Plakophilins in Desmosomal Adhesion and Signaling

Abstract

The regulation of adhesion and growth is important for epithelial function and dysfunction. β-catenin (armadillo in Drosophila) is the prototype of a multifunctional molecule that regulates cell adhesion via adherens junctions and cell signaling via LEF/TCF transcription factors. Desmosomal armadillo proteins comprise plakoglobin and the plakophilins 1, 2, and 3. These proteins are essential for building up the desmosome and linking the desmosomal cadherins to keratin filaments. High expression of plakophilins in desmosomes correlates with strong intercellular cohesion and is essential for tissue integrity under mechanical stress. However, like β-catenin, these proteins have diverse non-desmosomal functions, for example, in regulating actin organization, protein synthesis, and growth control. In line with these functions, their de-regulated expression with up- as well as down-regulation has been connected to cancer and metastasis. Now, recent evidence sheds light on the post-translational regulation and provides an explanation for how de-regulation of plakophilins can contribute to cancer.

Keywords::

• desmosome
• armadillo protein
• plakophilin
• scaffold
• Rho-signaling
• translation
• phosphorylation
• post-translational modification

INTRODUCTION

Desmosomes are dynamic intercellular junctions that are essential for cell cohesion and tissue integrity. They are composed of transmembrane proteins, the desmosomal cadherins (desmogleins and desmocollins), and intracellular plaque proteins that link the transmembrane proteins to the intermediate filament network. The desmosomal cadherins interact with molecules of the armadillo family such as plakoglobin (PG) and plakophilins (PKPs) 1–3. This membrane complex provides a platform for association with desmoplakin (DP) (Thomason et al., 2010; Dubash & Green, 2011). DP couples the keratin network to sites of desmosomal adhesion. In recent years, it has been recognized that the desmosomal armadillo proteins are multifunctional: When located at the desmosomes they promote cell adhesion, but in the cytoplasm or nucleus they participate in cell signaling.

When desmosomal adhesion is compromised, cells lose cohesiveness leading to severe conditions affecting primarily tissues that are subjected to mechanical stress, such as the heart and the skin (Bazzi & Christiano, 2007; Lai-Cheong et al., 2007; Thomason et al., 2010; Brooke et al., 2012; Al-Jassar et al., 2013). Symptoms typically include epidermal fragility and blistering, thickened skin on the palm or soles (palmoplantar keratoderma), and/or cardiomyopathy. Of the armadillo-related desmosomal plaque proteins, PG and PKP2 are essential and their ablation in the mouse leads to embryonic death (Ruiz et al., 1996; Gallicano et al., 1998; Ruiz & Birchmeier, 1998; Bierkamp et al., 1999). In humans, mutations have been detected in PG, PKP1, and PKP2 leading to heart or skin diseases or a combination of both (Bazzi and Christiano, 2007; McGrath & Mellerio, 2010; Brooke et al., 2012; Al-Jassar et al., 2013).

In this review, we will summarize the structural and signaling functions of the PKP family of armadillo proteins in epithelia and discuss how these proteins are regulated. Finally, we will evaluate the potential role of these proteins in cancer. For a discussion of PKPs in genetic diseases of the skin or the heart, the reader is referred to other reviews in this issue.

PKPS ARE EVOLUTIONARILY CONSERVED PROTEINS

PKPs are members of a subfamily of armadillo proteins named according to the founder member the p120-catenin (p120ctn) family of armadillo proteins. This family comprises seven members, four of which are found primarily in adherens junctions (p120ctn, NPRAP/ δ-catenin, ARVCF (Armadillo-Repeat gene deleted in Velo-Cardio-Facial syndrome) and p0071/PKP4), whereas PKPs1–3 localize primarily at desmosomes. Several family members with homology to p120ctn are found in vertebrates, although invertebrates such as C. elegans and Drosophila express only one representative of this family. Ablation of the sole p120ctn in these species is not lethal, whereas a knockout of p120ctn in mice leads to early embryonic death, indicating that these proteins have acquired essential functions in higher organisms. In contrast to the p120ctn family, PKPs are vertebrate-specific and PKPs1–3 are found in zebrafish, Xenopus, chicken, and mammals (Tewari et al., 2010). Whereas their coding sequence is highly conserved between the species, the untranslated region (UTR) reveals considerable divergence in length and sequence, suggesting some heterogeneity in their post-transcriptional regulation (Figure 1).

Figure 1. Sequence conservation of PKPs.

Phylogenetic analysis of the PKPs is based on cDNA-sequences of the respective coding sequences (CDS) or 3´-UTR. The phylogenetic tree was created using the “One click” mode of the Phylogeny.fr platform (Castresana, 2000; Guindon and Gascuel, 2003; Edgar, 2004; Anisimova and Gascuel, 2006; Chevenet et al., 2006; Dereeper et al., 2008; Dereeper et al., 2010). The length of the horizontal connections correlates with the degree of nucleotide substitution.

THE STRUCTURE OF PKPS DETERMINES THEIR INTERACTIONS

How can PKPs combine adhesive and signaling activities and interact with numerous binding partners? PKPs consist of a central domain of nine armadillo (arm) repeat motifs that are flanked by rather long N- and very short C-terminal domains, respectively. Each arm repeat comprises approximately 42 amino acids that form three short α-helices arranged in a triangular shape (Choi & Weis, 2005). β-catenin and PG contain 12 arm repeats flanked by rather short N- and C-terminal domains. In β-catenin, these arm repeats form a superhelix with a long, positively charged groove which serves as an interaction platform for many binding partners and is therefore considered as a scaffold for protein complex formation (Huber et al., 1997). Since most protein interactions involve the same region of β-catenin/armadillo the spatial segregation of these binding partners in the cell appears to be an important regulatory mechanism. The PKP arm domain is interrupted between repeats five and six by a sequence that introduces a kink into the overall structure (Choi & Weis, 2005).

