Cancer immunology. Bench to bedside immunotherapy of cancers. Nima Rezaei (Ed.). Springer-Verlag (2015)
Metastasis, the path through which a tumor cell exits the primary tumor and colonizes in distant organs, is a complex process with many questions still unanswered .
This multistep process  can be summarized in two major phases:
- First, the translocation of a cancer cell from the primary tumor to a metastatic site has to occur: cancer cells have to escape from the primary tumor mass and enter the microvasculature of the lymphatic and blood systems. Cancer cells moving through the circulation exhibit anchorage-independent survival features and are called circulating tumor cells (CTCs) . CTCs usually head to organs where an environment suitable for their survival has been previously primed; they can be attracted by chemoattractive signals coming from certain tissues and also adhere to the wall of blood vessels through specific surface adhesion molecules. Sometimes they even settle in the bed of the capillaries of target organs, with a diameter too small to let them pass through . Then, cancer cells have to exit the circulation and survive in the new microenvironment of the metastatic site.
- Colonization of the target tissue by translocated cells: cells have to adapt to the different environmental conditions and proliferate, in order to engraft and form a secondary tumor [193, 194].
Mechanisms controlling colonization still remain largely unknown; nonetheless, experimental and clinical data support the hypothesis that disseminated cancer cells adapting to the new microenvironment can be found as solitary viable cells in a dormant, nonproliferative state. These dormant cells can then become more responsive to proliferative signals arising from the primed microenvironment, thus forming micrometastases—whose size is maintained small by a balance between proliferation, apoptosis, and phagocytosis by the target tissue immune system—or even proliferating macrometastasis, after recruitment of an adequate blood supply .
CSCs, EMT, and metastasis
It is now largely believed that only a small fraction of cancer cells have the capability to form metastases .
The definition of CSCs as the only self-renewing tumor cells capable of initiating a new tumor implies that CSCs likely have the major responsibility in invasion and metastasis . This hypothesis is supported by the observations that CSCs are the only cells capable of giving rise to distant metastases and to the growth of new tumors following tail vein injection in mice [194, 195] and that cancer cells that had spread to the bone marrow display a CSC marker profile [196, 197]. The notion that CSCs are responsible for initiation of metastasis is then strengthened by the established association between CSCs and EMT [146, 147, 198]. It has been recently shown that EMT plays a key role in tumor progression and metastasis [199, 200] and that cancer cells have to activate the EMT pathway and acquire a migrating CSC phenotype in order to disseminate and metastasize [156, 171]. The acquirement of this mesenchymal-like phenotype requires cues from the tumor stromal components that secrete EMT-inducing factors (such as TGF-β) . Indeed, EMT can be induced by autocrine or paracrine secretion of mediators such as cytokines and growth factors, due to the cross talk between tumor cells and the microenvironment [156, 185, 199, 200].
When they arrive at the metastatic site, cancer cells have then to undergo the reverse process of mesenchymal–epithelial transition allowing them to initiate the growth of a new tumor [201, 202].
In addition to transforming differentiated cancer cells in migrating cells with self-renewing features and enhanced proliferative capacity, EMT induces proliferation and spread of the existing CSC population, thus further increasing the chances for seeding at distant sites and forming metastases [146, 203, 204]. Moreover, cells generated by EMT show an enhanced resistance to apoptosis that certainly potentiates their capability to survive to the adverse conditions met during the translocation from the primary tumor to metastatic sites . In line with these results, EMT has been associated with poor prognosis in several tumors [156, 206]. However, the role of EMT in facilitating the metastatic spread still remains to be fully demonstrated, especially since there are technical difficulties in detecting this transitory process in human cancer patients .
Signaling pathways involved in metastasis
CSCs are believed to share physiological SC Trafficking mechanisms; thus, migration of CSCs is probably regulated by several redundant and overlapping pathways, similarly to SC homing .
One of the critical regulators of metastatic spread is hypoxia that has been shown to increase the expression of Snail—a pivotal EMT-inducing transcription factor  shown to be expressed at the tumor–stroma interface in several cancers —via the Wnt signaling activation. The Wnt/β-catenin pathway is thought to be a critical factor in the regulation of metastatic process [185, 198, 207]. Accordingly, colorectal cancer cells, residing at the host–tumor interface and thus suggested to be CSCs on the verge of metastasizing , exhibit a high nuclear β-catenin expression .
In addition, hypoxia is able to increase the expression of c-MYC, OCT4, and NANOG, important stem cell factors, in differentiated cancer cells [185, 199, 200].
Furthermore, expression of HIF2a facilitates the metastatic spread both by enhancing the tumorigenic potential in differentiated cells and by inducing proliferation and dissemination of the preexisting CSCs [146, 203, 204].
Interestingly, HGF, which has been recently shown to induce CSC properties and high tumorigenic potential in differentiated colon cancer cells , was used to induce cell scattering of MDCK cells in the initial studies on EMT .
In addition, HGF enhances migration and metastasizing capability of bone marrow hematopoietic cells, by activating the receptor tyrosine kinase MET , involved in metastases establishment. MET expression has been shown to be induced in marrow cells from highly metastatic melanomas by the tumor-derived factors exosomes that thus may stimulate the formation of metastases .
Other factors involved in CSC metastatic and invasive behavior include growth factors, VEGF receptor1 signaling, as well as cytokines and chemokines, such as the SDF-1/CXCR4 migration axis [194, 211].
