Skin contains three major layers: the epidermis, dermis, and subcutaneous layers (Figure 1). Different cell types reside in each layer. The epidermis is the topmost layer of the skin, and is exposed to the outside environment. Keratinocytes are the major cell type in the epidermis. Melanocytes, Merkel cells, and dendritic dells (Langerhans cells) are also found in the epidermis. The dermis lies underneath the epidermis and contains hair follicles, sweat glands, nerves, and blood vessels. Fibroblasts in this layer produce collagen, and the dermis has abundant extracellular matrix proteins, such as collagen and elastin. The subcutaneous layer is the innermost skin layer, and contains adipocytes, which store lipids.
The major types of skin tumors are basal cell carcinomas (BCC), squamous cell carcinomas (SCC), and melanomas. Basal cell carcinomas and SCCs are occasionally grouped and termed non-melanoma skin cancers (NMSCs), and each type originates from keratinocytes. In contrast, melanomas originate from melanocytes. Both keratinocytes and melanocytes reside in the epidermis, where tumors occur. One of the most common cancers is BCC and it accounts for approximately 80% of newly diagnosed skin cancers in the United States (Bowden, 2004). They often grow slowly and rarely metastasize. SCCs are more malignant compared with BCCs, and certain types of SCCs are highly metastatic. Melanomas rapidly proliferate in comparison with the other skin cancer cells, and are highly invasive and terribly malignant. Melanomas are resistant to many anti-cancer treatments, such as chemotherapy, and are associated with poor patient prognosis. These three types of the tumor account for nearly 99% of skin tumors: BCCs 80%, SCCs 16%, and melanoma up to 4%, in the United States (Bowden, 2004).
Skin is an organ directly exposed to sunlight. BCCs and SCCs occur primarily on the skin exposed to sunlight. UV light is a primary cause of these skin tumors (Figure 1) and it is a mutagen which causes thymine dimers, contributing to skin tumors. Meanwhile, the effect of UV light on melanoma remains uncertain because these tumors do not exhibit UV-induced mutations, and also occur in parts of the body which are not often exposed to sunlight. Nevertheless, recent reports suggest that constant exposure to UV light causes melanoma in mouse and fish models (De Fabo et al., 2004; Setlow et al., 1989). In addition to environmental factors, inherited mutations or alterations in gene expression, which can alter cellular responses such as DNA repair mechanisms, can also contribute to oncogenesis. Ubiquitination is a post-translational protein modification. This process is involved in multiple cellular responses, such as cell growth, cell death, and DNA damage repair. Its biological significance in both the prevention and promotion of cancer has been described (Bashir and Pagano, 2003). This review will discuss the ubiquitin pathways in the three major skin cancer types and the mechanisms of ubiquitin pathway alteration in skin cancers.
Ubiquitin and ubiquitin-related pathways
Ubiquitination is a post-translational protein modification. Ubiquitin is a 76-amino acid protein which is covalently conjugated to a lysine residue in proteins. This can occur with a single ubiquitin conjugated to each lysine (mono-ubiquitination), or multiple ubiquitins can be conjugated to form a chain on the lysine residue (poly-ubiquitination). A number of proteins undergo ubiquitination that serves as a degradation signal for proteins to be recognized by the proteasome, a large protease complex. An ubiquitinated protein is degraded into peptides in the proteasome. It is now clear that ubiquitination has a broader function for regulating protein localization, sorting, and protein–protein interactions.
Ubiquitin is bonded to the protein by a series of enzymes: namely, the ubiquitin activating enzyme (E1), the ubiquitin conjugating enzyme (E2), and the ubiquitin ligase (E3) (Figure 2) (Nalepa et al., 2006). Ubiquitin is first linked to E1 and activated by using the energy of ATP. The activated ubiquitin is then transferred to E2 and in conjunction with E3, the ubiquitin is conjugated to the target proteins (Figure 2). Ubiquitin ligase E3 recognizes a substrate for ubiquitination, and is considered to be the key factor to determine ubiquitination specificity. E3s are grouped according to the domain structure such as the Really Interesting New Gene (RING)-finger type, the Homologous to E6-Associated Protein C-Terminus (HECT) type, or the Skp 1-Cul 1-F box protein (SCF) complex type. The SCF complex consists of a Cullin-based scaffold and adaptor proteins, such as the F-box protein and the RING-finger-type ubiquitin ligase. The F-box protein determines the substrate specificity of the ligase. Ubiquitins can also be removed from ubiquitinated proteins. This de-ubiquitination process is mediated by the de-ubiquitinating enzyme (DUB) family of proteins (Figure 2) (Reyes-Turcu et al., 2009).
There are subsets of proteins which are functionally or structurally similar to ubiquitin, and are termed ubiquitin-like proteins (UBLs). UBLs include SUMO, NEDD8, ISG15, and FAT10 (Welchman et al., 2005). These proteins are conjugated to target proteins by an analogous enzyme cascade to ubiquitin E1, E2, and E3. SUMOylation uses AOS1/Uba2 as an E1 protein and Ubc9 as an E2 protein, and NEDDylation uses APPBP1/Uba3(E1-like) and Ubc12 (E2-like). UBL protein modifications are usually not associated with protein degradation, thereby contributing instead to subcellular localization, transportation, or protein–protein interactions.
