The dystroglycan complex: From biology to cancer

Authors

  • Alessandro Sgambato,

    Corresponding author
    1. Centro di Ricerche Oncologiche “Giovanni XXIII”, Istituto di Patologia Generale, Catholic University, Rome, Italy
    • Istituto di Patologia Generale-Centro di Ricerche Oncologiche “Giovanni XXIII”, Università Cattolica del Sacro Cuore, Largo Francesco Vito 1, 00168 Rome, Italy.
    Search for more papers by this author
  • Andrea Brancaccio

    1. Istituto di Chimica del Riconoscimento Molecolare-Sezione di Roma (CNR) C/o Istituto di Biochimica e Biochimica Clinica, Catholic University, Rome, Italy
    Search for more papers by this author

Abstract

Dystroglycan (DG), a non-integrin adhesion molecule, is a pivotal component of the dystrophin–glycoprotein complex, that is expressed in skeletal muscle and in a wide variety of tissues at the interface between the basement membrane (BM) and the cell membrane. DG has been mainly studied for its role in skeletal muscle cell stability and its alterations in muscular diseases, such as dystrophies. However, accumulating evidence have implicated DG in a variety of other biological functions, such as maturation of post-synaptic elements in the central and peripheral nervous system, early morphogenesis, and infective pathogens targeting. Moreover, DG has been reported to play a role in regulating cytoskeletal organization, cell polarization, and cell growth in epithelial cells. Recent studies also indicate that abnormalities in the expression of DG frequently occur in human cancers and may play a role in both the process of tumor progression and in the maintenance of the malignant phenotype. This paper reviews the available information on the biology of DG, the abnormalities found in human cancers, and the implications of these findings with respect to our understanding of cancer pathogenesis and to the development of novel strategies for a better management of cancer patients.

STRUCTURAL AND FUNCTIONAL ASPECTS OF THE DYSTROGLYCAN COMPLEX

Dystroglycan (DG) is an adhesion complex composed of two subunits, α and β, firstly identified in skeletal muscle and brain (Ibraghimov-Beskrovnaya et al., 1992). It is encoded by a single gene (DAG1) located on human chromosome 3 and consisting of two exons and one intron. DG subunits are liberated by an early post-translational proteolytic cleavage of the precursor translation product within the endoplasmic reticulum (Holt et al., 2000; Esapa et al., 2003). Subsequently, the non-covalent α/β complex is ferried through the Golgi apparatus for an extensive maquillage with sugar molecules and finally it is targeted to the plasma membrane where it builds up a number of molecular contacts with several partners belonging to the extracellular matrix (ECM) and the cytoskeleton (Winder, 2001). DG is the pivotal member of a transmembraneous complex of glycoproteins associated with the cytoskeleton (dystrophin-glycoprotein complex, DGC) which, together with integrins, likely represents the major molecular bridge involved in the formation and stabilization of contacts at the cell-ECM interface during embryogenesis and in a wide variety of adult tissues (Ervasti and Campbell, 1993; Williamson et al., 1997; Durbeej and Campbell, 1998; Gesemann et al., 1998; Henry and Campbell, 1998; Straub et al., 1999; Saito et al., 2003).

The α-DG subunit is highly and heterogeneously glycosylated and binds with high affinity and in a calcium-dependent fashion to several extracellular partners sharing LNS domains (laminins, agrin, perlecan, and neurexins) (Rudenko et al., 2001). α-DG is extensively decorated with N- and O-linked sugar molecules, the latter especially present within its elongated central region (rich in Ser, Thr, and Pro residues) which accordingly has been defined as a mucin-like region (Brancaccio et al., 1995) (Fig. 1). Such extensive carbohydrate branching likely functions as a protecting shelter for the core protein towards chemical dangers coming from the outer matrix space (i.e., exogenous proteases). It is also responsible of the high variability in the apparent molecular mass of the protein which ranges between 120 and 180 kDa depending on the degree of glycosylation. Carbohydrate moieties, while still not completely characterized at the molecular level (Chiba et al., 1997; Smalheiser et al., 1998) have been also demonstrated to play a major role for DG function (Winder, 2001). Although no structural data on a protein-carbohydrate complex have been collected so far, it is likely that specific sugar epitopes protruding from the DG central mucin-like region are involved in the formation of contacts with its binding partners (Kanagawa et al., 2004).

