Community Research and Development Information Service - CORDIS


TUMIC Report Summary

Project ID: 201662
Funded under: FP7-HEALTH
Country: Germany

Final Report Summary - TUMIC (An integrated concept of tumor metastasis: implications for therapy)

Executive Summary:
At the outset of the TuMIC project, it was clear that our concept of the process of metastasis is inadequate and needs to be revised. In the TuMIC project we aimed to use novel experimental approaches to integrate newly emerging principles and ideas such as cancer stem cells and metastatic niches with the different hypotheses that have until now tried to explain the process of metastasis. Specifically we wanted to understand how cancer stem cells (CSCs) behave in and contribute to metastasis, and how networks and pathways that are known to regulate metastasis affect their properties. Further objectives were to determine how a permissive niche microenvironment for metastasis formation is established in given organs, how this contributes to determining patterns of metastasis, and how these microenvironments interact with cancer stem cells. We also set out to perform preclinical studies that build on TuMIC findings with the aim of developing novel anti-cancer therapies.
Our data link CSC properties with metastatic proclivity. For example, a stem cell marker was shown to be functionally required for metastasis of squamous cell carcinomas. Furthermore, a gene expression signature specific to breast cancer CSCs proved to be a robust and reliable indicator of poor prognosis for breast cancer patients. Several lines of evidence we produced link the epithelial-mesenchymal transition (EMT) with CSC properties and metastasis, and the underlying transcriptional regulatory mechanisms have been elucidated. For example, a gene expression signature induced by an EMT-inducing transcriptional regulator proven to be a potent indicator of poor prognosis in breast cancer.
We have made important progress in understanding how the formation of metastatic niches is regulated. The S100 family member S100A4 has emerged from our work as a central regulator of metastatic niche formation, acting to create an inflammatory milieu that constitutes a pivotal component of the metastatic niche, as well as suppressing expression of factors in metastatic niches that counteract metastatic formation. Other factors such as VEGF-C, CCL2 and the c-Kit/KitL axis have arisen from our research as other factors that positively regulate niche formation. Together our data functionally demonstrate the importance of CD11b+ bone marrow-derived cells in the metastastic niche, and also validate the metastatic niche as a therapeutic target for inhibition of metastasis. We also found that EMT endows tumor cells with stemness properties that is associated with angiogenesis induction that is in turn required for metastasis formation, consistent with the notion that CSCs can functionally interact with the metastatic niche by inducing angiogenesis.
Based on these findings we have set up high throughput screens, developed inhibitory antibodies and rationally designed novel compounds with the aim of therapeutically applying our result to the treatment of cancer patients. Novel inhibitors of EMT have been discovered, and several ways of inhibiting metastatic niche formation have been identified. Ongoing studies beyond the lifetime of the TuMIC funding period will develop these promising preclinical approaches to the point where clinical trials can be started.
•To ensure that the findings of the TuMIC project has maximum impact, we have engaged in a number of dissemination and exploitation activities, including scientific publications, organization and participation in conferences, intellectual property protection in the form of patents, and establishment of a project website. We have been actively involved in technology transfer, and have extensively interacted with EU-funded consortia and other key stakeholders.

Project Context and Objectives:
•Summary description of project context and objectives.
It is rare for a cancer patient to die due to the local effects of their primary tumor. Rather, it is the metastatic spread of tumor cells that is ultimately responsible for the vast majority of cancer morbidity and deaths. Understanding the cell and molecular biology of invasion and metastasis and the genetic changes that drive these processes represents one of the last great frontiers of exploratory cancer research. Therapies directed against metastatic cells hold the promise of clearing the body of tumor cells and curing the patient.
Despite the central importance of tumor metastasis for the clinical management of cancer, we are still a long way from understanding how tumor dissemination is orchestrated, be that at the molecular, cellular or organismic level. The mechanisms leading to the metastatic dissemination of tumor cells appear to be similar for many different types of cancer and are associated with multiple cellular processes. These include the transition of tumor cells from an epithelial, adhesive phenotype to cells with mesenchymal morphology and migratory and invasive capabilities, invasion into surrounding tissue, intravasation into blood or lymphatic vessels, survival and dissemination through the blood or lymphatic circulation, colonization of distant organs by adhesion to the vessel wall, extravasation and invasion into distant organ parenchyma, and finally metastatic outgrowth in the distant organ. Thus, metastasis is a highly complex problem with many facets. Several hypotheses have been developed over recent years to explain how this process, or aspects of it, is regulated. However, none of these explanations fully reconcile the available experimental data.
Recent ideas about the cellular basis of tumor growth (cancer stem cells) and the establishment by remote tumors of special permissive microenvironments in target organs prior to metastasis formation (metastatic niches) have the potential to radically change our view of the metastatic process. It has been established for many types of cancer that the bulk of cells that make up a tumor are derived from a small subpopulation of cancer stem cells (CSCs). CSCs are distinguished from the bulk population of tumor cells by their ability to successfully seed new tumors when implanted in low numbers into experimental animals. In contrast, the non-CSC population cannot initiate tumor growth in vivo even when implanted in high numbers. The concept of CSCs should have major consequences for our understanding of the dissemination and metastasis of solid tumors, as CSCs probably represent the only subpopulation of cells able to seed metastases successfully. However, this research area remains little explored and has not been integrated into current concepts that attempt to explain the process of metastasis. Furthermore, an emerging paradigm is that tumors are able to produce factors that induce the formation of so-called pre-metastatic niches in organs where metastases will ultimately develop. These pre-metastatic niches are thought to provide an appropriate and requisite environment for the development of secondary tumors. Again, the molecular and cellular mechanisms underlying the establishment of pre-metastatic niches remain poorly investigated.
Taken together, these observations indicate that our concept of the process of metastasis is still incomplete and needs a major overhaul. In particular, novel experimental approaches are needed that integrate newly emerging principles and ideas with the different hypotheses that have thus far been developed to explain the process of metastasis. This will facilitate the development of an improved and more accurate concept about the process of metastasis. In turn, this will have fundamental ramifications for the way in which novel anti-cancer therapies are designed, and most importantly should provide important new insights into how cancer and in particular metastatic disease can be successfully treated.
Objective 1: To understand how cancer stem cells behave in and contribute to metastasis. While CSCs have been defined and studied in the context of primary tumors, virtually nothing is known about their role in the metastatic process. To address this we planned to study the basic biology of CSCs with particular emphasis on metastasis. New markers and ways of defining CSCs were to be identified, and the function of known CSC markers, some of whose expression is associated with metastasis, in determining CSC properties was to be explored. Using these tools we wanted to examine what happens to CSC populations during progression to metastatic competence, and to determine for example whether the numbers of CSCs increase, and whether their properties change. Furthermore, CSCs from primary tumors were to be compared with those from the metastases they seed to determine whether only particular CSC subpopulations have metastatic propensity. Moreover, the metastatic properties of CSCs derived from different organ metastases were to be compared to find out if particular CSC subpopulations target specific organs. We also aimed to examine the plasticity of CSC and non-CSC populations, for example to determine whether non-CSC populations can acquire stemness under given circumstances.
Objective 2: To determine how networks and pathways that regulate metastasis affect the properties of cancer stem cells. Many genes and regulatory molecular pathways have been shown to promote metastasis, some by modulating cellular programmes that regulate cell polarity, motility and EMT. We aimed to examine how these genes and pathways contribute to metastasis by altering the properties of CSCs. Particular attention was to be paid to pathways such as the Wnt and TGF-beta pathways that not only play a role in regulating the above-mentioned cellular programmes, but also stemness properties. Using the wide range of tools, cells and animal models available to us, we planned to use gain and loss of function approaches to examine the effects of manipulating components of signaling pathways on the properties of CSCs and their role in metastasis.
Objective 3: To understand how a permissive microenvironment for metastasis formation is established in given organs, and how this contributes to determining patterns of metatasis. Our current understanding of the metastatic niche is that primary tumors secrete factors that induce changes in the microenvironment of organs to which they will subsequently metastasize, in part by recruiting immune cells and/or bone marrow precursor cells. In turn, these niches may secrete factors that attract disseminating tumor cells. The metastatic niche is suggested to form a permissive environment that supports the growth of metastases. Using animal models, foundational studies were planned to examine localized changes that primary tumors induce in the microenvironment of organs in which metastases will form, and the extent to which these changes determine the sites within organs where metastases will develop. These studies were expected to allow us to address the questions of what constitutes a metastatic niche, whether the organ distribution of pre-invasive niches faithfully correlate with patterns of organ-specific metastasis, and whether there are different types of metastatic niche. A major aim was to determine the extent to which networks and pathways that are known to regulate the metastatic properties of tumor cells operate by creating permissive microenvironments for metastasis formation in distant organs. Using the wide range of tools, cells and animal models we had in hand, we planned to use gain and loss of function approaches to examine the effects of manipulating these genes and regulatory molecular pathways in primary tumors and then assess the effect on the formation and function of metastatic niches. The role of immune cells and the S100 family of proteins in establishing the metastatic niche was to be another focus of this work.
Objective 4: To understand how microenvironments that are permissive for metastasis formation interact with cancer stem cells. The interaction of CSCs with metastasic niche microenvironments in target organs is likely to be a key step in metastasis formation. We therefore planned a series of studies to examine this step. We wanted to investigate whether metastatic CSCs specifically home to/occupy pre-metastatic niches, or whether they can also lodge elsewhere. The ability of all or only certain subpopulations of circulating CSCs to colonise metastic niches was to be examined. We aimed to determine whether subpopulations of CSCs exhibit specific organ preferences, or whether CSCs that colonise pre-metastatic niches rather occupy niches in all organs in which pre-metastatic niches are to be found. These studies were also to address the issue of whether the mere presence of niches in organs is sufficient to determine organ-specific patterns of metastasis, or whether the interaction of CSCs with a subset of niche microenvironments found only in particular organs determines in which organs metastases grow. An additional focus was to analyse whether the niche microenvironment regulates CSC numbers and properties in established metastases. Another important aspect of this work was to compare the niches occupied by CSCs in metastases with those in primary tumors, to determine whether changes in the niches occupied by CSCs during tumor progression occur, and if so whether this results in mobilization and subsequent dissemination of CSCs from the primary tumor.
Objective 5: To integrate TuMIC findings with the existing literature to produce an improved concept of how metastasis occurs and is regulated. The studies in Objectives 1-4 were predicted to generate radical new insights into the process of metastasis. Importantly, the questions addressed by these studies integrate several currently disparate observations concerning the metastatic process. An important aim was therefore to combine these new data with the existing relevant literature to produce a modernized concept of how metastasis occurs and is regulated. This concept should be more accurate in its understanding and unified in its scope than the current hypotheses that try to explain metastasis.
Objective 6: To perform preclinical studies that build on TuMIC findings with the aim of developing novel anti-cancer therapies. We aimed to translate the findings we were to make concerning metastasis by engaging in a robust series of preclinical studies aimed at finding novel ways of treating metastatic cancer. The knowledge generated as a result of pursuing the previous objectives was expected to identify key molecular targets that play an important role in the process of metastasis. Based on the results from our extensive exploratory and functional experimentation, we aimed to establish assay systems for high throughput screening (HTS) of chemical and biological libraries against novel target molecules. Particular attention was to be paid to targeting CSCs directly that are involved in metastatic spread, as well as to trying to interfere with the interaction between CSCs and the metastatic niche. The end point of these studies was to be proof of principle demonstrations in animal models. Clinical validation was considered beyond the scope of this project.

Project Results:
Description of the main S&T results/foregrounds.