Ubiquitin and ubiquitin-related signaling pathways in basal cell carcinoma
Hedgehog signaling pathway
The hedgehog (HH) signaling pathway plays a crucial role in the development and growth of BCC (Figure 3A) (Epstein, 2008). BCCs originate from keratinocytes and are one of the most frequent types of skin cancer, whereas the metastatic rate of these tumors is low. Hedgehog binds to the Patched (PTCH1) receptor on the cell surface and releases a Smoothened (SMO)-mediated signal, which in turn activates intracellular signaling events (Epstein, 2008). Mutations in PTCH1, which activates HH signaling, were originally identified in BCC patients (Gailani et al., 1992; Hahn et al., 1996; Johnson et al., 1996). These mutations are thought to be the major cause of HH pathway activation in BCCs. A transgenic animal model also indicates that constitutive activation of the HH pathway, whether by ectopically expressing sonic HH, the transcription factor Gli1, or Gli2, is sufficient to cause BCC (Hutchin et al., 2005; Nilsson et al., 2000; Oro et al., 1997). Gli is a zinc-finger-type transcription factor and there are three forms in mammals, numbered Gli1 through Gli3 (Figure 3A) (Kinzler et al., 1988). HH activates the Gli family of transcription factors, and the activity of Gli is regulated by ubiquitination.
Regulation of Gli – processing and degradation
The activity of the Gli proteins is regulated by processing and degradation, both of which are mediated by ubiquitination. Whereas the degradation of Gli proteins reduces the amount of the protein, the processing generates a cleaved Gli form which acts as a negative regulator (Jiang, 2006). The three Gli proteins exhibit different activities on the target promoter sequence. Gli1 has the highest promoter-inducing activity, Gli2 exhibits moderate activity, and Gli3 is mainly inhibitory. These differences are partly explained by the presence of the cleaved form of the Gli proteins. Gli3 is largely found in a cleaved form in cells, whereas only a limited subset of Gli2 is subjected to cleavage (Pan et al., 2006). The activity of the Gli transcription factor is correlated with the ratio of the cleavage and the amount of the inhibitory form existing in cells. Gli1 is not cleaved to form the inhibitory molecule (Dai et al., 1999; Kaesler et al., 2000). The processing step was originally identified for the Drosophila Gli homolog Ci, which is processed by the Slimb ubiquitin ligase (Aza-Blanc et al., 1997; Jiang and Struhl, 1998). The 190 kDa Gli3-190 is cleaved and forms an 83 kDa Gli-83 upon co-expression with β-Trcp, a mammalian homolog of Slimb (Pan et al., 2006). Gli3 is phosphorylated by PKA, CK1, and GSK-3 (Tempe et al., 2006; Wang and Li, 2006), and β-Trcp only interacts with the phosphorylated form of Gli3. Gli3 is subjected to ubiquitination, presumably by the SCFβ-Trcp ubiquitin ligase complex, and is processed in the proteasome. Inhibition of proteasome activity suppresses the formation of Gli-83 form (Figure 3B) (Wang and Li, 2006). A similar processing mechanism is also present for Gli2, which generates Gli2-78 from Gli2-185 (Figure 3B). However, the efficiency of Gli2 processing is lower in comparison with Gli3, suggesting that Gli3 is the preferential target for the SCFβ-Trcp complex. Ubiquitination of Gli2 is also mediated by SCFβ-Trcp (Figure 3B) (Bhatia et al., 2006), although how the same ligase regulates the processing and degradation mechanisms is unknown. The processing mechanisms of Gli1 have not yet been reported; however, Gli1 is also subjected to ubiquitination and subsequent degradation. Gli1 has two degradation domains: the N-terminal degron DN and the C-terminal degron DC (Huntzicker et al., 2006). The degron DC is conserved in the Gli2 degradation domain and interacts with β-Trcp (Figure 3B). The degron DN is also highly conserved among the three Gli proteins. The degron DN induces the degradation of Gli1 independently of the DC. Complete stabilization of the Gli1 is achieved by mutations in both the DN and DC regions. Constitutive expression of Gli1 in keratinocytes is sufficient to induce BCC, albeit at a later stage, at up to 6 months of age (Huntzicker et al., 2006). However, the expression of the most stable form of the deletion mutant (the degron DN and DC double mutant) in the skin of mice leads to the formation of BCCs at the perinatal stage and causes death. The ubiquitin ligase responsible for degron DN-mediated degradation is still uncharacterized. Another ubiquitin ligase which targets Gli for degradation, HIB/SPOP, has recently been identified. HIB/SPOP is a BTB domain protein, and is present in the Cullin3 complex. HIB/SPOP recognizes the serine/threonine-rich sequence in the N- and C- terminal regions of the Ci protein in Drosophila (Zhang et al., 2009). The expression of HIB/SPOP is upregulated by the HH pathway, and the HIB/SPOP-mediated degradation of Gli1 appears to constitute a negative feedback loop (Figure 3A). Therefore, the Gli1 protein is regulated by the two complex types of ubiquitin ligases, SCF β-Trcp and Cul3-HIB/SPOP. Moreover, the HECT-type ubiquitin ligase Itch targets Gli1 for ubiquitination-dependent degradation (Figure 3B) (Di Marcotullio et al., 2006). In this case, Numb serves as an adaptor between Itch and Gli1 and induces the degradation of Gli1. Therefore, Numb functions as a negative regulator of HH signaling.
Suppressor of fused ubiquitination
Suppressor of fused (SuFu) is a negative regulator of the HH pathway. SuFu is a cytoplasmic protein and interacts with Gli to prevent its nuclear translocation, thus inhibiting HH-Gli dependent transactivation. Decreased expression of SuFu is found in several cancer cell lines, such as NCI H322M lung cancer (Yue et al., 2009). Treatment with a proteasome inhibitor restores the expression of SuFu. Furthermore, SuFu is actively ubiquitinated in cells treated with an activator of HH signaling. The activation of the HH pathway by a Smo agonist or by deletion of Patched1 reduces the half life of Sufu, while the Smo inhibitor prolongs it (Yue et al., 2009). The SuFu-K257R ubiquitination site mutant is more stable in comparison with the wild-type SuFu, and thus displays a stronger activity for inhibiting cancer cell growth.
