150th Anniversary Series: Desmosomes and the Hallmarks of Cancer

Abstract

Desmosomes represent adhesive, spot-like intercellular junctions that in association with intermediate filaments mechanically link neighboring cells and stabilize tissue architecture. In addition to this structural function, desmosomes also act as signaling platforms involved in the regulation of cell proliferation, differentiation, migration, morphogenesis, and apoptosis. Thus, deregulation of desmosomal proteins has to be considered to contribute to tumorigenesis. Proteolytic fragmentation and downregulation of desmosomal cadherins and plaque proteins by transcriptional or epigenetic mechanisms were observed in different cancer entities suggesting a tumor-suppressive role. However, discrepant data in the literature indicate that context-dependent differences based on alternative intracellular, signal transduction lead to altered outcome. Here, modulation of Wnt/β-catenin signaling by plakoglobin or desmoplakin and of epidermal growth factor receptor signaling appears to be of special relevance. This review summarizes current evidence on how desmosomal proteins participate in carcinogenesis, and depicts the molecular mechanisms involved.

Keywords::

• cell–cell contacts
• desmosomes
• adherens junctions
• plakoglobin
• plakophilin
• desmoplakin
• cancer

Abbreviations
ADAM	=	a disintegrin and metalloproteinase
AJ	=	adherens junction
Dsc	=	desmocollin
Dsg	=	desmoglein
DP	=	desmoplakin
GFR	=	growth factor receptor
IF	=	intermediate filament
MMP	=	matrix metalloproteinase
PKP	=	plakophilin

INTRODUCTION

The locally and timely coordinated establishment of cell–cell junctions is a driving force for cell positioning, morphogenesis, and differentiation during development. In the adult organism, maintenance of tissue integrity or repair of tissue defects after wounding in a similar way depends on specific cell–cell contacts. In contrast, a loss of cell–cell contacts in response to either deregulated protein expression or the expression of mutated variants of the corresponding junctional molecules is associated with diseases. Especially Ca²⁺-dependent cell–cell adhesion molecules of the cadherin family both in adherens junctions (AJs) and desmosomes play a pivotal role in this respect. With their single transmembrane domain, the cadherins are integrated into the plasma membrane and interact with cadherins on the surface of opposing cells (van Roy, 2014). A specific set of proteins associated with the cadherin cytosolic domains tethers them to actin or intermediate filaments (IFs), respectively, thus establishing a mechanically stable network of connections between cells in epithelial and endothelial cell layers (Wu & Yap, 2013). In addition to their adhesive function, some of these cadherin-associated proteins are involved in signaling. Specifically, members of the armadillo repeat family of proteins (β-catenin, p120^ctn, plakoglobin, plakophilins [PKPs]) adopt a second function in the regulation of gene transcription and thus were classified as NACos (Nuclear and Adhesion Complexes) proteins (Balda & Matter, 2003). In this context, cell–cell contacts do not only represent sites where cells just stick together but also have to be considered as highly dynamic platforms for outside-in and inside-out signaling. Disturbance of these processes can be linked to tumor formation.

In mammals, the desmosomal cadherins desmoglein1–4 (Dsg1–4) and desmocollin1–3 (Dsc1–3) through interactions of their extracellular domains form spot-like adhesion sites between opposing cell membranes. To establish and maintain desmosome-mediated mechanical stability of tissues it is essential that desmosomal cadherins are associated with IFs (Nekrasova & Green, 2013). This requires assembly of a characteristic complex of proteins binding to their cytosolic domains (Figure 1) detectable as an electron-dense plaque in electron micrographs. Within this protein complex, plakoglobin and PKPs directly associate with the desmosomal cadherins and thereby link desmoplakin (DP) or other plakins including plectin, envoplakin, periplakin, and epiplakin to the complex (for an overview on plakins see Bouameur et al., 2014). The desmosomal armadillo-repeat proteins, in addition to their role in clustering of desmosomal components within the adhesive complexes, exhibit signaling activities in the cytosol and nucleus. The specific role of plakoglobin and PKPs in this respect is summarized in recent reviews (Aktary & Pasdar, 2012; Hatzfeld et al., 2014). From these facts it is clear that impairment of desmosomal components by mutations or autoantibodies dramatically affects desmosome functions becoming especially prominent in skin and cardiac muscle. Readers interested in these desmosomal diseases are referred to recent exemplary reviews (Al-Jassar et al., 2013; Amagai & Stanley, 2012; Brooke et al., 2012; Cirillo, 2014; Cirillo & Al-Jandan, 2013; Patel & Green, 2014; Simpson et al., 2011; Spindler & Waschke, 2014).