Whereas β-catenin binds to many distinct proteins by its arm domain, PKPs interact with binding partners by their N-terminal domains and interactions that depend on their arm repeat region have so far not been described. The N-terminal domains of PKPs are much larger than in β-catenin and are devoid of any known secondary structures with the exception of a short conserved α-helical stretch close to the N-terminus. Therefore, they are considered as largely unfolded. The function of natively disordered proteins has been related to their flexibility and their ability to adopt suitable conformations upon interacting with their binding partners. Binding to distinct partners can lead to distinct structures that correlate with multiple functions (Zhuang et al., 2011). Thus, the N-terminal domains of PKPs may adopt distinct structures when associated with various desmosomal, cytoplasmic or nuclear binding partners. Such protein interactions may be required to induce folding and protect PKPs from degradation. Post-translational modifications could serve a similar purpose by inducing conformations suitable for the interaction with a specific binding partner. In this context, the high number of phosphorylation sites in the N-terminal domains of PKPs is striking and suggests that these modifications play a role in determining the structure–function relationships. A third option to induce a secondary structure could be an association of the N-terminal domain with the arm repeat region either in cis or in trans within a dimer.

FUNCTION OF PKPS IN REGULATING ADHESION

PKPs were first described as desmosomal proteins that interact with many other desmosomal proteins including the desmosomal cadherins, PG, and DP. They exhibit differential but overlapping tissue distribution patterns (Schmidt & Jager, 2005). PKP1, originally described as band 6 protein (Hatzfeld et al., 1994; Heid et al., 1994) is found in desmosomes of stratified epithelia and localizes primarily to desmosomes in suprabasal layers. PKP2 is expressed in all simple, complex, and stratified epithelia but is also found in non-epithelial junctions from cardiac myocytes and germinal centers of lymph node follicles. PKP3 is more widely expressed and was detected in most simple and stratified epithelia except hepatocytes. In the epidermis, no differentiation-associated changes in its distribution have been noted. These overlapping but distinct expression patterns suggest tissue-specific functions but at the same time allow for some redundancy, for example, in facilitating cell cohesion (Bass-Zubek et al., 2009). In agreement, loss of function mutations in the PKP1 gene in humans (McGrath & Mellerio, 2010) or targeted deletion of PKP3 in mice (Sklyarova et al., 2008) lead to skin disease with reduced epithelial stability but are not lethal. In contrast, loss of PKP2 which is the only PKP of cardiomyocytes, induces embryonic lethality in a knockout mouse model due to heart failure (Grossmann et al., 2004).

Based on their interaction with many desmosomal proteins PKPs appear to act as scaffolds in desmosome formation to provide a link among the transmembrane cadherins, PG, DP, and keratins. Additionally, they can laterally cluster with DP to increase desmosome size (Kowalczyk et al., 1999; Hatzfeld et al., 2000; Chen et al., 2002; Bonne et al., 2003). Although a direct association with keratins was demonstrated in vitro (Kapprell et al., 1988; Smith & Fuchs, 1998; Hatzfeld et al., 2000; Bonne et al., 2003), the in vivo relevance of this interaction has been disputed since PKPs were detected in the outer desmosomal plaque, whereas keratins are anchored at the inner desmosomal plaque (North et al., 1999). The distance between the inner and the outer plaque is bridged by DP, which anchors the keratin filaments via its C-terminus and at the same time binds to PKPs via its N-terminal domain. Therefore, PKPs are considered to anchor keratin filaments at the desmosome indirectly via DP.

Although all PKPs localize to desmosomes only PKP1 considerably enhances the number and/or size of cell–cell contacts when overexpressed in cultured cells (Kowalczyk et al., 1999; Hatzfeld et al., 2000; South et al., 2003; South, 2004). This correlates with the finding that PKP1 is highly expressed in the suprabasal cells of the epidermis which contain numerous large desmosomes. In agreement, loss-of-function mutations in PKP1 are accompanied by small and fewer desmosomes in the epidermis (McGrath et al., 1997). Furthermore, PKP1 overexpression transformed desmosome adhesion from a calcium-dependent to a calcium-independent, hyperadhesive state and protected keratinocytes from Pemphigus vulgaris autoantibody-induced loss of cell–cell adhesion (Tucker et al., 2013). These data suggest that PKP1 is specifically suited to impart strong intercellular cohesion and protect the epidermis from mechanical and other stresses.

Although PKPs are considered as core desmosomal proteins, they can under certain circumstances associate with proteins from adherens junctions. This is observed for PKP2 in cardiomyocytes which form mixed junctions termed “area composita” (Pieperhoff et al., 2008). However, PKP2 has also been shown to co-localize with and co-immunoprecipitate β-catenin when ectopically expressed in cultured keratinocytes (Chen et al., 2002). These findings may indicate a role for PKP2 in sorting adherens junctions and desmosome components (Bass-Zubek et al., 2009). Moreover, it was suggested that E-cadherin together with PG is required to recruit PKP3 to cell borders to initiate desmosome formation (Gosavi et al., 2011). An association of the adherens junction proteins p0071 and p120ctn with desmosomal proteins has also been reported (Hatzfeld et al., 2003; Kanno et al., 2008a, 2008b). Taken together, these findings suggest that the armadillo proteins, although restricted to either adherens junctions or desmosomes in the mature cell–cell contacts, can transiently associate with the “wrong” type of junctional proteins and may participate in the segregation of junction complexes or in limiting the size of the individual junctions.