Regulation of FoxM1 expression
FoxM1 is a transcription factor which belongs to the Winged-helix/Fox protein family (Myatt and Lam, 2007) and is involved in cell cycle progression. The inhibition of FoxM1 expression significantly impairs cell growth. FoxM1 is upregulated in BCCs, but not in normal keratinocytes or SCCs (Teh et al., 2002). Immunostaining of BCC specimens revealed strongly positive signals in tumor islands. The staining pattern of FoxM1 highly overlaps with that of Gli1. Indeed, the ectopic expression of Gli1 in cultured keratinocytes induces FoxM1 expression, thus implying that FoxM1 is a target gene of Gli1 (Figure 3A). The expression of FoxM1 shows cell cycle dependency; it is highly expressed in the S to G2/M phase and decreases during the M to G1 phase. The changes in FoxM1 expression are mediated by the ubiquitin ligase complex APC/C-Cdh1 (Park et al., 2008). The anaphase promoting complex (APC complex) specifically binds to and ubiquitinates FoxM1 at the M to G1 transition phase. FoxM1 has two tandem D-boxes and a KEN box, which are the typical motifs in proteins regulated by APC/C-Cdh1. Inhibition of Cdh1 expression abolishes the cell cycle dependency of FoxM1, and allows cycle progression by prompting entry into the S-phase. A similar cell cycling profile is observed when the FoxM1 KEN/D box deletion mutant is expressed in these cells. FoxM1 induces Skp2 and Cks1 expressions, which are two components required for formation of the SCF complex. The inhibition of FoxM1 expression by either gene knockout or knockdown is sufficient to halt cell growth as the cyclin-dependent kinase (CDK) inhibitors p21 and p27 accumulate. The SCF ubiquitin ligase does not function properly without Skp2 and Cks1 (Wang et al., 2005). Therefore, FoxM1 expression, which is elevated in BCCs, regulates the ubiquitination of other proteins by inducing the expression of SCF ligase components and promotes cell growth. It is not currently known whether BCCs expressing high levels of FoxM1 contain any defects in the APC/C system.
Ubiquitin and ubiquitin-related signaling pathways in squamous cell carcinoma
Squamous cells are present in the skin as well as in the multi-layered epithelia of tissues, such as esophagus or oral mucosa. The SCCs described in this review also include esophageal, oral, head and neck, and oropharyngeal SCCs, because ubiquitin-related activity is increasingly found in these SCCs.
Transforming growth factor-β (TGF-β) signaling inhibits cell proliferation and often functions as a tumor suppressor (Akhurst and Derynck, 2001). Smad2 is a transcription factor which mediates TGF-β signaling (Massague, 2000). The inactivation of Smad2 leads to the inhibition of TGF-β signaling, which can contribute to cancer development. Examples include the Smad2-P445H mutation, found in colorectal cancer, and the D450E mutation in lung cancer, both of which lead to Smad2 inactivation (Eppert et al., 1996; Prunier et al., 2001; Uchida et al., 1996). Decreases in the activated form of Smad2 correlate with the esophageal SCC progression (Fukuchi et al., 2006).
Smurf2 is a HECT-type ubiquitin ligase which targets Smad proteins for ubiquitin-dependent degradation (Figure 4) (Lin et al., 2000). Smurf2 preferentially targets Smad2 for the degradation over other Smad family members. Upregulation of Smurf2 is correlated with a poor prognosis of SCC patients (Fukuchi et al., 2002). It has been shown that primary esophageal SCCs are strongly positive for Smurf2 expression. There is an inverse correlation between the expression of Smurf 2 and active Smad2; when a tumor is positive for Smurf2 expression, the Smad2 expression is often low or is not present at all.
Role of PGP9.5/ubiquitin carboxy-terminal hydrolase L1 (UCH-L1)
PGP9.5/UCH-L1 is a bifunctional protein that has ubiquitin hydrolase and ligase activities. It hydrolyzes the bond between unfolded polypeptides and ubiquitin, and generates ubiquitin monomers by processing ubiquitin gene products, in which ubiquitins are tandemly aligned (Larsen et al., 1998). On the other hand, it requires dimerization to function as an ubiquitin ligase (Liu et al., 2002). α-Synuclein is one of the known targets. The brain exhibits a high expression of PGP9.5/UCH-L1 and elevated levels are also found in certain types of cancer (Hibi et al., 1999; Tezel et al., 2000; Wilson et al., 1988; Yamazaki et al., 2002). Non-functional PGP9.5/UCH-L1 causes accumulation and aggregation of α-synuclein in cultured cells, which contributes to Parkinson’s disease (Liu et al., 2002). Indeed, PGP9.5/UCH-L1 is characterized as one of the Parkinson’s disease genes, PARK5. PGP9.5/UCH-L1 was identified in a screening as a gene whose promoter region is highly methylated in head and neck squamous cell carcinoma (HNSCC), together with Cyclin A1, BMP2A, and others (Tokumaru et al., 2004). Similarly, the PGP9.5/UCH-L1 promoter region is methylated in pancreatic cancers and esophageal SCCs (ESCC). The ectopic expression of PGP9.5/UCH-L1 in HNSCC cell lines inhibits cell growth (Figure 4) (Tokumaru et al., 2008). Because PGP9.5/UCH-L1 inhibits the growth of these cancers, the methylation of such tumor suppressor genes would be advantageous for cancer cells to proliferate. On the contrary, PGP9.5/UCH-L1 expression is found in the highly invasive SCC cell line H157. While overexpression of PGP9.5/UCH-L1 in poorly invasive WI38 cells increases invasion potential, knockdown in H157 cells markedly reduces cell migration ability and lung metastasis in the tail vein injection model (Kim et al., 2009). PGP9.5/UCH-L1 appears to affect the epidermal growth factor (EGF)-activated JNK, p38, and Akt pathways, and introduction of an Akt-negative mutant reduces PGP9.5/UCH-L1-induced cell migration (Kim et al., 2009). Whether PGP9.5/UCH-L1 is related to ubiquitin hydrolase or ubiquitin ligase activity in these tumors has yet to be explored.
