RASSF1C oncogene elicits amoeboid invasion, cancer stemness, and extracellular vesicle release via a SRC/Rho axis
2021; Springer Nature; Volume: 40; Issue: 20 Linguagem: Inglês
10.15252/embj.2021107680
ISSN1460-2075
AutoresMaria Laura Tognoli, Nikola Vlahov, Sander Christiaan Steenbeek, Anna M. Grawenda, Michael Eyres, David Cano-Rodríguez, Simon Scrace, Christiana Kartsonaki, Alex von Kriegsheim, Eduard Willms, Matthew J. A. Wood, Marianne G. Rots, Jacco van Rheenen, Eric O’Neill, Daniela Paňková,
Tópico(s)Protist diversity and phylogeny
ResumoArticle17 September 2021Open Access Source DataTransparent process RASSF1C oncogene elicits amoeboid invasion, cancer stemness, and extracellular vesicle release via a SRC/Rho axis Maria Laura Tognoli Maria Laura Tognoli orcid.org/0000-0002-0640-0748 Department of Oncology, University of Oxford, Oxford, UK These authors contributed equally to this work Search for more papers by this author Nikola Vlahov Nikola Vlahov Department of Oncology, University of Oxford, Oxford, UK These authors contributed equally to this work Search for more papers by this author Sander Steenbeek Sander Steenbeek Molecular Pathology, Oncode Institute, The Netherlands Cancer Institute, Amsterdam, The Netherlands Search for more papers by this author Anna M Grawenda Anna M Grawenda Department of Oncology, University of Oxford, Oxford, UK Search for more papers by this author Michael Eyres Michael Eyres Department of Oncology, University of Oxford, Oxford, UK Search for more papers by this author David Cano-Rodriguez David Cano-Rodriguez orcid.org/0000-0002-2155-9159 University of Groningen, University Medical Center Groningen, Groningen, The Netherlands Search for more papers by this author Simon Scrace Simon Scrace Department of Oncology, University of Oxford, Oxford, UK Search for more papers by this author Christiana Kartsonaki Christiana Kartsonaki Department of Oncology, University of Oxford, Oxford, UK Search for more papers by this author Alex von Kriegsheim Alex von Kriegsheim orcid.org/0000-0002-4952-8573 Cancer Research UK Edinburgh Centre, MRC Institute of Genetics & Molecular Medicine, The University of Edinburgh, Western General Hospital, Edinburgh, UK Search for more papers by this author Eduard Willms Eduard Willms Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK La Trobe Institute for Molecular Science, La Trobe University, Melbourne, Vic., Australia Search for more papers by this author Matthew J Wood Matthew J Wood Department of Paediatrics, University of Oxford, Oxford, UK Search for more papers by this author Marianne G Rots Marianne G Rots University of Groningen, University Medical Center Groningen, Groningen, The Netherlands Search for more papers by this author Jacco van Rheenen Jacco van Rheenen orcid.org/0000-0001-8175-1647 Molecular Pathology, Oncode Institute, The Netherlands Cancer Institute, Amsterdam, The Netherlands Search for more papers by this author Eric O'Neill Corresponding Author Eric O'Neill [email protected] orcid.org/0000-0002-0060-6278 Department of Oncology, University of Oxford, Oxford, UK Search for more papers by this author Daniela Pankova Corresponding Author Daniela Pankova [email protected] orcid.org/0000-0002-3478-8065 Department of Oncology, University of Oxford, Oxford, UK Search for more papers by this author Maria Laura Tognoli Maria Laura Tognoli orcid.org/0000-0002-0640-0748 Department of Oncology, University of Oxford, Oxford, UK These authors contributed equally to this work Search for more papers by this author Nikola Vlahov Nikola Vlahov Department of Oncology, University of Oxford, Oxford, UK These authors contributed equally to this work Search for more papers by this author Sander Steenbeek Sander Steenbeek Molecular Pathology, Oncode Institute, The Netherlands Cancer Institute, Amsterdam, The Netherlands Search for more papers by this author Anna M Grawenda Anna M Grawenda Department of Oncology, University of Oxford, Oxford, UK Search for more papers by this author Michael Eyres Michael Eyres Department of Oncology, University of Oxford, Oxford, UK Search for more papers by this author David Cano-Rodriguez David Cano-Rodriguez orcid.org/0000-0002-2155-9159 University of Groningen, University Medical Center Groningen, Groningen, The Netherlands Search for