Figure 1.

Scheme of dystroglycan (DG) domains. The DG precursor in higher vertebrates is formed by 895 amino acids. An early post-translational cleavage releases the two subunits (α and β) followed by an extensive decoration with sugars. Although some glycosylation sites are spread all along the core protein scaffold, the majority of carbohydrate groups are O-linked to the central region of α-DG (≈316–484). As a result, α-DG shows a dumbbell-like molecular shape in which two globular (and less glycosylated) domains are separated by an eloganted and extensively glycosylated mucin-like region rich in Ser, Thr, and Pro residues (Brancaccio et al., 1995). The DG complex is formed via a non-covalent interaction between the C-terminal domain of α-DG (linear binding epitope 550–565, (Bozzi et al., 2003)) and the ectodomain of β-DG (linear binding epitope 691–719, (Sciandra et al., 2001)). The cytodomain of β-DG includes at its very C-terminal end the dystrophin binding site (881–895) harboring a Tyr residue (phosphorylation site in position 892).

β-DG is the transmembrane subunit of the complex; its cytodomain binds the WW-domain of dystrophin (Huang et al., 2000) and a variety of other cytoskeletal or cytosolic protein factors, while the ectodomain forms non-covalent contacts with the C-terminal domain of α-DG (Sciandra et al., 2001) (Fig. 1). The association with cytosolic components involved in signal transduction (Cavaldesi et al., 1999; Russo et al., 2000) and the presence of a consensus sequence for Tyr phosphorylation, whose phosphorylation state has been proposed to influence its affinity for dystrophin (Ilsley et al., 2001), clearly alludes at an eventual role of the C-terminal end of β-DG as part of a signaling system (Chockalingam et al., 2002; Langenbach and Rando, 2002; Spence et al., 2004a).

As for many other multidomain protein molecules, the recombinant dissection of DG proved to be quite important. A good deal of knowledge about DG structural and functional relationships has been achieved by investigating its isolated domains obtained using recombinant expression approaches (Fig. 1). The quantitative expression of a peptide spanning the N-terminal region of α-DG in E. coli, or of its autonomous sub-domains, has allowed to shed light on its modular architecture (Brancaccio et al., 1997; Bozic et al., 2004), while the expression of α-DG C-terminal and β-DG ectodomain has allowed to identify the reciprocal inter-subunit linear binding epitopes involved in the formation of the α/β complex (Sciandra et al., 2001; Bozzi et al., 2003). In particular, the ectodomain of β-DG has been catalogued, based on its pronounced structural flexibility, within the growing family of natively unfolded proteins (Bozzi et al., 2004), displaying in vitro a relatively low affinity (lower µM range) towards the α subunit; in vertebrates such features could have been evolutionary selected in order to facilitate the action of DG at the sarcolemma of large skeletal muscle fibers, undergoing continuous contraction cycles, or also in order to favor an efficient connection to the actin-cytoskeleton, via the cytodomain of β-DG, for transmission of a signaling cascade (Spence et al., 2004b).

The N-terminal region of α-DG represents its extracellularly exposed side and therefore much attention has been focused on its binding properties. Several lines of evidence pointed out that glycosylation is strictly required for laminin binding (Ervasti and Campbell, 1993; Winder, 2001). Sugar epitopes protruding from the highly glycosylated mucin-like region of α-DG are likely to represent the major laminin-binding epitopes, although a partial contribution of its core protein cannot be totally ruled out (Wizemann et al., 2003; Bozic et al., 2004). Recent data obtained in DG-transfected cell lines show that α-DG undergoes a N-terminal cleavage after targeting at the plasma membrane (Kanagawa et al., 2004; Singh et al., 2004) involving a furin consensus site that would liberate a fragment (CN-DG, cleaved NHZ-terminal α-DG) corresponding to the N-terminal region of α-DG (Brancaccio et al., 1997; Bozic et al., 2004; Singh et al., 2004). Further experiments will be needed to verify the hypothesis that such peptide fragment could play an autonomous role within cells (Singh et al., 2004); however, it remains to be determined whether such an additional maturation step would also take place in tissues.