Involvement of hypoxia-inducible factor
Tumor environments are often associated with hypoxic conditions, although the degree of hypoxia depends on the tumor type and the surrounding environment. HNSCCs are tumors exposed to severe hypoxia (Hockel and Vaupel, 2001), and hypoxia-inducible factor (HIF)-α is expressed under such conditions (Figure 4). HIF is a basic helix-loop-helix/PAS family transcription factor consisting of α and β subunits. The α subunit has a hypoxic-dependent expression pattern, which is mediated by the ubiquitin system (Semenza, 2003). The α subunit is efficiently ubiquitinated under normoxic conditions by an SCF-type ubiquitin ligase-pVHL-Cullin2/elongin B/elongin C complex, which degrades the α subunit. The pVHL complex will not ubiquitinate the α subunit under hypoxic conditions, thus displaying hypoxia-dependent expression. An elevated HIF-α expression is also found in esophageal, oropharyngeal, and oral SCCs (Semenza, 2002; Uehara et al., 2009). HIF-α presence in SCC induces the expression of a set of HIF target genes, including VEGF, IL-6, IL-8, PDK-1, and PHD2 (Chen et al., 1999; Jokilehto et al., 2006; Wigfield et al., 2008). 2-Methoxyestradiol (2-ME) is a potent inhibitor of HIF-1 activity. The treatment of SCC cell lines with 2-ME promotes cell death (Ricker et al., 2004). Moreover, 2-ME treatment induces a downregulation of the VEGF gene expression to similar levels obtained with HIF-1α siRNA. The administration of 2-ME to an HNSCC xenograft mouse significantly reduces tumor size together with a loss of tumor blood vessels. The anti-tumor effect was more pronounced when the tumor was treated with 2-ME in combination with Paclitaxel, which is an agent also known to inhibit HNSCC growth by preventing microtubule depolymerization (Ricker et al., 2004). The inhibition of the HIF pathway is therefore a potent method for treating HNSCC.
Involvement of other ubiquitin-related pathways in SCC
The aberrant expression of several ubiquitin-related molecules has also been reported (Figure 4). The E2-EPF ubiquitin carrier protein, which is one of the E2 proteins, is expressed at a high level in ESCCs (Chen et al., 2009). High expression of E2-EPF is associated with poor patient prognosis. The knockdown of E2-EPF in tumor cells inhibits the growth, migration, and invasion of these cells, and renders the tumor more sensitive to Cisplatin treatment or γ-irradiation. Inhibition of E2-EPF induces the expression of von-Hippel Lindau protein (pVHL), which in turn decreases HIF-1α expression. This may be one mechanism responsible for tumor inhibition when E2-EPF is knocked down (Chen et al., 2009).
The SUMOylation process occurs in a manner analogous to ubiquitination. E1 (AOS1/Uba2), E2 (Ubc9), and E3 are the key enzymes for SUMOylation. There are also de-SUMOylating enzymes, called SENPs, which remove the SUMO residues from the SUMOylated proteins. Two molecules involved in the SUMO pathway, Ubc9 and SENP5, are upregulated in SCCs (Ding et al., 2008; Ronen et al., 2009). Ubc9 expression is elevated in malignant HNSCCs and is predominantly present in the nucleus, while Ubc9 is found cytoplasmic in the cells of peri-tumoral region (Ronen et al., 2009). The upregulation of SENP5 is found in oral SCCs, and strong SENP5 expression is correlated with poor prognosis. However, in contrast to Ubc9, the majority of the SENP5 is localized to the cytoplasm (Ding et al., 2008). Therefore, it is possible that while Ubc9 is involved in the progression of SUMOylation in the nucleus, SENP5 removes the SUMO from cytoplasmic proteins through its endopeptidase activity and contribute to SCC progression. A more consistent comparison by evaluating the molecules in the same specimens of the same SCC types will address this possibility. It will also be important to identify the molecule(s) regulated by these proteins, and determine whether any changes exist in the broad pattern of SUMO- and de-SUMOylated proteins.
NEDDylation is another type of modification by an UBL, NEDD8. Proteins such as p53, Mdm2, pVHL, or Cullins are subjected to NEDDylation and therefore display altered protein functions (Rabut and Peter, 2008). For example, Cullin NEDDylation induces conformational changes which activate its ubiquitin ligase activity as a complex, while removal of NEDD8 inactivates the enzyme. Altered functions of these proteins frequently result in cancer. The SCC-related oncogene (SCCRO, also called DCUN1D1) is overexpressed in SCCs of mucosal origin (such as esophageal, head and neck, and oral SCCs) (Figure 4) (Sarkaria et al., 2006). The short hairpin RNA-based inhibition of SCCRO induces apoptosis in a cancer cell line, whereas overexpression of SCCRO leads to the transformation of NIH3T3 cells. Recently, SCCRO was identified as a NEDDylation-associated E3 ligase (Kim et al., 2008). SCCRO binds to the Cullin ubiquitin ligase complex and efficiently recruits Ubc12-NEDD8 to the complex. Therefore, NEDDylation is also likely to be involved in SCC progression.