As outlined above, the TuMIC project set out six S&T objectives in order to address the salient issues concerning the basis of metastasis and its treatment that we had identified. The work and results achieved for each of these objectives is first summarised, then the contributions of the individual teams involved in TuMIC to meeting the objectives is outlined in more detail below. The publications arising from the work are then listed.
Objective 1: To understand how cancer stem cells behave in and contribute to metastasis.
A prerequisite for studying CSCs is to be able to isolate and characterize them. We therefore compared the reliability of methods used for this purpose, as well as the utility of stem cell markers for identifying and targeting CSCs. We found that surrogate assays for CSCs do not invariably reflect tumor initiation in vivo. Nevertheless, isolation of CSC subpopulations using surface markers revealed that CSCs are heterogeneous populations containing high and slow cycling stem-like cells
The stem cell markers Oct4 and Nanog were found not to be expressed in breast tumors and melanoma. Nevertheless, FRM4DA, a gene expressed in normal adult epidermal stem cells, is a novel potential marker of CSCs. Expression of the gene is upregulated in human head and neck SCC tumor samples. In functional assays, knockdown of FRM4DA reduced SCC proliferation, decreased colony forming efficiency, impaired growth in vivo, and suppression of metastasis formation in lungs and liver. Together these data indicate that a gene associated with normal adult epidermal stem cell identity contributes to the tumorigenicity of SCC.
In further experiments we determined how the 3D ECM microenvironment regulates the ability of tumor cells to initiate the growth of a new tumor when implanted in vivo (the functional definition of CSCs), and found that Id1 and Id3, genes that play a pivotal role in tumor initiation in vivo.
As a proof of principle that CSCs are relevant for metastasis, we found that breast cancer CSCs isolated using FACS sorting for expression of breast CSC markers gave rise to enhanced metastsatis in vivo compared to the non-selected bulk population of tumor cells. These findings are highly relevant to human disease, because we have defined a gene expression signature that is specific for mammary CSCs, that is earmarked by genes involved in EMT and p53 signalling, and that predicts poor prognosis for all breast cancer patient cohorts tested so far. The predictive power is in the 125 upregulated genes (down-regulated genes do not contribute), and is specific for breast cancer but not other tumor types. There is virtually no redundancy with other previously published CSC signatures, and thus associates expression of many new genes with the CSC phenotype.
Objective 2: To determine how networks and pathways that regulate metastasis affect the properties of cancer stem cells.
The main focus of this work was to determine how EMT is regulated and how this contributes to the stemness properties of tumor cells. Although EMT is associated with increased stemness, we found that EMT did not influence all properties typically associated with CSC identity, but EMT-induced cells formed spheroids more efficiently than epithelial cells. Nevertheless, EMT strongly increased the tumorigenicity of tumor cells in a manner that turned out to be dependent on increased tumor-induced angiogenesis subsequent to EMT induction. These data raise the possibility that increased tumorigenesis and thus stemness is dependent on angiogenesis induction.
We have made extensive progress in delineating transcriptional control networks and their temporal regulation that regulates EMT both positively and negatively. Key transcriptional effectors of EMT-inducing signal transduction pathways that have emerged include Dlx2, Klf4, Sox4, Tcf/?-catenin and Snail1. Importantly, this work has also cast light on the major epigenetic reprogramming that occurs during EMT, and has identified factors involved in its regulation.
The transcriptional regulator ZEB2 is potent inducer of EMT. Gene expression profiling was used to generate a 16-gene ZEB2 activity index for breast cancer cells. By applying the ZEB2 activity index to primary human breast cancers, it was found that the index is a potent indicator of poor prognosis, and superior to evaluation of ZEB2 immuno-histochemical staining. These data provide a first proof of the clinical relevance of ZEB2 in breast cancer, and provide a rational for targeting ZEB2 in this context. The data also indicate the relevance of EMT in breast cancer progression, and support that notion that ZEB2 contributes to CSC subpopulations.
We found that the EMT-inducing transcription factor Snail is overexpressed in a significant proportion of different human skin cancers. To examine the role of Snail in skin cancer, we introduced skin-specific Snail expression into a mouse tumor model. Snail1 stimulates skin tumor formation and metastasis by increasing the CD34+ CSC population, promoting genomic instability and increasing resistance to genotoxic and other forms of stress. These data strongly link the induction of EMT with CSC properties, Similarly, conditional knockout of Snail1 in pancreatic ß-cell tumors resulted in tumors that were less invasive, exhibited reduced angiogenesis, and showed an impaired transition from adenoma to carcinoma.
We also examined how individual genes that promote metastasis formation contribute to CSC properties, and found that not all genes examined promote metastasis by endowing tumor cells with stemness properties. For example, the loss of the polarity protein Par3 in the skin showed that Par3 can have both pro-oncogenic and tumor suppressive activity, depending on the context. Nevertheless, h-Prune was found to promote Wnt-1 signalling, which has been implicated in endowing tumor cells with stemness properties.
Objective 3: To understand how a permissive microenvironment for metastasis formation is established in given organs, and how this contributes to determining patterns of metastasis.
Much of our work for this objective has focused on the extracellular form of S100A4. We found that S100A4 enhances EGFR/ErbB2 signalling, and generates a favourable niche microenvironment for metastasis formation by stimulating massive infiltration of CD4+ T-cells this protein acts as a pro-inflammatory mediator that stimulates the formation of metastatic niches. Mechanistically we found that S100A4 induces expression of SAA1 and SAA3 proteins, key regulators of metastatic niche formation, in an organ-specific manner. In turn, SAA proteins are powerful stimulators of metastasis formation. In addition, S100A4 upregulates expression of a number of pro-inflammatory cytokines and initiates a number of positively-acting feed-forward loops that together create an inflammatory milieu.
We have also discovered ways in which factors secreted by tumor cells can downregulate expression of genes in fibroblasts that normally act to suppress metastasis formation. The matricellular protein Fibulin 5 was found to be downregulated in stromal fibroblasts by tumor-derived S100A4. Expression of Fibulin 5 suppressed metastasis in the lungs and liver through a mechanism that included inhibition of MMP9 expression. These data demonstrate that tumor-derived factors can suppress expression of proteins in fibroblasts in distant organs that suppress metastasis, which in turn promotes metastasis formation in these organs.
We also found that growth factors that induce lymphangiogenesis such as VEGF-C contribute to a metastasis-supporting metastatic niche in the lymph node by increasing the density of perivascular sites. VEGF-C also acts systemically to mobilize CD11b+ bone marrow-derived cells that are recruited to pre-metastatic sites in the lung.
The activity of the metastasis-inducing protein h-Prune is inhibited by dipyridamole. Dipyridamole was found in vivo to inhibit infiltration of tumor-associated macrophages and myeloid-derived suppressor cells, key components of metastatic niches, as well as to reduce levels of inflammatory cytokines. These data suggest that h-Prune may contribute to metastatic niche formation.
Objective 4: To understand how microenvironments that are permissive for metastasis formation interact with cancer stem cells.
The interaction of CSCs with niche components such as ECM proteins and endothelial cells has begun to be investigated, for example through the development of novel co-culture techniques that allow the interaction between endothelial cells and tumor cells to be assessed. We have also shown that EMT is associated with stemness properties and increased VEGF-A expression. The increased angiogenesis that is thereby induced is required for efficient growth of primary tumors and metastases. This provides an elegant example of how CSCs interact with the metastatic niche microenvironment to promote metastasis formation.
Objective 5: To integrate TuMIC findings with the existing literature to produce an improved concept of how metastasis occurs and is regulated.
The TuMIC Coordinator acted as Guest Editor for a special issue of Seminars in Cancer Biology entitled “Novel Concepts in Cancer Metastasis”. This special issue acted as a forum for publication a unified and extended concept of metastasis. The concept paper, which was entitled “Concepts of metastasis in flux: The stromal progression model” presented a new model of metastasis called the “Stromal Progression Model” which unifies and extends virtually all of the current models of metastasis, and which incorporates several findings made during the TuMIC research activities.
The central tenet of the stromal progression model is that the formation of primary tumors is dependent not only on progressive genetic changes in cancer cells, but also on the progressive development of an inflammatory tumor stroma. These two processes are mutually dependent on each other in a continual co-evolution. The dependency of primary tumor cells on their stroma means that when they disseminate, the absence of the requisite stroma at secondary sites determines that most if not all of the cancer cells are unable to survive or simply remain dormant, depending on the microenvironment they encounter. The formation of metastatic niches either pre-metastatically or after the settlement of DTCs serves to re-establish the stromal environment that DTCs require for growth as tumors, and can be coupled to further genetic changes in the tumor cells themselves. Thus the evolution of metastatic niches recapitulates stromal progression in the primary tumor
The TuMIC special issue of Seminars in Cancer Biology is available online ( The concept document is being made open access to ensure that the article is available to all interested stakeholders. The print version of the special issue will be forwarded to the Commission once it becomes available from the publishers.
Objective 6: To perform preclinical studies that build on TuMIC findings with the aim of developing novel anti-cancer therapies.
A number of exciting and promising novel therapeutics have been developed based on the concept and findings of the TuMIC project. Derivatives of dipyridamole, the inhibitor of h-Prune, were developed which have much superior activity and more potently suppress metastasis formation in vivo compared to dipyridamole. Bindarit, an anti-inflammatory indazolic derivative that inhibits the synthesis of MCP-1/CCL2, was found to impair metastatic disease in breast and prostate cancer models by decreasing the infiltration of tumor-associated macrophages and myeloid-derived suppressor cells, validating the niche as a therapeutic target. We have also developed antibodies against the metastasis-promoting S100A4 protein that block its ability to promote lung metastasis.
In other work, we have performed high throughput screens to identify promising therapeutic lead compounds. For example, as Lrg5/GPR49 is a marker for CSCs and plays a crucial role in cancer development, we set up a number of screens with a view to developing antagonists against this molecule. We have and continue to screen siRNA and chemical libraries for novel compounds that interfere with EMT, for example using a novel high content screen we have developed. Therapeutic targets have been identified in an RNAi screen for transcription factors that modulate the EMT-inducing effects of ZEB1. Knockdown of one of these (EMTF1) reversed ZEB1-induced EMT, upregulated miR200 family members that are involved in suppressing EMT, and blocked invasion in vitro. Importantly, knockdown of EMTF1 also blocked lung metastasis formation in vivo.
Irradiation of normal mammary tissue induces a hypoxic microenvironment that promotes metastasis through HIF1-dependent expression of KitL. KitL induces mobilization of CD11b+ cells from the bone marrow that express c-Kit, the receptor for KitL. These CD11b+ cells home to the primary tumor as well as to pre-metastatic sites. Inhibition of either HIF1, KitL or c-Kit prevented mobilization of the CD11b+ bone marrow cells and attenuated metastasis. These data cast important light on how irradiation of normal tissue (the so-called tumor bed effect) can play a role in the relapse of cancer patients by promoting the formation of pre-metastatic niches. Furthermore, suppression of PlGF expression in tumor cells also attenuated the pro-angiogenic activity of the CD11b+ cells, resulting in reduced tumor growth and metastasis. These data show that targeting either the mobilization of activity of CD11b+ cells is an effect way of suppressing metastasis formation, and clearly identified strategies for therapeutic implementation of the results.

