MESENCHYMAL FEATURES MEDIATED BY TWIST1 IN COLORECTAL CANCER CELLS AND MICROENVIRONMENT

Background. Colorectal cancer (CRC) is a major cause of death for cancer in western countries, ranking third in both sexes. Therapeutic developments in the past decades have extended life expectancy in patients with advanced disease (i.e., stage III), and even for those with distant metastasis (i.e., stage IV). Most treatments for advanced disease nowadays include a combination of chemotherapy with target therapy. Despite advances, the fact that metastatic colorectal cancer remains largely incurable, pushes to pursue a better understanding of the factors underlying cancer progression. Nowadays, a major field of investigation is the relationship between epithelial tumor cells and the surrounding compartment, namely tumor microenvironment, and in particular its contribution to cancer progression. The tumor microenvironment essentially comprises tumor infiltrating cells, vasculature, extracellular matrix, plus other matrix associated molecules. Transformed cells can modulate the functions of stromal cells, likely to facilitate their own growth and survival. In this Darwinian perspective, outgrowth of cancer cells goes together with local changes. Such changes, like clonal ones, are likely progressive, from the stage of local invasion, up to regional lymph-node colonization and finally to the development of distant metastasis. Infiltrating cells are a mix of populations having myeloid or mesenchymal origin, including tumor-associated macrophages, myeloid -derived suppressor cells, mast cells, monocytes, neutrophils, CD3+ T cells, natural killers, dendritic cells, endothelial cells, mesenchymal stem cells and cancer-associated fibroblasts. Taken together, all these players are involved in a double-faced game, in which an anti-tumor effect (such as that exerted by CD3+ cells in early CRC, [1]) is counteracted by a cancer promoting one (e.g., that exerted by macrophages [2]. Currently, this cancer-microenvironment match implies contradictory and controversial data, largely depending upon the investigated cell population and upon the experimental setting. This Ariadnes' thread seems unravelled, mainly because of the contemporaneous evolution of both tumor and microenvironment cells during multi-stage tumor progression. What is certainly perceived today, is that a switch from a genetic to a non-clonal prospective is required to understand tumor evolution. Clearly, the dynamic architecture of the stromal compartment and the interactions therein, need to be reconciled with the evidences concerning genetic irreversible changes in stromal cells. The latter include loss of heterozygosis, microsatellite instability (MSI) [3], trisomy of Chromosome 7 in connective tissues elements of CRC [4], and p53 mutations (in the stroma of breast carcinoma[5]). These surprising results rise the possibility that the stromal compartment contains not only an admixture of non-neoplastic cells, but also cancer cells with an aggressive and invasive phenotype which became able to invade the surrounding tissues by mimicking fibroblast morphology. This metamorphosis would be possible through the re-use of an embryonic program by cancer cells, that is the epithelial to mesenchymal transition (EMT). Originally, in a murine model of spontaneous metastatic breast cancer, Weinberg and Coll. demonstrated that the highest aggressive potential of cancer cell was reached after their transition from an epithelial to a mesenchymal morphology (i.e., EMT), driven by Twist1 gene [6]. Twist1 expression and EMT are strictly associated with the acquisition of a spindle-like fibroblast morphology, the down-regulation of the epithelial marker E-cadherin and the expression of mesenchymal ones (N-cadherin and vimentin), with metastases development, and with the inhibition of the key pathway of senescence p16 driven [7]. Currently, many studies demonstrate that this embryonic program is activated by a set of transcription factors which act pleiotropically. These include Snail, Slug, Zeb1/2, and obviously Twist1. These regulators are expressed in various combination in a number of malignant tumor types and have been shown in experimental models of carcinoma formation to be casually important for programming invasion [8], [9], [10], [11]. The issues. The multi-step process of metastasis development, recapitulated by local invasion, intravasation, extravasation, colonization and growth in distant organs, postulates that cells undergoing EMT should be able to complete each individual step. However, no experimental evidence of the EMT process has been provided for the first step of local invasion (nor for the following ones) in tissues specimens of human malignancies. Similarly, proof of EMT in human epithelial cancer cells remains partial, largely derived from murine models, and based on ectopic expression of EMT regulators, or on stimulation to achieve transient mesenchymal features. Although several studies demonstrate the expression of EMT transcription factors in both the tumoral and stromal compartments of different cancers [12], [13], [14], they do not directly link this expression to the presence of cancer infiltrating cells. The experimental work. We moved from the unexplained evidence of genetic and chromosomal abnormalities in the stromal compartment of solid tumors, and from the role of Twist1 as EMT regulator in cancer cells. Moving by microarray data of a pool of CRC cells, we first show that tumor cells with a permanent mesenchymal signature, in a stable EMT state, can arise from epithelial CRC cells, both in humans and mice. Then we clarify that, within the mesenchymal signature, Twist1 plays a crucial role in the migration and invasion of CRC tumor cells. By a combination of immuno-based and cytogenetic methods we detected in human CRC a Twist1+subpopulation of stromal cells, with a mesenchymal phenotype. This subpopuation was more represented in MS-stable CRC than in less metastatic MS-unstable CRC, and was associated with advanced CRC stage and with worse survival. Additionally, we showed that Twist1 transcript is degraded in MSI CRC, due to a frameshifted 3’-UTR, and propose that this mechanism contributes to the low capability of MSI CRC to exploit EMT to undergo metastasis. Finally, we identified Twist1+, stromal cells which share genetic changes with epithelial cancer cells, establishing a genetic link between epithelial and mesenchymal components of CRC. In summary, we provide data from an original cellular model to tissue studies to prove the occurrence of EMT in human CRC. We demonstrated the presence of tumor cells in the stroma of CRC tumors and we give a method to identify those cells undergone to EMT and able to invade the surrounding tissues. Thus, our results readdress the study approach to the stromal compartment from that of a recipient of non-neoplastic cells to an incubator of cancer EMT cells able to disrupt tissues by mimicking activated fibroblast. The experimental demonstration of human CRC in stable EMT, coupled to the identification of previously postulated tumor cells with mesenchymal morphology in the stromal compartment, add substantial evidence to EMT, translating a model it into a real phenomenon contributing to the metastatic process of human CRC. In a clinical perspective, our study highlights the importance to re-evaluate the target therapy of solid tumors from the epithelial compartment to include the mesenchymal one.