more papers by this author Simon Scrace Simon Scrace Department of Oncology, University of Oxford, Oxford, UK Search for more papers by this author Christiana Kartsonaki Christiana Kartsonaki Department of Oncology, University of Oxford, Oxford, UK Search for more papers by this author Alex von Kriegsheim Alex von Kriegsheim orcid.org/0000-0002-4952-8573 Cancer Research UK Edinburgh Centre, MRC Institute of Genetics & Molecular Medicine, The University of Edinburgh, Western General Hospital, Edinburgh, UK Search for more papers by this author Eduard Willms Eduard Willms Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK La Trobe Institute for Molecular Science, La Trobe University, Melbourne, Vic., Australia Search for more papers by this author Matthew J Wood Matthew J Wood Department of Paediatrics, University of Oxford, Oxford, UK Search for more papers by this author Marianne G Rots Marianne G Rots University of Groningen, University Medical Center Groningen, Groningen, The Netherlands Search for more papers by this author Jacco van Rheenen Jacco van Rheenen orcid.org/0000-0001-8175-1647 Molecular Pathology, Oncode Institute, The Netherlands Cancer Institute, Amsterdam, The Netherlands Search for more papers by this author Eric O'Neill Corresponding Author Eric O'Neill [email protected] orcid.org/0000-0002-0060-6278 Department of Oncology, University of Oxford, Oxford, UK Search for more papers by this author Daniela Pankova Corresponding Author Daniela Pankova [email protected] orcid.org/0000-0002-3478-8065 Department of Oncology, University of Oxford, Oxford, UK Search for more papers by this author Author Information Maria Laura Tognoli1, Nikola Vlahov1, Sander Steenbeek2, Anna M Grawenda1, Michael Eyres1, David Cano-Rodriguez3, Simon Scrace1, Christiana Kartsonaki1, Alex von Kriegsheim4, Eduard Willms5,6, Matthew J Wood7, Marianne G Rots3, Jacco van Rheenen2, Eric O'Neill *,1 and Daniela Pankova *,1 1Department of Oncology, University of Oxford, Oxford, UK 2Molecular Pathology, Oncode Institute, The Netherlands Cancer Institute, Amsterdam, The Netherlands 3University of Groningen, University Medical Center Groningen, Groningen, The Netherlands 4Cancer Research UK Edinburgh Centre, MRC Institute of Genetics & Molecular Medicine, The University of Edinburgh, Western General Hospital, Edinburgh, UK 5Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK 6La Trobe Institute for Molecular Science, La Trobe University, Melbourne, Vic., Australia 7Department of Paediatrics, University of Oxford, Oxford, UK *Corresponding author. Tel: +44 1865 617321; E-mail: [email protected] *Corresponding author. Tel: +420325873939; E-mail: [email protected] The EMBO Journal (2021)40:e107680https://doi.org/10.15252/embj.2021107680 PDFDownload PDF of article text and main figures. Peer ReviewDownload a summary of the editorial decision process including editorial decision letters, reviewer comments and author responses to feedback. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions Figures & Info Abstract Cell plasticity is a crucial hallmark leading to cancer metastasis. Upregulation of Rho/ROCK pathway drives actomyosin contractility, protrusive forces, and contributes to the occurrence of highly invasive amoeboid cells in tumors. Cancer stem cells are similarly associated with metastasis, but how these populations arise in tumors is not fully understood. Here, we show that the novel oncogene RASSF1C drives mesenchymal-to-amoeboid transition and stem cell attributes in breast cancer cells. Mechanistically, RASSF1C activates Rho/ROCK via SRC-mediated RhoGDI inhibition, resulting in generation of actomyosin contractility. Moreover, we demonstrate that RASSF1C-induced amoeboid cells display increased expression of cancer stem-like markers such as CD133, ALDH1, and Nanog, and are accompanied by higher invasive potential in vitro and in vivo. Further, RASSF1C-induced amoeboid cells employ extracellular vesicles to transfer the invasive phenotype to target cells and tissue. Importantly, the underlying RASSF1C-driven biological processes concur to explain clinical data: namely, methylation of the RASSF1C promoter correlates with better survival in early-stage breast cancer patients. Therefore, we propose the use of RASSF1 gene promoter methylation status