Contribution of individual teams

Beneficiary 1 – University of Heidelberg (initially Beneficiary 12 – FZK)
Reliability of in vitro assays of CSC properties.
A number of properties attributed to CSCs have been used to define the CSC population, the most important of which is the ability of selected subpopulations of cancer cells to initiate reproducibly the growth of new tumors in vivo. Other assays such as spheroid formation, expression of particular markers and label retention have also been widely used for defining CSCs, although the degree to which these assays invariably reflect the ability to form tumors in vivo remains to be carefully evaluated. Given the importance of correctly defining and isolating CSCs to be able to make valid conclusions about their characteristics, we used syngeneic animal models to compare these different assays. In standard spheroid assays, cell aggregation rather than spheroid growth from single cell suspensions ensued, but aggregation was circumvented by the inclusion of methylcellulose in the medium. Label-retaining subpopulations did not reliably exhibit an enhanced ability to form spheriods, and were enriched for senescent cells. Spheroid forming ability was found to correspond to expression of established CSC markers, although not invariably. Furthermore, spheroid-forming ability was not always reflected in tumor-initiating properties in vivo. Long-term culture of primary mammary tumor cells as adherent monolayers dramatically increased their tumor-initiating ability in vivo. Together these data indicate that assays that are widely used to define CSC subpopulations do not invariably reflect tumor-initiating properties in vivo.
Expression of stem cell markers in CSC as a means of purifying them
The transcriptional regulators Oct4 and Nanog regulate stem cell characteristics, have been reported to be increased in CSCs, and have been suggested to regulate CSC identity. We have crossed mouse lines that develop breast cancer or melanomas with “knockin” mice that express a fluorescent marker protein (GFP) under the control of the promoters of either Oct4 or Nanog. If Oct4 and Nanog are expressed in CSCs and regulate their identity, then tumors arising in the resulting compound transgenic mice should express the fluorescent GFP protein exclusively in the CSC population, allowing their isolation by FACS. Breast tumors or melanomas taken from compound Nanog-GFP or Oct4-GFP transgenic mice were examined for GFP expression. In all cases, no Nanog-positive GFP-expressing cells in the breast or melanoma tumors could be detected, despite clear expression of GFP in the stem cells of the testis. This was true to both MMTV-PyMT and MMTV-neu transgenic breast cancer models. Together, these data suggest that the association of Oct4 and Nanog expression with CSC characteristics may be a feature of cultivated cell lines that does not reflect the situation in autochthonus tumors.
Role of 3D ECM in determining CSC properties
Tumor initiation in vivo is currently the gold standard in defining CSCs, yet the take rate of tumors in vivo can be manipulated, most importantly by coinjecting tumor cells with matrigel or fibroblasts. The majority of published studies on CSCs use co-injection of tumor cells with matrigel to determine in vivo tumor initiation rates. We found in syngeneic animal models that matrigel, laminin and collagen all have pronounced enhancing effects on take rate in vivo, permitting breast and melaoma tumors to grow with only 5 cells when co-injected with these ECM components. These and other data showed that a 3D ECM-rich environment is critical for determining growth of tumor cells in vivo. Together with Charles DeCraene we showed that three genes were found to be strongly upregulated in response to 3D but not 2D culture: Id1, Id3 and Smad6. Interestingly, Id1 and Id3 are genes that have been shown to play a pivotal role in allowing tumor cells to grow in vivo. On the basis of these and other data, our working hypothesis is that tumor initiation assays do not reflect CSC properties, but rather the ability of tumor cells to interact with ECM components in 3D microenvironments that support tumor growth in vivo.
Role of metastasis-inducing genes in regulation of CSC properties
We have examined whether Ier2, a gene we have shown functionally contributes to metastasis formation, exerts its metastasis-promoting effect by regulating EMT and thereby endowing tumor cells with stemness properties. We found that Ier2 is upregulated in response to TGF-beta-induced EMT in mammary epithelial cells. This upregulation was abrogated in NMuMG cells in which Smad2 is knocked down, demonstrating that TGF-beta signalling regulates Ier2 expression. However, knockdown of Ier2 expression in response to TGF-beta did not impair TGF-beta-induced EMT. Furthermore, ectopic expression of Ier2 also did not suffice to induce EMT. Together these data indicate that Ier2 exerts its metastasis-promoting effects independently of EMT and subsequent endowment of cells with enhanced stemness properties.
Formation of the metastatic niche
A perivascular niche has been reported to be required for maintenance of the CSC subpopulation. Using breast cancer metastasis models, we found that lymphangiogenesis is induced in regional lymph nodes prior to metastatic colonisation of the lymph nodes by disseminating tumor cells, and was further increased once metastases started to form. We therefore investigated the hypothesis that the lymphatic vasculature may provide a perivascular niche for CSCs, and that expansion of the lymphatic microvasculature in lymph nodes pre-metastatically fosters metastasis formation by providing such a niche. We manipulated lymph node lymphangiogenesis in a syngeneic rat breast cancer model using intra-dermal delivery of either recombinant VEGF-C or VEGFR-3 blocking antibodies to induce or suppress lymph node lymphangiogenesis, respectively. Our results show that for poorly metastatic NM-081 breast tumours, the induction of lymph node lymphangiogenesis is not sufficient to promote metastasis formation in lymph nodes or other organs in the absence of primary tumour lymphangiogenesis. However, while inhibition of the lymph node lymphangiogeneis induced by highly metastatic MT-450 breast tumours suppressed the outgrowth of lymph node metastases, the formation of lung metastases was not inhibited. Together these data demonstrate that metastatic outgrowth in lymph nodes can be dependent on VEGF-C-induced lymph node lymphangiogenesis, consistent with the notion that expansion of the lymphatic vascular bed in lymph nodes may provide metastatic niches for incoming tumor cells. However, the data also suggest that VEGF-C may induce lung metastasis independently of its effects on lymph node metastasis.
Further work suggests that increased systemic levels of VEGF-C consequent to tumor growth promotes metastasis through mobilizing bone marrow-derived cells that contribute to metastatic niche formation. Interstitial fluid from VEGF-C-producing MT-450 breast tumors contain enhanced levels of VEGF-C. Furthermore, MT-450 tumor-bearing animals have increased levels of VEGF-C in their blood. Moreover, we have found statistically significant increases in pro-lymphangiogenic factors in the blood of breast cancer patients compared to healthy individuals. Systemic delivery of VEGF-C in experimental animals to levels observed in tumor-bearing individuals results in dramatic mobilisation of a number of bone marrow-derived cells, including mature neutrophil granulocytes, monocytes and eosinophils. Importantly, we also observed deposition of CD11b+ bone marrow-derived cells in the lungs around the terminal bronchi, a common site of metastasis formation. Given the role of these bone marrow-derived populations in metastasis formation, and the deposition of CD11b+ cells in the pre-metastatic niche, current work is focusing on examining the hypothesis that VEGF-C not only promotes metastasis by regulating the lymphatic vasculature, but also by acting on the bone marrow.
During these studies, we also observed that increased systemic levels of VEGF-C significantly reduced thrombocyte levels. By following up these observations, we found that VEGFR-3, the prime receptor for VEGF-C, is expressed on megakaryocyte precursors, and that its activation by VEGF-C regulates megakaryopoiesis through suppressing differentiation and maintaining the progenitor pool. Importantly, we found that systemic delivery of VEGF-C following 5-FU treatment, a widely used chemotherapeutic, stabilises thrombocyte levels, pointing to a possible therapeutic application.
The S100 protein A8 has been reported to induce premetastatic niche formation, in part through activiation of SAA3 expression that recruits CD11b+ cells to sites where metastases will form. Together with the Lukanidin lab we have investigated whether S100A4, a protein that promotes metastasis, is also involved in this process. Using microarray analysis we found that cells treated with S100A4 strongly upregulated SAA3. Using purified S100A4 and S100A12 protein produced by the Lukaindin group, we showed that S100A4 induces organ-specific SAA3 expression in the liver but not the lung. In subsequent experiments, S100A4 has emerged as a central regulator of pro-inflammatory pathways. Importantly, SAA1 and SAA3 expression is sufficient to promote widespread metastasis formation, providing a major insight into how S100A4 promotes metastasis formation. These data also identify SAA proteins as potential therapeutic targets.
Publications #10, 12, 22, 44, 52 ,65, 67, 69, 70, 71, 73, 83 – 91, 95, 106