The observation that SDF-1 is often highly expressed in typical sites of metastasis as lung, liver, bone marrow, and lymph nodes suggests that it is associated with a metastatic process . CSCs, that express the SDF-1-specific receptor CXCR4, can migrate along a gradient of SDF-1, thus facilitating metastasis . The involvement of SDF-1/CXCR4 axis in metastasis has been, indeed, described in various tumor models, such as lung, breast, colorectal, and pancreatic cancers [12, 213–217]. Several studies reported that SDF-1/CXCR4 signaling increases cancer invasion and metastasis also by enhancing the expression of metalloproteinases (i.e., MMP-2 and MMP-9) and integrins (i.e., a5-, β1-, β3-integrins), enzymes known to promote tumor dissemination by extracellular matrix degradation . More recently, an alternative receptor for SDF-1 has been identified and named CXCR7, that shows a high affinity for SDF-1 and for another chemokine, I-TAC . The correlation between CXCR7 expression and tumor aggressiveness, adhesion, invasion, and survival increase was observed in both in vivo and in vitro studies .
Another cytokine appearing to be connected to metastasis is IL-8, which, together with its receptor CXCR1, has been demonstrated to be related to ALDH+ CSCs invasion in breast cancer cell lines . However, the involvement of this chemokine in the metastatic process has not yet been completely clari?ed. Furthermore, a recent work  suggests that the capability of CSCs to express the matrix protein tenascin C (TNC) demonstrates a direct relationship between CSCs and the metastatic process, since TNC expression by CSCs seems to be required to form metastases in the lungs of an animal model. TNC is described to regulate two key pathways for metastatic spread, Notch and Wnt [83, 195, 223–225], through the increase of the expression of musashi homolog 1 (MSI1) and leucine-rich repeatcontaining G protein-coupled receptor 5 (LGR5), respectively . The matrix protein osteopontin (Arg-Gly-Asp (RGD)-containing sialoprotein) was also reported to be involved in metastasis , since it interacts with CD44, a4-integrins, and a5β1-integrins, markers of cell adhesion typically expressed by stem cells [227–229].
CD44 is considered to support the recruitment of cancer cells into secondary tumor sites in lymph nodes, lungs, and bone marrow , and CD44+ breast cancer cells (displaying a breast CSC phenotype) appeared to have enhanced metastatic properties in xenograft models [10, 231].
However, data on CD44 are contradictory: in colorectal cancer, tumor progression has shown to be associated to a loss rather than a gain of membranous CD44 .
In addition to its role in controlling CSC stemness and proliferation and saving CSCs from depletion [232–234], microenvironment is suggested to be of crucial importance in metastatic spread as well. Growing evidences suggest that disseminated cancer cells are able to initiate a secondary tumor only if they migrate in a favorable microenvironment (“seed and soil” hypothesis). Thus, metastasis is likely not a random process, but it selectively takes place in specific organs, such as lungs, liver, brain, and bones, whose microenvironment appears to be more responsive to migrating cancer cells, as compared to other organs . Accordingly, the engraftment of cancer cells in different organs seems to be enhanced by the establishment of a so-called pre-metastatic niche that allows secondary tumors’ initiation and growth. The cells of this pre-metastatic niche have shown to release factors (i.e., SDF-1, S1000A8, S100A9) that can attract disseminating cancer cells, thus enabling their successful homing to distant organs and metastasis development [156, 192]. It has been hypothesized that these pre-metastatic niche cells can be educated within primary tumor and then migrate to form distant metastases or activated locally or in the circulation by tumor-secreted factors . The capability of the primary tumor to prime premetastatic niches, in preparation for and before the disseminating tumor cell arrival, by secreting systemic factors (i.e., cytokines, VEGF-A, PlGF, PSAP), has been observed in several experimental models [235–238]. The presence of bone marrowderived cell clusters in pre-metastatic sites before the arrival of green fluorescent protein (GFP)labeled cancer cells has been proved by flow cytometry and immunofluorescence studies . These bone marrow-derived hematopoietic progenitor cells (HPC) have been demonstrated to express VEGFR1 and other hematopoietic markers, such as CD34, CD11b, c-kit, and Sca-1 [236, 240–242], thus making pre-metastatic microenvironment more hospitable for metastases. Moreover, the VEGFR1 agonist, placental growth factor (PlGF), secreted by the primary tumor, is reported to enhance the production of fibronectin by tumor stroma in pre-metastatic niches ; since fibronectin appears to bind to VLA-4 (a4β1), a fibronectin receptor expressed on HPC, its increased expression likely primes HPC to these sites, to establish the clusters in preparation for metastasis [210, 236]. Furthermore, an experimental model of melanoma showed that tumor fibroblasts participated in pre-metastatic niche formation by inducing the stroma remodeling that was necessary for the establishment of liver metastases .
Despite the increasing studies on CSC metastatic process, it is still debated whether the premetastatic niche only keeps metastasized CSC properties or it is also able to induce a CSC identity in differentiated cells . As already mentioned, the central role of the CSC microenvironment in tumor growth and progression and metastasis formation, combined with its CSC protection against genotoxic damages, strongly highlighted CSC niche and mediators of the cross talk between CSCs and CSC niche as important therapeutic targets . Thus, a better knowledge of CSCs is necessary, in order to develop improved therapies without the risk of tumor recurrence.