Figure 1. Desmosomal proteins contribute to tumorigenesis. In normal epithelial cells, AJs (cadherin–catenin complex) linked to the actin cytoskeleton and desmosomes (desmosomal cadherins: Dsg and Dsc, and plaque proteins) associated with IFs form stable cell–cell contacts. The tetraspanin protein PERP appears to be essential for desmosome formation, but its role is not fully understood. Activation of GFR tyrosine kinases disrupts desmosomes and induces endocytosis of desmosomal cadherins and dissociation of desmoplakin from the desmosomal plaque. Moreover, MAPK family members are activated including ERK1/2 which normally can be kept inactive by Erbin. Cytosolic tyrosine kinases of the Src-family also destabilize desmosomes. In targeting specific residues in plakoglobin, they define its binding to specific interaction partners. Plakoglobin can be included into AJs or compete with β-catenin degradation and transcriptional activity in association with TCF/LEF transcription factors. Probably, plakoglobin like β-catenin can also regulate the transcriptional activity of further transcription factors. Plakophilin released from the desmosome is involved in the regulation of translation (not shown) or has nuclear functions in association with β-catenin or RNA-polymerase III (not shown, see Hatzfeld et al. 2014). α, α-catenin; β, β-catenin; DP, desmoplakin; Dsg, desmoglein; Dsc, desmocollin; E-cad, E-cadherin; GFR, growth factor receptor tyrosine kinase; IFs, intermediate filaments; LEF, lymphocyte enhancer factor; p120, p120-catenin; PERP, p53 apoptosis effector related to PMP-22; Pg, plakoglobin; PKP, plakophilin; TCF, T-cell factor; TF, transcription factor.

Desmosomes are composed of a tissue-specific set of cadherin transmembrane proteins and are especially abundant in tissues subjected to heavy mechanical and contractile stress. Dsg2 and Dsc2 are common to all desmosome-containing tissues, whereas the other Dsgs and Dscs are expressed in a differentiation-dependent graded pattern in certain stratified epithelia (Kljuic et al., 2003; Mahoney et al., 2006; North et al., 1996; Nuber et al., 1995). In human skin, Dsg1, Dsg4, and Dsc1 exhibit most prominent expression in the granular differentiated layer, whereas Dsg3, Dsc2, and Dsc3 show the highest expression in the basal proliferative layer (for more detail see previous reviews in this series Berika & Garrod, 2014; Harmon & Green, 2013). In cells expressing an overlapping pattern of desmosomal cadherins, different Dsgs and Dscs can also be detected in individual desmosomes (North et al., 1996; Nuber et al., 1996). In general, there is evidence that all desmosomes in vivo contain at least one member of the Dsgs and Dscs, and only in the presence of one cadherin of each family strong adhesion occurs (Getsios et al., 2004). In this context, different studies provided evidence, that both homo- and heterophilic interactions between desmosomal cadherins can exist (Chitaev & Troyanovsky, 1997; Getsios et al., 2004; Marcozzi et al., 1998; Nie et al., 2011; Runswick et al., 2001; Syed et al., 2002; Waschke et al., 2007). Interestingly, a Ca²⁺-independent “hyperadhesive” state was reported for mature desmosomes, which is characterized by a complete resistance to chelation of Ca²⁺ ions with ethylene glycol tetraacetic acid (EGTA) (Garrod & Tabernero, 2014; Garrod, 2013; Garrod et al., 2005).

Electron tomography of desmosomes from neonatal mouse skin embedded in plastic revealed that the N-terminal extracellular cadherin (EC) domains of desmosomal cadherins form a series of 10–20 disordered knot-like assemblies with an architecture of desmosomal cadherin extracellular domains highly similar to classical cadherins (He et al., 2003). A more recent cryo-electron tomographic analysis of vitreous sections of human skin desmosomes revealed a highly ordered zipper-like assembly of alternating cis- and trans-interacting desmosomal cadherins without interdigitations (Al-Amoudi et al., 2007). Both approaches were not able to differentiate between homophilic and heterophilic interactions of desmosomal cadherins. Despite these structural similarities with E-cadherin, the contribution of desmosomal cadherins to tumorigenesis is by far less clear.

In two seminal papers, Hanahan and Weinberg defined characteristic hallmarks of cancer (Hanahan & Weinberg, 2000; Hanahan & Weinberg, 2011). Here, we want to discuss how the known functions of desmosomes or desmosomal components can be classified into these hallmarks of cancer and how they may be involved in cancer. This overview does not have the intention to be exhaustive.