Beneficiary 2 – University of Basel
Epithelial to mesenchymal transition (EMT) involves the loss of epithelial-cell markers and gain of mesenchymal-cell markers at the invasive front of various solid tumors and constitutes a central step during malignant tumor progression. Acquiring mesenchymal properties liberates these cells from the primary tumor and allows them to invade neighboring tissues, to enter the circulation and to colonize to distant organs. The dramatic changes in cell morphology and behavior during the process of EMT and metastasis are accompanied by substantial changes in gene expression. Recently, a number of transcription factors have been identified that play critical roles in the initiation and execution of EMT and in the metastatic process, including Snail1 (Snail), Snail2 (Slug), Zeb1 (?EF1), Zeb2 (Sip1), E47, Twist, goosecoid, FoxC2, Dlx2, RBPj?, Yap/Taz, and NF?B. However, the activities of these transcriptional regulators do not explain the full range of changes in gene expression during the multistage process of EMT, and the epistatic hierarchy of the transcriptional regulatory networks has not been worked out. Transcription factors with master control functions and their functional interactions remain to be identified and delineated.
Transcriptional control of EMT and malignant tumor progression
We have established a list of genes that change in their expression during the consecutive morphological states of TGF?-induced EMT in normal mammary epithelial cells and in murine breast cancer cells. In collaboration with Prof. Erik van Nimwegen’s laboratory at the Biocenter of the University of Basel we have employed motif activity response analysis (MARA) to identify binding motifs for various transcription factors at the promoters of the regulated genes, and predicted a number of transcription factors that possibly regulate a subset of genes in the early, intermediate and later stages of EMT. MARA identified all known transcription factors already known to play a role in EMT, including Snail-1 and 2, Zeb-1 and 2, Twist, FoxC2, NF?B, Hmga2, Goosecoid and several others. In addition, we identified a large number of additional transcription factors that may potentially are required for the regulation of EMT, including including Klf4, Sox4, Tcf4, Lhx2, FoxC1 and 2, FoxF2, Tead2, Rbpj? and Dlx2.
We have analyzed a defined number of signaling pathways and their transcriptional effectors and we here report the most advanced findings on the functional contribution of the transcription factors Dlx2, Klf4, Sox4, Tcf/?-catenin and Snail-1 and some of their target genes to EMT and metastatic tumor progression. Moreover, we have taken a genome-wide comprehensive approach to delineate the functional contribution of Polycomb-mediated epigenetic control to EMT.
We have found that the transcription factor distal-less homeobox 2 (Dlx2) exerts critical functions during EMT. Dlx2 counteracts TGF?-induced cell cycle arrest and apoptosis in mammary epithelial cells by at least two molecular mechanisms: Dlx2 acts as a direct transcriptional repressor of TGF? receptor II (TGF?RII) gene expression and reduces canonical, Smad-dependent TGF? signaling and thus cell cycle arrest and apoptosis. On the other hand, Dlx2 directly induces the expression of betacellulin, a ligand of epidermal growth factor receptor, which in turn promotes cell survival by stimulating EGF receptor signaling. These results establish Dlx2 as one critical player in shifting TGF? from its tumor-suppressive to its tumor-promoting functions.
Kruppel-like factor 4 (Klf4) has been identified as one of the transcription factors that are significantly down-regulated in their expression and activities during EMT in mammary epithelial cells and in breast cancer cells. Loss and gain of function experiments demonstrate that reduced Klf4 expression is required for the induction of EMT and for metastasis. Genome-wide chromatin immunoprecipitation/deep sequencing (ChIP-Seq) experiments reveal Klf4 binding to the promoters of key EMT genes, including N-cadherin, vimentin, ?-catenin, VEGF-A, endothelin-1 and Mapk8 (Jnk1). The expression of these genes is up-regulated upon the loss of Klf4 expression during EMT. Notably, Jnk1 appears required for TGF?-induced cell migration and EMT. These observations reveal a critical role of Klf4 as a tumor suppressor by directly repressing key EMT genes.
We have furthermore identified Sox4 among the transcription factors that are significantly upregulated in expression and activity during TGF?-induced EMT. Loss and gain-of-function experiments have revealed that Sox4 is required for TGF?-induced EMT. Importantly, Sox4 is also required for breast cancer primary tumor growth and metastasis in tumor transplantation experiments in vivo. Genome-wide ChIP-Seq experiments reveal that Sox4 directly binds to the promoter and activates expression of the gene encoding Ezh2, a Polycomb Group (PcG) complex methyltransferase that trimethylates histone 3 Lysine 27 (H3K27), a major repressive epigenetic modification. Highly comparable to Sox4, Ezh2 is required for EMT and for tumor cell migration and invasion. Together, our results provide a novel mechanism of EMT regulation by the transcription factor Sox4, where in addition to controlling the expression of crucial EMT genes it also contributes to EMT by inducing the expression of Ezh2, a critical epigenetic regulator of chromatin structure and transcription.
We have further investigated the role of two prominent epigenetic modifications - H3K27me3 and DNA methylation - during TGF?-induced EMT of murine mammary epithelial cells. Genome-wide ChIP-Seq analysis of consecutive stages of EMT revealed the dynamics of H3K27me3 at the promoters of key EMT genes and the corresponding changes in their expression. On the other hand, genome-wide analysis of promoter methylation using Methylated DNA Immunoprecipitation (MeDIP) has not revealed significant changes during EMT. The data provide strong evidence that Polycomb-mediated epigenetic remodeling critically contributes to transcriptional reprogramming during EMT.
Characterization of cancer stem cells during EMT and malignant tumor progression
Our group has also investigated the hypothesis that EMT promotes an increase in the number and/or phenotype of CSCs. Recently, several groups have reported that upon induction of EMT in transformed human mammary epithelial cells the whole cell population gains stem cell-like properties, such as the ability to form mammospheres and to cause tumors upon transplantation. We have employed a number of different transformed and non-transformed cellular EMT systems and transgenic mouse models to study the functional contribution of CSC to EMT, tumor progression and metastasis and, conversely, we have aimed at the elucidation whether EMT supports the generation and maintenance of CSC.
In order to establish procedures to identify and isolate CSC during EMT and tumor progression, we have tested a number of established cellular assays commonly employed for the identification and characterization of cancer stem cells in various cellular EMT systems to study the number and quality of potential CSC during EMT and tumor progression. However, flow cytometry analysis failed to reveal any correlation between the expression of commonly used markers of CSC and the EMT process. In addition to these analyses, we tested the presence of drug resistant cells that are able to export specific dyes, such as Hoechst, cells of the so-called side-population. While we observed a moderate increase of drug resistance in cells undergoing EMT, the number of side population cells did not change. We also tested whether cells undergoing EMT have an increased ability to form mammospheres, a hallmark of stem cells. While epithelial cells were able to grow in clusters or adhesive clumps, cells undergoing EMT indeed formed round and smooth spheres that were on average much larger and more homogenous. Next, we tested whether cells undergoing EMT are increased in so-called label-retaining cells, i.e. slowly proliferating cells, another hallmark of stem cells. Using the membrane dye CSFE, we did not observe an increase in label-retaining cells after EMT in the various EMT systems tested. Moreover, we also did not find any difference in the expression of aldehyde dehydrogenase by the Aldefluor assay, another hallmark of stem cells.
In addition, to the commonly employed cellular assays of stem cell identification, we have employed genetic tagging experiments in which label-retaining cells are identified in the MMTV-PyMT transgenic mouse model of breast cancer and in the Rip1Tag2 transgenic mouse models of pancreatic ? cell cancer in vivo. For this purpose, transgenic mice were crossed to express a histone H2B-GFP fusion protein in a tetracycline-inducible manner specifically in tumor cells. Short pulses of tetracycline-induced expression of H2B/GFP are followed by various periods of time before tumor cells are sorted by flow cytometry for cells with high nuclear GFP staining (label-retaining cells). Label-retaining and non-label-retaining cells will be tested for tumorigenicity in transplantation experiments to evaluate whether indeed label-retaining, slowly proliferating cells exhibit higher tumorigenicity as compared to fast proliferating cells. The results from these genetic experiments indicate that it is crucial to employ an inducible system that is absolutely tight and non-leaky in order to be able to identify label-retaining cells in vivo. By employing a variety of different mouse lines expressing the tetracycline-activator specifically in the tumor cells, we have now optimized the tightness of the inducible labeling system and are currently isolating prospective CSC by flow cytometry for further analysis for gene expression and tumorigenicity.
Altogether, our current results suggest that during the process of EMT cells do not obviously turn into CSC. Rather, the extent of such conversion is subtle and the number of CSC seems to be low.
Tumorgenicity of EMT cells is based on increased angiogenic activities
Based on the results mentioned above, we have combined the enrichment of CSC by spheroid formation with the analysis of tumorigenicity and other stem cell tests. We have tested tumorigenicity, that is the number of tumor-initiating cells in cell populations undergoing EMT, by transplanting decreasing numbers of epithelial murine breast cancer cells before EMT and mesenchymal breast cancer cells after having undergone EMT into syngeneic FVB/N mice. The results indicate a marked increase of tumorigenicity of the cells after EMT. This ability is even more pronounced when cells were cultured as mammospheres, indicating that “stemness” and EMT were correlating events. The resulting tumors showed a mesenchymal and invasive phenotype and efficiently formed lung metastasis.
Notably, both primary tumors and metastasis formed by mesenchymal cancer cells were characterized by increased tumor angiogenesis. Gene expression profiling analysis of several cellular models of EMT raised the possibility that with ongoing EMT cells gained the expression of a number of inflammatory cytokines, cytokine receptors, and, notably, a large number of angiogenic growth factors, including VEGF-A, VEGF-C, PDGF, angiopoietin-2, and FGF-2. To explain the substantial difference in tumor vascularization that correlated with the stronger tumor initiation ability of mesenchymal breast cancer cells as compared to epithelial breast cancer cells, we have tested the levels of VEGF family members in cells and tumors. VEGF-A and VEGF-C as well as VEGFR1 are upregulated in cells that undergo EMT. To elucidate the mechanism that could be responsible for such high frequencies in tumor take upon EMT, we have performed experiments inhibiting VEGF signaling either via VEGF-A knockdown in the tumor cells or by pharmacological inhibition of VEGF receptor signaling in transplanted mice. The results indicate that VEGF-A and VEGFR signaling are required for the early tumor onset and the high tumorigenicity of breast cancer cells that have undergone EMT.
These observations were confirmed in a mouse model of multistage pancreatic ?-cell carcinogenesis (Rip1Tag2). Tumor cell-specific ablation of E-cadherin expression in these mice leads to a dramatic increase of EMT in tumor cells and a subsequent increase in lung and liver metastasis. In these mice, tumors that have undergone EMT exhibit a markedly increased tumor vessel density and lack any necrotic areas, while tumors that are still in an epithelial differentiation state are poorly vascularized and show many necrotic areas. Conversely, tumor cell-specific knockout of Snail-1 expression, a potent transcriptional repressor of E-cadherin gene expression, results in diminished tumor invasiveness and with it in reduced tumor vessel densities. These results support the conclusion that an angiogenic program is activated during EMT and tumor progression which is critical for faster primary tumor growth and the formation of tumor metastasis.
Together, these data raise the intriguing possibility that the increased tumorigenicity and, thus, the stemness of cells that have undergone EMT is due to increased angiogenesis, an intriguing possibility that can be therapeutically targeted.
Publications #3, 4, 8, 15, 16, 28, 29, 34, 41, 56, 57, 86, 93, 96 – 101, 105, 108 – 111, 113-115, 118-122

Beneficiary 3 - Erasmus MC
The work of our laboratory within the TuMIC EU consortium focussed on the following objectives:
1. The improvement of methods for defining and isolating CSCs
2. The implementation of these novel methods to answer basic questions about the behavior of CSC populations during metastasis
3. The function of CSC markers in defining CSC properties that are relevant to metastasis
4. The bioinformatic comparision of gene expression profiles and signatures CSCs (and normal SCs) with metastatic potential
5. The functional analysis of S100A4 in affecting Apc-drive intestinal and mammary tumor onset and malignant behavior
6. The role of myofibroblasts in eliciting invasive and metastatic behavior by modulating stemness and EMT through paracrine Wnt and TGF-beta/BMP signaling.
Objectives 1 and 2. Our work on two preclinical mouse models for intestinal and mammary cancer developed in our laboratory, namely Apc1638N/+/KRASG12D and Apc1572T/+, led to the identification and characterization of the corresponding cancer stem cells (CSCs) subpopulations. The CD24hiCD29+ and CD24+CD29hi surface antigen markers were found to enrich the CSC population in Apc1638N/+/KRASG12D intestinal adenocarcinomas and Apc1572T/+ mammary metaplastic squamous carcinomas (a model of triple-negative breast cancer), respectively. Genome-wide expression profiling, label-retaining pulse-chase assay, organoid formation, and subcutaneous/orthotopic/tail-vein transplantation were applied to these specific tumor cell subpopulations and compared with more differentiated tumor cells. Notably, the results indicate that CSCs are in fact complex and heterogeneous cell populations with high and slow cycling stem-like cells coexisting with niche cells, thus reflecting the composition of the normal stem cell niche. These observation have strengthened our definition of CSCs which is largely based on the analogy with the normal stem cell niche: the hierarchical structure of stem cell niches and their functionality during homeostasis and tissue injury is in general conserved in the tumor mass, especially with regard to the dichotomy between fast and slowly (quiescent) cycling stem cells and their dependence on niche cells.
Objective 3. As for the functionality of the markers commonly employed to isolate CSCs from common malignancies, our collaboration with the Sleeman lab led to the conclusion that CD24 is not a functional marker as its genetic ablation does not affect tumorigenesis in the above mentioned murine models.
Objective 4. Expression profiling of the CSCs from Apc1638N/+/KRASG12D and Apc1572T/+, i.e. the two preclinical mouse models for intestinal and mammary cancer respectively, led to important observations and to the development of a mouse CSC-based signature that turned out to have prognostic value among human breast cancer patients. In collaboration with Dr. Decraene, we have derived a CSC signature from the CSC profile of Apc1572T/+ mammary tumors. The signature has a unique predictive value for human breast cancer, demonstrated across all breast cancer cohorts published to date. Accordingly, the mouse mammary CSCs, upon injection into the tail vein, were shown to be able to form malignant lesions which closely resemble the primary tumors in multiple organs. As for the intestinal cancer mouse model (Apc1638N/+/KRASG12D), bioinformatic analysis of the CSCs enriched in the CD24hiCD29+ subpopulation revealed expression of genes both characteristic of cycling stem cells (e.g. Lgr5) but also of Paneth cell (defensins etc.) a cellular type that we found to serve as a source of stem cells in situations of tissue injury. Again, the structure of the normal stem cell niche appears to be conserved in the CSC niche.
Objective 5. The functional role of the S100A4 protein has been thoroughly elucidated by an extensive genetic analysis where the above-mentioned Apc-mutant models for intestinal and mammary tumorigenesis (Apc1638N/+, Apc1638N/+/KRASG12D, and Apc1572T/+) have been bred with mice either null (S100A4-/-) or overxpressing (Tg-Mts1) the S100A4 gene and their progeny analyzed. Unfortunately, no effect could be found on the mammary and intestinal tumor phenotype of these mice when compared with the Apc-only control animals. However, ablation of S100A4 in Apc1638N/+ animals dramatically reduces the incidence of desmoids (aggressive fibromatosis) characteristic of this mouse model and of individuals with a germline APC mutation (affected by the FAP syndrome). We are currently investigating whether S100A4 expression in primary desmoids and established desmoids cell lines overlaps with that of mesenchymal stem cell markers, previously employed to identify the cell of origin of desmoids disease.
Objective 6. Concerning the analysis of the role played by myofibroblasts in eliciting invasive and metastatic behavior by modulating stemness and EMT through paracrine signaling, notwithstanding a jump start in the very first year with promising results (see first period report), we did not pursue anymore this specific line of investigation. This was due to an article published in 2010 from a competing laboratory on the promoting effect of myofibroblasts on Wnt signalling in colon CSCs.
Overall, the highlights of the results obtained by our laboratory in the framework of the TuMIC consortium are relative to the dosage- and context-dependent fashion by which Wnt signalling modulates stemness both in homeostasis and cancer. The latter, together with the role played by infrequently cycling and quiescent stem cells during tissue injury and cancer will form the basis of the investigations to be pursued by our laboratory in the coming years.
Publications #9, 30, 31, 37, 42, 64, 68, 76 – 80, 82, 86, 104, 107