SIGNALING FUNCTIONS

Rho-signaling

Although the function of p120ctn and its direct homologs in regulating the actin cytoskeleton via small Rho- GTPases is well established (Wolf et al., 2006; Anastasiadis, 2007; Keil et al., 2007; Dohn et al., 2009; Schackmann et al., 2013), much less is known about the role of PKPs in organizing actin filaments and regulating Rho-GTPases. Keratinocyte cell–cell adhesion and differentiation depends on calcium-mediated and calcium-sensing receptor-mediated RhoA-signaling (Tu & You, 2013). Calcium stimulates formation of filopodia that embed into neighboring cells in a Rho/ROCK-dependent way. E-cadherin localizes at filopodia tips generating a zipper of E-cadherin positive puncta. The contacts achieved by interdigitating filopodia induced adherens junction initiation and maturation and subsequently desmosome assembly at flanking sites (Vasioukhin et al., 2000). This suggests that RhoA is involved in regulating not only adherens junctions but also in desmosome assembly. PKP2 was shown to regulate cortical actin rearrangement during junction formation (Godsel et al., 2010). These authors showed that in PKP2 knockdown cells activated RhoA failed to localize at intercellular interfaces after the cell–cell contact induction, implicating a role of PKP2 in the spatio-temporal control of RhoA during junction formation (Figure 2). At the same time, actin stress fibers were increased resulting from elevated levels of cellular RhoA. Consistent with this observation, focal adhesions were larger and more stable in PKP2-deficient cells, and vinculin dynamics were reduced (Koetsier et al., 2013). Based on these data, the authors speculated that PKP2 might play a role in integrating the cell–cell and cell–substrate contact signaling in basal keratinocytes.

Figure 2. Regulation and functions of PKPs1–3.

How PKPs1 and 3 are involved in Rho-signaling has not been established. PKP1 associates with actin filaments when ectopically expressed in cells with little or no desmosomes such as HeLa or NIH3T3 cells (Hatzfeld et al., 2000). In keratinocytes, an association with actin was observed in filopodia not connected to neighboring cells. Moreover, ectopic expression of the arm repeat domain induced the formation of filopodia and long cellular protrusions strongly suggesting a de-regulation of Rho-GTPases. Waschke et al. provided evidence that pemphigus vulgaris autoantibodies interfere with RhoA activation, thereby inducing loss of desmosomal adhesion (Waschke et al., 2006). Specific activation of RhoA abolished all pemphigus-triggered effects, including cell dissociation. Since the overexpression of PKP1 also abolished cell dissociation the control of Rho-signaling by PKP1 might participate in this phenomenon. However, how exactly PKP1 influences actin organization and how this correlates with a role in regulating Rho-GTPases remains to be determined.

PKP functions in RNA metabolism, translation and proliferation

PKPs1 and 3, but not PKP2, have been shown to associate with RNA-binding proteins (RNA-BPs) and modulate RNA metabolism. Upon cell stress, translation of housekeeping genes is arrested and translation becomes restricted to targets that are directly required for cell survival. However, mRNAs are not degraded under these conditions but instead are stored in stalled translation initiation complexes called stress granules together with many RNA-BPs. Application of oxidative stress to cultured cells induced the accumulation of PKPs1 and 3 in such stress granules, whereas PKP2 was excluded from these structures suggesting distinct functions of the PKPs in this context (Hofmann et al., 2006; Wolf and Hatzfeld, 2010; Wolf et al., 2010). In an attempt to characterize the role of PKP3 in RNA metabolism, Hofmann et al. (2006) analyzed proteins that co-purified together with PKP3 in immunoprecipitation experiments. This approach led to the identification of several RNA-BPs, Polyadenylate-binding protein 1 (PABPC1), Fragile X mental retardation-related protein 1 (FXR1), and GAP-SH3-binding protein (G3BP) as PKP3-interacting proteins. The functional relevance of these interactions has so far not been analyzed. PABPC1 is a component of the translation initiation complex. However, whether the PKP3–PABPC1 interaction facilitates translation remains to be determined. FXR1 is a member of the mammalian Fragile X protein family that comprises three members, Fragile X Mental Retardation Protein (FRMP), FXR1 and FXR2, respectively. These RNA-BPs are involved in controlling mRNA translation (Zarnescu & Gregorio, 2013). Loss of FMRP is the cause for the most common form of inherited intellectual disability, Fragile X syndrome (reviewed in [Kim & Ceman, 2012]). FMRP family proteins appear to regulate distinct sets of target mRNAs (Darnell et al., 2011). Although FXR1 is typically considered as a negative regulator of translation, it can also act as a translational activator by binding AU-rich elements (AREs) of the tumor necrosis factor alpha (TNFα) mRNA (Vasudevan & Steitz, 2007). AREs are one of the most common determinants of RNA stability and are found in many mediators of the inflammatory response. They usually target the mRNA for degradation. Binding of RNA-BPs can stabilize these mRNAs (Vasudevan & Steitz, 2007). Based on the finding that the targeted deletion of the PKP3 gene was associated with an increase in the inflammatory response (Sklyarova et al., 2008) it is tempting to speculate that PKP3 might contribute to limiting the inflammatory response via its interaction with FXR1. Alternatively, the PKP3–FXR1 interaction could play a role in controlling desmosomal adhesion. In cardiomyocytes, the translation of DP was shown to be controlled by FXR1, which binds the DP–mRNA and inhibits its translation (Whitman et al., 2011). Thus, by associating with FXR1–PKP3 could also control desmosomal adhesion. Interestingly, DP levels were elevated in the PKP3 knockout mice (Sklyarova et al., 2008). In order to understand, the relevance of the PKP3-RNA-BP associations it will be necessary to determine under which circumstances PKP3 interacts with these proteins and how it modulates their activities.

PKPs1 and 3 have been shown to stimulate translation in vitro as well as in cells (Wolf et al., 2010). Moreover, both proteins associated with the mRNA-cap- binding complex that consists of the eukaryotic translation initiation factors (eIFs) 4A, 4B, 4G, 4E, and PABPC. PKP1 directly interacted with and stimulated the enzymatic activity of eIF4A, a helicase that has been implicated in the unwinding of secondary structures in the 5’-UTRs of mRNAs to facilitate scanning for the start codon and translation initiation. This leads to an overall up-regulation of translation in PKP1 overexpressing cells, whereas the knockdown of PKP1 reduced protein biosynthesis. Since proliferation depends on protein synthesis the function of PKPs1 and 3 in controlling translation correlated with a function in regulating proliferation and cell size.

PKP functions in the nucleus

All PKPs have been observed in the nucleus. Nuclear localization of PKP1 is most prominent and endogenous as well as overexpressed PKP1 have been detected in the nucleus. Two splice variants that differ by an insertion in the arm repeat domain of PKP1 have been described with distinct subcellular localization patterns: PKP1a appears able to associate either with desmosomes or localize to the nucleus, whereas the minor variant PKP1b has been reported to be exclusively nuclear, suggesting that splicing is involved in regulating PKP1 localization and function (Schmidt et al., 1997).