DESMOSOMES IN THE CONTEXT OF THE CANCER HALLMARKS

Activation of invasion and metastasis

A critical step during cancer progression is local invasion of primary tumor cells into nearby blood and lymphatic vessels, dissemination throughout the body, and finally extravasation into distant tissues. Loss of contacts between cells of the primary tumor definitely represents an essential step during malignant neoplasia. This includes that not only AJ but also all other cell–cell contacts including desmosomes have to be resolved or modulated (Boyer et al., 1989; Savagner et al., 1997). Early findings showed that expression of desmosomal cadherins inhibits invasive behavior of nonadhesive fibroblasts (Tselepis et al., 1998). Reduced expression of desmosomal cadherins later on was observed in different tumors such as skin (Chen et al., 2012b; Harada et al., 1996; Tada et al., 2000; Xin et al., 2014), head and neck (Shinohara et al., 1998; Wong et al., 2008), lung (Cui et al., 2012a; Cui et al., 2012b), breast (Klus et al., 2001; Oshiro et al., 2005), prostate (Barber et al., 2014; Pan et al., 2014), cervix (Alazawi et al., 2003), uterus (Nei et al., 1996), pancreas (Hamidov et al., 2011), liver (Schüle et al., 2014), gastric (Biedermann et al., 2005; Yashiro et al., 2006), and colon (Cui et al., 2011; Funakoshi et al., 2008; Kamekura et al., 2013; Khan et al., 2006; Knösel et al., 2012; Kolegraff et al., 2011). Taken together, these data suggest a role of desmosomal cadherins as suppressors of tumor metastasis.

However, the situation is not as clear. Other studies report enhanced expression of desmosomal cadherins in cancer (Brennan & Mahoney, 2009; Brown et al., 2014; Chen et al., 2007; De Bruin et al., 1999; Huang et al., 2010; Kurzen et al., 2003; Savci-Heijink et al., 2009). Therefore, both loss and overexpression of desmosomal cadherin expression frequently correlate with advanced tumor grading, increased invasion and metastasis, and/or poor prognosis/survival. It is interesting to note that especially the expression of Dsgs appears to be upregulated in tumor cells.

Genetic mouse models represent interesting tools to address the role of desmosomal cadherins in vivo in more detail. Overexpression of Dsg2 or Dsg3 in suprabasal skin layers in a transgenic mouse model increased keratinocyte proliferation, anchorage-independent cell survival, and development of skin tumors (Brennan et al., 2007; Merritt et al., 2002). Expression of an N-terminally deleted Dsg3 in skin also resulted in hyperproliferation (Allen et al., 1996), whereas Dsg3 knockout mice revealed no abnormalities (Koch et al., 1997). In contrast, in a Dsc1 knockout situation, epidermis is hyperproliferative (Chidgey et al., 2001). When Dsg3−/− or Dsg3+/− keratinocytes transformed through activated H-Ras and p53 inactivation where assayed for tumor growth in Scid mice, tumors derived from Dsg3+/− keratinocytes showed significantly higher tumor volume compared with Dsg3−/−-derived tumors. This suggests that Dsg3 facilitates tumor growth (Baron et al., 2012). In an autochthonous squamous cell carcinoma model with chronic ultraviolet B (UVB)-treated Dsg3−/− mice no enhanced tumorigenesis was detectable probably due to compensatory upregulation of Dsg1 and Dsg2 (Baron et al., 2012). Consistently, Dsg1 was reported to compensate the genetic loss of Dsg3 in a transgenic mouse model (Hanakawa et al., 2002).

Currently, the molecular mechanisms for these desmosomal cadherin-specific differences in proliferative behavior are not understood. Apparently, mutations in desmosomal cadherin genes are not of major relevance in cancer (Biedermann et al., 2005). Recent studies indicate that individual desmosomal cadherins define distinct variations in adhesiveness, migration, differentiation, and/or morphogenesis. Here, the ratio and expression of specific desmosomal cadherins, Ca²⁺-dependent or -independent state of adhesion, and distinct mechanism of incorporation into desmosomes appear to play a role (Getsios et al., 2004; Hardman et al., 2005; Hartlieb et al., 2013; Ishii et al., 2001; Lowndes et al., 2014).

Regulation of desmosomal cadherin levels

According to the observations reported above, changes in desmosomal cadherin levels apparently contribute to tumorigenesis. Three major mechanisms have to be considered that can lead to altered desmosomal cadherin expression in cancer.