Beneficiary 6 – NKI
Cell polarization is crucial in many biological processes and is required for development and maintenance of proper tissue integrity. Loss of cell polarity results in tissue disorganization and is thought to facilitate both initiation and progression of cancer. Most human cancers are formed from epithelial cells, a cell type in which polarity proteins play crucial roles in maintaining epithelial structures. Dissemination of epithelial tumor cells is accompanied by epithelial-to-mesenchymal transition (EMT). Loss of apical-basal polarity and cell-cell adhesions causes EMT and is associated with metastasis. Nevertheless, functional evidence for an involvement of polarity proteins in mammalian cancer is limited. Given the connection between EMT and stemness properties, in the TuMIC project our task was to analyse the role of polarity signaling proteins in the metastasis-relevant biology of CSCs.
The different forms of cell polarity are regulated by a few conserved proteins that are clustered into three polarity protein complexes, termed the Par, Crumbs, and Scribble complexes. The Par complex consists of Par3, Par6 and an atypical protein kinase C (aPKC). The Crumbs complex consists of Crumbs, Pals1 and PATJ. The Par complex is localized apically in epithelial cells, and together with the Crumbs complex regulates apical domain maintenance. These two complexes act in a mutually antagonistic fashion with the third complex, the so-called Scribble complex, consisting of Scribble, Dlg and Lgl. In epithelial cells this complex is located basolaterally and is necessary for maintenance of the basolateral membrane and basal protein restriction. Evidence is accumulating that implicates deregulated signaling of the three polarity protein complexes in multiple steps of human tumorigenesis.
Cytoskeletal changes are required for all different modes of cell polarity and numerous studies indicate a link between signaling of small RhoGTPases and polarity proteins. Small RhoGTPases, including Cdc42, Rac1 and RhoA, control the cytoskeletal changes in cells by switching between an active GTP-bound state and an inactive GDP-bound state. Downstream signaling by these small GTPases triggers various processes including differentiation, cell-cell and cell matrix adhesions, and cell migration, all subject to intensive cytoskeletal reorganization. The polarity protein Par3 can bind the Rac-activator Tiam1 linking Rac activation to polarity signaling of the Par complex. During the TuMIC project we found that Tiam1 is required for polarized protrusional outgrowth of primary astrocytes by affecting the organization of the microtubule network.
Using mice in which Par3 is deleted in the epidermis, we found that upon induction of skin tumorigenesis, epidermal loss of Par3 resulted in strongly reduced formation and growth of papillomas and mislocalization of aPKC. The aPKC kinase is involved in the progression of various types of human cancer and overexpression or mislocalization of the protein contributes to poor prognosis. These functional in vivo studies indicate that loss of Par3 in the mouse epidermis results in reduced numbers and growth of papillomas in a Ras-driven tumor model, similar to deletion of Tiam1 and Rac1 in this model. Not only do these observations stress the conjunctional action of these proteins, they also provide evidence for a pro-oncogenic function of Par3. Par3 likely mediates its tumor-promoting activity through regulation of growth and survival, since Par3 deletion increases apoptosis and reduces growth both in vivo and in vitro. However, evidence for a tumor suppressive function of Par3 became also apparent from this study as loss of epidermal Par3 results in increased incidence of a Ras-induced cutaneous tumor type, called keratoacanthoma. These tumors arise from a different epidermal compartment and display reduced Par3 expression, indicating a tumor suppressive function of Par3 in this tumor type. They are frequently observed in humans. These data show tumor-promoting and suppressing functions of Par3 depending on the tumor type. Together, our results identify a dual function of Par3 in skin cancer, with both (pro)oncogenic and tumor-suppressive activity depending on the tumor type.
Publications #26, 27, 45, 46, 86

Beneficiary 7 – UFN
Within the TuMIC project we have addressed several aims connected with our main tasks: Analysis of the role of metastasis-promoting genes in the biology of CSCs and preclinical therapeutic studies. We have focused on the metastasis-promoting gene h-prune and examined its effect on the metastatic proclivity of CSCs using animal model of breast cancer. We could show that dipyridamole, one of the first inhibitors of h-prune identified that inhibits the h-Prune cAMP phosphodiesterase (PDE) activity, impairs Wnt signaling pathways in vivo. Moreover, dipyridamole significantly decreased the infiltration of tumor-associated macrophages and myeloid-derived suppressor cells in tumors, and reduced the inflammatory cytokines levels in the sera of the treated mice, findings which are highly relevant to metastatic niche formation. These data suggest that dipyridamole is a promising agent for breast-cancer triple negative patient treatment, as wekk as other cancers that rely on similar signaling pathways.
A major activity of our unit was to develop new derivatives of dipyridamole characterized by the same side groups at the 2,6-positions as dipyridamole, as well as the 4-methoxybenzylamino groups at the 4,8-positions. Screening the activity of these compounds followed by subsequent optimization of the chemical structure resulted in the novel dipyridamole derivative AA7.1. This compound impairs both the cAMP-PDE activity of h-Prune and h-Prune-driven cellular motility in breast cancer cells. AA7.1 has a much higher efficacy compared to dipyridamole. It also inhibits h-prune expression in vitro and impairs the NF-Kb signalling pathway. Animals bearing tumors derived from human and murine breast cancer cell lines (MDA235-T-Luc and 4T1-Luc) that had been implanted in the mammary gland were treated with AA7.1 or dipyridamole. The results showed that AA7.1 has a better efficacy compared to dipyridamole in inhibiting tumor growth and metastasis. These findings are currently being patented.
We generated transgenic mice in which a mouse mammary tumour virus (MMTV) promoter drives the expression of h-prune. While these mice do not develop tumors, they do develop hyperplasia of the mammary epithelium. The MMTV Wnt-1 transgenic mouse is a useful animal model to study breast carcinogenesis. We therefore generated double transgenic animals (MMTV-h-prune-WNT1). These animals showed enhancement of WNT signaling (activated ?-catenin) and enhancement of tumor growth. Currently we are investigating effects of the transgenes on the metastatic microenviroment. These animals are a useful model for testing new drugs that impair mammary tumor growth and metastasis driven by H-prune and WNT-1.
To manipulate the metastatic niche microenvironment using defined factors in vivo, we have used “Bindarit”, an anti-inflammatory indazolic derivative that can inhibit the synthesis of MCP-1/CCL2. We found that Bindarit can modulate cancer cell proliferation and migration, mainly through negative regulation of TGF-? and AKT signaling, and it can impair the NF-kB signaling pathway through enhancing the expression of the NF-kB inhibitor IkB-?. In vivo administration of bindarit resulted in impaired metastatic disease in prostate cancer xenograft mice (PC-3M-Luc2 cells injected intra-cardially) and impairment of local tumorigenesis in syngeneic Balb/c mice injected under the mammary gland with murine breast cancer cells (4T1-Luc cells). In addition, bindarit treatment significantly decreased the infiltration of tumor-associated macrophages and myeloid-derived suppressor cells in 4T1-Luc primary tumors. Thus this compound is able to inhibit tumor progression and metastasis formation in animal models of breast and prostate cancer through inhibiting formation of the niche microenvironment. These results validate the metastatic niche as a therapeutic target.
We have structured the H-prune C-terminal region by Nuclear Magnetic Resonance methodology and defined the region that is able to bind and regulate GSK-3? function. The structural information about the C-terminal portion of h-prune can be used in the near future to develop new drugs that impair its interactions with other proteins that mediate its metastatic effects. We have further identified how h-prune is sustains and enhances WNT signaling.
Publications #5 -7, 13, 17, 23, 35, 36, 60, 66, 69, 93, 94, 117

Beneficiary 8 - DCS
In the TuMIC project, our man goal was to analyse factors that determine the formation and function of the pre-metastatic niche. In particular, we focused on the S100A4 protein, which we hypothesized was likely to play a central role in determining the metatatic niche. Our activities in the TuMIC project are outlined below.
1. The extracellular cytokine-like form of S100A4 promotes cell survival, differentiation, proliferation, motility, invasion and angiogenesis. It affects important intracellular signaling cascades, stimulating, for example, the phosphorylation of mitogen-activated protein kinases (MAPK), activation of NF-kB and calcium signaling. However, the molecular mechanisms and the potential surface receptor(s) transmitting these effects remain unidentified. To identify new S100A4 extracellular targets we screened a phage display peptide library against the extracellular active forms of S100A4. The screening revealed a number of independent consensus sequences, which specifically bound S100A4 with high affinity and inhibited S100A4-induced neuron differentiation. The most abundant sequence motif (KCC) showed similarity to the tertiary structure of mature ligands of the EGF receptor, implicating S100A4 in ERGR signaling. In vitro binding studies confirmed the S100A4 interaction with different ligands of the EGFR family with the highest affinity to Amphiregulin. Moreover, S100A4 was able to enhance Amphiregulin-induced proliferation of mouse embryonic S100A4-/- fibroblasts. Our findings suggest S100A4 enhances EGFR/ErbB2 signaling, and might point out a pathway whereby extracellular S100A4 protein promotes cell proliferation, tumor progression and metastasis.
2. Tumor and stroma cells interactions are essential for the progression of primary tumors and metastasis to distant organs. Recent data strongly linked CD4+ T cells with stimulation of metastasis. T cells secrete growth factors and cytokines, which support tumor angiogenesis and recruit macrophages, thus accelerating tumor growth and metastasis. We used the MMTV-PyMT tumor model to study the influence of S100A4 expression at early stages of tumor development on metastasis formation. We obtained evidence that S100A4 released into the tumor microenvironment at a stage of malignant transition from adenoma/MIN to early carcinoma induces massive infiltration of T cells and release of specific cytokines, leading to increased pulmonary metastases. Infiltration of T cells in pre-metastatic lungs was induced by S100A4+/+ fibroblasts, generating a favorable niche microenvironment for metastasis formation.
3. We found that fibroblast-derived factors, different from those produced by fibroblasts at the site of primary tumor, could play a key role during metastatic organ colonization. In collaboration with the Sleeman laboratory we have identified five genes whose expression in fibroblasts is modulated by tumor-derived soluble factors and which could potentially play a role in metastatic organ colonization. One of these genes, Fibulin 5 (Fbln5), was analyzed in detail. We found that tumor cells co-injected with fibroblasts that constitutively express Fbln5 were unable to form tumor colonies in peripheral organs. At the same time, these fibroblasts failed to invade 3D Matrigel. We suggest that FBLN5 might block the ability of tumor cells to colonize secondary organs by modulating the structure of ECM and suppressing the invasive activity of fibroblasts. Accordingly, we found that Fbln5 expression in fibroblasts suppressed Mmp9 expression, inhibited fibroblast invasion, and correlated with reduced metastasis formation. Our data suggest that factors produced by disseminating tumor cells trapped in the peripheral organs, or even released from the primary tumor itself, serve to down regulate Fbln5 in fibroblasts within the peripheral organs. In turn, this results in upregulation of metalloproteinases, ECM remodeling, stimulation of angiogenesis, invasion of fibroblasts and metastasis formation.
4. In collaboration with the Sleeman laboratory we have investigated the effect of S100A4 on tumor cells to stimulate expression of genes associated with inflammation. We showed that two of these genes, namely SAA1 and SAA3, are transcriptional targets of S100A4. SAA proteins stimulate transcription of cytokines and MMPs, migration and adhesion of tumor cells. Overexpression of SAA in tumor cells profoundly influences the metastatic properties of the cells, both in cell culture assays and in animal models. Consistently, coordinate expression of S100A4 and SAA in human colorectal patient samples was strongly predictive of poor prognosis. The data collectively suggest that S100A4 stimulates metastasis through triggering an inflammatory response, mediated at least in part by SAA proteins, and imply that the induction of SAA expression is a major means by which S100A4 exerts these effects.
Publications #1, 32, 40, 43, 44, 48 – 50, 65, 67, 72, 116