DG AND HUMAN DISEASES

DG versatility is outlined by its multiple biological functions. Aside from skeletal muscle, DG is expressed and plays an important role in building up mature post-synaptic elements within the central and peripheral nervous system (Gesemann et al., 1996; Grady et al., 2000; Zaccaria et al., 2001). Moreover, DG is quite a viable molecular bridge since it is exploited by dangerous infective agents such as arenaviruses and Mycobacterium leprae to invade and infect human cells (Cao et al., 1998; Rambukkana et al., 1998).

In mice the concerted action of DG and laminin is believed to trigger the initial phase of embryogenesis when the first contacts between cells and basement membranes (BMs) are established (Williamson et al., 1997). However, its role during embryogenesis remains controversial. While no mutations have been identified so far in human populations, thus confirming DG crucial primary role during peri-implantation in mammals, knock-out experiments in zebrafish showed that early development remained unaffected by the absence of DG while a severe dystrophic phenotype emerged during adulthood (Parsons et al., 2002). In fact, the connection between BMs and the cytoskeleton appears to be profoundly altered in a number of human neuromuscular pathologies. Severe muscular genetic disorders, such as Duchenne, limb-girdle, or congenital muscular dystrophies, originate from mutations hitting dystrophin, members of the DGC (sarcoglycans, laminin-2), or other membrane proteins (dysferlin, caveolin-3) but also other genes whose products are not necessarily targeted to the sarcolemma (as in the case of nuclear lamins or cytosolic calpain-3; for an extensive review see Cohn and Campbell, 2000). Although a common molecular scenario underlying all known and diverse neuromuscular diseases cannot be depicted, it has been observed that destabilization of sarcolemma can be frequently related to a certain degree of disassembly of the DGC, thus indicating a major role of DGC and DG for muscle stability (Bies et al., 1997; Allikian et al., 2004; Dudley et al., 2004; Lapidos et al., 2004).

The emergence of dystrophic phenotypes related to DG in differentiated skeletal muscle has been demonstrated by analyzing chimaeric mice harboring DG−/− muscle fibers (Cote et al., 1999) or via the conditional DAG1 gene disruption in skeletal muscle and brain (Cohn et al., 2002; Moore et al., 2002). Moreover, a number of congenital dystrophies have been recently identified as genetic disorders associated to genetic abnormalities of a group of glycosyltransferases (e.g., LARGE or the O-mannosyltranserases POMT1) which would be specifically involved in the α-DG glycosylation process (Longman et al., 2003; Willer et al., 2004). Pathological hypoglycosylation of α-DG, affecting its laminin-binding properties, is likely to represent a major molecular trigger in such diseases that therefore could be considered as secondary dystroglycanopathies (Haliloglu and Topaloglu, 2004; Muntoni et al., 2004). Interestingly, some new recent data suggest that hypoglycosylation and proteolytic degradation of α-DG N-terminal would also take place in normal cells or tissues generating a population of α-DG isophorms (Kanagawa et al., 2004; Pavoni et al., 2005). Transgenic experiments in mice have shown that the conservation of the post-translational α/β cleavage site, Gly-Ser, is crucial for protease recognition and for proper maturation of DG as indicated by the resulting dystrophic phenotype (Jayasinha et al., 2003). A site-directed mutagenesis approach has been carried out in a cellular system confirming an interesting relationship between the glycosylation pathway and the proper proteolytic processing of the DG precursor (Esapa et al., 2003). It should be pointed out that aside from the already mentioned secondary dystroglycanopathies, an altered glycosylation pattern seems to emerge also in cancer cell lines (see below and Singh et al., 2004).

The increased knowledge about DG domain organization (see Fig. 1 and Bozic et al., 2004) may be of crucial help also for investigating, and perhaps preventing, infectious diseases affecting human populations. In fact, it has been demonstrated that the 2nd domain within the α-DG N-terminal region represents a prominent region of the target site for arenaviruses via the formation of protein–protein contacts established with the viral capsid protein GP-1 (Kunz et al., 2001) while β-DG would play a marginal role within the infection pathway (Kunz et al., 2003).