Transcriptional regulation of desmosomal cadherins

In general, little is known about the mechanisms controlling transcription of desmosomal cadherin genes, although promoter regions of genes have been identified early after cloning of the corresponding coding genes (Adams et al., 1998; Johns et al., 2005; Marsden et al., 1997; Silos et al., 1996). Recent years identified a number of transcriptional factors that modulate desmosomal cadherin expression including Stat6, Klf5, Smad4, grainy head-like 1, Foxn1, Cdx1/Cdx2, c-Rel, lymphocyte enhancer factor/transcription factor (LEF/TCF), and p53/p63 (Ferone et al., 2013; Funakoshi et al., 2008; Johnson et al., 2014; Kenchegowda et al., 2012; Mlacki et al., 2014; Omori-Miyake et al., 2014; Oshiro et al., 2003; Owens et al., 2008; Tokonzaba et al., 2013; Wilanowski et al., 2008). Interestingly, plakoglobin in a complex with LEF-1 is able to bind to the proximal promoter regions of both Dsc2 and Dsc3, however with opposite effects. Dsc2 expression is activated, whereas Dsc3 transcription is repressed resulting in a shift in isoform expression (Tokonzaba et al., 2013). This effect shows some similarity to the classical cadherin switch from E- to N-cadherin during epithelial–mesenchymal transition (EMT). In this context, it has to be noted that expression of the EMT-inducing transcription factor Snail was reported to induce degradation of Dsg2 (Kume et al., 2013). In addition, epigenetic alterations turned out to play a dominant role in the regulation of desmosomal cadherin expression in cancer. Promoter methylation (Cui et al., 2012a; Cui et al., 2011; Oshiro et al., 2005; Pan et al., 2014; Wang et al., 2014) as well as histone methylation (Ke et al., 2009) and histone acetylation (Gould et al., 2010; Johnson et al., 2014) have been reported to modulate desmosomal cadherin expression. Similar observations have been made for the desmosomal plaque proteins plakoglobin (Breault et al., 2005; Shafiei et al., 2008; Shiina et al., 2005; Winn et al., 2002), PKPs (Oka et al., 2009), and DP (Yang et al., 2012) and have been summarized in recent reviews (Aktary & Pasdar, 2012; Bouameur et al., 2014; Hatzfeld et al., 2014).

Impaired transport, targeting, and assembly into mature desmosomes

Impaired desmosome transport and assembly is a highly complex process that requires multiple coordinated steps. According to the present view, desmosomal cadherins in association with plakoglobin are transported by kinesin-mediated vesicular traffic (Burdett & Sullivan, 2002). Interestingly, vesicular transport of different desmosomal cadherins depends on distinct kinesins (Dsg2—kinesin-1; Dsc2—kinesin-2) (Nekrasova et al., 2011). DP and PKP2 play an important role in desmosomal cadherin association with IFs and assembly into cell–cell contact sites (Bass-Zubek et al., 2008; Godsel et al., 2005). Impaired DP expression or function affects desmosome assembly at the cell surface and thus may contribute to tumorigenesis. In line with this, genetic deletion of DP in a pancreatic neuroendocrine tumor model resulted in increased local invasion (Chun & Hanahan, 2010). Finally, at the cell surface specific factors may stabilize desmosomal cadherins as shown for Dsg2 in association with the lectin galectin-3 (Gal3). In the absence of Gal3, Dsg2 is internalized (Jiang et al., 2014). However, it cannot be excluded that half-desmosomes pre-assemble in the membrane and attach with counterparts on the opposing cell surface to establish a mature desmosome (Demlehner et al., 1995).

Moreover, the actin cytoskeleton has been shown to play a role for desmosome assembly and desmosomal cadherin dynamics (Roberts et al., 2011). In this context, it was reported that the actin-organizing protein adducin regulates Dsg3 membrane incorporation and mobility in a phosphorylation- and RhoA-dependent manner and its knockdown impairs cohesion (Rötzer et al., 2014; Waschke et al., 2006). On the other side, Dsg3 acts as a regulator of Rac-1 and Cdc42 activity, thereby modulating the structure of the actin cytoskeleton (Tsang et al., 2012a). For more details on desmosome assembly and dynamics see Harmon & Green (2013) and Nekrasova & Green (2013). However, the impact of these mechanisms on tumor formation and progression has to be analyzed in more detail.

Inactivation by proteolytic cleavage

Active depletion of desmosomal cadherins in cancer by proteolytic degradation can either occur directly at the cell surface by shedding proteases or via the endosomal– lysosomal degradation pathway. Shedding proteases are frequently upregulated in tumor cells and promote the release of growth factors and cytokines but also target a number of adhesion molecules. Proteomic analyses indeed could detect released extracellular fragments of desmosomal cadherins in tumor secretomes (Bech-Serra et al., 2006; Greening et al., 2013; Shi et al., 2009). Different proteinases have been identified that release desmosomal cadherin extracellular domains from cells including a disintegrin and metalloproteinase (ADAM), matrix metalloproteinases (MMPs), or kallikreins (Bech-Serra et al., 2006; Borgono et al., 2007; Brooke et al., 2014; Cirillo et al., 2007; Jiang et al., 2011; Klessner et al., 2009; Weiske et al., 2001). There is evidence that cleavage is cadherin- and/or cell-type-specific. It is currently not clear, if this proteolytic shedding of extracellular domains only reduces the number of molecules available for adhesion, or whether these soluble fragments themselves have a further physiological function, for example, in interference with intercellular adhesion (Noe et al., 2001; Wheelock et al., 1987).