Beneficiary 9 - VIB
The poor prognosis in epithelial neoplasia is associated with the acquisition of motile or invasive properties by the cancerous cells. Epithelial-mesenchymal transition (EMT), an essential process during morphogenesis, represents a natural process to reshape epithelia for cellular movement. This process is also induced through the abnormal activation of different signaling pathways during cancer progression and organ fibrosis. A crucial role in this phenotypic conversion is played by a subset of transcription factors such as Snail, Slug, SIP1/ZEB2 and ?EF-1/ZEB1. During EMT these transcription factors are able to reprogram the epithelium at the molecular level with new biochemical instructions associated with invasive and metastatic cells. One such important gene downregulated by Snail and SIP1/ZEB2 is the invasion tumour suppressor E-cadherin. During the TuMIC project we have specially focused on how these EMT-inducing transcription factors enable non-invasive cancer cells to acquire the features needed to execute the entire invasion-metastasis cascade. We have also studied new functional interactors and target genes of these transcription factors, as they likely possess key activities to activate motility, to inhibit apoptosis, to enhance invasion and to dismantle the local basement membrane.
ZEB2/SIP1 has identified by us as a potent repressor of the invasion suppressor protein E-cadherin. In further studies, we set out to identify the relevance of ZEB2 to human cancer progression. We found that ZEB2/SIP1 expression is enhanced in ER- and PR- negative breast tumours. Repression of ZEB2 expression in basal breast cancer results in enhanced adhesion and reduced migration, invasion and metastasis formation. Functional analysis of ZEB2 knockdown in basal breast cancer cells using gene expression profiling resulted in the identification of a ZEB2 metagen signature of ZEB2-regulted genes, which predicts poor outcome of breast cancer patients. Importantly, further analysis revealed that established breast cancer stem cell markers are affected upon ZEB2 expression modulation, suggesting that ZEB2 endows stemness properties on cancer cells. These research data have resulted in 2 patent applications.
We also experimentally assessed a module network inferred from both miRNA and mRNA tumour expression data, using a novel developed network inference algorithm based on probabilistic optimization techniques. We could show that several miRNAs are predicted to be statistically significant regulators for various modules of tightly co-expressed genes. We further designed a set of experiments to test the assignment of miR-200a as the top regulator of a small module of nine genes. Our results strongly suggest that miR-200a regulates this gene module via the EMT inducing transcription factor ZEB1. Interestingly, this particular gene module is most likely involved in epithelial homeostasis and its dysregulation might contribute to malignant process in cancer cells.
Skin cancer is the most common cancer in man. We found that the EMT-inducing transcription factor Snail is overexpressed in a significant proportion of different human skin cancers. To examine the role of Snail in skin cancer, we introduced skin-specific Snail expression into a mouse tumor model based on skin-specific knockout of p53. Combined transgenic expression of Snail and p53 loss resulted in accelerated development of invasive and metastatic skin carcinomas. In addition, Snail mice are more susceptible to cancer formation through chemical carcinogenesis. Most surprisingly, Snail transgenic expression results in spontaneous skin cancer formation. Our novel data suggest that Snail changes the normal homeostasis of hair follicle stem cells to drive differentiation towards sebaceous gland amplification. Together our results suggest that transgenic Snail expression contributes to both skin tumor initiation and metastasis.
Finally we have explored the regulatory network that controls EMT and thereby drives malignant progression. We set up a genetic screen using non-metastatic, human breast tumour cells with a conditional EMT phenotype that was screened with an siRNA library. Thereby we have identified a series of proteins that can block EMT-stimulated cancer cell migration and invasion. Several aggressive cancer cell lines derived from different epithelial tissues depend on the endogenous activity of these new EMT regulators to migrate and invade efficiently. We would like to understand better how these EMT regulators control malignant behavior. To identify signaling pathways and direct targets of these novel EMT-inducing factors, genome-wide expression analyses and chromatin immunoprecipitation sequencing experiments have been planned. As EMT has been suggested to contribute to cancer stem cell properties we are focusing on how these factors can contribute in cancer stem cell formation.
Publications #11, 24, 25, 33, 38, 74, 75, 86, 102, 103

Beneficiary 10: Actelion
Our research activities in the TuMIC project focused on drug discovery and development based on TuMIC findings. We aimed to target Lrg5 as a putative regulator of CSC properties. We also carried high throughput screens to interfere with the induction and maintenance of EMT.
Lgr5/GPR49 as a drug target and screening for agonists/ antagonists
Intestinal cancer is initiated by Wnt pathway-activating mutations in genes such as adenomatous polyposis coli (APC) or ?-catenin. The cell of origin of such tumors has remained elusive until recently when lgr5 was established and validated as a marker for CSC of colon adenocarcinoma. Hence, in mice, activation of the APC pathway can only give rise to full intestinal tumors when performed in lgr5 expressing cells. Furthermore, lgr5 expression is tightly associated with APC mutation and accumulation of cytoplasmic ?-catenin, two characteristics of 80% of colon cancers. There, lgr5 is found over-expressed in the core of the tumor and in patches that comprises two third of tumors, the expected pattern for a CICs marker. In addition to colon tumors, increased expression of lgr5 has been observed in ovarian tumors (53%), hepatocellular carcinoma (47% total and 90% of HCC with mutation in the b-catenin pathway) and in basal cell carcinoma (95%).
In the last three years, in addition to its role as a marker, lgr5 function has been shown to play a crucial role in cancer development: for example, while over-expression of lgr5 in tumor cells increases their growth when transplanted sub-cutaneously in several models, siRNA-mediated knock-down of lgr5 leads to strong decrease in cell proliferation and increased apoptosis. Altogether these data indicate that lgr5 offers a unique chance to target chemotherapy-resistant CSCs in colorectal carcinoma
Since lgr5 is an orphan G protein-coupled receptor (GPCR), we set out a strategy to identify agonists which would allow then to screen for antagonists or may be potent enough to internalize receptors are therefore act as functional antagonists. Three approaches were undertaken: 1) test effects of putative ligands (based on literature, patents and other GPCR family members), 2) phage display library screen to identify active peptides, 3) High Throughput Screening (HTS) with a chemical compound collection to identify small molecules as agonists.
As part of our TuMIC activities, we successfully set up several HTS compatible assays to monitor lgr5 activation: ?-arrestin assay using the Tango system from Invitrogen (several independent cell lines generated), a HTRF cAMP assay (cisbio, France) and an impedance assay (ACEA/ Roche Applied Science). Of the 15 putative ligands tested, none led to activation of lgr5. The second approach (phage library screen) led to the identification of 19 peptides that specifically bind to lgr5. However, these peptides do not activate the receptor in all the assays tested. For the third approach, more than 300 000 small molecules were screened using a ?-arrestin recruitment assay. 97 hits leading to >2 fold activation of lgr5 were identified. However, after numerous secondary assays where the specificity of these hits was evaluated, no small molecule could be selected for further use in drug discovery and development.
Screening for EMT regulators
Epithelial-mesenchymal transition (EMT) has been described as a reversible phenotypic trait of cancer cells associated with metastasis, stem cell-like behaviour and therapy resistance. Epithelial cancer cells can undergo EMT as a result of environmental stimuli and the transition is associated with specific alterations in gene and protein expression as well as cell behaviour.
As a model system to identify genes and chemical compounds which are essential modulators of the EMT process we have selected non-transformed mouse mammary gland epithelial cells (NMuMG cells). These cells undergo EMT when exposed to TGF-? and the process underlies molecular mechanisms identical to human cells. We developed a high content screening assay system using the high content reader, ImageXpress from Molecular Devices to quantify several EMT markers such as stress fibres, focal adhesion patches and fibronectin deposition as well as cell number. The assay system allows the identification of genes and compounds which modulate one or several EMT parameters in the same assay well.
The robustness and stability of the assay was tested in order to allow screening of large compound sets. For all 3 parameters, focal adhesion, stress fibres and fibronectin deposition z’ values of >0.6 were obtained indicating suitability for high throughput screening. An siRNA library consisting of about 1640 siRNAs and a collection of 2700 small molecule compounds from commercial sources have been selected for a first screening campaign. Screening is presently ongoing.
Publications #59

Beneficiary 11: Institut Curie
Using microarray technology, a large amount of data can be collected, but the downstream processes developed to analyze the expression profiles remain complex. It is crucial to have access to validated, controlled and updated annotations files for all interrogated probes. Moreover, the amount of biological material from cell sorting or laser capture microdissection is much too small to perform microarray studies. To address this issue, RNA amplification methods have been developed to generate sufficient targets from picogram amounts of total RNA to perform microarray hybridisation. In the context of the TuMIC project we focused on the use and analysis of these methods to achieve the objectives of the project. Specifically, we have carried out the following tasks:
1. Comparison of RNA amplification methods
2. Development of new bioinformatics tools to (re)annotate the microarrays dedicated to gene expression analysis
3. Collaboration with other partners to collect and analyse gene expression data.
Comparison of RNA amplification methods
To validate the use of RNA amplification methods we have evaluated and compared four commercial protocols for amplification of picograms amounts of input RNA for microarray expression profiling. The quantitative and qualitative performances of the methods were assessed. Microarrays were hybridized with the amplified targets and the amplification protocols were compared with respect to the quality of expression profiles, reproducibility within a concentration range of input RNA, and sensitivity. The results demonstrate significant differences between these four methods and the WT-Ovation pico system proposed by Nugen appears to be the most suitable for RNA amplification.
Development of new bioinformatics tools
We have created a new microarray re-annotation protocol called Validated Chip Annotation (VCA) allowing heterogeneous datasets comparison (intra- and inter-technologies; intra- and inter-species). This tool has been developed for all microarray technologies (e.g. Affymetrix, Agilent, Illumina). For each microarray, we have produced a specificity score in order to control which gene, transcript(s), exon(s) and CDS are targeted by the probes/probesets. As a result, native or custom annotation files have been re-annotated including a quality control. This new microarray annotation tool gives the possibility to improve the quality of the data analysis and to explore the biology of the cell using the large number of available bioinformatics tools.
Collaboration with TuMIC partners
We have analysed a novel mouse model, Apc1572T/+, generated by the Riccardo FODDE laboratory and characterized by a highly penetrant predisposition to metaplastic mammary adenocarcinomas and pulmonary metastases. The heterogeneous histological composition of the Apc1572T tumors makes this model an ideal one to study the role of cancer stem cell (CSCs) self-renewal and differentiation in tumor progression and metastasis. To this end, we employed the CD29hiCD24+ cell surface antigens to identify and characterize a subpopulation of CSCs capable of recapitulating tumorigenesis when transplanted at low multiplicities in vivo, and of forming self-renewing organoids when cultured in vitro. The gene expression analysis of the stem-like and differentiated subpopulations from normal and neoplastic mammary tissues revealed that normal stem cells are more similar to tumor cells than to their own differentiated progenies and an “epithelial-mesenchymal transition” can be observed in the normal cells but not in the tumor cells.
Finally, a particularly relevant and important result is the identification of a specific CSC gene signature that significantly correlates with overall survival, relapse and distant metastasis-free survival of breast cancer patients when published data sets are analysed. The predictive power is in the 125 upregulated genes (down-regulated genes do not contribute), and is specific for breast cancer but not other tumor types. The signature is earmarked by genes involved in EMT and p53 signalling. There is virtually no redundancy with other previously published CSC signatures, and thus associates expression of many new genes with the CSC phenotype.
Publications #18, 74, 86