DNAse I digestion of chromatin resulted in loss of PKP1 from the nucleus of cultured cells suggesting an interaction with DNA. In agreement, the N-terminal domain of PKPs1 and 2 was reported to bind to single-stranded DNA in vitro. This was correlated with a putative function in the DNA damage response. Induction of DNA damage induced a partial redistribution of PKP1 to the nucleolus, and depletion of PKP1 resulted in increased survival in response to DNA damage (Hatzfeld, 2010; Sobolik-Delmaire et al., 2010). However, protein-binding partners of PKP1 that would link PKP1 to the DNA damage response and provide some mechanistic insight have not been reported.

PKP2 was found to co-immunoprecipitate with a subunit of RNA polymerase (pol) III (Mertens et al., 2001). Pol-III plays a pivotal role in the regulation of protein synthesis and growth control as it catalyzes the synthesis of a variety of short, untranslated RNA molecules, many of which have essential functions in cellular metabolism including tRNA and 5S rRNA (Hatzfeld, 2007; Bass-Zubek et al., 2009). Pol-III transcription is regulated by nutrient availability, cell stress, and cell cycle stage and is elevated in tumor cells. However, the role of PKP2 in the pol-III complex remains elusive. Another nuclear function of PKP2 might be in transcriptional control. PKP2 can indirectly regulate transcription by stimulating β-catenin/TCF-mediated transcriptional activity implicating a role in Wnt signaling (Chen et al., 2002). The functions of PKPs reported to date are summarized in Figure 2.

PKP FUNCTION IN CANCER: LOSS OR GAIN OF FUNCTION?

In contrast to the well-documented role of adherens junctions in tumorigenesis, the contribution of desmosomes to cancer is not well understood. The loss of E-cadherin expression is a hallmark in carcinogenesis and activation of the signaling function of β-catenin promotes proliferation by stimulating the expression of pro-tumorigenic factors such as myc and this is frequently observed in colon and other cancers (Birchmeier et al., 1995; Conacci-Sorrell et al., 2002; Gloushankova, 2008).

Since desmosomes have traditionally been viewed as static protein complexes that reinforce adhesion between epithelial cells, it was expected that a loss of desmosomes would facilitate migration and proliferation and thus tumor growth and/or metastasis. Many descriptive studies have been published addressing the expression of desmosomal proteins including PKPs in cancer and trying to correlate the expression pattern with tumor progression (Dusek & Attardi, 2011). However, the results of these studies are contradictory with up-regulation, down-regulation, or maintenance of desmosome components being reported. While the expression of PKPs1–3 was increased in certain cancers of various origin (Furukawa et al., 2005; Cheung et al., 2007; Demirag et al., 2012; Takahashi et al., 2012), the loss or reduction of PKPs1–3 was observed in the development and/or progression of the same or other cancers (Sobolik-Delmaire et al., 2007; Narayana et al., 2010; Demirag et al., 2011; Takahashi et al., 2012). Both conditions were correlated with advanced tumor grade, increased metastasis and/or poor prognosis. Figure 3 summarizes PKP expression in cancer of various origins. PKPs1 and 2 appear up- or down-regulated with similar frequencies (PKP1 up-regulation described in 41, down-regulation in 32 studies; PKP2: up-regulation described in 30, down-regulation in 23 studies; Table 1), whereas an up-regulation of PKP3 was more frequently described than its down-regulation (53 vs. 18 reports). Finally, a number of analyses found no changes in PKP-expression during cancer progression (Kurzen et al., 2003; Demirag et al., 2011; Demirag et al., 2012). In a few studies, immunohistological or immunofluorescence data suggested that changes in intracellular localization rather than expression level might be important (Kurzen et al., 2003; Sobolik- Delmaire et al., 2007; Gomez-Morales et al., 2013). Studies in cultured tumor-derived cell lines were likewise confusing. In some cases, the overexpression of a PKP suppressed tumor-promoting behavior such as invasion and anchorage-independent growth (South et al., 2003), whereas in other experiments the overexpression promoted proliferation and increased invasion (Kundu et al., 2008; Wolf and Hatzfeld, 2010; Wolf et al., 2010). In light of the multiple functions of PKPs in cell adhesion and signaling, their contribution to cancer might be context-dependent and determined by the status of distinct signaling pathways that affect PKP functions.

Figure 3. Changes in PKP expression in various tumors.

The NextBio Disease Atlas (Kupershmidt et al., 2010) was used to identify studies showing significantly altered expression of the PKPs in various cancers compared to the corresponding normal tissues. Red bars correspond to the number of studies showing up-regulation, green bars indicate the number of studies showing a down-regulation of PKP-mRNA levels.

Table 1. PKP expression in cancer.