DG AND CANCER

The interest for DG has been for long time limited to its role in muscle diseases. However, given its widespread distribution and important role in cell–ECM interaction, DG is expected to play also a crucial role in the function(s) of epithelial cells. Indeed, knock out mice for DG undergo premature death early in embriogenesis, before myogenesis has begun, mainly due to their inability to form and develop the Reichert's membrane, suggesting defects in early BM formation (Williamson et al., 1997). This finding is in agreement with the fact that genetic defects of DG have not been reported as causes of hereditary diseases in humans. Moreover, several evidence suggest that DG function is essential for a correct epithelial development in vivo (Henry and Campbell, 1998) and for epithelial morphogenesis in vitro (Durbeej et al., 2001) thus supporting an important role of this molecule in regulating the interaction of epithelial cells with the ECM and for the assembly of BMs.

Cellular interactions with the ECM are an important factor in the development and progression of many types of cancer and defects in ECM organization with perturbations of the BMs separating the epithelial and stromal compartments are considered a hallmark of malignant transformation. As mentioned above, DG is expressed at the interface between the BM and cell membrane. It forms a continuous link from the ECM to the actin cytoskeleton, providing structural integrity and perhaps transducing signals, in a manner similar to integrins. Thus, loss of DG expression might perturb the interactions between cells and the surrounding matrix and might contribute to the deregulation of cell ability to interact with BM and/or with the surrounding cells frequently observed in tumor cells.

The highly and heterogeneously glycosylated cell surface-associated α-DG has a size which ranges from 120 to 180 kDa, depending on different amount of glycosylation. β-DG has been identified as a 43 kDa band, sometimes accompanied by a 31 kDa band which likely originates from proteolytic fragmentation. We recently demonstrated that the 31 kDa band is expressed at an increased level in human breast cancer cell lines and in rat mammary tumors but not in the normal counterparts (Losasso et al., 2000). Increased expression of the 31 kDa band was usually associated with a reduced expression of the 43 kDa band. Interestingly, expression of α-DG was reduced or absent in most of the cell lines in which β-DG was found as a 31 kDa band but was also reduced or absent in some cell lines expressing a normal level of the 43 kDa β-DG band (Losasso et al., 2000). The accumulation of the 31 kDa band was consequence of post-transcriptional events as demonstrated by RT-PCR experiments which allowed to exclude other phenomena, such as alternative splicing of the mRNA. This band is retained into the cell membrane and lacks the extracellular domain of β-DG (Losasso et al., 2000). Yamada et al. (2001) demonstrated that the 31 kDa form of β-DG is the product of a proteolytic processing of the extracellular domain of β-DG. This processing is due to the membrane-associated matrix metalloproteinase (MMP) activity and disintegrates the DG complex (Yamada et al., 2001).

We hypothesized that an aberrant processing of DG might play a role in the development of epithelial cancers and in the definition of tumor cell phenotype by altering the interactions between cells and the surrounding matrix. DG may act as a tumor suppressor gene and a reduction/loss of its function could influence the formation of strong contacts between BMs and the cytoskeleton of cells, thus eventually favoring tumor development and invasiveness. From this point of view, it is of interest that the human DG gene maps to chromosome 3p21, a locus that has been involved in tumor suppression (Ibraghimov-Beskrovnaya et al., 1993).

Indeed, Henry et al. (2001) reported a reduction in the expression levels of DG in human primary prostate and breast cancers which was most pronounced in high-grade disease. To further address the role of DG in human tumorigenesis, we evaluated by Western blot analysis the expression of DG in a series of human cancer cell lines of various histogenetic origin and in a series of human primary colon and breast cancers. Decreased expression of DG was observed in most of the cell lines and analysis of the mRNA levels suggested that expression of DG protein is likely regulated at a post-transcriptional level (Sgambato et al., 2003). We also evaluated by Western blot analysis, the expression of both DG subunits in a series of human primary colon and paired adjacent normal mucosa samples and found that its reduced expression is a frequent event in colon cancer, is mainly due to post-transcriptional events and correlates with increasing tumor grade and stage (Sgambato et al., 2003). A reduced expression of DG has been also recently reported in skin and in other squamous cell carcinomas cancers and a potential involvement of MMPS in the loss of α-DG in these tumors has been demonstrated (Herzog et al., 2004; Jing et al., 2004). Evaluation of α-DG expression by immunostaining in a series of archival cases of primary breast carcinomas confirmed that α-DG expression is lost in a significant fraction of tumors (66%). Loss of DG staining correlated with higher tumor stage and high proliferation index and in a univariate analysis loss of α-DG was associated with an increased risk of death for disease. Loss of α-DG expression confirmed to be an independent predictor of overall survival when a Cox proportional hazards model was constructed that included α-DG expression, lymph node involvement, and tumor grade. Taken together these findings confirm that reduced expression of DG may play an important role in tumor development by causing an abnormal cell–ECM interaction and thus contributing to progression to metastatic disease, rather than merely being a consequence of neoplastic transformation. Indeed, the results obtained in breast cancer suggest that loss of α-DG expression may be an early event in the multistep process of breast carcinogenesis since it was already detected in pT1 tumors and in cases of in situ carcinoma (Sgambato et al., 2003). Preliminary studies also suggest that loss of DG expression might be an early event in other tumors being already detectable in preneoplastic lesions (unpublished observations).