as a biomarker for patient stratification. SYNOPSIS The known tumour-suppressor RASSF1A encodes an alternative transcript, RASSF1C, that acts as oncogene in human breast cancer, triggering pro-invasive cellular features and aggressiveness via SRC/Rho/ROCK signaling. RASSF1C expression promotes mesenchymal-to-amoeboid transition in breast cancer cells in vitro and xenograft models in vivo. RASSF1C increases SRC kinase activity, Rho/ROCK signalling and actomyosin contractility, independently of RASSF1A. RASSF1C-induced Rho/ROCK signaling promotes expression of cancer stem-like markers such as Nanog, ALDH1, and CD133. RASSF1C/Rho/ROCK-dependent amoeboid motility results in release of extracellular vesicles that transfer invasive traits to less aggressive cell types. RASSF1C promoter methylation improves stratification and prognostication of early-stage breast cancer patients. Introduction Cell invasion and migration are essential processes during cancer progression and metastatic dissemination. In different tissue contexts, cancer cells use distinct modes of invasion, moving either as collective groups or single cells, adopting mesenchymal or amoeboid motility (Friedl & Bröcker, 2000; Friedl et al, 2012). Mesenchymally invading single cells use integrins and metalloprotease-dependent degradation of the extracellular matrix to invade and metastasize to distant organs. Downregulation of integrins or inhibition of metalloproteases results in mesenchymal-amoeboid transition (MAT) and adoption of amoeboid type of invasion. Amoeboid cancer cells are characterized by upregulation of Rho/ROCK signaling pathway, which drives reorganization of the actin cytoskeleton and leads to generation of protrusive forces. It has been reported that enhanced Rho/ROCK signaling directly promotes phosphorylation of myosin light chain 2 (pMLCII) in vitro (Kimura et al, 1996), which in turn induces reduction of stress fibers and formation of cortical actomyosin (Kimura et al, 1996; Álvarez-González et al, 2015). These major changes in actin architecture enable cancer cells to remodel components of the extracellular matrix and to squeeze into the surrounding tissue. Interestingly, invading cells can employ a hybrid mode of motility, switching between MAT and amoeboid-mesenchymal transition (AMT), depending on the tissue specificity of the matrices (Egeblad et al, 2010; He et al, 2011). This cellular plasticity allows cancer cells to initiate a lesion and adopt appropriate invasive modes for each step of the metastatic process. Cancer dissemination is a complex mechanism, and the exact sequence of molecular events that lead to metastatic colonization of distant sites of the body is not well understood. Extracellular vesicles (EVs) have recently been demonstrated to influence epithelial-mesenchymal transition (EMT) and migration in both an autocrine and non-cell autonomous manner (Luga et al, 2012; Sung et al, 2015). Extracellular vesicles are small lipid bilayer vesicles released from internal compartments such as multivesicular bodies (exosomes) or via blebbing from the plasma membrane (ectosomes, microvesicles) and have been implicated in promoting metastasis (Peinado et al, 2012; Hoshino et al, 2015). Extensive membrane blebbing facilitated by Rho-driven motility has been associated with increased shedding of vesicles (Li et al, 2012; Paluch & Raz, 2013; Sedgwick et al, 2015). It has recently been reported that RhoA-driven tumor progression is attributed to loss of the Hippo pathway scaffold and tumor suppressor RASSF1A (Lee et al, 2016). Epigenetic suppression of the RASSF1-1α promoter has been associated with poor cancer survival (Pefani et al, 2014, 2016) due to suppression of RASSF1A transcript expression. However, this event also represents an epigenetic switch and expression of an alternative isoform, RASSF1C, from the internal RASSF1-2γ promoter (Vlahov et al, 2015). Although methylation of the RASSF1 gene has been investigated in a large number of clinical studies and has been considered as a potential biomarker for breast cancer progression (Grawenda & O'Neill, 2015), how methylation of RASSF1-2γ influences disease outcome has not been addressed. RASSF1C has been shown to activate SRC kinase and induce an invasive phenotype, both in vitro and in vivo (Reeves