Name of study	PKP1	PKP2	PKP3	References
Intestine
Expression data from colorectal cancer patients	↑	↓	↑	(Sheffer et al., 2009)
Colon cancer tumor and normal biopsies from a series of patients	↔	↔	↑	(Ancona et al., 2006)
Midgut carcinoid primary tumors and liver metastasis expression profiling	↓		↓	(Leja et al., 2009)
Epithelial and stromal tissues in colorectal cancer	↑	↔	↓	(Abba et al., 2012)
Expression profiling in early onset colorectal cancer	↔	↓	↓	(Hong et al., 2007)
Colon biopsies of benign adenomas and malignant adenocarcinomas	↑	↓	↔	GEO accession GSE37364
Colorectal cancer gene expression profiles in tumors and matched adjacent tissue	↔	↓	↔	(Hinoue et al., 2012)
Expression data from human colonic epithelial cells	↑	↓	↔	(Galamb et al., 2010)
Multiclass cancer diagnosis using tumor gene expression signatures	↔	↑	↔	(Ramaswamy et al., 2001)
Colorectal carcinoma regulated gene expression in Middle Eastern patients	↑	↓	↑	(Uddin et al., 2011)
Colon adenocarcinomas and matched normal tissue	↔	↓	↔	(Lin et al., 2006)
Modeling oncogenic signaling in colon tumors by multidirectional analyses of microarray data	↑	↓	↔	(Skrzypczak et al., 2010)
Expression data from human colonic biopsy sample	↔	↓	↔	(Galamb et al., 2008)
Expression data from normal colon tissues and colon tumors	↔	↓	↔	(Irizarry et al., 2009)
Colon cancer samples compared with adjacent normal colon	↑	↔	↔	(Callari et al., 2012)
Epithelial and stromal tissues in colorectal cancer	↑	↓	↔	(Abba et al., 2012)
Time dependent changes of gene expression in surgical specimen of colorectal cancer	↑	↔	↔	GEO accession GSE50746
Mutated or wildtype KRAS in rectal cancer patients	↓	↑	↔	(Gaedcke et al., 2010)
Stomach
Gastric Cancer Study	↔	↑	↓	(Chen et al., 2003)
Expression data from primary gastric tumors and adjacent normal samples	↓	↑	↑	(D’Errico et al., 2009)
Gastric carcinoma and GIST gene expression profiling	↑	↑	↑	(Cho et al., 2011)
Expression data from gastric cancer	↓	↓	↑	(Peng et al., 2006)
Gastric stromal tumor and non-malignant gastric tissue expression profiles	↔	↔	↑	(Wu et al., 2013)
Bladder
DNA methylation and expression profiling study for bladder cancer	↔	↔	↑	(Kim et al., 2013b)
Comparative Gene Expression Profiling Analysis of Urothelial Carcinoma of Renal Pelvis and Bladder	↑	↑	↑	GEO accession GSE24152
Urethelial carcinoma of bladder	↑	↔	↑	(Zhou et al., 2013)
Clinical bladder cancer classification	↑	↔	↑	(Mengual et al., 2009)
Classification of carcinoma in situ lesions in human bladder cancer	↑	↔	↑	(Dyrskjot et al., 2004)
Multiclass cancer diagnosis using tumor gene expression signature	↑	↑	↔	(Ramaswamy et al., 2001)
Skin
Malignant transformation and progression of metastatic melanoma	↔	↔	↑	(Riker et al., 2008)
Novel Molecular Markers for non-melanoma skin cancer	↑	↔	↑	(Nindl et al., 2006)
Head and Neck
Hypopharyngeal cancer transcriptome	↑	↔	↓	(Cromer et al., 2004)
Oral tongue cancer gene expression profiling: Identification of novel potential prognosticators	↑	↔	↑	(Estilo et al., 2009)
Tongue carcinomas with TNM staging and normal control comparisons	↓	↑	↓	(Rentoft et al., 2012)
Oral Cavity Cancer Compared to Adjacent “Normal” Tissue	↓	↑	↓	(Lohavanichbutr et al., 2013)
Tumor invasion in Oral Squamous Cell Carcinoma	↓	↔	↓	(Toruner et al., 2004)
Expression data from biopsies of Nasopharyngeal carcinoma (NPC) and non-malignant controls	↑	↔	↔	(Bose et al., 2009)
mRNA expression profiling of nasopharyngeal carcinoma	↔	↑	↔	(Dodd et al., 2006; Sengupta et al., 2006)
Gene expression profiling of oral squamous cell carcinoma	↑	↑	↓	(Chen et al., 2008)
Breast
Breast tissue comparative analysis of infiltrating ductal carcinoma	↑	↑	↑	(Hawthorn et al., 2010)
Molecular Signature of Pregnancy Associated Breast Cancer	↓	↑	↔	(Harvell et al., 2013)
Metastatic and primary breast tumors	↔	↓	↔	(Hu et al., 2009)
Human body index - transcriptional profiling	↓	↓	↔	GEO accession GSE7307
Transcription profiling of breast tumors and surrounding tissue	↓	↑	↑	(Cheng et al., 2008)
A gene expression signature identifies two prognostic subgroups of basal breast cancer	↓	↓	↑	(Sabatier et al., 2011; Sircoulomb et al., 2010)
Expression profiling of human DCIS and invasive ductal breast carcinoma	↓	↔	↑	GEO accession GSE21422
Proliferative genes dominate malignancy-risk gene signature in histologically-normal breast tissue	↓	↔	↑	(Chen et al., 2010)
Expression in normal and cancerous breast cell lines and tissues	↓	↔	↑	(Novak et al., 2006)
Gene expression signatures in breast cancer	↓	↑	↑	(Pedraza et al., 2010)
Identifying breast cancer biomarkers	↓	↔	↔	GEO accession GSE29431
Study of Multiple Solid Cancers	↓	↓	↑	(Yu et al., 2008)
Invasive lobular and ductal carcinomas of the breast	↓	↓	↔	(Zhao et al., 2004)
Expression data from human breast tissue	↓	↑	↔	(Richardson et al., 2006)
Early Dysregulation of Cell Adhesion and Extracellular Matrix Pathways in Breast Cancer Progression	↓	↔	↔	(Emery et al., 2009)
Lung
Lung tumors in human and mouse—human subset	↔	↔	↑	(Stearman et al., 2005)
Rewiring of human lung cell lineage and mitotic networks in lung adenocarcinomas	↔	↑	↑	(Kim et al., 2013a)
Glycine decarboxylase is a metabolic oncogene critical for tumor initiating cells and tumorigenesis in lung cancer	↑	↔	↑	(Zhang et al., 2012)
Lung gene expression of non-smoking female lung cancer	↓	↔	↑	(Lu et al., 2010)
Expression data from Lung cancer	↔	↔	↑	(Su et al., 2007)
Non-Small Cell Lung Cancer	↑	↑	↑	(Dehan et al., 2007)
Expression data for early stage NSCLC	↑	↑	↑	(Hou et al., 2010)
Lung adenocarcinoma stage I, II, and II and healthy lung tissue	↑	↑	↑	GEO accession GSE40791
Gene expression profile of lung tumors	↔	↑	↑	(Wrage et al., 2009)
Study of Multiple Solid Cancers	↔	↔	↑	(Yu et al., 2008)
Lung tumors from 293 patients with clinical outcome	↑	↑	↑	(Rousseaux et al., 2013)
Gene expression analysis of human lung cancer	↑	↑	↑	(Sanchez-Palencia et al., 2011)
Primary lung adenocarcinoma expression profiles	↔	↔	↑	GEO accession GSE31908
Whole-Genome Gene Expression Profiling of Formalin-Fixed, Paraffin-Embedded Tissue Samples	↑	↑	↔	(April et al., 2009)
Classification of High-grade neuroendocrine tumors of the lung	↑	↔	↔	(Jones et al., 2004)
Gene expression data of primary lung cancer	↑	↔	↔	(Moriya et al., 2009)
Squamous Lung Cancer and adjacent normal tissue	↑	↔	↔	(Wachi et al., 2005)
Gene expression data for pathological stage I-II lung adenocarcinomas	↓	↑	↑	(Okayama et al., 2012)
Gene expression profiling of Non-small cell lung cancer	↓	↔	↑	(Wei et al., 2012)
Prostate
Gene expression profiling identifies clinically relevant subtypes of prostate cancer	↓	↓	↑	(Lapointe et al., 2004)
Gene expression down-regulation in prostate tumor-associated stromal cells	↔	↓	↑	(Pascal et al., 2009)
Prostate cancer and normal prostate tissue	↔	↔	↑	GEO accession GSE30522
Expression Data from Normal and Prostate Tumor Tissues	↓	↔	↑	(Chandran et al., 2007; Yu et al., 2004)
Gene expression profiling of prostate cancer	↔	↔	↑	(Arencibia et al., 2009)
Gene signatures in prostate cancer linked to the Gleason score	↓	↔	↑	(Kuner et al., 2013)
Comprehensive gene expression analysis of primary and metastatic prostate cancer	↑	↔	↔	(LaTulippe et al., 2002)
Esophagus
Study of Multiple Solid Cancers	↔	↔	↓	(Yu et al., 2008)
Esophageal epithelium in patients with Barrett’s esophagus or squamous cell carcinoma	↓	↑	↓	GEO accession GSE26886
Esophageal squamous cell carcinoma and normal esophageal tissue expression profiles	↓	↔	↓	(Yan et al., 2012)
Esophageal squamous cell carcinoma gene expression profiling	↔	↔	↓	GSE17351
Esophageal adenocarcinoma and Barrett’s esophagus global gene expression signatures	↓	↑	↓	(Kim et al., 2010)
Gene expression profiling in esophageal squamous cell carcinoma	↔	↔	↓	(Su et al., 2011)
Barrett’s esophagus—metaplasia progression to adenocarcinoma	↓	↑	↓	(Grotenhuis et al., 2010)
Barrett’s esophagus and adenocarcinoma compared to normal esophageal	↓	↑	↓	(Hao et al., 2006)
Barrett’s esophagus—metaplasia progression to adenocarcinoma	↓	↑	↓	(Kimchi et al., 2005)
Psoriasis/Eczema
Gene expression data of skin from psoriatic patients and normal controls	↑	↓	↑	(Nair et al., 2009; Swindell et al., 2011)
Skin lesion gene expression in efalizumab-responsive psoriasis patients	↑	↔	↑	(Johnson-Huang et al., 2012)
Affected and unaffected skin of psoriasis patients and psoriasis free individuals	↑	↔	↑	(Kulski et al., 2005)
Lesional and non-lesional skin from psoriasis patients	↑	↓	↑	(Yao et al., 2008)
Expression data from skin biopsy samples from patients with moderate-to-severe psoriasis	↑	↓	↑	(Suarez-Farinas et al., 2012)
Comparison of Gene Expression in Psoriatic Skin from Different Sources	↑	↓	↑	(Bigler et al., 2013)
Expression data from human skin biposies from psoriatic patients	↑	↔	↑	(Reischl et al., 2007)
Psoriasis-associated gene expression in lesional and non-lesional epidermis and dermis	↑	↔	↑	(Mitsui et al., 2012)
Skin biopsies from atopic eczema and healthy controls	↑	↔	↑	(Olsson et al., 2006)