DG AS A POTENTIAL TUMOR SUPPRESSOR GENE

Following the in vivo evidence of a potential involvement of DG in the process of tumor development, several studies have investigated the role of DG in the development and maintenance of the malignant phenotype mainly addressing the significance of the loss of α-DG in tumor cells and the responsible mechanism.

To investigate the effects of DG expression levels on the phenotype of breast normal and cancer cells we engineered the MCF7 human breast cancer and the MCF10F non-tumorigenic mammary epithelial cell line to stably express an exogenous DG cDNA. Both α- and β-DG subunits were increased in the MCF-10F cells expressing an exogenous DG cDNA. On the other hand, the MCF7 cells, that do not express α-DG, were not able to restore the expression of α-DG and DG overexpressing derivatives only displayed an increased expression level of β-DG.

The DG-overexpressing derivatives of the MCF-10F cells displayed an increase in the percentage of cells in the G0/G1 phase of the cell-cycle associated with a reduction in the percentage of cells in the S phase, a lengthening of the doubling time and a reduction of the saturation density (Sgambato et al., 2004). The DG-overexpressing derivatives of the MCF7 cells displayed similar effects in term of anchorage-dependent growth. Moreover, DG overexpression markedly inhibited both the anchorage-independent growth and the tumorigenicity of the MCF-7 cells. However, while DG overexpression resulted in an increase of cell substratum adhesion of the MCF10F cells, the DG-overexpressing derivatives of the MCF7 cells displayed a decrease of their strength to adhere to the substratum, as indicated by the decrease in their resistance to detachment. This effect was shown to be associated with a release of DG-related molecules in the extracellular medium, which might disturb cell-ECM interaction mediated by DG and/or other surface receptors. Release of α-DG from cell surface has been previously reported in other cell models, including bovine aortic endothelial cells (Shimizu et al., 1999) and rat schwannoma cells (Matsumura et al., 1997). We hypothesize that breast cancer cells have lost their ability to retain α-DG on cell membrane, that overexpression of an exogenous DG cDNA is not able to override this inability and that this mechanism might explain the loss of α-DG frequently observed in cancer cells.

It is of interest that, in agreement with our results, overexpression of an exogenous DG cDNA was previously reported to be not able to restore α-DG expression in a series of breast cancer cell lines (Muschler et al., 2002). In fact, Muschler et al. (2002) only succeeded to overexpress both DG subunits in the T4-2 breast cancer cell line but not in any of several other breast cancer cell lines. DG overexpression in the T4-2 breast cancer cell line was able to restore the ability of cells to undergo cytoskeletal changes, to polarize, and to restrict growth in response to BM proteins and was associated with a reduction of the tumorigenic potential.

Since DG subunits are encoded by a single gene and are formed upon cleavage of a precursor protein (Holt et al., 2000; Winder, 2001), we believe that breast cancer cells frequently develop a specific post-transcriptional mechanism responsible for the lack of expression of α-DG. Overexpression of an exogenous α-DG was obtained in the T4-2 breast cancer cell line and it is of interest that this cell line lacks expression of both α and β-DG subunits (Muschler et al., 2002). Thus, unlike most of the other breast cancer cell lines, a genetic or transcriptional alteration might be responsible for the loss of the entire endogenous DG complex in the T4-2 cell line.