et al, 2010; Vlahov et al, 2015). SRC plays a crucial role in cancer cell plasticity and has been described as a key player in EMT in solid tumors, via its association with FAK and β-integrins (Canel et al, 2010). SRC can also regulate small Rho-GTPases by phosphorylation of RhoGDI (DerMardirossian et al, 2006). Here we identified a novel mechanism where SRC promotes amoeboid invasiveness via RASSF1C-SRC-RhoGDI signaling and Rho/ROCK/pMLCII activation. We show that SRC activation by RASSF1C leads to inhibitory phosphorylation of RhoGDI, followed by ROCK/Rho upregulation, which in turn drives MAT. We also demonstrate that molecular events driven by MAT are associated with cancer stemness and that amoeboid cells express the cancer stem cell markers ALDH1, CD133, and the pluripotency marker Nanog. Additionally, our data illustrate that RASSF1C-SRC-Rho activity results in release of EVs that transfer stemness, invasive ability, and metastatic behavior to recipient cells in vitro and in vivo. Detailed analysis of the RASSF1 promoter (RASSF1-1α vs RASSF1-2γ) in early-stage breast cancer supports the association of RASSF1C with adverse prognosis and offers an optimized biomarker for patients. Results The RASSF1C oncogene promotes mesenchymal-amoeboid transition RASSF1C oncogene is one of the two main isoforms encoded by the RASSF1 gene (EV1A). The major isoform, RASSF1A, is a bona fide tumor suppressor epigenetically inactivated in numerous cancers (lung and breast cancer, among others (Grawenda & O'Neill, 2015)). Surprisingly, RASSF1C expression is maintained in tumors and emerging studies indicate a pro-oncogenic role for this isoform of the RASSF1 gene (Estrabaud et al, 2007). We have previously linked RASSF1C to SRC kinase activation and aggressiveness in breast cancer, contributing to explain the biological events leading to the aggressive phenotype observed in RASSF1A-methylated tumors (Vlahov et al, 2015). To study the effect of RASSF1C oncoprotein (EV1A) in breast cancer, we employed MCF7 and MDA-MB-231 cell lines, both lacking expression of the tumor suppressor isoform RASSF1A, as shown in reports from our and other groups (Montenegro et al, 2012; Vlahov et al, 2015; Calanca et al, 2019; Chatzifrangkeskou et al, 2019). We first assessed RASSF1C transcript endogenous levels in both breast and H1299 lung cancer cell lines (Fig EV1B, left). The mRNA levels of RASSF1C were significantly higher in MDA-MB-231 cells compared to RASSF1A transcript levels (Fig EV1B, middle), in agreement with reports indicating loss of RASSF1A expression in this cell line. Additionally, we confirmed expression levels of RASSF1C mRNA after MDA-MB-231 cells were transfected with a plasmid encoding for a control sequence (pcDNA3) or RASSF1C (Fig EV1B, right graph). Importantly, upon RASSF1C over-expression, cells remained viable, despite showing lower proliferation rates than controls (Fig EV1C), and displayed markers of proliferation (Ki-67, Fig EV1D) but not markers of apoptosis (FITC-VAD, Fig EV1D) or active mitosis (pH3, Fig EV1D). Click here to expand this figure. Figure EV1. RASSF1C oncogene is responsible for MAT Schematic representation of the RASSF1 gene locus (top) and protein domains of the two main RASSF1 isoforms (bottom). Top: black boxes represent exons, black arrows represent CpG island promoters. Bottom, RASSF1A transcript contains six exons (1α, 2αβ, 3, 4, 5, and 6) and is translated to a protein with 340 aa. The RASSF1C variant is transcribed from an intragenic CpG island and consists of five exons (2γ, 3, 4, 5, and 6) and is translated to a protein of 270 aa. Bottom: at the protein level, RASSF1A and RASSF1C both encode for a RA (Ras Association) domain and a C-terminal SARAH (Sav/RASSF/Hpo) domain. RASSF1C however lacks the C1 (protein C kinase conserved) region at the N-terminal, which is present in RASSF1A. Left: fold change of qRT–PCR for endogenous RASSF1C expression in the cells lines used in this study is shown. Middle: fold change of qRT–PCR for RASSF1C and RASSF1A mRNA in MDA-MB-231 cells are shown. Right: MDA-MB-231 cells were transfected with a Control (pcDNA3) or FLAG-RASSF1C plasmid and mRNA levels of RASSF1C were analyzed via qRT–PCR. Data show a 2.5-fold increase in RASSF1C mRNA upon over-expression. MDA-MB-231 cells stably expressing