Epidermolysis bullosa (EB), a group of mechanobullous blistering disorders, is characterized by extensive clinical heterogeneity. Certain subtypes, such as the recessive dystrophic variants (RDEB), are associated with considerable morbidity and premature demise of the affected individuals within the first four decades of life. Although recent advances in the clinical management of RDEB have significantly extended the lifespan of the affected individuals, RDEB patients increasingly magnifest a major life-threatening complication related to the development of squamous cell carcinomas (SCCs). RDEB-associated SCCs manifest early in life, and are distinguished by a particularly aggressive clinical course. Very high rates of metastatic spread have been observed, rendering skin cancer to be a major cause of death in RDEB patients (1). Despite the life-threatening nature of SCC in RDEB patients, reports on the pathogenesis of RDEB-associated SCCs are scarce. For example, Arbiser et al. (2) described mutations in the p53 tumor suppressor gene in three out of eight RDEB SCC samples. Consistent with compromised p53 function (3,4), RDEB-associated SCC also expresses reduced levels of the insulin-like growth factor binding protein IGFBP-3 (5). In addition, both RDEB-associated and sporadic SCC exhibit enhanced expression of the transmembrane glycoprotein MUC1 (6). Yet, it remains largely unknown whether RDEB-associated tumors represent a distinct entity or share pathways to malignancy with SCCs in the general non-RDEB population. This lack of information has precluded the rational administration of targeted therapies increasingly considered in the management of SCCs.
A recent study has highlighted a potentially interesting feature of RDEB-associated SCC related to the expression of a collagen VII fragment. Collagen VII is the predominant, if not the exclusive, component of anchoring fibrils, attachment structures extending from the lamina densa of the dermo-epidermal basement membrane to the upper papillary dermis and providing stability to the cutaneous basement membrane zone (BMZ) (Fig. 1). In a substantial number of patients with RDEB the collagen VII gene, COL7A1, harbors nonsense mutations, giving rise to premature termination codons, and consistent with expression of truncated collagen VII (7). In fact, expression of a collagen VII fragment that corresponds to the N-terminal non-collagenous domain (NC1) has recently been shown to be necessary for tumorigenic conversion of keratinocytes cultured from RDEB patients xenotransplanted to immunodeficient mice (8). Conversely, RDEB keratinocytes which did not express the NC1 domain did not develop SCCs.
The notion that NC1 expression is required for SCC development in RDEB patients was recently challenged by the isolation of keratinocytes from two RDEB patients with SCC that lack procollagen VII expression (9). It is also worth noting that NC1-dependent tumor-formation has as yet only been described in keratinocytes that were immortalized by the coexpression of Ha-RasV12 and mutant IκBα to inhibit NF-κB activity. Ha-Ras mutations are relatively infrequent in sporadic SCC (10–14) and, at least in fibroblasts, Ha-Ras itself has significant effects on the production of extracellular matrix components, including fibronectin and collagen I (15). This complicates the interpretation of results in Ha-Ras-V12 expressing keratinocytes, as they relate to extracellular matrix composition and tumor-formation. Thus, it will be important to resolve whether NC1 expression is required for tumorigenesis in models of RDEB SCC other than the Ha-RasV12/IκBα model.
Ortiz-Urda and colleagues (8,16) also proposed a molecular mechanism by which NC1 could participate in tumor development based on studies using antibodies reactive with NC1 sequences and predicted to disrupt the interaction of NC1 with laminin 3321. These antibodies inhibited tumorigenicity of Ha-RasV12-transformed NC1-expressing RDEB keratinocytes and inhibited PI-3-kinase activity in these cells. PI-3-kinase activation has been previously shown to provide robust survival signals to multiple epithelial cell types (17,18), including keratinocytes (19). Furthermore, in keratinocytes, integrin-mediated matrix adhesion significantly contributes to PI-3-kinase/Akt activation (19,20). Incidentally, aberrant high-level expression of laminin 332 has also been shown to induce anchorage-independent survival of malignant mammary epithelial cells dependent on α6β4 integrin expression and Rac/NF-κB signaling (21). Collectively, these results suggest the intriguing possibility that the NC1/laminin 332 interaction serves to trigger integrin-mediated signals essential for the survival of transformed keratinocytes in the anchorage-independent state. Implicitly, these results also suggest that the collagenous (C7) and C-terminal (NC2) domains of procollagen VII are not required for α6β4 integrin-mediated survival signaling. This provokes the question whether other means of blocking NC1 expression or function, for example antisense approaches, similarly disrupt tumorigenicity and signal transduction in NC1-expressing keratinocytes.
The extensive and recurrent blistering in RDEB patients attests to the fact that the C7/NC2 domains are essential for anchoring keratinocytes to the BMZ. C7/NC2-dependent BMZ attachment may also serve a critical role in preventing the dissemination of keratinocytes to inappropriate sites, as observed during the metastatic process. Thus, it is possible that the convergence of two mechanisms contributes to the clinical aggressiveness of SCCs in RDEB patients expressing NC1 (Fig. 1). Whereas loss of C7/NC2-dependent cell anchorage enables keratinocyte dissemination, retention of the NC1 domain enables signal transduction in support of cell survival during transit to metastatic sites. Finally, RDEB patients do not develop SCCs immediately after birth but in later life. Thus, oncogenic mechanisms other than NC1-dependent events are likely to contribute to the clinically aggressive, metastatic phenotype of RDEB-associated SCCs and potentially related to the extensive scarring associated with RDEB. Whereas the precise nature of cooperating oncogenic events in RDEB-associated SCC is unknown, it would be worthwhile to investigate whether such mechanisms affect molecular pathways in sporadic SCCs as well.