Deregulated endocytosis was defined as multifaceted characteristic of cancer cells (Mosesson et al., 2008) and represents another efficient route to reduce the levels of cell surface proteins. It is now well accepted that tumor-associated growth factors induce endocytic internalization of junctional molecules (also see “Sustainment of proliferative signaling” section). Recent studies indicate that cadherins can choose different routes of internalization depending on the cellular environment, growth factors, and migratory needs of the respective cells (Delva & Kowalczyk, 2009). Dsg3 uses a lipid raft-mediated pathway for internalization in cross talk with epidermal growth factor receptor (EGFR) (Bektas et al., 2013; Calkins et al., 2006; Delva et al., 2008; Stahley et al., 2014). Interestingly, tail–tail interactions of the Dsg2 C-terminal unique region inhibit internalization (Chen et al., 2012a). In binding to a specific motif in Dsg2, a role for caveolin-1 in the process of endocytosis was recently reported (Brennan et al., 2012).

Future studies have to further unravel the regulatory signals that induce endocytosis. In cancer cells, GFR-induced posttranslational modifications may contribute to destabilization of desmosomes in releasing desmosomal cadherins from mature junctional complexes and subsequent uptake into endocytotic vesicles (Aoyama et al., 2010; Jennings et al., 2011).

One important consequence of changes in desmosomal cadherin expression in cancer is that the signaling functions of the desmosomal plaque-associated armadillo repeat proteins are modulated. The interrelation of desmosomal plaque proteins with β-catenin signaling is of specific interest in this context. Changes in the localization of the desmosomal plaque proteins, their expression levels or stability, for example, in response to posttranslational modifications, have been shown to modulate TCF/LEF-dependent transcription in a context-specific manner, with plakoglobin mainly acting as negative regulator suppressing oncogenic β-catenin–TCF/LEF signaling. For more details see recent comprehensive reviews (Aktary & Pasdar, 2012; Chidgey & Dawson, 2007; Dusek & Attardi, 2011; Hatzfeld, 2007; Hatzfeld et al., 2014).

Resistance to cell death

Loss of cell–matrix or cell–cell contact in normal epithelial cells is a cell-death-inducing signal (Gilmore, 2005; Valentijn et al., 2004); however, tumor cells often develop specific properties to resist cell death. During apoptosis, a large set of substrate proteins including adhesion molecules including desmosomal cadherins, cytoskeletal proteins, and their regulators were identified as caspase substrates. Interestingly, after induction of apoptosis not only a caspase-mediated cleavage within the cytosolic domain but also a release of the extracellular domain of desmosomal cadherins was observed (Cirillo et al., 2007; Cirillo et al., 2008; Dusek et al., 2006; Lanza & Cirillo, 2007; Weiske et al., 2001). In addition, proteins of the desmosomal plaque have been identified as caspase targets (Aho, 2004; Kalinin et al., 2005; Weiske et al., 2001).

The interesting question in respect to tumor progression now is, do the observed changes in desmosomal cadherin expression contribute to regulation of cell death and how? Overexpression of Dsg2 increased anchorage-independent cell growth in an EGFR and nuclear factor kappa B (NFκB)-dependent pathway resulting in upregulation of c-myc and anti-apoptotic Bcl-X_L (Brennan et al., 2007). In contrast, knockdown of Dsg1 in keratinocytes or of Dsg2 in enterocytes was associated with decreased rates of apoptosis in response to UV irradiation (Dusek et al., 2006; Nava et al., 2007). Interestingly, analogous effects were not observed with Dsc2 in enterocytes (Nava et al., 2007). This again suggests that there are cell-type- and cadherin-specific effects. However, overexpression of the apoptotic Dsg2 intracellular domain fragment induced apoptosis and furthermore increased expression of endogenous Dsg2. This increase is specific for the Dsg2 intracellular domain fragment, since expression of full-length Dsg2 did not result in upregulated expression of endogenous Dsg2 (Nava et al., 2007). Thus, the intracellular Dsg2 fragment appears to feed forward the apoptotic signal. Currently, the role of an upregulation of endogenous Dsg2 is not clear. In line with a feed-forward mechanism, more Dsg2 will generate further substrate for cleavage. Taken together, desmosomal cadherins contribute to regulation of apoptosis in keeping cells sensitive to apoptosis and in further enhancing the process. In contrast, loss of Dsg in tumor cells may represent a mechanism to evade apoptosis.