Summary of studies identified by NextBio Disease Atlas (Kupershmidt et al., 2010) showing significantly altered expression of the PKPs in various tumors compared to the corresponding normal tissue. ↑ mRNA up-regulation; ↓ mRNA down-regulation; ↔ mRNA expression not affected or not determined.

An up-regulation of PKPs1 and 3 has also been reported in prevalent skin diseases associated with inflammation and hyperproliferation such as psoriasis, whereas PKP2 was not affected in this condition (Figure 3, Table 1).

REGULATION OF MULTIFUNCTIONAL PKPS BY PHOSPHORYLATION

The spatio-temporal control of many multifunctional proteins including PKPs during cell and tissue differentiation remains a challenging topic. As discussed before, PKPs can interact with various binding partners in specific cellular compartments where they have distinct and partially opposing functions. Thus the decision on where PKPs localize and with which proteins they interact has a major impact on cell fate and an imbalance in PKP regulation may lead to disease such as cancer. Post-translational modifications can modulate protein interactions and as a consequence control protein stability, transport, localization, and finally determine protein function. Phosphorylation represents the most frequent form of modification, and serine/threonine or tyrosine kinases are activated by signaling pathways to control the functions of their targets. PKPs contain numerous phosphorylation sites, which are highly clustered in their N-terminal domains. Many of these sites have been detected as being phosphorylated in large-scale mass spectrometry approaches. PKP1 contains 52 serine/threonine and 15 tyrosine residues that were found to be phosphorylated, in PKP2 43 serine/threonine- and 23 tyrosine-phosphorylated sites have been identified, and PKP3 contains 32 phosphorylated serine/threonine and 11 phosphorylated tyrosine residues (http://www.phosphosite.org/). Moreover, methylation has been detected in all three PKPs, an acetylation in PKP3 and 15 ubiquitinated lysine residues were reported in PKP2 (Kim et al., 2011; Wagner et al., 2012). This wealth of modifications needs further characterization to distinguish between “basal” modifications that could, for example, serve to stabilize the unfolded head domain and those that are induced by specific signals and thus depend on the context of the cell or tissue. Moreover, the functional consequences with respect to PKP associations, localization, and functions need to be determined.

