SKI is critical for repressing the growth inhibitory function of TGF‐β in human melanoma

Dear Sir, We read with interest the review of Javelaud et al. (2008) entitled ‘Transforming growth factor-b in cutaneous melanoma’. The purpose of this letter is to point out some issues for readers that were apparently misinterpreted or overlooked in this review. We showed several years ago that the oncogenic protein SKI is expressed in melanoma tumors in a disease progression manner (Reed et al., 2001). The biological significance of such over-expression was established by demonstrating that down-regulation of SKI levels by antisense strategies (RNAi was not readily available at that time) inhibited clonogenic growth, up-regulated p21 and impaired CDK2 kinase activity in two independent melanoma cell lines. Our data were confirmed by others in non-melanocytic human tumors including esophageal squamous cell carcinoma (Fukuchi et al., 2004) and pancreatic cancer (Heider et al., 2007). The review by Javelaud et al. questions our data in a way that can confuse and mislead readers. The authors state that ‘... it has been established that a number of TGF-b target genes are up-regulated in melanoma cells exposed to TGF-b, in particular those involved in invasion and metastasis’. They erroneously conclude that such findings are an indication that SKI does not repress TGF-b signaling in melanoma (the reference provided was a published abstract not available in PubMed). Such a claim is very misleading because TGF-b has dual functions as both tumor suppressor and tumor promoter. Indeed, we agree that TGF-b is up-regulated during melanoma progression (Reed et al., 1994). However, our work and that of others has established that SKI curtails specifically the TGF-b tumor suppressor function, which leads to cell cycle arrest and ⁄ or apoptosis. Javelaud et al. appear to ignore these findings. In fact, we were the first to suggest that by curtailing the tumor suppressor activity of TGF-b, SKI allows this cytokine to stimulate autocrine and paracrine mechanisms leading to stroma remodeling, invasion, angiogenesis and metastasis (Medrano, 2003). Also, we and many others have demonstrated that SKI acts downstream of the TGF-b receptor and does not act by interfering with TbRI phosphorylation of Smad2 ⁄ 3. It is also important to remember that (a) SKI does not disrupt Smad3–Smad4 heteromer formation; (b) recruitment of SKI to the Smad3 ⁄ 4 complex through binding to either Smad3 or Smad4 is both necessary and sufficient for repression (Ueki and Hayman, 2003). The Javelaud review also failed to discuss the fact that SKI localizes to the cytoplasm in some metastatic melanoma tumors, thereby entrapping activated Smads and preventing their nuclear translocation (Reed et al., 2001). The authors must be aware of these findings because one of them co-authored a paper demonstrating that the SKI-related protein SnoN can also localize to the cytoplasm of cells (Krakowski et al., 2005). Javelaud et al. also claim that TGF-b induces rapid degradation of the SKI protein in melanoma cell lines. This statement is also misleading. At low TGF-b concentrations, SKI is resistant to proteasomal degradation, a result also observed by others in cultured fibroblasts (Liu et al., 2008). TGF-b is also unable to degrade SnoN in esophageal cancer cells (Edmiston et al., 2005), consistent with endogenous or transfected SKI being resistant to TGF-b degradation in melanoma tumor cells. The Javelaud et al. review does not discuss how the SKI protein stability is affected by its own protein levels, or how TGF-b dosage affects the kinetics of SKI down-regulation and recovery. TGF-b is secreted by melanoma tumor and stroma cells; therefore, it is abundant in the melanoma microenvironment. So, if the SKI protein were so sensitive to degradation by TGF-b in vivo, it would have been practically impossible to detect SKI by immunohistochemical methods, is well documented in paraffin sections of melanomas, esophageal, pancreatic and gastric cancers. SKI interacts with other important partners, including the retinoblastoma protein RB, HDAC1, mSin3, MeCP2, the kinase HIPK-2, and the lim-only protein FHL2 (Reed