However, it has to be kept in mind that expression levels of specific cadherins may define whether cells finally divide or die. In addition, there is evidence that plakoglobin and PKPs are involved in regulation of apoptosis. Changes in the levels of desmosomal cadherins may increase or reduce binding sites for plakoglobin or PKPs at sites of cell–cell contact thereby switching their signaling functions in apoptosis regulation off or on, respectively. In this context, plakoglobin-deficient keratinocytes isolated from plakoglobin knockout mice are impaired in induction of apoptosis (Dusek et al., 2007). An opposite situation was observed during analysis of mice with epidermal knockout of plakoglobin (Li et al., 2012). This suggests that multiple factors such as the local environment or mechanic tension influence behavior of cells.

It is interesting to note that plakoglobin, PKP2, and DP expression itself is repressed by the EMT-inducing transcription factors Slug or Sip1/zinc finger E-box-binding homeobox-2 or ZEB2 (Bailey et al., 2012; Vandewalle et al., 2005). Re-expression of plakoglobin in human squamous carcinoma cells revealed dose-dependent differences with low expressing cells showing contact inhibition of growth, whereas in high expressors, apoptosis is inhibited correlating with an increased expression of the anti-apoptotic factor Bcl-2 (Hakimelahi et al., 2000). Plakoglobin apparently does not control Bcl-2 expression directly, but indirectly in modulating -catenin activity (Li et al., 2007a) and this in consequence may regulate Bcl-2 expression via c-myc and E2F1 (Li et al., 2007b). Both plakoglobin and β-catenin can bind to the c-myc promoter and act as repressors or activators, respectively (He et al., 1998; Müller-Tidow et al., 2004; Williamson et al., 2006). Furthermore, it was recently shown that plakoglobin but not β-catenin forms a complex with p53 and increases p53 transcriptional activity on the promoter of the tumor suppressor 14-3-3σ (Aktary et al., 2013), and in addition can repress the oncogenic chromatin remodeling factor special AT-rich sequence-binding protein-1 or SATB1 (Aktary & Pasdar, 2013). Loss of Rnd3/RhoE leads to an enhanced expression of desmosomal proteins and protects keratinocytes from cisplatin-induced apoptosis in a plakoglobin-dependent manner (Ryan et al., 2012). A recent study suggested that PKP2 also is involved in regulation of apoptosis (Kim et al., 2013). Furthermore, re-expression of DP in cells with an epigenetic inactivation of the DP gene exhibited increased sensitivity to apoptosis by upregulation of plakoglobin and concomitant inhibition of β-catenin–TCF/LEF-dependent transcription (Yang et al., 2012). For further reading on desmosomes and apoptosis also see Huber & Weiske (2009).

Sustainment of proliferative signaling

Continuous expression of growth factors and cytokines in tumors provides an environment promoting unrestricted proliferation predominantly by signaling through receptors with tyrosine kinase activity. In respect to desmosomes, the role of EGF and EGFR has been studied in most detail. Plakoglobin and β-catenin were identified as targets of EGFR. They are tyrosine phosphorylated within minutes after EGF treatment and both interact with EGFR (Gaudry et al., 2001; Hoschuetzky et al., 1994; Kanai et al., 1995). EGFR-dependent phosphorylation of tyrosine residue(s) in the C-terminus of plakoglobin shifts it to the Triton X-100-soluble fraction in association with Dsg and induces dissociation of DP resulting in decreased adhesive strength (Gaudry et al., 2001; Yin et al., 2005). In line with these findings, inhibition of EGFR activity induces an epithelial phenotype, promotes assembly of desmosomes, recruits IFs, and strengthens cell adhesion. There is evidence that in addition to plakoglobin, Dsg2 can be phosphorylated (Lorch et al., 2004). Another important consequence of EGFR activation is the upregulation of metalloproteinases involved in the shedding of desmosomal cadherins (Ellerbroek et al., 2001; Santiago-Josefat et al., 2007), thus contributing to enhanced migratory and invasive capacity of cells (see “Inactivation by proteolytic cleavage” section). In this context, it was shown that EGFR signaling regulates Dsg2 shedding and endocytotic trafficking in cooperation with ADAMs (Klessner et al., 2009). Moreover, the intramembrane serine protease iRhom2 is involved in maturation of ADAM17 and as a consequence in the activation of the EGF–EGFR axis and desmosomal cadherin processing (Brooke et al., 2014).

