Epithelial-Mesenchymal Transition-mediated regulation of gefitinib resistance and invasion through three dimensional (3D) collagen gels

Abstract

Normal epithelial cells are attached with one another or to the extracellular environment by cell-cell or cell-extracellular Matrix (ECM) interactions, respectively. By accumulated genetic alterations or other factors, the epithelial cells can become tumorigenic, with forming heterogeneous tumor masses. Among the tumor cell mass, certain metastatic cells lost their cell polarity and cell-cell adhesions, being converted to mesenchymal-like cells. This epithelial-mesenchymal transition (EMT) is critically involved in the cancer metastasis. This EMT process also causes resistance of cancer cells to anti-cancer reagents, leading to an attenuation of drug efficiency and enhances migration and invasion for metastatic potential of cancer cells. In this study, I investigated how EMT causes gefitinib resistance of NSCLC by virtues of TM4SF5 (transmembrane 4 L6 family member 5)-mediated EMT induction in NSCLC cells, and how EMT-rendered mesenchymal properties of cancer cells could regulate invasion of TM4SF5-positive and E-cadherin negative to be mesenchymal cells-like, through 3D collagen I gel systems. One of the most important pathways in NSCLC is the epidermal growth factor receptor (EGFR) pathway during tumor progression. In many cases, NSCLC patients can be initially treated with the EGFR-TKI (Tyrosine Kinase Inhibitor), gefitinib. However, continued gefitinib therapy does not benefit the survival of patients due to acquired resistance through additional EGFR mutations, c-MET amplification, or EMT. It is of further interest to determine whether mesenchymal-like, but not epithelial-like, cancer cells can become resistant to gefitinib by bypassing EGFR signaling and acquiring alternative routes of proliferative and survival signaling. Here I examined whether gefitinib resistance of cancer cells can be caused by TM4SF5, which has been shown to induce EMT via cytosolic p27Kip1 stabilization. Gefitinib resistant cells exhibited higher and/or sustained TM4SF5 expression, cytosolic p27Kip1 stabilization, and mesenchymal phenotypes, compared with gefitinib-sensitive cells. Conversion of gefitinib-sensitive to -resistant cells by introduction of the T790M EGFR mutation caused enhanced and sustained expression of TM4SF5, phosphorylation of p27Kip1 Ser10 (responsible for cytosolic location), loss of E-cadherin from cell-cell contacts, and gefitinib-resistant EGFR and survival signaling activities. Additionally, TM4SF5 overexpression lessened the sensitivity of NSCLC cells to gefitinib. Suppression of TM4SF5 or p27Kip1 in gefitinib-resistant cells via the T790M EGFR mutation or TM4SF5 expression rendered them gefitinib-sensitive, displaying more epithelial-like and less mesenchymal-like characteristics. These results indicate that TM4SF5-mediated EMT may have an important function in the gefitinib resistance of cancer cells. I have then investigated whether TM4SF5 induced EMT for enhanced cell migration and invasion to investigate how mesenchymal cell properties following EMT process may regulates the invasive properties such as invadopodia formation and ECM degradation in 3D collagen I-surrounded condition. The two dimensional (2D) cell culture systems can stratify neither the environments around in vivo cancer cells nor evaluation of efficacy of anti-cancer drug candidates with regards to cancer cell invasion and metastasis. For this reasons, I have used three dimensional (3D) cell culture systems using type 1 collagen matrices to study invasive behaviors and mechanisms of tumor cells. In this 3D cell culture system, revealing the TM4SF5-positive cell behaviors involved a technical problem of cell precipitation toward the bottom of the collagen I gels, leading to no-real 3D environment. Therefore, alternatively using highly-invasive and TM4SF5-positive MDA-MB-231 breast cancer cells, their invasive properties in 3D collagen I gels with normal serum-containing media were monitored for the underlying mechanisms. Although an in vitro 3D environment cannot completely mimic the in vivo tumor site, embedding tumor cells in a 3D extracellular matrix (ECM) allows for our study of cancer cell behaviors and the screening of anti-metastatic reagents with a more in vivo-like context. I explored the behaviors of MDA-MB-231 breast cancer cells embedded in 3D collagen I. Diverse tumor environmental conditions (including cell density, extracellular acidity, or hypoxia as mimics for a continuous tumor growth) reduced JNKs, enhanced TGFβ1/Smad signaling activity, induced Snail1, and reduced cortactin expression. The reduced JNKs activity blocked efficient formation of invadopodia labeled with actin, cortactin, or MT1-MMP. JNKs inactivation activated Smad2 and Smad4, which were required for snail1expression. Snail1 then repressed cortactin expression, causing reduced invadopodia formation and prominent localization of MT1-MMP at perinuclear regions. MDA-MB-231 cells thus exhibited less efficient invasion in 3D collagen I upon JNKs inhibition. These observations support a signaling network among JNKs, Smads, Snail1, and cortactin to regulate the invasion of MDA-MB-231 cells embedded in 3D collagen I, which may be targeted during screening of anti-invasion reagents. Altogether, this study reveals that mesenchymal properties acquired by EMT process, which can be induced by membrane proteins such as TM4SF5 can cause drug resistance and enhanced metastatic potentials of cancer cells, suggesting that EMT process can be targeted for anti-metastatic reagents.