Clinical and experimental evidence suggests a parallel between increasing cancer aggressiveness and procoagulant tendencies, which are often attributable to high levels of tissue factor (TF). TF is upregulated on the surface of cancer cells and their derived microvesicles due to a combined impact of oncogenic events and tumor microenvironment [1–4] and thereby becomes involved in cancer progression, both as the principal initiator of the blood coagulation cascade and as a signaling receptor [1,2,5,6]. These TF activities are implicated in tumor growth, angiogenesis and metastasis [1,2,5,6], a finding congruent with the anticancer effects of anticoagulation revealed in recent clinical trials [1,7]. It remains unclear which cancer cells harbor biologically relevant TF.
It has recently come to light that the capacity of cancer cells to initiate tumor growth is not universal, but rather can only be executed by a small minority of specialized cancer cells (fewer than 1%), often referred to as cancer stem cells (CSCs) . Identification of CSCs in several solid tumors has been made possible by the recent discovery of their molecular markers, of which expression of CD133 (prominin-1) represents one of the best-known paradigms [9–12]. CD133 belongs to a family of five-transmembrane cell-surface glycoproteins commonly localized to membrane protrusions of various progenitors [9,13]. While CD133-positive cells have been identified as CSCs in several solid tumors, e.g. of the brain , colon  and in melanoma , it is unclear whether CD133 expression plays a causative, contributive or correlative role in the formation of the CSC population. Still, CD133-expressing CSCs have been implicated in a number of key processes, including tumor repopulation, resistance to therapy , increased aggressiveness  and angiogenesis .
It is intriguing to note that various recent reports point either to cancer cells with increased expression of TF or to cancer cells harboring the CSC marker CD133 as particularly central to cancer progression [4,16]. We reasoned that this may signify a deeper interrelationship between these two properties. In order to examine this in more detail, we tested the expression of CD133 and TF in the highly tumorigenic squamous cell carcinoma cell line A431, which is known to express considerable procoagulant activity . We used Caco-2 cells as a positive control due to their curiously high CD133 expression [coupled with a paradoxically poor tumor-forming capacity and undetectable TF expression (our unpubl. obs.)].
Interestingly, whole cell lysates of the pooled A431 cell population were found to contain appreciable amounts of TF , but surprisingly little detectable CD133 (Fig. 1A). However, flow cytometry analysis revealed that A431 are heterogeneous in that a small subset of these cells (0.5%) stained strongly with the CD133 antibody (Fig. 1B). In order to better understand the significance of this heterogeneity, A431 cells were separated immunomagnetically into CD133-positive and CD133-negative fractions (using the CD133 MACS system, Miltenyi Biotech, Auburn, CA, USA), each of which was then tested for TF content. Interestingly, CD133-positive A431 cells expressed a 5- to 6-fold greater amount of TF antigen on their surfaces than their CD133-negative counterparts did (Fig. 1C) . This was paralleled by a corresponding increase in the TF-dependent procoagulant activity (TF-PCA) [3,18] of CD133-positive A431 cells relative to cells lacking this stem cell marker. This latter assay measures the ability of the TF/factor (F) VIIa complex to generate FXa-dependent, prothrombin-activating proteolytic activity  on the surface of A431 cancer cells.
Thus we found that in a subset of A431 cancer cells CD133 is coexpressed with high levels of TF. As the CD133-positive (CSC) cancer cell subset is thought to drive tumor initiation/formation events, we asked whether TF was required for the manifestation of these properties in vivo. Immunodeficient (SCID) mice were injected with A431 cells and treated with CNTO 859, a neutralizing TF-directed antibody that blocks FX activation . Indeed, this therapy resulted in a marked inhibition of tumor growth (Fig. 1E).
In light of our observations, we propose that TF expression (and TF-PCA) probably contributes to the tumor growth-initiating/-regulating properties of CD133-positive CSCs. This could occur through recruitment of growth-promoting effectors of the coagulation system such as thrombin, fibrin and platelets, all of which could provide matrix and growth factor support for the emerging tumor, i.e. could act as a provisional stem cell niche.