has an important role in the brain and is mutated in myoclonus-dystonia syndrome (MDS, DYT11) [86]. In addition, the most recently described member of the sarcoglycan family, ζ-sarcoglycan, is also highly expressed in the brain [87].

ε-Sarcoglycan mutations and MDS

In 2001, Zimprich and colleagues identified heterozygous mutations in the SGCE gene, encoding ε-sarcoglycan, in five families with MDS [86]. MDS is a dystonia-plus syndrome characterized by myoclonic jerks, which predominantly affect the neck and upper limbs, in combination with focal or segmental dystonia. Symptoms often develop during childhood or early adolescence with myoclonus as the predominant feature, frequently accompanied with mild dystonia. Alcohol ingestion is frequently found to suppress myoclonic symptoms. In addition to the motor features, psychiatric abnormalities such as obsessive-compulsive disorder (OCD), alcohol misuse, depression, anxiety, and panic attacks have also been reported in some individuals [88], [89].

SGCE is a maternally imprinted gene thus, in most tissues, only one allele is expressed [86]. The majority of MDS cases are caused by nonsense mutations in the SGCE gene, with a lower incidence of missense mutations [90], [91]. Thus, although MDS is inherited in an autosomal dominant manner (with reduced penetrance on maternal transmission), most patients with heterozygous mutations in SGCE fail to produce any functional ε-sarcoglycan. Functional analysis of a small set of SGCE missense mutations has shown that mutations in the extracellular domain of ε-sarcoglycan impair membrane trafficking of the mutant proteins in cultured cells [92]. These missense mutations are less stable than the wild-type protein and are degraded by the ubiquitin proteasome system [92].

Whilst the function of ε-sarcoglycan has not been determined, defects in brain development and synaptic function have been implicated in the pathophysiology of the dystonias [93]. ε-Sarcoglycan is found in many brain regions and is associated with dopaminergic neurons in the substantia nigra and ventral tegmental area [94], [95]. Altered dopaminergic neurotransmission has been implicated in other forms of dystonia such as DOPA (l-3,4-dihydroxyphenylalanine)-responsive dystonia (DRD, GCH1) caused by mutations in the gene for guanosine triphosphate (GTP) cyclohydrolase I [96]. Preliminary neurochemical analysis of ε-sarcoglycan-deficient mice showed increased levels of dopamine and its metabolites within the striatum [97]. Thus, ε-sarcoglycan may be involved in dopaminergic neurotransmission.

A sarcoglycan complex in the brain?

A plethora of in vivo and in vitro studies have analysed the trafficking and assembly of the sarcoglycan complex. Despite evidence for assembly and trafficking of incomplete sarcoglycan complexes [98], a tetrameric assembly seems to be functionally important for endogenous sarcoglycans in all tissues and cell lines studied to date [99]. All currently described complexes consist of a βδ-sarcoglycan core complex, which then associates with α-/ε-sarcoglycan and γ-/ζ-sarcoglycan to form a tetrameric complex [98], [100]. Although α- and γ-sarcoglycan are predominantly expressed in striated muscle, the other sarcoglycans are more widely expressed. It is therefore possible that the tetrameric εβδζ-complex could be present in the brain. This complex has already been described in Schwann cells of peripheral nerve [101]. However, although theoretically possible, the existence of a prototypical tetrameric sarcoglycan complex in the brain is unlikely. No motor phenotypes have been described in animal models of LGMD that lack β- and δ-sarcoglycan [102], [103]. Conversely, Hjermind and colleagues have recently found that MDS patients with mutations in the SGCE gene do not appear to have any muscle involvement [104]. These data suggest that ε-sarcoglycan may function independently of the other sarcoglycans in the brain. Interestingly, ε-sarcoglycan does not require the presence of other sarcoglycans in order to reach the plasma membrane in heterologous cells [92], [105]. The existence of brain-specific ε-sarcoglycan isoforms generated by alternative splicing may suggest that ε-sarcoglycan has acquired additional functions independent of the conventional sarcoglycan complex. Further studies are required to determine sarcoglycan function and synaptic physiology in animal models of MDS.

Conclusion

D/BMD are complex diseases that in many cases can have both cognitive and neuropsychiatric components. Although little is known about the molecular mechanisms that cause cognitive impairment in D/BMD patients, considerable progress has been made in identifying the proteins and biochemical pathways that may contribute to this abnormality. Recent studies implicate dystrophin and the DGC in GABA_A receptor clustering. It is therefore tempting to speculate that GABAergic dysfunction may contribute to some of the behavioural and cognitive abnormalities in patients with D/BMD. Beyond dystrophin, there is now a considerable body of evidence showing that DGC-like complexes have an important role in the brain. For example, dystroglycan plays a role in neuronal migration during brain development, whereas other DGC-like complexes are involved in channel anchoring in the adult brain. Finally, SGCE mutations in MDS define an important and unexpected role for ε-sarcoglycan and the putative neuronal sarcoglycan complex in the brain. Whilst some of these interactions may contribute to the neuropathology of DMD and allied disorders, it remains to be established whether proteins such as the syntrophins and ε-sarcoglycan can function independently of the DGC. A detailed understanding of the functional role of dystrophin and the DGC in the brain will augment our understanding of the complex pathogenesis of muscular dystrophy.

Acknowledgements

Due to space constraints, it has sadly not been possible to cite all of the primary literature in this review. The work was generously supported by grants from the Wellcome Trust (#061225, Senior Research Fellowship to DJB) and the Medical Research Council. AW and ML are funded by an MRC Studentship and an Usher Cunningham Scholarship from Exeter College, Oxford respectively. Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.