This observation suggests that multiple mechanisms might be responsible for the lack of α-DG expression in different cancer cell lines and primary tumors. Indeed, Singh et al. (2004) recently reported an evidence supporting the existence of multiple mechanisms responsible for the lack of a correct α-DG function in breast cancer cells including: (i) MMP activity which might cleave the extracellular domain of β-DG or other DG-associated molecules, essential to stabilize α-DG attachment to the cell surface; (ii) altered glycosylation of α-DG which abolishes its ability to interact with ECM components and to be recognized by the commercially available antibodies that mainly interact with carbohydratic epitopes on the α-DG molecule; and (iii) cleavage of α-DG by the proprotein convertase furin which would remove a NH2-terminal peptide. However, other mechanisms have been also proposed and it would not be surprising if the list will extend in the future (Table 1). Laminin, perlecan, and agrin are the major binding partners of α-DG in non-muscle tissues and they might also play a role in guarantee the stability of the DG complex. Although the significance of some of these events and the contribution of each of them to the loss of DG function in cancer cells remain to be defined, these findings suggest a complex regulation of DG complex in non-muscle cells and the existence of multiple mechanisms for its disturbance during the process of tumor development. Thus, it is evident that multiple mechanisms might occur in the same cells (Singh et al., 2004) and that different mechanisms might be responsible for the loss of α-DG in different type of tumors but also amongst tumors of the same histogenetic origin (Yamada et al., 2001; Herzog et al., 2004; Jing et al., 2004; Singh et al., 2004).

Table 1. Multiple mechanisms proposed to explain the loss of α-DG expression and or function in cancer cells
MechanismsReference
Membrane shedding of α-DG 
 Proteolytic processing of β-DG(Losasso et al., 2000; Jing et al., 2004; Singh et al., 2004)
 Cleavage of β-DG by MMPs(Yamada et al., 2001; Herzog et al., 2004; Jing et al., 2004; Singh et al., 2004)
 Loss of perlecan(Herzog et al., 2004)
 Cleavage of other α-DG stabilizing molecules by MMPs(Singh et al., 2004)
 Cleavage of α-DG by furin(Singh et al., 2004)
Lack of ECM binding 
 Altered glycosylation of α-DG(Singh et al., 2004)

Moreover, the results obtained with the DG-overexpressing derivatives of the MCF-7 cells suggest that the DG complex might negatively regulate cell growth independently by the interaction with ECM components, at least in these cells. Indeed, β-DG contains three well characterized sites for protein interaction, recognized by SH2, SH3, and WW-domain proteins and can bind several molecules (Table 2) including MEK2, ERK, and Grb2, which are involved in the activation of several signaling pathways (Russo et al., 2000; Spence et al., 2004a). Moreover, inhibition of DG binding to laminin has been shown to disrupt the PI3/AKT pathway in muscle cells (Langenbach and Rando, 2002) and it has been suggested that in certain cells DG can act antagonistically to other adhesion molecules, such as integrins, suppressing the activation of downstream kinase cascades (Ferletta et al., 2003). Thus, DG might be involved in signal transduction pathways and an excess of β-DG might inhibit proliferation not only by a direct effect (i.e., activating inhibitory signals) but also by an indirect effect by blocking or interfering with stimulatory signals (i.e., sequestering molecules important for the activation of signaling pathways). DG might also play an important role in modulating the changes of cell shape and adhesion occurring in cultured mammalian cells during the cell-cycle (Suzuki and Takahashi, 2003). Future studies looking at downstream molecules that mediate DG-dependent intracellular signaling pathways will be necessary to definitively assess the role(s) of the DG complex on the regulation of non-muscle cells growth and proliferation.