Ubiquitin and ubiquitin-related signaling pathways in melanoma
Regulation of MiTF by ubiquitination, SUMOylation, and caspase cleavage
Microphthalmia-associated transcription factor (MiTF) is a key regulator of melanocyte differentiation (Levy et al., 2006). MiTF is a basic helix-loop-helix leucine zipper (bHLH-Zip) transcription factor which binds to the E-box motif of the promoters. MiTF target genes are also involved in cell survival and cell cycle progression of melanocytes. Some examples of MiTF target genes include tyrosinase, tyrosinase-related proteins 1 and 2, which regulate pigmentation, B-cell CLL/lymphoma 2(BCL2), an anti-apoptotic factor, and HIF-1α, which is a master regulator of the hypoxia response (Bentley et al., 1994; Busca et al., 2005; McGill et al., 2002; Yasumoto et al., 1994, 1997). While MiTF plays a key role in melanocyte differentiation, it is also amplified and highly active in certain melanoma cells (Garraway et al., 2005).
Microphthalmia-associated transcription factor is subjected to ubiquitination, which leads to protein degradation (Figure 5A) (Wu et al., 2000). This ubiquitination requires phosphorylation on the serine 73 residue (S73), which is mediated by MAP kinase, and the serine 409 residue (S409), which is mediated by ribosomal S6 kinase. These two kinases are activated by c-kit signaling, which plays an important role in MiTF transcriptional activity (Wu et al., 2000). When the two serine residues are mutated to alanine, MiTF becomes highly stable, indicating that the serine phosphorylation sites are essential for the MiTF ubiquitination and degradation (Wu et al., 2000). Importantly, MiTF stabilization does not increase its activity as a transcription factor; rather, the stabilized form of MiTF has reduced transcriptional activity, indicating that MiTF-dependent transcription needs to be coupled with ubiquitination to achieve maximum efficiency. Therefore, the melanomas, which have an increased MiTF copy number and presumably have high MiTF activity, are likely to induce the ubiquitination of MiTF as well. The ubiquitin ligase for MiTF ubiquitination remains unclear. However, the E2 enzyme Ubc9 was identified as a MiTF-interacting protein by yeast two-hybrid screening (Xu et al., 2000). While Ubc9 is generally known as an E2 enzyme for SUMOylation, it induces the ubiquitination of MiTF, thus resulting in the degradation of the protein. The site for the MiTF ubiquitination has been mapped to the lysine 201 residue (K201).
MiTF is also subjected to SUMO modification (Figure 5A). The analyses of a series of MiTF lysine mutants identified K182 and K316 as SUMOylation sites (Miller et al., 2005; Murakami and Arnheiter, 2005). While the SUMOylation of MiTF does not alter its DNA binding activity, MiTF lacking SUMOylation sites has significantly higher transcriptional activity (Miller et al., 2005; Murakami and Arnheiter, 2005). This effect is possibly due to the role of SUMOylation to inhibit the cooperative action of MiTF with other transcription factors, such as MiTF or another melanoma-related transcription factor Sox10, because the inhibitory effect of SUMO was furthermore enhanced if the promoter region contained multiple MiTF binding sites or both MiTF and Sox10 sites (Miller et al., 2005; Murakami and Arnheiter, 2005). The MiT family transcription factors TFE3 and TFEB are also targeted for SUMOylation (Figure 5B) (Miller et al., 2005). SUMOylation does not affect homo- or heterodimerization among the MiT family of proteins. Interestingly, Sox10, which regulates MiTF expression, is also subjected to SUMOylation, thus suggesting the sequential regulation of transcription factors mediated by SUMOylation (Figure 5B) (Girard and Goossens, 2006). Several questions, regarding precisely when, how, and by which ligase SUMOylation is regulated, remain to be clarified.
In addition to the regulation by ubiquitin-related pathways, MiTF is cleaved by activated caspases-3, -6, and -7 upon TRAIL treatment in human melanoma cells (Figure 5A) (Larribere et al., 2005). Caspases cleave MiTF into 17 and 45 kDa fragments. A point mutation of the caspase cleavage site decreases the number of cells undergoing apoptosis (Larribere et al., 2005). The C-terminal portion of MiTF (17 kDa) plays a critical role for the induction of cell death in melanoma.
The Notch pathway enhances the growth of melanoma cells, in contrast to BCC and SCC, whose growth is inhibited by Notch signaling (Balint et al., 2005; Nicolas et al., 2003). Multiple primary melanomas as well as melanoma cell lines exhibit a high expression level of activated Notch1, whereas melanocytes have almost no Notch activity (Balint et al., 2005). The expression of Notch1 in melanomas is regulated by the Akt pathway through NF-κB activation. The ability of melanomas to form colonies on soft agar is significantly inhibited when melanoma is treated with the γ-secretase inhibitor DAPT to inhibit Notch activation (Balint et al., 2005). On the other hand, the expression of the active form of Notch1 in cells promotes cell growth. Furthermore, cells expressing the active form of Notch1, which were injected into SCID mice, form not only large skin tumors, but also exhibit increased rates of metastasis. This indicates that Notch1 activity regulates both the growth and metastatic capabilities of melanomas (Balint et al., 2005). Importantly, melanomas expressing the active form of Notch1 exhibit a high level of β-catenin expression, which is more prominent in primary melanomas, suggesting that Notch-dependent expression of β-catenin plays a key role in the proliferation of melanoma cells in early stages (see also below) (Balint et al., 2005). Furthermore, the siRNA inhibition of β-catenin expression reduces the melanoma metastases to the lung, indicating that β-catenin is involved at least in part in the metastatic process. An increase in β-catenin levels is mediated by stabilization of this protein, which implicates Notch signaling as one of the upstream regulators of β-catenin, whose expression is regulated by ubiquitination. Notch1-intracellular domain (IC) is degraded by the SCFFbw7 ubiquitin ligase complex via the ubiquitin-proteasome system (Gupta-Rossi et al., 2001; Oberg et al., 2001; Wu et al., 2001). Notch1-IC degradation is enhanced by integrin-linked kinase (ILK) (Mo et al., 2007). ILK phosphorylates Notch on the serine 2173 residue, thereby enhancing the interaction of Fbw7 and Notch-IC, which in turn leads to enhanced degradation. Thus, there is an inverse correlation between the ILK and Notch1-IC expression levels. In melanomas and BCCs, ILK is expressed at relatively high levels and Notch1 at lower levels. In contrast, SCC expresses low level of ILK and high level of Notch1-IC (Mo et al., 2007). The role of Notch activity in either inducing or suppressing the tumor depends on the cell type and the stage of the tumor.