either a ZsGreen-Control or ZsGreen-RASSF1C construct were FACS sorted for greenhigh expression, re-seeded, and kept in 2D culture for the following 7 days. Confluency was measured by taking a picture every other day for 7 days and quantifying the amount of green fluorescent cells. Confocal images of MCF7 cells transiently transfected with ZsGreen or Flag constructs and stained after 72 h for Ki-67 (red) expression, FITC-VAD (green), and Flag (red), phospho-Histone 3 (red) and DAPI (blue). Scale bars represent 10 μm (Ki-67 and FITC-VAD panels) and 20 μm (pH3 panel). Data information: Data are analyzed by Student's t-test and represented as mean ± SEM. ****P ≤ 0.0001, n = 3. Source data are available online for this figure. Download figure Download PowerPoint The process of MAT is associated with changes in cell morphology, where actin cytoskeleton and stress fibers are reorganized into a contractile actomyosin cortex. Cancer cells adopting amoeboid mode of motility have then typical rounded morphology in three-dimensional matrices. To determine whether RASSF1C is involved in MAT, we transiently transfected MCF7 cells with a fluorescently tagged ZsGreen-RASSF1C (ZsRASSF1C) or control construct (ZsGreen). We observed that ZsRASSF1C-expressing cells grown on 2D surface displayed a progressive decrease of diameter within 12 h of expression of the construct compared to control (Fig 1A, Appendix Fig S1A, Movie EV1). As rounded phenotype is a morphological change associated with reorganization of actin cytoskeleton and actin cortical localization, this indicated a potential correlation between RASSF1C expression and mesenchymal-amoeboid transition (Fig 1A). In order to verify whether these phenotypic changes could also be observed in three-dimensional matrices, the necessary 3D environment for upregulation of Rho/ROCK signaling, we cultured RASSF1C-expressing cells in 3D extracellular collagen matrices. We observed increased numbers of RASSF1C-expressing cells that exhibited rounded morphology in 3D collagen matrix in mesenchymal (MDA-MB-231) (Fig 1B, bottom) and also in epithelial (MCF7 from 80% of epithelial-polygonal to 85% rounded morphology and H1299 from 81% epithelial-polygonal to 88% rounded shape) cell lines (Fig 1B top, Appendix Fig S1B and C), suggesting that RASSF1C may be involved not only in mesenchymal-amoeboid but also in epithelial-amoeboid transition (EAT). Next, we visualized the co-localization of F-actin (via phalloidin stain) and pMLCII, a marker of high myosin II activity correlated to rounded morphology (Georgouli & Herraiz, 2019), which indicated phenotypic shift from actin stress fibers in the mesenchymal control cells (visible in the phalloidin stain, Fig 1C top panel), to cortical actomyosin in RASSF1C-expressing MDA-MB-231 cells (visible in both pMLCII and phalloidin stain, Fig 1C bottom panel). Therefore, as RASSF1C-mediated morphological changes are associated with cortically localized actin and myosin, this indicated MAT (Keller & Eggli, 1998; Wyckoff et al, 2006). To additionally confirm whether RASSF1C cells also actively adopt rounded phenotype during MAT process, we produced MDA-MB-231 spheroids and embedded them into a 3D collagen matrix. We observed that single cells in both control (ZsGreen) and ZsRASSF1C+ve 3D spheroids were able to actively detach and migrate from the spheroids. However, while control cells maintained a typically elongated, fibroblast-like, shape, cells transfected with RASSF1C showed rounded morphology (Appendix Fig S1D). We reasoned that expression of RASSF1C is responsible for EAT, in MCF7 and H1299 cell lines, and MAT in MDA-MB-231 cell line, and for the switch from elongated to rounded morphology induced in 3D collagen matrix. To corroborate these results, we tested whether loss of RASSF1C can revert the phenotype from rounded to elongated, fibroblast-like shape. To this end, suppression of endogenous RASSF1C expression via siRNA-mediated knock-down in MDA-MB-231 cells resulted in a significant decrease in the baseline number of rounded cells and a significant increase in the number of elongated cells in 3D matrix (Fig 1D). Figure 1. The RASSF1C oncogene promotes mesenchymal-amoeboid transition Top: graphs showing the change in cell diameter for MCF7 cells transfected with ZsGreen (left) or ZsRASSF1C (right) plasmids at 0 h and 12 h, connections indicate individual cell tracking. Bottom: example images diameter measurements (red line). Scale 20 μm. Single-cell morphology of MCF7 and MDA-MB-231 cell lines, expressing or not a RASSF1C construct, cultured in 3D rat tail collagen I and analyzed for their rounded or elongated morphology 24 h after seeding in 3D matrix. Confocal images of MDA-MB-231 cells transfected with ZsGreen or ZsRASSF1C grown in 3D collagen matrix. Cells were imaged and stained with Phalloidin-568 (red), pMLCII/Alexa 633 (false-colored green), or imaged for ZsGreen (false-colored white). Scale 20 μm. Single-cell morphology assay of MDA-MB-231 cells cultured in 3D collagen and analyzed for morphology 24 h after transient knock-down using either a control sequence (NT, non-targeting) or a sequence targeting the RASSF1 locus. RASSF1 knock-down marks a reduction in rounded cells. Gelatine zymography assay shows downregulation of MMP-9 metalloprotease in MDA-MB-231 cells over-expressing RASSF1C. 3D Matrigel Boyden chamber invasion with or without metalloproteases inhibitor GM6001 and 3D Hydrogel invasion of MDA-MB-231 control cells or cells over-expressing RASSF1C (24 h). Quantification of Boyden chamber invasion after 24 h in 3D Matrigel of MDA-MB-231 cells, transfected with ZsGreen (control) or ZsRASSF1C (RASSF1C) and treated with inhibitors against ROCK (Y27632, 10 μmol/l) or Rho (C3, 2 μg/ml). Cartoon summarizing intravital imaging experiments. Lung colonization of MDA-MB-231 CFP+ cells (MDA-MB-231CFP) stably expressing control or HA-RASSF1C in mammary gland tumors from 5 mice each, measured from intravital imaging time-lapse images. Average migration speed (expressed as μm/min) of n = 30 randomly selected CFP-positive cells in intravital images of Control and RASSF1C-expressing MDA-MB-231 tumors. Quantification of CFP+ MDA-MB-231 cells displaying different types of morphology in vivo, expressed as average number of events, observed from a total of 30 positions in n = 5 mice/group (Control or RASSF1C). Mesenchymal, amoeboid, and hybrid morphology were scored manually based on diameter and circularity (distance between the two furthest points). The higher distance between cell edges was defined as mesenchymal morphology, while the lower distance was considered amoeboid morphology. Data information: All data are from n = 3 independent experiments. Data are analyzed by Student's t-test and represented as mean ± SEM.*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001. Source data are available online for this figure. Source Data for Figure 1 [embj2021107680-sup-0007-SDataFig1.pdf] Download figure Download PowerPoint As reported in the literature, mesenchymal invasion is dependent on production of metalloproteases that mediate proteolytic degradation of the extracellular matrix (Brooks et al, 1996). Conversely, amoeboid invasion does not rely on proteolytic degradation of surrounding tissue (Pandya et al, 2017), but is dependent on either squeezing through pre-existing pores in the extracellular matrix or deforming surrounding tissue by tension generated by cortical actomyosin (Sahai & Marshall, 2003a; Wyckoff et al, 2006). To investigate the production of metalloproteases we performed gelatin zymography assay in situ, where we analyzed MMP enzymatic activity from the cell medium (Snoek-van Beurden & Von Den Hoff, 2005). We detected increased activity of MMP-9 metalloprotease in control MDA-MB-231 cells compared to cells expressing RASSF1C (Fig 1E, Appendix Fig S1E for quantification). To confirm this observation, we performed Boyden chamber invasion assay with a broad spectrum metalloprotease inhibitor, GM6001. As expected, in control cells GM6001 inhibited Matrigel invasion (Fig 1F, left bar graph). However, in the presence of RASSF1C, metalloprotease inhibition did not influence the ability of cells to invade Matrigel (Fig 1F, left bar graph). These data suggested that RASSF1C-mediated mode of invasion was not dependent on metalloproteases. To further investigate this, we performed an invasion assay using hydrogel matrix, which is not degradable by metalloproteases (Fig 1F, right bar graph, and Appendix Fig S1F). Hydrogel invasion assay showed that control mesenchymally invading cells were not able