Dystroglycan versatility in cell adhesion: a tale of multiple motifs
© Moore and Winder. 2010
Received: 29 December 2009
Accepted: 17 February 2010
Published: 17 February 2010
Skip to main content
© Moore and Winder. 2010
Received: 29 December 2009
Accepted: 17 February 2010
Published: 17 February 2010
Dystroglycan is a ubiquitously expressed heterodimeric adhesion receptor. The extracellular α-subunit makes connections with a number of laminin G domain ligands including laminins, agrin and perlecan in the extracellular matrix and the transmembrane β-subunit makes connections to the actin filament network via cytoskeletal linkers including dystrophin, utrophin, ezrin and plectin, depending on context. Originally discovered as part of the dystrophin glycoprotein complex of skeletal muscle, dystroglycan is an important adhesion molecule and signalling scaffold in a multitude of cell types and tissues and is involved in several diseases. Dystroglycan has emerged as a multifunctional adhesion platform with many interacting partners associating with its short unstructured cytoplasmic domain. Two particular hotspots are the cytoplasmic juxtamembrane region and at the very carboxy terminus of dystroglycan. Regions which between them have several overlapping functions: in the juxtamembrane region; a nuclear localisation signal, ezrin/radixin/moesin protein, rapsyn and ERK MAP Kinase binding function, and at the C terminus a regulatory tyrosine governing WW, SH2 and SH3 domain interactions. We will discuss the binding partners for these motifs and how their interactions and regulation can modulate the involvement of dystroglycan in a range of different adhesion structures and functions depending on context. Thus dystroglycan presents as a multifunctional scaffold involved in adhesion and adhesion-mediated signalling with its functions under exquisite spatio-temporal regulation.
As a cell adhesion molecule dystroglycan performs two basic functions: as a physical connection between extracellular matrix and cytoskeleton, and as a transducer of signals from outside to inside, so called 'outside in signalling' (see below). Whatever ones belief in the contribution of either of these functions to normal or disease processes, it remains difficult to dissect functionally the mechanical adhesion component from the signalling component, indeed they probably are inseparable. Nonetheless different aspects of disease tend to be ascribed to mechanical or signalling functions. As alluded to above, a mechanical role for the DGC in muscle, of which dystroglycan is an essential core element, initially held sway, but this has been supplanted to a certain extent by signalling roles (reviewed in [12–14]). We shall discuss some of the signalling function that is directly associated with dystroglycan in later sections, but it is also worth noting that the DGC as a whole acts as a signalling scaffold for pathways that are not directly mediated through dystroglycan. Examples of these include nitric oxide synthase (NOS) signalling, Pi3-kinase and Akt-mediated survival signalling and pathways involving small and heterotrimeric G-proteins, mediated variously through dystrophin, dystrobrevin and syntrophins, for examples see [15–19]. Although in muscle there appears to be only one DGC, there must be spatio-temporal regulation of these signalling functions through mechanisms that remain to be elucidated, as it seems unfeasible that they could all operate simultaneously through the same complex. But besides these signalling roles that would clearly be perturbed if the integrity of the DGC were compromised due to changes in dystroglycan, there are more direct changes in dystroglycan function that would intuitively appear to have a more mechanical function. Disease associated with this latter scenario are termed the dystroglycanopathies, and arise not through mutations in dystroglycan itself, but in a number of genes that are known or presumed to have function in the glycosylation of dystroglycan [[Muntoni, 2007 #2825]]. These glycosyltransferase genes are responsible for the essential post-translational modifications of α-dystroglycan that are required for LG module binding, and without this glycosylation, dystroglycan is functionally unable to connect to the extracellular matrix. This can result in very severe disease involving both muscle and neurological phenotypes such as in Fukuyama and congenital muscular dystrophy 1D, muscle eye brain disease or Walker Warburg syndrome. These phenotypes also highlight another area of significant involvement of dystroglycan function, namely in the neuronal system, with functions in neuronal cell migration and patterning as well as neuronal cell adhesion, brain architecture and neuronal signalling. A discussion of this is beyond the scope of this review and readers are directed to more authoritative sources .