Skeletrophin (also known as mindbomb homolog 2) is a RING-finger protein which serves as a ubiquitin ligase for the Notch ligands, Jagged2 and Delta. Skeletrophin adds poly-ubiquitin chains to Delta, leading to endocytosis (Itoh et al., 2003) but not degradation. Efficient internalization of the Notch ligands is required for the activation of Notch signaling (Le Borgne and Schweisguth, 2003), thus Skeletrophin positively regulates Notch signaling. The expression of Skeletrophin is suppressed in melanomas by promoter hypermethylation. The ectopic expression of Skeletrophin in melanomas inhibits cell invasion. This phenomenon is only found with the wild-type Skeletrophin, but not with its RING mutant form, which lacks ubiquitin ligase activity. Skeletrophin decreases expression of the c-met oncogene, which could be one mechanism for the inhibition of cell invasion (Takeuchi et al., 2006). Activation of the Notch pathway facilitates the progression and metastasis of melanomas (Balint et al., 2005), whereas Skeletrophin, which also positively regulates Notch signaling, inhibited the metastasis of melanoma. A specific timing in regard to precisely when Skeletrophin can effectively regulate the invasion of melanoma may therefore exist.
Similar to other tumors, many melanomas show constitutive activation of the NF-κB pathway (Amiri and Richmond, 2005). NF-κB is a transcription factor which induces a number of genes involved in growth, inflammation, and anti-apoptosis (Perkins, 2007). The activity of NF-κB is inhibited by IκB, which binds to NF-κB and inhibits nuclear localization. Upon NF-κB activation, IκB is subjected to phosphorylation followed by ubiquitination-dependent degradation which results in free NF-κB that translocates into the nucleus and thereafter transactivates its target genes. IκB ubiquitination is mediated by the SCF-type ubiquitin ligase complex, SCFβ-Trcp1/2. β-Trcp1/2 is an F-box protein which recognizes IκB and plays an essential role in the ubiquitination process. BRAF signaling, which is often activated in melanomas, induces the expression of β-Trcp1/2 (Liu et al., 2007). Thus, melanomas express higher levels of β-Trcp in comparison to normal melanocytes. While the activation of BRAF signaling in melanomas increases the expression of β-Trcp, inhibition of BRAF signaling either by a Raf kinase inhibitor, a MAPK inhibitor, or BRAF knockdown reduces β-Trcp mRNA expression (Liu et al., 2007). Therefore, BRAF signaling plays a key role in maintaining β-Trcp mRNA levels. BRAF also activates I kappa B kinase (IKK), which is a kinase involved in the phosphorylation of IκB. BRAF efficiently induces the degradation of IκB by maintaining β-Trcp expression and activating IKK, which in turn activates NF-κB in melanomas. Inhibition of BRAF activity by knockdown or by a Raf inhibitor reduces NF-κB activity in melanomas and renders them more susceptible to pro-apoptotic stimuli (Liu et al., 2007). Similarly, inhibition of the IκB degradation pathway with the dominant negative form of β-Trcp2 (also called HOS), which lacks the F-box domain in melanomas, increases the number of cells undergoing apoptosis upon TNF-α/cycloheximide treatment, ionizing irradiation, UV radiation, and anti-cancer drug treatment (Soldatenkov et al., 1999). Moreover, β-catenin-signaling, which is often activated in melanomas, leads to an increase in β-Trcp levels. This increase occurs at the mRNA level, and is mediated by the coding region determinant-binding protein (CRD-BP) (Noubissi et al., 2006). CRD-BP expression is induced by β-catenin and binds to β-Trcp mRNA to stabilize the mRNA. Therefore, increased β-Trcp may lead to the activation of the NF-κB pathway. The expression pattern of β-catenin, CRD-BP, and the p65 subunit of NF-κB are correlated in melanomas, and inhibition of CRD-BP expression significantly reduces NF-κB activity and the growth of melanomas (Elcheva et al., 2008). There is also evidence suggesting that the NF-κB p105 subunit is actively SUMOylated in melanomas, although its biological significance has yet to be determined (Ganesan et al., 2007).
