The role of MSX2 in embryonic development has been studied thoroughly throughout the years, while its importance in tumourigenesis is only starting to emerge. Previous studies in pancreatic and breast cancer have painted a diverse picture in respect to this protein. In human pancreatic cancer (Satoh et al, 2008
), as well as in a mouse mammary epithelial cell line (di Bari et al, 2009
), ectopic expression of MSX2 was shown to trigger EMT. Histopathological studies on pancreatic cancer samples (Hamada et al, 2007
; Satoh et al, 2008
), as well as examination of four infiltrating breast ductal carcinomas (di Bari et al, 2009
), showed an association between MSX2 and more aggressive disease characteristics. However, in a recent study from our group, MSX2 expression was shown to correlate with good prognosis and over-expression of the protein-induced apoptosis in both a breast cancer and a human mammary epithelial cell line (Lanigan et al, 2010
In an effort to explore the functionality of MSX2 in melanoma, we employed a number of in vitro
functional assays using the melanoma cell line WM793 and its metastatic derivative 1205Lu (Juhasz et al, 1993
). Ectopic expression of MSX2 resulted in a loss of cell viability over time and reduced clonogenic activity. An evaluation of Caspase 3/7 activity and cell-cycle distribution suggested a pro-apoptotic role of MSX2 in melanoma cells. No significant impact on cell-cycle progression was detected. In a three-dimensional spheroid model, the pro-apoptotic function of MSX2 appeared to be reduced for WM793 cells and abolished in the case of 1205Lu cells. Instead, a novel aspect of MSX2 protein function, the potent inhibition of cell invasion, became apparent. Similar resistance to apoptosis in three-dimensional cultures has previously been described in the context of drug treatments and can be attributed to enhanced pro-survival signals mediated by the spheroid microenvironment (Smalley et al, 2006
; Haass et al, 2008
Using western blot analysis, MSX2 could be linked to a number of signalling pathways involved in the regulation of cell survival and invasion. The anti-apoptotic proteins, BCL2 and Survivin, were found to be markedly reduced in response to MSX2 over-expression. BCL2 is expressed in up to 90% of all melanomas (Cerroni et al, 1995
) and its downregulation using antisense oligonucleotides was shown to be a potent sensitizer for apoptosis in a range of cell lines (Jansen et al, 1998
; Koty et al, 1999
). Blocking the activity of the inhibitor of apoptosis protein, Survivin, using a dominant-negative mutant has also been associated with the induction of apoptosis in melanoma cells (Liu et al, 2004
). Interestingly, Survivin has recently been associated with the regulation of melanoma cell invasion (McKenzie et al, 2010
) and its downregulation might contribute to the reduction in invasive capacity, observed in the present study.
MSX2 is often mentioned in connection with BMP signalling. Bone morphogenetic proteins are multi-functional growth factors that belong to the transforming growth factor β
) superfamily of proteins (Chen et al, 2004
). Similar to MSX2, they have been ascribed both oncogenic and tumour suppressor activities, depending on their cell type or tissue context (Hsu et al, 2005
). For instance, BMP4- (Hamada et al, 2007
) and BMP2- (Ma et al, 2005
) induced MSX2 expression stimulated EMT in pancreatic cancer and cardiac cushion cells, respectively. On the other hand, BMP4 and MSX2 have been associated with the occurrence of apoptosis in sympathetic neuroblasts (Gomes and Kessler, 2001
) and ventricular zone progenitor cells (Israsena and Kessler, 2002
). In both cases, the apoptotic effect was not only dependent on the expression of MSX2 but also on p21 (CIP1/WAF1), a well-described downstream target of BMPs. Coincidentally, upregulation of p21 in response to MSX2 over-expression was observed both in breast cancer (Lanigan et al, 2010
) and in our study on malignant melanoma. This renders a possible role for MSX2 in BMP-mediated apoptosis in these cell types feasible. Interestingly, BMP-induced apoptosis, similar to what has been described earlier for MSX2 over-expression in melanoma, has also been associated with a downregulation of BCL2 and Survivin (Lagna et al, 2006
). This could be seen as additional, albeit circumstantial, evidence for such an involvement.
Furthermore, BMP4 has been shown to stimulate cleavage of N-Cadherin during neural crest delamination (Shoval et al, 2007
). In melanoma cells, N-Cadherin is known to promote cell survival and migration (Li et al, 2001
). Inhibition of N-Cadherin-mediated intercellular interactions using a blocking antibody has been shown previously to increase the level of apoptosis in 1205Lu cells (Li et al, 2001
). Thus, downregulation of N-Cadherin might be a contributing factor in the induction of apoptosis following MSX2 over-expression. Both N-Cadherin and Survivin have been shown to activate AKT signalling (Li et al, 2001
; McKenzie et al, 2010
). However, MSX2-induced downregulation of these proteins did not result in decreased levels of p-AKT, indicating the activation of an MSX2-specific signalling pathway or compensation through other proteins. Similarly, ectopic expression of MSX2 in breast cancer cells was shown to elevate p-ERK levels (Lanigan et al, 2010
), yet no such correlation could be observed in melanoma. Interestingly, in breast cancer, ERK activation was less pronounced in MDA-MB-231, a cell line harbouring KRAS G38A and BRAF G464V mutations, compared with the non-transformed epithelial cell line MCF10a. Both WM793 and 1205Lu cells express mutant BRAF V600E that might mask additional, MSX2-mediated activation of ERK. Moreover, future studies will clarify a direct or indirect role of the above-mentioned proteins in MSX2-induced apoptosis and the regulation of cell invasion.
In addition, a comprehensive cohort of 218 primary melanoma samples was screened for the clinical relevance of MSX2 protein expression. Cytoplasmic MSX2 expression was significantly associated with lower T-stages and prolonged recurrence-free survival. Nuclear MSX2 expression showed no significant correlations. This suggests that the primary, clinically beneficial function of MSX2 is not coupled with DNA binding and direct transcriptional regulation in the nucleus. Instead, a mechanism based on protein–protein interactions is proposed. This is supported by previous findings that demonstrate that the transcriptional activity of MSX2 does not depend on its homeodomain DNA binding sites but can be directed to heterologous promoters, most likely through interactions with other proteins (Catron et al, 1996
). In similar regard, the formation of nuclear heterodimers with the homeobox proteins MIZ1 and DLX5 and their impact on DNA binding are well described (Satoh et al, 2004
). Here, we propose a slightly different mechanism where MSX2 forms heterodimers in the cytoplasm to sequester its binding partner from another location or to form an active signalling complex. However, future studies are needed to shed light onto this concept and the spatial relevance of MSX2 expression. During in vitro
characterisation studies, MSX2 was primarily located to the nucleus of melanoma cells with some cytoplasmic staining being detectable. Further in vitro
models with defined local MSX2 expression will help to clarify the relevance of in vitro
results for clinical outcome.
Overall, in vitro over-expression of MSX2 led to the induction of apoptosis and a clear reduction in melanoma cell invasiveness. A number of signalling proteins were shown to be altered by MSX2 over-expression, including BCL2, Survivin and N-Cadherin. In addition, MSX2 was shown to be associated with good prognosis in melanoma. Cytoplasmic expression of the protein correlated significantly with longer recurrence-free and overall survival. In addition, multivariate Cox regression analysis could establish cytoplasmic MSX2 expression as an independent prognostic factor for increased recurrence-free survival.