A number of studies utilising non-muscle cells, myoblasts and myotubes had identified tyrosine 890 of β-dystroglycan (Y890 in mouse, Y892 in humans) as a target for phosphorylation by Src and other Src family kinases, and furthermore that Y890 phosphorylation regulates the interaction between dystroglycan and key cellular binding partners, most notably dystrophin and utrophin [28–32]. Considering dystroglycan as a cell adhesion molecule, with regulation by tyrosine phosphorylation; we have investigated dystroglycan function in light of the integrin paradigm. Binding of extracellular matrix proteins to integrins causes allosteric changes across the plasma membrane resulting in the binding of signalling and adaptor proteins, leading to tyrosine phosphorylation and the establishment of multiprotein signalling complexes on the intracellular face, so called outside-in signalling (see  for recent review). Integrins constitute a large family of cell adhesion receptors found in many cell types and tissues. Integrins make connections with extracellular matrix proteins and cytoskeletal components, and they are all capable of signalling via tyrosine kinases (reviewed in [34, 35]). Depending on the cell, the extracellular matrix ligand and the particular combination of α-/β-integrin heterodimers, different signalling complexes are assembled leading to different intracellular responses. Unlike the multitude of integrin heterodimer combinations with different cell and tissue distributions however, there is only one ubiquitously expressed α-/β-dystroglycan heterodimer. Therefore integrins are represented in multiple genes giving rise to multiple functions, whereas dystroglycan is represented by one gene but is still capable of mediating multiple functions. In the case of dystroglycan in particular we have focussed our attentions on the cytoplasmic domain of β-dystroglycan and the myriad of predicted protein-protein interaction sites that could be involved in mediating these multiple functions (Figure 2). We have therefore investigated dystroglycan function and the potential interactions mediated through these sites in the context of adhesion-mediated signalling. Dystroglycan engagement by laminin results in the phosphorylation of the cytoplasmic region of β-dystroglycan with the site being unequivocally identified as tyrosine 890 [29, 31]. Not only was dystroglycan a substrate for adhesion-mediated tyrosine phosphorylation in a manner analogous to integrins, but also the tyrosine phosphorylation of dystroglycan at Y890 resulted in the disruption of the C-terminal WW domain interaction site for both utrophin and dystrophin (Figure 2C) [28–30]. Interestingly however, caveolin-3 binding to this same site via a WW-like domain  appeared insensitive to the tyrosine phosphorylation . Therefore extracellular signals were sufficient to generate a phosphotyrosine signal which acts as a molecular switch subsequently altering the binding of dystroglycan to two of its major ligands connecting it to the actin cytoskeleton, no different to outside-in signalling. The kinases responsible for the phosphorylation of dystroglycan was subsequently demonstrated to be Src and other Src family members , also key integrators of integrin-mediated adhesion signalling. In addition to the interaction between dystroglycan and dystrophin or utrophin, other cytolinker proteins such as plectin have also been demonstrated to make interactions with dystroglycan, and may also have a compensatory and protective role in situations where dystroglycan function is partly compromised such as in the mdx mouse . Whether there is direct competition for binding to dystroglycan between plectin and dystrophin in vivo is not clear, but biochemically, binding sites on dystroglycan for the two proteins do overlap, and in the absence of dystrophin, such as in the mdx mouse, the organisation of plectin and dystroglycan in the sarcolemma does change . A further competitive interaction is that between the signalling adapter Grb2 and dystrophin. Grb2 binds to the SH3 domain interaction motif (PxxP) that overlaps the WW domain binding motif (PPPY; Figure 2C) in dystroglycan [38–40], and although at the biochemical level the interaction is clearly competitive, the functional significance of any Grb2 interaction with dystroglycan in cells and tissues remains unclear. From studies in Drosophila there appears also to be competitive interactions for dystroglycan, but rather than the C-terminal WW domain motif being the focus of interactions a non-overlapping SH3 motif located more amino-terminally is functionally more important for such functions as polarity determination . Nonetheless the WW domain interaction site is essential, though due to the presence of a second WW domain interaction site that can also mediate interactions through dystroglycan, is partially redundant . It should be noted however, that compared to the almost 95% identity between vertebrate dystroglycan cytoplasmic regions, Drosophila dystroglycan is considerably more divergent .