Regulation of β-catenin-signaling
Melanomas, like other cancers, exhibit constitutive activation of the Wnt-β-catenin pathway (Omholt et al., 2001; Rimm et al., 1999). Nearly a third of human primary melanoma specimens and melanoma cell lines exhibit nuclear accumulation of β-catenin (Rimm et al., 1999). β-catenin stability is regulated by the SCFβ-Trcp1 ubiquitin ligase complex. β-catenin first becomes phosphorylated by its priming kinase GSK-3β on the serine 33 and 37 residues. The phosphorylated β-catenin is captured by β-Trcp1 and becomes ubiquitinated (Winston et al., 1999). Several β-catenin mutants, such as S37F (serine 37 residue substituted with phenylalanine) or S45P/Y/F (serine 45 residue substituted with proline, tyrosine or phenylalanine) which lead to the stabilization of β-catenin, are observed in melanoma cells (Rubinfeld et al., 1997). The overexpression of β-catenin promotes the proliferation and survival of melanomas (Widlund et al., 2002). This effect is mediated at least in part by MiTF, the expression of which is regulated by β-catenin. Inhibition of melanoma growth by the dominant negative form of T cell factor (TCF), one of the key transcription factors in Wnt signaling, is restored by the expression of MiTF, suggesting that MiTF acts downstream of Wnt-β-catenin signaling. Non-degradable β-catenin mutants immortalize melanocytes in a transgenic animal model (Delmas et al., 2007). The immortalized melanocytes lack expression of the p16-ink4a tumor suppressor gene, which is caused by β-catenin directly binding to the promoter region.
The nuclear accumulation of β-catenin has recently been reported to correlate with decreased melanoma growth (Chien et al., 2009). Nuclear β-catenin expression in human melanoma specimens is inversely related to the expression of Ki-67, a cell marker of proliferation. The constitutive expression of Wnt3A in murine B16 melanoma cells leads to the stabilization and accumulation of nuclear β-catenin, which coincides with a reduced cell growth and tumor suppression (Chien et al., 2009). Therefore, the role of β-catenin in melanoma progression cannot be explained as a positive or a negative factor, but perhaps depends on the status of other signaling pathways.
Siah2 regulation of melanoma growth and metastasis
Siah2 is a RING-finger-type ubiquitin ligase which has multiple roles in cellular signaling. Cellular responses, such as transcription, metabolism, hypoxia response, and cell death are regulated by Siah2 (Nakayama et al., 2009). The role of Siah2 in cancer has been demonstrated in pancreatic cancer, lung cancer, and melanoma (Ahmed et al., 2008; Qi et al., 2008; Schmidt et al., 2007). Inhibition of Siah2 significantly reduces tumorigenesis and metastasis in a melanoma xenograft model. Inhibition of the PHD3-HIF-1α pathway by expressing PHYL, a small protein derived from the Siah2 adaptor protein Phyllopod, competes with the Siah2 substrates and inhibits lung metastasis of xenograft tumors (Qi et al., 2008). Meanwhile, the inhibition of Siah2 ubiquitin ligase by the ring mutant form of Siah2 suppresses the Sprouty2-Ras signaling pathway, resulting in a smaller tumor mass with fewer metastases (Qi et al., 2008). Therefore, Siah2 is involved in the two key processes of melanoma progression; growth and metastasis by targeting Sprouty2 and PHD3 for degradation, respectively.
Regulation of insulin-like growth factor receptor (IGF-1R)
Insulin-like growth factor (IGF-1) is an important factor for melanocytes and melanoma cells to grow and survive (Stracke et al., 1989). A transgenic animal model of IGF-1 develops multiple tumors (DiGiovanni et al., 2000). One of the mechanisms for regulation of IGF-1 receptor expression is through ubiquitin-dependent degradation in an Mdm2-dependent manner (Girnita et al., 2003). Mdm2 interacts with and ubiquitinates the IGF-1 receptor in malignant melanomas, such as in Mel-5 or Mel-28. β-arrestin, which is a regulator of the G protein-coupled receptor, serves as an adaptor for the Mdm2–IGF-1R interaction and is required for IGF-1 receptor degradation (Girnita et al., 2005).
Common ubiquitination pathways involved in melanoma and NMSCs development
This review has focused thus far on the differential regulation of the three major types of skin cancer. There are also some common mechanisms in these cancer types. In this chapter, the role of the p27 and p53 pathways will be described as two examples of common mechanisms which exist among these cancers.
Regulation of p27 expression by Skp2
p27 is a CDK inhibitor and helps to regulate the G1/S transition during cell cycle progression. p27 binds to CDK-Cyclin complexes to inhibit CDK activity, and is actively degraded by the ubiquitin-proteasome system during S-phase (Frescas and Pagano, 2008). The SCF complex is the ubiquitin ligase which regulates the stability of p27. Skp2 serves as an F-box protein for p27. The downregulation of p27 is found in multiple cancer types, such as breast, colorectal, and gastric cancers. Similarly, downregulation of p27 is observed in melanoma and SCC (Bhatt et al., 2005; Kudo et al., 2005). The expression level of Skp2 is elevated in melanoma cells in comparison with non-transformed melanocytes (Hu and Aplin, 2008; Li et al., 2004; Woenckhaus et al., 2005). This increase is associated with activating mutations of BRAF, BRAFV600E; a common mutation found in 70% of the melanomas (Davies et al., 2002). BRAF V600E activates the ERK pathway and inhibits the expression of p27 by inducing expression of Skp2 mRNA and its associated protein Cks1 (Bhatt et al., 2007). Inhibition of BRAF V600E by siRNA increases the p27 expression and suppresses the growth of melanoma cells. Therefore, the BRAF mutation in melanomas is a critical factor in p27 regulation. The introduction of Skp2 siRNA restores p27 expression in melanomas and reduces the growth of melanoma cells in culture as well as in a xenograft model (Katagiri et al., 2006). Moreover, the simultaneous inhibition of Skp2 and BRAF in melanomas exerts an additive effect, thus resulting in the nearly complete inhibition of melanoma growth in cell culture (Sumimoto et al., 2006). These results highlight the efficacy of increasing p27 together with the inhibition of BRAF to suppress melanoma growth. Of note, the use of p27 as a melanoma marker is not accepted, because some studies have reported p27 overexpression in malignant melanomas (Bales et al., 1999). The altered expression and activity of Skp2/p27 appear to be main drivers of tumor progression in SCCs, such as in oral SCCs (Kudo et al., 2001). The expression levels of p27 and Skp2 are inversely correlated in oral SCCs, and high Skp2 levels are associated with poor prognosis. The Cks1 expression is also inversely correlated with p27 expression levels in oral SCCs (Kitajima et al., 2004). siRNA inhibition of Cks1 induces the accumulation of p27 and inhibits the growth of the tumor. In contrast to melanomas and SCCs, the significance of p27 pathway in BCCs remains largely unknown.