Beneficiary 13 –UCAM (initially Beneficiary 4 – CR-UK)
Clonal growth is a functional marker of cultured human interfollicular epidermal (IFE) stem cells and has been used to identify cell surface markers. Markers that enrich for cells with high colony-forming ability include ?1 integrin ECM receptors, the Notch ligand Delta-like 1 (DLL1), the transmembrane proteoglycan MCSP (NG2; CSPG4) and LRIG1, a transmembrane protein that negatively regulates the EGF receptor (EGFR). Cells that express these markers lie in clusters in the IFE basal layer that have a specific location relative to the topology of the underlying basement membrane. Whole mount labeling has revealed that cells within the clusters are less frequently proliferating than neighbouring cells and that the onset of differentiation occurs outside the clusters.
Although immunolocalization studies show that ?1 integrins, MCSP, LRIG1 and DLL1 are co-expressed, we observed that, at the single-cell level, this is not always the case. We set out to examine the significance of this heterogeneity by performing single cell gene expression profiling. We found that human epidermal stem cells fall into two distinct clusters (D+, D-), one of which contains cells that express DLL1, including all those that co-express LRIG1. The two clusters have distinct transcriptional profiles, indicating that intercellular differences in gene expression cannot be attributed solely to random stochastic fluctuations.
The different populations of cells defined on the basis of DLL1 and LRIG1 expression are not functionally equivalent. DLL1+LRIG1+ cells are particularly sensitive to the differentiation stimulus of restricted spreading and on differentiation give rise to DLL1-LRIG1- cells, suggesting that they are arranged in a hierarchy. Whether or not DLL1+ cells can reversibly transition between the single (DLL1+LRIG1-) and double positive state remains to be investigated.
Analysis of gene expression profiles in D+ and D- cells shows that expression of DLL1 does not impact on Notch signaling activity. Rather, DLL1 expression results in increased expression of ?1 integrins and genes associated with receptor tyrosine kinase signaling and endocytosis. The marker that discriminated best between the two clusters identified by global gene expression profiling was syntenin, which not only interacts with the cytoplasmic domain of DLL1, but also ?1 integrins and MCSP, two further stem cell markers. Our data support a model in which the intrinsic transcriptional profile of D+ stem cells controls niche responsiveness by rendering them more adhesive to the underlying ECM and less responsive to EGF and, potentially, other growth factors. This protects them from differentiation and reduces their proliferation rate. Thus the intrinsic and extrinsic regulators of epidermal stem cells are not independent, but are tightly interconnected.
In our previous single cell gene expression profile studies we identified FRMD4A as a human epidermal stem cell marker. We selected it for further analysis because it was consistently upregulated in human SCC cell lines. We have now shown that in human epidermis, FRMD4A is only expressed in the basal cell layer, where the stem cells reside. In contrast, FRMD4A is upregulated in HNSCCs, regardless of differentiation status, and high levels correlate with poor prognosis.
Our functional studies place FRMD4A downstream of CD44 in the Hippo pathway. CD44 ligation with hyaluronan resulted in loss of nuclear FRMD4A. Conversely, knockdown of FRMD4A resulted in nuclear accumulation of YAP.
To investigate whether targeting FRMD4A is an effective treatment for oral SCCs we developed a xenograft model using low passage cells cultured from tumours. The cells are transduced with a YFP-luciferase lentiviral vector and implanted into the tongues of immunocompromised mice. Primary tumour growth and metastasis to the liver and lungs has been evaluated. Our in vivo studies indicate that targeting FRMD4A is a promising new approach for treating SCC, both primary tumours and metastases, and also suggest possible therapeutic options, such as injection of hyaluronan.
One issue of interest is whether FRMD4A, a marker of normal epidermal stem cells, is also a cancer stem cell marker. FRMD4A is upregulated in HNSCCs and knockdown stimulates terminal differentiation. It is widely expressed by cells throughout the tumor mass, and while this could potentially reflect expansion of stem cells bearing oncogenic mutations, it is surprising that it does not correlate with differentiation status. The same has been observed for another marker of normal epidermal stem cells, the ?1-integrin subunit. Thus, rather than defining specific subsets of tumor cells, two stem cell markers, FRMD4A and ?1-integrin, show a generalized upregulation in HNSCCs.
Publications #19, 20, 21, 39, 51,

Beneficiary 14 - Université de Fribourg (initially Beneficiary 5 – CePO)
During the TuMIC funding period, we have achieved three main results relevant to breast cancer diagnosis, treatment and metastasis prevention. The first two achievements might be immediately translated into clinical studies, and we are investing efforts into this direction. The third achievement (mechanisms of brain metastasis) is still at an experimental level. However, we are now testing drugs that may used to inhibit some of the identified targets to prevent breast cancer brain metastasis formation or to treat established metastases.
PlGF mobilizes and educates bone marrow-derived monocytes to acquire tumor-promoting properties
Tumor angiogenesis is key to tumor growth and metastasis. Tumor-mobilized bone marrow derived cells significantly contribute to tumor progression. In particular, tumor-mobilized bone marrow-derived CD11b+ myeloid cells promote tumor angiogenesis and metastasis, but how and when these cells acquire these properties is not fully elucidated. Using a syngeneic and orthotopic model of breast cancer we have discovered that CD11b+ myelomonocytic cells develop proangiogenic properties during their differentiation from hematopoietic progenitors, and that Placenta Growth Factor (PlGF) is critical in promoting this education. Hemeotopoietic progenitor cells supplemented with conditioned medium derived from breast cancer cell lines or PlGF, but not from normal breast epithelial lines, generate CD11b+ cells capable of inducing angiogenesis. Inhibition of PlGF or its receptor abolished these effects. Silencing of PlGF in tumor cells prevented the generation of proangiogenic activity in circulating CD11b+ cells, inhibited tumor blood flow and slowed tumor growth and progression. Peripheral blood of breast cancer patients at diagnosis, but not of healthy individuals, contained elevated levels of PlGF and circulating proangiogenic CD11b+ myelomonocytes.
Taken together our results show that cancer cells can program proangiogenic activity in CD11b+ myelomonocytes during differentiation of their progenitor cells in a PlGF-dependent manner. These findings impact breast cancer biology, detection and treatment. In particular, since the CD11b+ myelomonocytes have a proangiogenic profile while they circulate in the peripheral blood, they could be used for the early detection of breast cancer, or monitoring during therapies.
The KitK/c-Kit loop promotes metastasis of breast cancers relapsing after radiotherapy
Adjuvant radiotherapy provides survival advantages compared to surgery alone and it is nowadays a standard treatment in the management of several cancers, including breast. Despite progress in the delivery mode, loco-regional post-radiotherapy relapses still occur in a fraction of treated patients. Relapses occurring within a pre-irradiated area are associated with an increased risk of local invasion, metastasis formation and a poor prognosis. In experimental models, tumors developing in pre-irradiated tissues are more invasive and metastatic compared to tumors growing in non-irradiated tissues, a condition also referred to as tumor bed effect. In spite of its clinical relevance, the cellular and molecular mechanisms underlying the tumor bed effect are still not fully elucidated. We used the well-characterized 4T1 syngeneic, orthotopic, breast cancer model to study the effect of mammary tissue irradiation on tumor progression and to identify novel mechanisms of post-radiation metastasis. We observed that tumors growing in previously irradiated mammary tissue have reduced angiogenesis, are more hypoxic, invasive and metastatic to lung and lymph nodes compared to tumors growing in non-irradiated mammary tissue. We demonstrated that increased metastasis involves the mobilization of a specific subclass of myeloid bone marrow-derived cells through though HIF1-dependent expression of Kit ligand (KitL). KitL, also known as stem cell factor (SCF) or Steele factor, exists as a cell surface and as a secreted protein. KitL binds to c-Kit, a tyrosine kinase receptor of the PDGF-R family that play an important function during development, and late on in hematopoiesis. KitL secreted by tumor growing in a preirradiated tissue caused the mobilization from the bone marrow of CD11b+ cells expressing c-Kit, the receptor of KitL. These cells home to primary tumors and pre-metastatic lungs, where they contribute to the formation of a premetastatic niche. Pharmacological inhibition of HIF1 or KitL in tumor cells and inhibition of c-Kit with an anti-c-Kit blocking antibody or with a tyrosine kinase inhibitor prevented the mobilization of CD11b+Kit+ cell and attenuated metastasis.
This work defines KitL/c-Kit as a previously unidentified axis that is critically involved in promoting metastasis of 4T1 tumors growing in previously irradiated mammary tissue. This work has two immediate translational implications. Firstly, circulating CD11b+Kit+ cells might act as biomarkers to identify patients at risk for post-radiation metastasis. Secondly, inhibition of c-Kit could be an attractive novel therapeutic approach to prevent metastasis in breast cancer patients with local relapses after radiotherapy. Because of their immediate clinical relevance, translational studies aimed at validating these results in patients are warranted.
Mechanisms and genes of breast cancer metastasis to the brain
Between 10 and 15%% of patients with metastatic breast cancer develop brain metastasis late during the disease course, and after metastases have appeared in the lung, liver, and/or bone. This frequency has recently risen to approximately 25-30%, as a consequence of improved disease control at other sites, prolonged survival and therapy-induced changes in tumor biology. Treatment options for brain metastases are palliative and have limited efficacy. The mean one-year survival rate from the time of diagnosis is approximately 20%. In view of the rising frequency of breast cancer metastasis to the brain and the dismal prognosis of affected patients, there is an urgent need to develop novel and more effective therapeutic modalities. In order to contribute to this goal we established an orthotopic, breast cancer metastasis model to the brain in immunocompetent mice based on 4T1 murine breast cancer cells. This is the first model that robustly recapitulates all metastatic events leading to brain metastasis from the primary tumor site. Using this model we have already identified over 200 genes that are differentially expressed in metastatic vs non-brain metastatic cancer cells. By interrogating human databases we have identified clinically-relevant genes associated with reduced brain metastasis survival in human patients. Inhibition of 7 of these genes lead to reduced brain metastasis formation in our model. The functional roles of the genes in promoting brain metastasis, except for 6GALNAC4/ST6GALNAC5, which has been investigated by the group of J. Massagué, has not been recognized to date. As these genes are associated with reduced brain metastasis-free survival, they represent bona fide novel candidate therapeutic targets to inhibit brain metastasis.
In short, we have established the first spontaneous orthotopic model of breast cancer metastasis to the brain in immunocompentent mice, and have identified a number of clinically relevant genes associated with reduced brain metastasis-free survival in cancer patients, and essential for brain metastasis formation in our model. This model represents a major improvement of currently available models of brain metastasis and will serve as the basis for the experiments proposed here. We are currently seeking intellectual property protection (patenting) of the identified genes for diagnostic/predictive purposes.
Publications #14, 47, 53 – 58, 61 - 63, 81, 86, 92, 118,

Potential Impact:
The potential impact, main dissemination activities and exploitation of results.