Antibodies directed against α-dystroglycan and β-dystroglycan readily recognise both proteins in the sarcolemma and more precisely in costameres of striated muscles. However in the analogous focal adhesion structures of fibroblast or myoblast cells for example, only the extracellular epitopes of α-dystroglycan are detectable by antibody staining [44, 45]. As α- and β-dystroglycan are believed to be obligate heterodimers it has therefore been assumed that whilst β-dystroglycan was not detectable in focal adhesions by antibody staining or using GFP-fusions, that it is nonetheless present. It has been hypothesised that the cytoplasmic region of β-dystroglycan is in an environment subject to considerable 'molecular crowding' or 'epitope masking', a phenomenon seen with other focal adhesion proteins such as α-actinin , rendering it undetectable by routine visualisation techniques. Despite its 'presumed presence' the inability to detect or visualise dystroglycan in other classes of adhesion has hampered further investigation in this area. Recently significant inroads have been made into this problem, with the finding that dystroglycan is a component of podosome adhesion structures, and techniques have been developed to visualise dystroglycan in focal adhesions making the study of dystroglycan in these adhesion types more amenable [47, 48].
From work investigating the role of dystroglycan in adhesion structures it is clear that DG can play a part in regulating the type and abundance of such structures within a cell. Over-expression of tagged dystroglycan results in a relative increase in the levels of focal complexes and a decrease in fibrillar adhesions compared to other cellular adhesions whereas shRNA knockdown of dystroglycan causes the reverse effect . The mechanism of action of dystroglycan in this case is not known, but a candidate to mediate this effect may be vinexin recruitment. Dystroglycan has been shown to interact with vinexin, a vinculin binding partner that resides in focal adhesions. This binding occurs between the 3rd SH3 domain of vinexin and the proline rich C-terminal region of β-dystroglycan, specifically the most C-terminal SH3 binding domain  (Figure 2C). Dystroglycan shRNA knockdown myoblast cells spread poorly on E3 laminin (a dystroglycan specific ligand), but can be rescued through overexpressing dystroglycan. However, overexpression of a construct mutated in the vinexin-binding site is unable to rescue this cell spreading defect. Therefore, the interaction between vinexin and dystroglycan, and possibly vinculin too, can modulate cell spreading. The interaction of a vinexin SH3 domain with the C-terminal SH3 motif in dystroglycan also highlights a further potential competitive interaction. Peptide SPOT array data indicated that the vinexin SH3 interaction with dystroglycan was sensitive to Y890 phosphorylation , just as the binding of utrophin or dystrophin to this region are prevented by tyrosine phosphorylation [29, 30]. However as both vinexin and utrophin bind the same region when not phosphorylated, there could be direct competition between the two proteins for the SH3 motif to which vinexin binds and the overlapping type 1 WW domain motif to which utrophin binds (dystrophin is not present in myoblasts; see Figure 2C), rather than Src modulated phosphorylation-dependent switching for this site.
Transient overexpression of dystroglycan-GFP in the majority of cells in culture is associated with the appearance of numerous filopodia and microvilli like structures. Concomitant with the formation of these structures is a dramatic rearrangement of the actin cytoskeleton from the typical orthogonal array of stress fibres to a more peripheral band of actin cables and actin-containing microvilli and ultimately in the most extreme cases an apparent total dissolution of the stress fibre network and the appearance of numerous fine filopodia and dorsal microvilli-like structures rich in dystroglycan-GFP . Biochemical analyses of the dystroglycan cytoplasmic domain in isolation, revealed that it had intrinsic actin binding and bundling properties, which could in part account for the ability of dystroglycan to induce the formation of actin bundles at the cell membrane and drive filopodia formation . A series of deletion constructs of dystroglycan which removed all of the major domains of α- or β-dystroglycan in turn and in combination, revealed that the cytoplasmic domain of β-dystroglycan was both necessary and sufficient for filopodia formation providing that it was targeted to the plasma membrane either by secreted alkaline phosphatase-tags or by myristoylation sequences targeting it to the inner membrane leaflet [49, 50].