p53 regulating pathways
p53 is a transcription factor which has potent tumor-suppressing activity through regulating cell cycle progression and inducing apoptosis (Vazquez et al., 2008). Approximately 50% of cancers have a mutation in p53, causing aberrant function or expression of this protein (Lain and Lane, 2003). In addition to the mutations, p53 is regulated by Mdm2, a RING-finger-type ubiquitin ligase. Mdm2 directly interacts with p53 and alters its localization, activity, and stability by ubiquitination of p53, which results in degradation (Fuchs et al., 1998). p53 is maintained in an inactive form through Mdm2 binding, and is expressed under certain stress conditions, such as UV irradiation. Mdm2 acts as an oncogene if its ability to inhibit p53 is not properly regulated.
The expression levels of p53 are relatively high in melanomas (Sparrow et al., 1995). Mdm2 expression is also high, which may be due to p53-dependent induction of Mdm2 (Polsky et al., 2001). However, it is unclear how p53 is maintained at certain levels in cells which have a high level of Mdm2. Although the p53 mutation is rare in melanomas, the activation of H-Ras in melanocytes strongly promotes tumor progression in p53 null background mice, implying its role as a tumor suppressor in melanomas (Bardeesy et al., 2001).
Mutations or altered expression of p53 are also reported in different types of SCCs, such as HNSCCs or oral SCCs (Caamano et al., 1993; Lim et al., 2005). A high expression level of Mdm2 is found in oral SCC cell lines as well as in oral SCC tissue specimens (Katayama et al., 2007; Lim et al., 2005). An elevated level of Mdm2 in SCC tumors is partly caused by hypoxia, which in turn decreases p53 levels (Zhang and Hill, 2004). Mutations in the p53 gene are also observed in BCC patients (Han et al., 2006) which are presumably UV-induced transition mutations, but the effects on the expression of p53 is not known.
This review described herein the role of ubiquitin pathways which are altered in the three major types of skin cancers: BCCs, SCCs, and melanomas. Although they are categorized as skin cancers, the origin of the cell types and their characteristics differ. Therefore, it is not surprising that these cancers potentially utilize different signaling pathways in addition to several common ones. There are common stresses or stimuli which all of these cancer cells would be exposed to because they all originate from skin cells: for example, UV light. UV is thought to be one of the major risk factors, especially for the development of BCCs and SCCs. Genotoxic stress stabilizes p53 and induces genes involved in DNA repair. p53 may not be properly stabilized if the cells overexpressing Mdm2 are exposed to these conditions, and thus mutations will accumulate. Such alterations in the ubiquitin ligase might eventually lead to the transformation of cells.
When would the ubiquitination pathway be activated and inactivated in skin cancers? Phosphorylation is one of the key processes to alter the ubiquitination status of the transcription factors playing major roles in melanoma, such as MiTF, β-catenin, or IκB (Wu et al., 2000) (Perkins, 2007; Winston et al., 1999). Phosphorylation of these molecules is induced by ligands such as Steel factor, IGF-1, or Wnt, thus the existence of such ligands is one of the factors to alter the ubiquitination status in skin cancer. Hypoxic conditions are also associated with skin tumors (Hockel and Vaupel, 2001). Hypoxia stabilizes the HIF transcription factor, which functions in various steps of cancer progression, including growth, invasion, and metastasis (Semenza, 2003). Such environmental cues are also important regulators of the ubiquitination processes in skin cancer.
It is unknown whether ubiquitination-dependent degradation is a positive or negative factor for the progression of the skin cancers. Recently, the proteasome inhibitor PS-341 (Velcade) was approved to treat advanced multiple myeloma patients (Orlowski and Kuhn, 2008). Importantly, studies in other types of cancer also demonstrate that PS-341 has an anti-cancer effect. PS-341 effectively suppresses the growth of SCCs by inhibiting the NF-κB pathway, which leads to a decrease in the VEGF and KC chemokine levels (Sunwoo et al., 2001). Proteasomal activity is also important for melanoma survival, since the PS-341 treatment results in cell death (Amiri et al., 2004). Stabilization of the pro-apoptotic protein NOXA appears to be one of the major effects driven by the PS-341 treatment (Qin et al., 2005). It is not clear if there is any effect of PS-341 on BCCs. The efficacy of treating skin cancers with the proteasome inhibitor appears to be dependent on the signaling pathway the cells use for their growth or survival, and which are regulated by ubiquitination.
Finally, there is another set of proteins commonly used in the ubiquitin system, such as E2 and DUB, which are altered in skin cancers. Those proteins usually regulate a broad range of signaling molecules by either adding or removing ubiquitin or UBLs. Therefore, focusing on one pathway is not sufficient and integrative studies will be required to understand entire network of ubiquitin-related signaling in skin cancers. Such studies will provide vital insights into how specific or how general the ubiquitin pathway is in regulating the three major kinds of skin tumors.