Activation of EGFR induces multiple downstream signaling events. EGFR-dependent activation of Erk5 was shown to induce expression of the EMT promoting transcription factor Slug resulting in altered desmosomal organization and mobility of cells (Arnoux et al., 2008). The Ras–Raf–mitogen-activated protein kinase (MAPK) pathway was found to be necessary and sufficient for EGF-induced desmosome breakdown (Edme et al., 2002). Interestingly, Dsg1 with an N-terminal deletion which switches off its adhesive function still was able to suppress EGFR–extracellular-signal-regulated kinase 1 (ERK1)/ERK2 signaling in recruiting the EGFR into more insoluble membrane fractions, indicating that Dsg1 has adhesion-independent signaling functions (Getsios et al., 2009). The ability of Dsg1 to capture the ERK regulator Erbin to its cytosolic domain thereby inhibiting ERK activation was identified as a potential mechanism (Harmon et al., 2013). Similarly, re-expression of Dsc3 in lung cancer cells reduced ERK1/2 phosphorylation (Cui et al., 2012a). Moreover, Dsg3 appears to be involved in the regulation of p38MAPK in a cross talk with EGFR (Bektas et al., 2013; Berkowitz et al., 2005; Berkowitz et al., 2006).

In addition to EGF, other growth factors and cytokines including tumor necrosis factor alpha (TNFα), hepatocyte growth factor (HGF), fibroblast growth factor (FGF), insulin-like growth factor-1 (IGF-1), interferon gamma (IFNγ), interleukin 1β (IL-1β), IL-6, IL-13, or IL-17 (Asimaki et al., 2011; Kampf et al., 1999; Li et al., 2001; Savagner et al., 1997; Wolf et al., 2013; Zuckerman et al., 2008) have been reported to impair desmosomal structure and function. However, they were not studied in detail as reported for EGF.

In addition to GFR-associated tyrosine kinases, cytosolic tyrosine kinases are frequently upregulated in cancer. Keratinocytes from plakoglobin –/– mice exhibit increased Src activity correlating with enhanced cell motility. Both the Src inhibitor PP2 and a dominant-negative Src kinase attenuated cell migration in plakoglobin-negative cells (Yin et al., 2004). Src signaling in these plakoglobin-negative cells appears to be regulated through Rho and dependent on extracellular matrix components such as fibronectin and vitronectin (Franzen et al., 2012; Todorovic et al., 2010). Furthermore, Src is involved in a cross talk between Dsg3 and E-cadherin (Tsang et al., 2012a; Tsang et al., 2012b; Tsang et al., 2010). Plakoglobin was identified as a target of Src, Fyn, and Fer kinases. Interestingly, although plakoglobin and β-catenin are highly similar in structure, phosphorylation by these kinases or EGFR has opposing effects on the interaction of plakoglobin with common binding partners such as α-catenin, classical cadherins, and TCF/LEF as well as on association with DP. Even effects induced by the cytosolic tyrosine kinases Src and Fer, respectively, are different (Miravet et al., 2003).

Evasion of growth suppressors

The retinoblastoma proteins Rb and p53 represent two central regulators of cell cycle and apoptosis (Sherr & McCormick, 2002). A link between p53 and its close relative p63 to desmosomal structure and function is definitely given. Both not only directly control expression of desmosomal genes including Dsc3, Dsg1, and DP (Cui et al., 2012a; Cui et al., 2011; Ferone et al., 2013; Oshiro et al., 2003), but also p53 apoptosis effector related to peripheral myelin protein-22 (PERP). PERP encodes a desmosome-located 21-kDa protein with four transmembrane domains that acts as an essential factor for epithelial integrity (Ihrie & Attardi, 2005; Ihrie et al., 2005) and in apoptosis regulation (Ihrie & Attardi, 2004). Knockout of PERP in mice results in a loss of desmosomes, impaired wound healing, and enhanced tumorigenesis (Beaudry et al., 2010a; Beaudry et al., 2010b; Dusek et al., 2012), emphasizing the role of desmosomes in cancer progression. Nevertheless, it always has to be kept in mind that the role of desmosomal proteins in cancer is context-dependent and can result in cell-cycle arrest or induction of proliferation (Chen et al., 2013).