A manual search of the sequence of the cytoplasmic domain of dystroglycan revealed a stretch of basic residues in the juxtamembrane region (Figure 2B) that resembled similar sequences in a number of other cell adhesion receptors (CD44, ICAM-1, ICAM-2 and L-selectin) that are able to interact with ezrin, radixin, moesin (ERM) family proteins [51–54]. In most cases the recruitment of activated ezrin to these cell adhesion receptors has been reported to induce the formation of microvilli structures [51, 53, 54], dystroglycan would also appear to be no exception in this regard. Deletion or alteration of the character of these sequences in dystroglycan was sufficient to prevent the association of ezrin with dystroglycan and prevented dystroglycan-induced actin-rich protrusions [50, 55]. The co-localisation of endogenous dystroglycan and ezrin in microvilli-like structures and co-immunoprecipitation and GST-pulldown of dystroglycan and ezrin from cell extracts suggest a direct role for dystroglycan in the formation of microvilli-like structures [49, 55]. Moreover, the dystroglycan-dependent and ezrin-dependent formation of filopodia also required the activation of Cdc42, as dominant negative Cdc42 prevented the dystroglycan-dependent microvilli formation and the dystroglycan-dependent recruitment of ezrin to the actin cytoskeleton [49, 55]. Depletion of dystroglycan levels by RNAi also had an inhibitory effect on the ability of active Cdc42 to induce filopodia, indicating a central role for dystroglycan in the formation of filopodia in response to Cdc42 .
ERM family proteins have been demonstrated previously to interact with components of the Rho GTPase signalling machinery, notably RhoGDI  and the GDP/GTP exchange factor (GEF) for Cdc42 and Rho: Dbl [57, 58]. An interaction between dystroglycan, ezrin and regulatory elements of the Rho GTPase signalling pathway could explain the actions of dystroglycan on filopodia formation. We identified a dystroglycan-Dbl-ezrin complex by a combination of immunoprecipitation, tagged protein pulldown and immunofluorescence microscopy. Furthermore we demonstrated that the targeting of a dystroglycan-Dbl-ezrin complex to the membrane promoted the formation of filopodia and microvilli by the local activation of Cdc42 at the membrane. Mislocalisation of the dystroglycan-Dbl-ezrin complex to the cytoplasm by using a cytoplasmic dystroglycan construct had a dominant-negative effect on filopodia formation . Thus the peripheral membrane localisation of dystroglycan is again essential for its function as a scaffold for components of the actin-signalling machinery to productively generate new actin-based structures such as filopodia and microvilli which could also be important prerequisites for adhesion formation.
In addition to being a substrate for adhesion-mediated signalling, dystroglycan was also found to have an antagonistic role in other signalling cascades mediated by integrins. Most notably it has been demonstrated that integrin α6β1 and dystroglycan have antagonistic roles in signalling to the Ras-Raf-MEK-ERK (ERK) pathway, with dystroglycan having an inhibitory effect on integrin-mediated activation of the ERK pathway . Dystroglycan function is also required, independently of integrin function, for laminin 311 mediated stretch-activated ERK signalling in alveolar epithelial cells . The dystroglycan cytoplasmic region has predicted ERK MAPK binding sequences in the juxtamembrane region (Figure 2C) and through a combination of yeast-two hybrid and proteomic analyses components of the ERK pathway were found to interact with the cytoplasmic region of β-dystroglycan; both MEK and activated ERK were found as interactors. The activity of neither protein was affected by binding to dystroglycan, nor was dystroglycan a substrate, but surprisingly the localisation of dystroglycan with MEK or active ERK was quite different within the cell . Active ERK localised to focal adhesions as had been demonstrated previously , where it colocalised with α-dystroglycan, whereas MEK was colocalised with β-dystroglycan in membrane ruffles . This differential spatial localisation of two components of a canonical signalling cascade could provide a potential explanation for the antagonistic role of dystroglycan in opposing integrin-mediated ERK activation as above. A further development of this idea has been put forward, whereby scaffolds such as dystroglycan are important integrators of ERK signalling, in that they direct ERK signalling to or away from the nucleus by acting as dimerisation sites for ERK activation  reviewed in . Such a mechanism could explain the role of dystroglycan in integrin-mediated ERK signalling , whilst at the same time allowing a dystroglycan-dependent activation of ERK signalling  depending on whether the downstream target was nuclear or cytoplasmic .