Potential impact
The long-term strategic goal established by the Lisbon European Council in March 2000 is for Europe to become the most competitive knowledge-based economy in the world by 2010. In January 2002, the Commission adopted a Strategy for Europe on Life Sciences and Biotechnology (COM(2002)27) which sets out a comprehensive roadmap up to 2010 to take the European Union towards the Lisbon objectives. The work programme Cooperation Theme 1 Health is a key component of the practical outworking of this strategy, in which the Commission has identified key research areas and topics that should be supported in order to achieve these goals. The TuMIC project directly addressed one of the topics in the work programme Cooperation Theme 1 Health, namely call (part) identifier HEALTH-2007-2.4.1-6: “Understanding and fighting metastasis” The TuMIC project has therefore contributed and will continue to contribute directly to the impact of the work programme by fulfilling some of its key objectives. Specific areas of impact include:
1. Improvement of the health of European citizens. Cancer is a major health issue, as exemplified by breast cancer and colorectal cancer, two of the types of cancer we have studied in the TuMIC project. Cancer can be highly gender relevant, as in the case of breast cancer. Breast cancer is the most often diagnosed female cancer, and the majority of cases are already invasive at diagnosis. More than 17% of cancer deaths result from breast tumors. Colorectal tumors are second only to lung cancer as the highest cause of death due to cancer in Europe. Breast and colorectal cancer are therefore a major societal problem. Treatment of breast cancer involves radical and disfiguring surgery, often with long-term side effects such as the development of lymphedema of the arm, and radiotherapy and chemotherapy, again associated with severe side effects. Surgery for colorectal cancer can lead to bowel and bladder dysfunction, and a permanent colostomy may be necessary, dramatically impairing the quality of life for patients. Metastases to organs such as bone and brain are major causes of suffering in terminally ill patients.
The incidence of breast cancer increases sharply between the ages of 30 and 50 meaning that many women in the prime of life are affected by this disease. Colorectal cancer tends to have a later onset, but nevertheless significant numbers of patients are diagnosed during the economically active period of their lives. Not only does this mean that many families are traumatised, but it also has severe economic consequences, removing economically active individuals, particularly women in the case of breast cancer, from society. Further economic consequences arise as a result of the high health care costs associated with treating cancer patients.
Clearly improvements in the treatment and management of cancer would have impact on both health and the economy. The TuMIC project has increased our understanding of the process of metastasis, and the role of CSCs and organ microenvironments that support metastasis formation. Drug discovery and preclinical studies has built on these findings, and has identified tools that will contribute to improved clinical decision-making, prognostic evaluation and therapy of cancer. In the long term, if the TuMIC project is to impact broadly on the health of European citizens, then its findings and technologies will need to become widely distributed and implemented in the clinical setting. This will require robust dissemination and licensing activities, commercialization, and education of key stakeholders involved in the clinical management of cancer as outlined above. It will also require clinical trials that are beyond the scope of the TuMIC project,
2. Increased competiveness and boosted innovative capacity of European health-related industries and businesses. Enhanced competitiveness in the European life science and biotechnology sector continues to be crucial if Europe is not to further fall behind main competitors such as the US. The way forward in this area is to command and expand the knowledge base and transform it into new products. A key measure for achieving this is the integration and strengthening of the European research area. The TuMIC project has had impact in this area in several ways:
• Cooperation and integration of research efforts are important stimulators of increased competitiveness. The development of durable networking within the TuMIC consortium will last well beyond the funding period. It has generated a large critical mass of complementary expertise, and continuing synergistic research efforts will increase our competitiveness in the international arena.
• Competitiveness has been increased by proper management of the knowledge base to ensure that knowledge is protected and converted into economically relevant products. The TuMIC consortium plans several measures to ensure that the project impacts competitiveness in this way beyond the lifetime of the funding period:
o Patenting of findings initiated in the TuMIC project.
o Continuing involvement of technology transfer experts in managing the knowledge.
o The development of medically relevant tools out of the knowledge generated beyond the planned TuMIC project.
o Involvement of an SME (Actelion) working in the topic area who are directly exploiting the research findings.
o Continued public dissemination of knowledge generated through publications, international conference, web site, education and training.

3. Other areas of impact
• TuMIC finding will impact on the area of the health of the aging population. Cancer is a disease whose incidence increases dramatically in old age. The insights, tools and methods we have developed in the TuMIC project will greatly improve the chances of successful treatment through preventing metastasis or clearing metastases out of patients. In turn this will impact on the quality of life of older people.
• Breast cancer is a disease that overwhelmingly affects women. This gender-relevant aspect has been fully taken into account in the TuMIC project. In terms of research, TuMIC has strived towards balanced representation of women and men at all levels in the research project, and implemented a robust programme to address gender-relevant issues. Together, this means that TuMIC hashad a strong impact in various gender-relevant areas.
• The involvement of an SME (Actelion) in the TuMIC project has had an impact in the area of SME-relevant research by providing significant financial support to the SME. The expertise of Actelion in drug discovey and development and access to its resources and pipeline of oncology projects has and will continue to play a major role in the exploitation of the basic research findings of the TuMIC project.
• Dissemination activities are an important instrument for ensuring impact of research findings. part of the “Approach” towards implementation of the objectives of the work programme, not only for the research community and key stakeholders but also with a broader public. As outlined above, TuMIC will continue a robust series of dissemination actions beyond the timeframe of the TuMIC funding period.

Response to ethical requirements.
• The consortium should establish an ethical committee and should consider the possibility of an ethical advisory board.
• We established an ethical committee that met each project management meeting. Given the extensive local infrastructure and support for ethical issues available to the participants, an ethical advisory board was not deemed necessary.
• The Consortium should mention the issue of incidental findings, and this can be problematic. In case of relevant incidental findings, the informed consent form must take into account this problem, e.g. in case a disease gene is discovered. A special operation procedure for how and when to inform the patient and to provide genetic counseling should be established.
• Again, these issues are adequately regulated at the local ethical review board level. Patients are informed during the informed consent procedure about the procedure used in case the research uncovers health issues incidentally. Informed consent is agreed in the form of a written document, in which all of the relevant issues are included.
• Regarding patient data storage it must be clarified what kind of identifiable data are going to be held and for what reason. It is therefore necessary to clarify on the policy of data handling, with full description of the kind of identifiable data that are foreseen to be kept and the reason why this is necessary. Compliance with ethical approvals must be demonstrated. This should also be part of the protocols that are going to be written for (a) sampling (blood or tumors) done during this project (b) clinical trials in which this project is going to part of, either as co-investigator or as study designer. This request for data handling clarification also extends to potential exchange of patient data within the consortium and the origin and use of patient samples that must comply with the ethical approvals issued by local authorities.
• Data handling issues are part and parcel of applications to local ethical review boards, as are study protocols. These issues are reviewed by the review boards and form an important component of the consideration as to whether permission for collection and use of samples is given. Thus (as per the principle of subsidiary) these issues are thoroughly and adequately dealt with at the local level by the relevant authorities. No exchange of patient data within the consortium occurred.
• Regarding ractical issues of work package 6, a task 5 should be added in this WP. Ethical issues of clinical trials that are going to be designed should be closely considered by both the specific partners and the to be established ethical committee. This applies also to novel collaborations as outlined in task2 of WP6 and the very diffuse statement of testing the anti-metastatic potential of cilengitide in pilot studies studies in patients. A deliverable of identified ethical issues should be added including clarification how each of these has been addressed. This deliverable must be provided to the EC, together with the study protocols including clarification that relevant local IRB or IRC has given clearance before the trial starts.
• No clinical trials were initiated during TuMIC, therefore this point was not followed up.
• On the topic of Use of Animals, the Consortium is clearly aware of the main ethical issues and Guidelines, and of the three R’s. Nevertheless, information should be provided endpoints and on how the animals shall be “sacrificed”, and by whom (competent personnel is essential).
• These issues are thoroughly and adequately dealt with at the local level by the appropriate authorities. Information regarding endpoints, how animals are sacrificed and by whom forms an important component of applications to the local authorities for permission to perform experiments. Evaluation of these issues is intrinsic to whether or not permission is given. Thus the points raised by the ethical review are stringently controlled at the level of the local review boards who have the authority to issue permission for animal experimentation.

Main dissemination activities and exploitation of results

TuMIC website. The TuMIC web site ( acts as an important instrument to present the TuMIC project to the outside world, as well as to aid internal communication in the TuMIC project. The web site has been continually updated, for example to include the latest publications from TuMIC participants. The internal web pages include lists of resources and reagents available to TuMIc members, experimental protocols, and copies of the powerpoint presentations from the Kickoff Meeting and the project management meetings.
Publications. TuMIC participants have published an impressive list of papers that whose content has been supported by TuMIC funding. In total, 122 publications in peer-reviewed journals have been produced. These are listed elsewhere in the report. In addition, TuMIC members have advertised their work in a review series published as a special issue of Seminars in Cancer Biology entitled “Novel Concepts in Cancer Metastasis”.
Conferences. TuMIC findings have been presented at a wide rage of international conferences. In addition, TuMIC cooperated with other EU-funded consortia (MetaFIGHT and SFMet) to run a “Metastasis Summer School” at Sesimbra, Portugal in September 2009. The TuMIC coordinator also organized a workshop on Metastasis where several coordinators of EU projects as well as other key individuals to discuss with Commission members important issues regarding the future of metastasis research. Moreover, TuMIC initiated an organized an international conference entitled “New concepts in cancer metastasis” that was held in Lisbon, Portugal in June 2011. The TuMIC conference was held at the Champalimaud Centre together with the Metastasis Research Society. Additional funding was provided through networking with the British Association for Cancer Research, the European Association for Cancer Research and the Association for International Cancer Research. Many EU-funded consortia were also represented at the TuMIC conference.
Press releases, radio presentations, other reports. The University of Heidelberg issues a press release concerning the objectives and aims of the TuMIC project. The Coordinator was interviewed on “Radio Regenbogen” about the TuMIC project. The TuMIC project has been further publicised through reports for the Metastasis Research Society, British Association for Cancer Research, European Association for Cancer Research and the Association for International Cancer Research.
Patenting. The intellectual property generated by TuMIC has been protected in the form of a patents. Two have already been files, and two are in preparation. Part of the costs have been offset by the TuMIC central innovation budget.
Technology Transfer activities. We organized a TuMIC workshop on single cell cDNA library construction and analysis that was organised by Fiona Watt’s lab. This constituted an important mechanism of technology transfer between the participating laboratories and fostered additional collaborative efforts amongst the TuMIC participants. Exploitation of the findings of the TuMIC consortium is supported by the activities of the technology transfer departments established at several of the participating institutions. These play a key role in exploitation through their technology transfer and marketing activities
Interaction with other EU-funded consortia and other key stakeholders. We have made considerable efforts to coordinate and integrate efforts in the metastasis research arena across Europe, particularly through interacting with other EU-funded consortia, as part of our dissemination activities. As mentioned above, we held a Summer School on Metastasis together with MetaFIGHT and SFMet. We also held joint Project Management Meetings with the FP6 consortium MCSCs. The Coordinator also organised a workshop in Brussels to inform key stakeholders about the problem of metastasis, and has initiated a meeting with other coordinators of EU-funded metastasis-related projects to discuss future directions in European metastasis research. TuMIC members have participated in conferences organized by other EU-funded consortia. Furthermore, we have fostering close ties with the Champalimaud Foundation in Lisbon, Portugal who have recently opened a dedicated centre for metastasis research, as well as with the Metastasis Research Society.
Reunion meeting. To ensure that we make maximum advantage of the potential impact of TuMIC findings, we have planned a “TuMIC Reunion” meeting at the beginning of 2013 at Frankfurt Airport paid by our institutional funds. Here we will discuss further progress since the end of the TuMIC funding period, as well as additional methods for dissemination and exploitation of the results.

List of Websites:

Related information


Norbert Huber, (EU Research Officer)
Tel.: +49 6221 54 2157
Fax: +49 6221 54 3599
Record Number: 196122 / Last updated on: 2017-03-16
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top