to invade the hydrogel matrix, whereas RASSF1C cell ability to invade the matrix was significantly greater (Fig 1F, right bar graph). These results strongly support the hypothesis that RASSF1C-mediated mode of invasion is metalloprotease-independent and we reasoned that it may be supported by traction forces generated via contractile actomyosin, that allow cells to squeeze via pores in a three-dimensional environment. Given that Rho/ROCK signaling is the major pathway involved in amoeboid motility (Sahai & Marshall, 2003b), we next asked whether the observed RASSF1C-promoted amoeboid invasion is a consequence of Rho/ROCK upregulation in 3D matrices. Using ROCK (Y27632) and Rho (C3) inhibitors, we could demonstrate that Matrigel invasion by ZsRASSF1C-expressing MDA-MB-231 cells was greatly impaired by both inhibitors, thus demonstrating Rho/ROCK dependency (Fig 1G and Appendix Fig S1G). However, the treatment of control cells with both drugs resulted in no effect on their ability to invade (Fig 1G and Appendix Fig S1G). These data confirmed our previous results, showing that MDA-MB-231 control cells use a mesenchymal mode of invasion which is dependent on metalloprotease degradation of the ECM (Fig 1F). To further support Rho/ROCK dependency, we could show that expression of ZsRASSF1C and a dominant-negative RhoA derivative (RhoA-DN) suppressed ZsRASSF1C-driven invasion in both MDA-MB-231 and MCF7, while a catalytically active RhoA derivative (RhoA-CA) enhanced invasion upon RASSF1C expression (Appendix Fig S1H and I). We next set out to determine whether RASSF1C could promote amoeboid invasion and metastatic spread in vivo. To this end, we employed CFP-labeled MDA-MB-231 cells (MDA-MB-231CFP; Zomer et al, 2015) and stably expressed RASSF1C to promote a switch to amoeboid motility (MDA-MB-231CFP;HA-RASSF1C). Tumors were initiated in the mammary gland of immuno-compromised mice, and lung colonization as well as migration of individual cells were tracked by time-lapse images via intravital microscopy (IVM) (Fig 1H). Importantly, we observed that the number of MDA-MB-231CFP;HA-RASSF1C cells that colonized distal sites in the lungs was significantly greater than the number of control cells, therefore supporting RASSF1C oncogenic potential in vivo (Fig 1I). Low-adhesion attachment during in vitro amoeboid invasion in 3D matrices allows amoeboid cells to translocate at relatively high velocities, between 2 and 25 µm/min, while relative speed of mesenchymally invaded cells is approximately 0.1–0.5 µm/min due to recruitment of proteolytic enzymes and slow turnover of focal adhesions (Palecek et al, 1997; Friedl et al, 1998; Sahai & Marshall, 2003b). We next investigated the speed adopted by RASSF1C cells during in vivo invasion. Manual analysis of the invading cell speed in MDA-MB-231CFP;HA-RASSF1C and MDA-MB-231CFP; Control tumors showed higher migration speed of cells in MDA-MB-231CFP;HA-RASSF1C tumors (Fig 1J). Interestingly, these cells adopted more amoeboid, rounded morphology during their invasion from MDA-MB-231CFP;HA-RASSF1C, compared to control tumors (Fig 1K, morphology was defined as amoeboid, hybrid, or mesenchymal). The data so far suggest that RASSF1C expression promotes a phenotypic switch from mesenchymal or epithelial to amoeboid morphology and motility both in vitro and in vivo, likely through upregulation of Rho/ROCK/pMLCII pathway. RASSF1C-mediated SRC activation promotes pro-amoeboid Rho/ROCKI/pMLCII signaling We next set out to explore the mechanism through which RASSF1C promotes MAT. RASSF1C directly supports activation of SRC kinase by preventing CSK inhibitory phosphorylation on Y527 (Vlahov et al, 2015). Notably, SRC activation is well-documented to promote mesenchymal motility; however, it can also facilitate contractility via inhibitory phosphorylation of RhoGDI (pY156), which results in activation of RhoA (DerMardirossian et al, 2006). Therefore, we hypothesized that RASSF1C activation of SRC may be implicated in MAT via upregulation of Rho/ROCK/pMLCII signaling pathway. As the Rho interacting domain of RASSF1A is identical to RASSF1C (Lee et al, 2016), we investigated whether RASSF1C promotes Rho activation. Rho-GTP pull-down assay on lysates obtained from cells expressing a control or a RASSF1C construct s
Referência(s)