URPP Fellow: Evelyn Lattmann
Malignant melanoma is the most aggressive skin cancer in the world (Miller and Mihm, 2006). And not only is it quite deadly upon metastasis, but the worldwide incidence of cutaneous malignant melanoma is increasing more rapidly than almost any other cancer type in fair-skinned populations. The majority of melanoma patients (i.e. 80%) are diagnosed at an early enough stage that surgery alone is largely successful (Dummer et al., 2005). However, the prognosis is quite poor for patients with metastatic melanoma.
Building on the extensive melanoma biobank at the University of Zürich hospital, the Dummer lab has applied in vitro proliferation and invasion tests as well as microarray expression analysis to identify transcriptional signatures predictive of phenotypic characteristics of >90% of all known melanoma cultures (Widmer et al., 2012). Hierarchical clustering experiments revealed that the cell cultures could be grouped into two distinct cohorts: those with proliferative characters, and those with invasive features (Hoek, 2007). We established a list of genes that discriminates between these subgroups and that can be used to predict the phenotype of melanoma cell lines with an online tool that is available to the melanoma community (Widmer et al., 2012). Importantly, these cohorts could be distinguished phenotypically by in vitro tests for proliferation, invasion, and growth factor resistance (Zipser et al., 2011). Finally, markers for each phenotype were shown to be anti-correlated in tissue microarrays (TMAs) of melanoma biopsies and to be capable of switching between each phenotypic state in xenograph models (Hoek et al., 2004).
Although some preliminary work has been done to characterize the role of specific genes in regulating these two phenotypic states, most of the transcripts that comprise the proliferative or invasive expression signatures of melanoma cells remain functionally, poorly studied. The C. elegans invasion model used by the Hajnal lab offers an exceptional opportunity to rapidly assess the necessity for a large number of these genes in a developmental cell invasion system that is highly stereotyped, easy to observe, and genetically tractable.
During C. elegans larval development, a specialized cell called the anchor cell (AC) invades the underlying epidermis formed by the vulval cells (Sternberg, 1988). AC invasion is a highly reproducible process, as it is always the same cell that invades at a specific site and at a predetermined time point (Rimann and Hajnal, 2007). During AC invasion, two basal laminae (BL) between the uterus and epidermis must be breached, which involves two distinct mechanisms: 1) guidance of the AC towards the epidermis and 2) breaching of the BL separating the two compartments (Sherwood, 2006). AC guidance depends on a netrin signal produced by the ventral nerve cord (VNC) and an unknown guidance cue from the epidermis. BL breaching, on the other hand, requires the two transcription factors FOS-1 (the ortholog of the human FOS oncogene), and EGL-43 (the homolog of the EVI1 oncogene (Altincicek et al., 2010; Rimann and Hajnal, 2007). These two transcription factors induce in the AC the expression of several invasion effectors, such as the ZMP-1 metalloprotease, or the CDH-3 protocadherin. Finally, studies of AC invasion have not only identified genes that induce cell invasion but also identified genes that actively repress invasion to prevent ectopic BL breaching (Altincicek et al., 2010).
During the first phase of this URPP, we carried out an RNA interference screen in C. elegans to identify genes associated with invasive melanoma that may also be functionally involved in developmental cell invasion. For this purpose, we used anchor cell invasion in C. elegans as an in vivo invasion assay. Through this approach we identified novel pro-invasive genes that have so far not been associated with increased tumor cell invasion. Among the genes we further characterized are specific G1 cell cycle regulators that are frequently genetically altered by mutations and amplifications in melanoma, and components of the protein sumoylation pathway as well as specific sumoylation targets. In addition, some of the genes up-regulated in invasive melanoma cells have already been known to be involved in anchor cell invasion in the worm (e.g. AP-1 transcription factors, or the RAC GEF TRIO). Thus, we have identified among a long list of genes showing elevated expression in invasive melanoma a much shorter list of novel candidate genes that may be functionally involved in regulating melanoma cell invasion. Moreover, we have established 4D imaging of C. elegans larvae expressing different fluorescent reporters (e.g. for cell polarity or basement membrane breaching) in order to observe cell invasion in vivo and characterize the selected candidate genes in gain- and loss-of-function experiments in real time with high precision. Currently, we are setting up a microfluidic chamber system, which permits us to observe cell invasion in C. elegans over longer time periods and under drug exposure. To study cell invasion in human cells, we have established Boyden chamber assays coupled to growth factor stimulation to test the invasive potential of candidate genes in cancer cells including patient-derived melanoma cells. With this assay we are testing the candidate genes identified in the C. elegans invasion screen described above for their roles in melanoma cell invasion. In summary, we have not only identified conserved regulators of cell invasion but also created a unique infrastructure to target melanoma invasion. Based on these very encouraging achievements, we will focus in the following period on further elucidating how these newly identified invasion regulators are integrated in invasive signaling pathways in human melanoma and perform in parallel a detailed functional characterization in C. elegans.