One intriguing possibility that arises from the role of the juxtamembrane region in what appear to be disparate functions: ERM-mediated actin cytoskeleton remodelling and ERK signalling, is the additional observation that the same region is also involved in nuclear targeting of dystroglycan . Whilst the function of dystroglycan in the nucleus and the precise mechanism whereby it is targeted to the nucleus is unresolved it is interesting to consider this in the context of ezrin and ERK, proteins which are also targeted to the nucleus. ERM proteins and the related protein merlin are trafficked to the nucleus in a cell density/and or cell cycle-dependent manner [65, 66]. Merlin can regulate the activity of ERK, and ERK and merlin nuclear translocation are stimulated by cell adhesion . Given the interdependence in regulation of ERK by dystroglycan and or merlin (and possibly other ERM proteins) and that the NLS in dystroglycan is also the site of ERM or ERK interaction (Figure 2B) it is possible that ERK or ERM proteins might target dystroglycan to the nucleus but not vice versa. But what is the function of dystroglycan in the nucleus? The observations from the Cisneros lab would suggest there are separate nuclear dystroglycan complexes, containing dystroglycan and the short dystrophin isoform Dp71 and other dystrophin-associated proteins (DAPs), and that these may contribute to the nuclear cytoskeleton or nuclear membrane/cytoskeleton interface [67–69] in a manner analogous to the role of dystroglycan in the sarcolemma. Whether dystroglycan and any complexes it may form in the nucleus have a general role in nuclear architecture as might be suggested by the presence of DAPs, or the presence of dystroglycan in the nucleus is more dynamic and has some sort of regulatory function has yet to be established. More recent evidence from the Cisneros group suggests that there may be dynamic control of dystroglycan nucleocytoplasmic shuttling mediated by importins (Bulmaro Cisneros personal communication), but function still remains to be elucidated.
A further role for the juxta-membrane region of β-dystroglycan is in binding to the neuromuscular junction (NMJ) protein rapsyn . Dystroglycan and rapsyn are essential for the agrin-mediated clustering of acetylcholine receptors at the NMJ [71, 72], where they form a specific complex with utrophin even in mature muscle fibres (see  and reviews in [74, 75]). The RING-H2 domain of rapsyn appears to be responsible for the interaction with dystroglycan , and clearly given the separation from the C-terminal WW binding motif to which utrophin could bind, these two interactions could be simultaneous. But how the rapsyn-dystroglycan interaction in this region is accommodated in the face of potential competing interactions between ezrin and ERK remains to be tested. Furthermore, whilst binding to rapsyn is not in question, a role for ERM proteins in mediating acetylcholine receptor clustering through the same site has not been ruled out. Another dystroglycan interaction that has been characterised at the molecular level also occurs at synapses, though in this instance in the hippocampus. S-SCAM, also known as MAGI-2, itself a scaffold protein in inhibitory synapses in the hippocampus , interacts with dystroglycan at the N-terminal type 1 WW domain interaction motif  (Figure 2). To date S-SCAM is the only protein to be identified to bind to this region of dystroglycan. The precise role of dystroglycan in binding to S-SCAM, and how this impacts on S-SCAM functions as a scaffold to activate RhoA protein in response to NMDA receptor signaling in dendrites , is not clear.
We are grateful to Grinu Matthew for critical reading of the manuscript, to anonymous reviewers who made positive and valuable suggestions and to Bulmaro Cisneros for sharing his data ahead of publication. SJW is funded by a grant from the MRC G0701129. This review is based on a presentation given at the FEBS Workshop on Protein Modules and Networks in Health and Disease in Seefeld, Austria September 2009 organised by the Protein Modules Consortium http://www.proteinmodules.org/.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.