A recent report shows how phosphorylation of PKP1 controls the distribution between its desmosomal and cytoplasmic pools and shifts the balance from an adhesion promoting function of un-phosphorylated PKP1 to a translation and proliferation promoting function of the phosphorylated form (Figure 4, [Wolf et al., 2013]). Insulin/IGF-1-signaling via the PI3K–AKT pathway induced the phosphorylation of the PKP1 head domain. This led to the stabilization of the cytoplasmic pool which in turn correlated with an increase in translation rates and as a consequence an increase in proliferation. At the same time intercellular cohesion was impaired. This regulation was highly specific and mediated by the AKT2 isoform of AKT kinases that interacted directly and specifically with PKP1, but not PKP2 or 3. Moreover, EGF was much less efficient in stimulating PKP1 phosphorylation and re-localization compared to insulin/IGF-1. Four phosphorylation motifs in the N-terminal domain were identified. Interestingly, the expression of a phospho-mimetic mutant of PKP1 not only stimulated translation and proliferation but also conferred the capacity of anchorage-independent growth. This suggests that the AKT2-mediated modification might play a role in tumorigenesis. In support of this, AKT2 is associated with tumors with a much higher frequency than AKT1 (Cohen, 2013). Moreover, several of the AKT2 phosphorylation sites of PKP1 were identified in a mouse model for squamous cell carcinoma (Zanivan et al., 2013). Precisely how phosphorylated PKP1 is protected from degradation remains to be determined. 14-3-3 proteins bind specifically to phosphorylated target proteins and modulate their localization, stability, and functions. Seven isoforms of 14-3-3 in humans and several hundreds of 14-3-3 binding partners have been identified to date. These proteins are involved in diverse processes like regulation of the cytoskeleton, GTPase function, membrane signaling, cell fate determination, response to insulin and TNFα, cell cycle progression, and apoptosis (Uhart & Bustos, 2013). Interestingly, two of the motifs phosphorylated by AKT2 are putative 14-3-3 binding sites raising the possibility that an interaction with 14-3-3 might be involved in regulating PKP1.

Figure 4. Regulation of PKP1 function by insulin/IGF-1-signaling.

PKP2 is a target of Cdc25C-associated kinase 1 (C-TAK1) also called microtubule affinity-regulating kinase 3 (MAP/MARK3). Its phosphorylation by C-TAK1 was shown to generate a 14-3-3 binding site (Muller et al., 2003). Mutagenesis experiments identified S82 as a phosphorylation site and PKP2 mutated at this site (S82A) exhibited increased nuclear accumulation, suggesting that C-TAK1 phosphorylation and subsequent interaction with 14-3-3 prevents PKP2 nuclear localization. Whether this affects junction formation and stability has not been analyzed. PKP2 appears to be able to interact with all 14-3-3 isoforms, since all isoforms were detected as interaction partners in large-scale interaction screens (e.g., http://www.ebi.ac.uk/intact/pages/interactions/interactions.xhtml? query = Q99959*). Whether the isoforms have differential effects on PKP2 localization and function and whether they all recognize the same motif or interact with different phosphorylation sites have not been analyzed so far. Moreover, it is not known whether C-TAK1 is the only kinase generating a 14-3-3 binding site in PKP2 or any other kinases could be involved. Taken together these data indicate that PKP2, like PKP1, is regulated by specific signaling pathways that affect the translocation of cytoplasmic PKP2 to the nucleus, where it may act to stimulate β-catenin/ TCF-mediated transcriptional activity or have additional so far-unknown functions.

PKP3 was recently shown to co-purify with 14-3- 3σ/stratifin. This interaction was specific for PKP3 and not detected for PKP1 or 2 in a yeast-2-hybrid approach. Moreover, the interaction was lost in a PKP3 mutant carrying a S285A exchange, suggesting that phosphorylation of S285 in the N-terminal domain is required for this interaction (Roberts et al., 2013). Stratifin interacted with cytoplasmic PKP3 and limited the exchange of cytoplasmic PKP3 with its desmosomal pool. Accordingly, decreased stratifin expression lead to increased exchange rates resulting in reduced desmosomal adhesion and increased cell migration. Phospho-S285 has so far only been detected in mouse, but not in human tissues (Zanivan et al., 2008; Huttlin et al., 2010), and a kinase or signaling pathway that would induce the phosphorylation of PKP3 at S285 has not been reported.

Tyrosine phosphorylation of PKPs, although frequently detected in large-scale mass spectrometry analyses, has not been systematically studied so far. However, a co-immunoprecipitation experiment and a tandem affinity purification approach identified the EGF- receptor and ERBB2 as putative PKP2 interaction partners (Thelemann et al., 2005). Whether these receptors phosphorylate PKP2 directly or via downstream tyrosine kinases remains elusive.

Other post-translational modifications such as methylation and acetylation are likely to play additional roles in the regulation of PKPs. The functional consequences of these modifications or their interplay with phosphorylation have not been characterized.

Several large-scale screens revealed that PKP2 is regulated by ubiquitination and a total of 15 ubiquitinated lysine residues have been identified (Kim et al., 2011; Wagner et al., 2012). Protein ubiquitination is a regulatory post-translational modification that controls numerous biological processes including proteasomal degradation of proteins. In a systematic proteomics approach, Bennett et al. identified cullin 3, a core component of E3 ubiquitin-protein ligase complexes which mediate poly-ubiquitination and subsequent proteasomal degradation of specific protein substrates, as a putative PKP2-interacting protein (Bennett et al., 2010). It will be interesting to unravel the pathways that regulate PKP2 ubiquitination and protein stability.