Table 2. Potential partners of the β-DG subunit in epithelial cells
Cell functionReference
Signal transduction 
 Grb2(Yang et al., 1995; Russo et al., 2000)
 MEK1(Spence et al., 2004a)
 ERK(Spence et al., 2004a)
 c-Src(Sotgia et al., 2001)
 Fyn(Sotgia et al., 2001)
 Csk(Sotgia et al., 2001)
 NCK(Sotgia et al., 2001)
 SHC(Sotgia et al., 2001)
 Caveolin-1(Sotgia et al., 2001)
Cytoskeleton binding and stabilization
 Actin(Ghen et al., 2003)
 Utrophin(James et al., 2000)
Organization of microvilli and filopodia
 Ezrin(Spence et al., 2004b)

CONCLUSIONS AND PERSPECTIVES

It has been hypothesized that DG may act as a tumor suppressor gene and that a loss of DG expression and/or function might play an important role in tumor development which has been mainly linked to its function as a cell surface receptor for components of the ECM (Losasso et al., 2000; Henry et al., 2001; Muschler et al., 2002; Werner et al., 2002; Sgambato et al., 2003; Jing et al., 2004; Sgambato et al., 2004; Singh et al., 2004). Thus, loss of DG would inhibit the formation of strong contacts between BMs and the cytoskeleton of cells, thus eventually favoring tumor development and invasiveness. DG expression may also influence the ability of cancer cells to interact with the endothelium of lymphovascular channels and/or to migrate through the ECM at primary as well as secondary sites and reduced or loss of expression of this protein might induce an increased ability of cancer cells to detach from the primary site in the metastatic process. As for integrins, in fact, it may result in less sticky tumor cells able to move unhindered in the ECM, thus predisposed to metastasize (Gui et al., 1997). However, the above mentioned findings support a role for the DG molecule in cell functions (i.e., regulation of cell growth and differentiation) other than the structural integrity of cells and suggest that its involvement in the process of cell transformation might be extremely complex. This hypothesis is further supported by recent findings on the interaction of α-DG with two human homologues of genes involved in differentiation, hAG-2 and h-AG3, that appear to play a role in breast tumorigenesis (Fletcher et al., 2003). Thus, elucidation of the DG-dependent signaling pathways and of its function(s) and regulation in normal non-muscle cells will be essential to fully understand the role of this molecule in the process of human tumorigenesis.

It will be especially important to unravel all of the several mechanisms regulating the expression and function of the DG complex in epithelial cells and the DG-associated signaling pathways. In fact, DG is likely involved in signal transduction pathways, in a manner similar to integrins, whose elucidation is indispensable for a better understanding of the role of the DG complex in the regulation of cell functions.

The identification of these pathways will also contribute to a better understanding of tumor pathogenesis and will be essential in view of a potential exploitation of DG as a prognostic factor of risk of recurrence and metastasis and/or as a target for the development of novel therapies to halt tumor cell growth and metastasis. As previously mentioned, processing of β-DG with loss of α-DG has been shown to be due to the membrane-associated MMP activity (Yamada et al., 2001) and has been proposed as one of the major mechanisms responsible for loss of α-DG in cancer cells (Yamada et al., 2001; Jing et al., 2004; Singh et al., 2004). It is of interest that inhibitors of such activity have been proved effective for the treatment of human cancers (Maekawa et al., 1999) and it cannot be excluded, on the basis of the available data, that some of their beneficial effects might be in part due to the reduced processing of DG following their treatment. Restoration of DG expression and function has been also suggested as a new potential therapeutic strategy to inhibit tumor cell growth and spreading (Losasso et al., 2000; Muschler et al., 2002; Sgambato et al., 2003). Our results, however, suggest caution since they demonstrate that attempts to restore DG expression through a gene-therapy approach, as the one used in our study, while potentially useful for cancer treatment since they can inhibit cancer cell growth and tumorigenicity, might ultimately promote tumor cells spreading and metastasis by reducing cell attachment to ECM if cells have lost their ability to retain the α-DG subunit onto the membrane and release it into the extracellular medium.

In conclusion, we believe that a better understanding of DG-associated signaling pathway and of its role(s) in the process of tumor development is an area of high priority in cancer research and might provide useful insights into the mechanisms of human tumor pathogenesis and for the development of novel strategies to fight cancer cell proliferation and spreading. If this hypothesis will be confirmed, DG has the potential to become a suitable target for the development of targeted therapies, which might benefit not only cancer patients but also patients with muscular dystrophies.

Acknowledgements

The authors are grateful to all the members of their labs for helpful and stimulating comments and advices. This study was supported in part by a grant from Associazione Italiana per la Ricerca sul Cancro (AIRC) to A.S.

Ancillary