Induction of angiogenesis

Efficient formation of new vasculature is a prerequisite for tumor growth. Since endothelial cells do not possess desmosomes, only the desmosomal proteins plakoglobin, PKP4/p0071, and DP associated with the vascular endothelial (VE)-cadherin may contribute to angiogenesis in a desmosome-independent manner. Loss of DP results in impaired capillary formation and leakiness (Zhou et al., 2004). Vascular endothelial growth factor-A (VEGF-A) treatment of endothelial cells upregulates PKP suggesting an involvement in angiogenesis (Rennel et al., 2007). Studies with HUVEC cells indicated that plakoglobin is important for the mechanical stabilization of junctions under sheer stress and for regulation of endothelial cell migration, proliferation, and angiogenesis (Nagashima et al., 1997; Nottebaum et al., 2008; Schnittler et al., 1997; Venkiteswaran et al., 2002). Recently, it was shown that extracellular MMP inducer (EMMPRIN/basigin/CD147), an Ig superfamily member with strong relation to tumorigenesis contributes to angiogenesis in a complex with plakoglobin and Nm23, thereby regulating actomyosin contractility at endothelial junctions (Moreno et al., 2014).

Genome instability and replicative immortality

Little is known if desmosomes or desmosomal proteins are involved in these processes. Overexpression of plakoglobin in HCT116 cells was reported to promote genomic instability in inducing aneuploidy (Pan et al., 2007). On the other side, it was reported that plakoglobin is subjected to loss of heterozygosity in certain cancers (Aberle et al., 1995). Loss of heterozygosity was also detected for DP; however, the patients show specific heart phenotypes (Murray et al., 2013) with no specific evidence of enhanced tumorigenesis. It was recently shown that human telomerase is a direct target of β-catenin in association with TCF or Kruppel-like factor 4 (KLF4) (Hoffmeyer et al., 2012; Zhang et al., 2012); however, it is not known if plakoglobin or PKP in competing with β-catenin may modulate telomerase expression.

Tumor promoting inflammation, avoidance of destruction by the immune system, and deregulation of cellular energetics

During recent years, compelling evidence accumulated that reactions of the immune system and tumor progression are linked (Grivennikov et al., 2010). Cancer cells have to evade anti-tumor activities of immune cells to avoid destruction. On the other side, immune cells, especially partially differentiated myeloid progenitors, appear to be the source of numerous growth factors and cytokines which augment tumor growth. Desmosomal protein expression is affected by these factors (see “Sustainment of proliferative signaling” section). In addition, it was observed that tumor cells frequently exhibit a switch in energy metabolism. Most well-known in this context is the Warburg effect describing the reprogramming of the glucose metabolism to aerobic glycolysis resulting in generation of increasing amounts of lactate (DeBerardinis et al., 2008). These aspects were defined as the so-called “emerging” hallmarks of cancer by Hanahan and Weinberg (2011). The only minor hint to such a relation may be given by the observation that enhanced lipogenesis occurs in induced pluripotent stem cell or iPSC-derived cardiomyocytes which express a mutant form of PKP2 (Kim et al., 2013), a metabolic phenomenon also observed in tumor cells.

CONCLUSION

Despite many discrepant findings, current data provide substantial evidence that desmosomes or their protein components are involved in different processes during development and progression of cancer (Figure 2). However, it always has to be kept in mind that desmosomes can differ in respect to their composition, expression levels of specific constituents, and their localization in a tissue- and context-dependent manner. Thus, changes of these conditions can have different outcomes. They either modulate adhesiveness or intracellular signaling pathways. According to our current understanding, plakoglobin and β-catenin signaling activities and their mutual regulation appear to be of central importance in this context. Currently, there is only limited evidence for a specific role of desmosomes or desmosomal proteins in the context of some of the hallmarks (see “Induction of angiogenesis,” “Genome instability and replicative immortality,” and “Tumor promoting inflammation, avoidance of destruction by the immune system, and deregulation of cellular energetics” sections) defined by Hanahan and Weinberg. The characterization of new players, the role of posttranslational modifications, and the identification of new cross talks will help to better understand the role of desmosomes in tissue homeostasis and malignant disease in the future.

Figure 2. Desmosomes and the hallmarks of cancer. The indicated connections may depend on the context such as cell type or tissue and expression of desmosomal cadherin family member(s). In many cases, the molecular mechanisms are not fully unraveled and it is not clear if the observed effects are direct or indirect. We also included links that are mediated by desmosomal proteins and do not depend on the presence of desmosome as is the case in respect to angiogenesis since endothelial cells in blood vessels under normal conditions do not express desmosomal cadherins and thus cannot assemble desmosomes.

ACKNOWLEDGEMENT

The authors apologize to all colleagues whose work we did not discuss or cite in this manuscript. We thank Dr. Laura Bloch for critical reading of the manuscript.

Declaration of interest: The authors declare no conflict of interest. The authors alone are responsible for the content and writing of the paper.