TRANSCRIPTIONAL CONTROL OF PKPS

Since an up-regulation or a down-regulation of PKP-mRNAs has been associated with disease such as cancer and psoriasis, it is important to understand not only how post-translational modifications influence PKP functions and stability but also how their expression is controlled at the transcriptional and post-transcriptional levels. For all three PKPs, splice variants have been described. The PKP1 variants 1a and 1b differ by exon 7 as the alternatively used exon in the 1b isoform. This exon codes for 21 amino acids inserted at position 412 within arm repeat 4. A differential intracellular localization was reported for the two splice variants (Schmidt et al., 1997). However, neither the functional consequences nor the expression patterns and regulation of these alternative splice variants have been characterized so far. Moreover, two PKP1-mRNAs that differ in their 3’-UTR are generated (Schmidt et al., 1997). PKP2b differs from the canonical isoform 2a by an extra exon encoding 44 amino acids inserted at position 460 within arm repeat 3. No information is available on differential expression patterns, distinct functions, or the regulation of the splice variants. For PKP3, a splice variant has been detected very recently. PKP3b differs from the canonical PKP3a by an alternative amino- terminal exon which results in an altered N-terminal amino acid sequence (Muhmer et al., 2013). Whereas PKP3a is broadly expressed in epithelial cells, PKP3b was predominantly found in desmosomes of stratified epithelia, but absent from simple epithelial cells. Tissue-specific expression was validated using a reporter assay showing that a fragment upstream of the alternative exon 1b promoted transcription only in HaCaT keratinocytes but not in CaCo2 simple epithelial cells. Therefore, the authors speculated that the two isoforms are controlled by distinct promoters. Finally, by using electromobility shift assays, they found a potential binding site in the PKP3b promoter for transcription factor C/EBP (CCAAT/enhancer binding protein) although the identity of nuclear factors that bind to the PKP3 promoter has not been validated (Muhmer et al., 2013). C/EBP proteins are a family of basic leucine zipper (bZIP) transcription factors that includes six members with related functions. C/EBPα and β are highly expressed in the skin in a differentiation-dependent manner. C/EBP proteins function as transcriptional activators but can also repress transcription of certain target genes.

Epithelial–mesenchymal transition (EMT) is a process whereby epithelial cells lose apicobasal polarity and cell–cell contacts and gain mesenchymal phenotypes with increased migratory and invasive capabilities. Several transcription factors have been identified as master regulators of EMT, including the Snail, ZEB, and Twist families of E-box transcription factors. In invasive, de-differentiated MDA-MB-231 breast cancer cells, PKP3 was transcriptionally repressed by the ZEB1 (zinc finger E-box binding homeobox 1) transcription factor which also represses E-cadherin (Aigner et al., 2007). ZEB1 associated with two conserved E-box elements in the PKP3 promoter and partially repressed the activity of PKP3 promoter fragments in reporter assays. Since ZEB1 is up-regulated in invasive cancer cells, it was speculated that PKP3 repression by ZEB1 contributes to the disintegration of intercellular adhesion and EMT.

In breast cancer, estrogen receptor alpha (ERα) expression is associated with differentiated tumors and a lower metastasis risk. Estrogens were shown to increase the formation of desmosomes in normal and malignant cells, and several desmosomal proteins including PKP3 were up-regulated in response to estrogens (Maynadier et al., 2012). Whether PKP3 is directly regulated by estrogens or indirectly via other desmosomal proteins required for its stabilization remains to be determined. Interestingly, two ERα binding sites are predicted in the PKP3 promoter region.

Serum response factor (SRF) is a transcription factor that regulates gene expression to control growth, differentiation, and cytoskeletal integrity in different cell types. Epidermal SRF-loss resulted in an abnormal epidermal architecture with hyperplasia but without visible skin lesions or erosions. SRF can be regulated through monomeric actin which inhibits its transcriptional co-activator MAL/MRTF. PKP2 was identified as one of the actin-regulated MAL targets and MAL/SRF were recruited to cis-regulatory elements of the PKP2 gene suggesting a direct regulation of PKP2 by MAL/SRF (Leitner et al., 2011).

The transcriptional regulation of PKP1 has not been characterized to date. PKP1 promoter methylation resulting in a loss of protein expression was observed in Barrett's esophagus and may facilitate the progression of Barrett's esophagus to esophageal adenocarcinoma in a subset of patients via decreased desmosome assembly and increased cell motility (Kaz et al., 2012).

FUTURE DIRECTIONS—IS IT ALL A QUESTION OF CONTEXT?

Like β-catenin, PKPs are multifunctional molecules and we are only just beginning to unravel their diverse roles and to grasp the complexities of their distinct functions. The situation is even more complex than in the case of β-catenin since we deal with three related proteins which are often co-expressed and have unique as well as partially shared functions. Moreover, there is very little information available on the differential expression patterns or functions of the identified PKP splice variants. While all three proteins play a role in mediating desmosomal adhesion how they act in concert and how the balance of their expression influences cell cohesion remains to be determined. With regard to translation, we find overlapping functions in stimulating protein biosynthesis for PKPs1 and 3, whereas PKP2 appears to be excluded from translation initiation complexes. Least is known about the nuclear functions of PKPs and it remains one of the future challenges to identify interaction partners and functions of these proteins in the nucleus. Another unresolved issue is the role of PKP3 in limiting the inflammatory response which was deduced from the phenotype observed in PKP3 knockout mice.

As outlined in this review, part of the solution will be found in dissecting the roles of (a) the proteins that interact with PKPs and (b) the proteins that regulate PKPs. By introducing a variety of post-translational modifications, a plethora of proteins with distinct characteristics can be generated that are optimally suited for specific interactions while preventing other interactions. Large-scale screens have revealed that PKPs are the targets of numerous modifications and these modifications can act either alone or in combination to regulate nearly all aspects of their functions. Thus, deciphering the post-translational modifications of PKPs and how they are coordinately regulated is of fundamental importance to understanding their functions and the interplay of PKPs with their binding partners. Identifying the signals that induce PKP-modifications and characterizing their impact on protein function will finally allow us to understand the role of PKPs and their de-regulation in disease.

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

The work in our group is funded by grants from the DFG (Ha 1791/7-1, Ha1791/8-1, GRK 1591) and the BMBF (ProNet T3) to MH.