The association between cancer and thrombosis has long been established and many of the factors linking these conditions have been characterised [
1,
16,
45]. However, parallel associations between procoagulant properties of cancer cells and the induction of blood coagulation do not appear to be sufficiently significant to permit the prognostic determination of the risk of thrombosis [
4,
6]. The ability of cancer cells to express high levels of TF is well established [
2] but the examination of the level of total and surface TF antigen, or cell-surface TF activity does not present a clear and definitive correlation with risk of clot formation [
5,
6,
19‐
21]. In addition, while there has been a recent upsurge in studies examining the association of TF-bearing microvesicles and incidence of thromboembolism, correlations arising from these data have been reported to be heterogeneous [
14,
45]. It has been demonstrated that released microvesicles are rapidly cleared by cellular uptake in vitro and from the bloodstream in vivo, with estimates ranging from minutes to a few hours [
33,
46‐
51]. Consequently, any TF released as tumour-derived microvesicles may not remain within the circulation for long enough to accumulate to levels capable of promoting clot formation while concurrently, TF-bearing microvesicles may be present at measurable levels at the time of sampling. One feature of many cells is the ability to release TF as microvesicles in bursts upon activation of the cell. Such bursts in TF release into the bloodstream may fluctuate in duration and magnitude and may also vary in the length of time before the onset of the release of TF-bearing microvesicle [
16,
52,
53]. Therefore, in this study we examined the ability of seventeen cancer cell lines spanning various tissues to release TF in response to PAR2 activation. We used PAR2 activation as a stimulus for the study since 1) the release of TF-bearing microvesicles occurs in a much shorter time than for example stimulation with TNFα [
54] and also, remained unaltered in the untreated control cells. 2) PAR2 activation represents a more controlled activation of cells without influencing other pro-inflammatory mechanisms within the cell as observed with LPS treatment. Together, these advantages permit the analysis of TF release without the influences arising from de novo expression of various genes which may complicate the analysis. Finally, 3) we hypothesise that the exposure of tumour cells to bloodstream may induce low level generation of factor Xa and TF-factor VIIa complex allowing for the activation of PAR2 on cancer cells, without substantial amounts of clot formation within the tumour cells’ immediate locality.
Activation of cells generally resulted in the upregulation of TF release within 30–60 min but varied hugely in magnitude in the cell lines tested. Interestingly, the change in the rate of TF release in activated cells was only moderately dependent on the rate of TF release while under resting conditions, or the amount of TF protein stored within the cells. Previously, we showed that the suppression of TF expression in five cell lines resulted in divergent rates of decline in the amount of cellular TF [
35]. This was attributed to the background rate of TF release from these cells indicating that TF reserves were depleted at a faster rate when the amount of microvesicle-associated TF was maintained by the cells. In fact the percentage increase in the TF release following activation appeared to be inversely correlated to the ability of cells to release TF under resting conditions. Therefore, the level of cellular TF protein stored within the cells may be a function of the turnover of TF, determined by both the expression of TF mRNA and TF protein release from the cells and hence, only partly correlates with TF mRNA expression. This also explains the heterogeneous correlation between the level of circulating TF containing microvesicles and the incidence of thrombosis [
14,
45] since this correlation may be positive or negative, depending on whether the tumour cells are activated at the time of sampling. In addition, a weak correlation between cell surface TF activity and microvesicle-associated TF activity was detected in resting cells. This is also in agreement with the notion that TF is transferred to the cell surface prior to release as microvesicle although, the control of TF activity at the cell surface may strongly be regulated by TF encryption [
4]. In agreement with the above hypothesis, the level of microvesicle-associated TF during the short-term burst in TF release would be dependent on the ability of cell to replenish TF reserves through mRNA expression and therefore strongly correlated with the TF mRNA expression levels. Furthermore, the magnitude of TF release correlated strongly with the PAR2 mRNA expression in the cells. Interestingly, the level of TF mRNA expression also strongly correlated with the expression of PAR2 mRNA but not PAR2 protein. However, the turnover of PAR2 includes internalisation, recycling and degradation which may alter the perceived level of available PAR2 protein. The induction of TF expression following PAR2 activation has been shown previously [
55]. Therefore, low level activation of PAR2 may also contribute to the ability of cells to replenish TF reserves and enhance the cellular “TF-release potential” through separate mechanisms involving the upregulation of TF gene expression. However, the change in TF release, as a percentage of the levels observed in resting cells appears to be related to PAR2 mRNA and PAR2 protein levels. Perhaps this is not surprising in our experiments since the cells were activated by incubation with PAR-activating peptide. However, it is surprising that 1) the strongest correlation was observed between TF release and PAR2 mRNA expression and 2) some cell lines such a ZR-75-1 did not respond proportionally to PAR2 activation. In contrast, the phosphorylation and the subsequent release of TF is not dependent of PAR1 activation [
24,
56]. Although, the release of microvesicles themselves may occur upon activation of PAR1 and PAR2 by separate mechanisms [
57]. In our present study, we did not observe a significant correlation between microvesicle release and TF release, although in general a higher TF release was often accompanied with moderate to high microvesicle release. Therefore, it is unlikely that PAR1 activation would have any direct influence on TF release potential although indirect mechanisms have not been ruled out. Additionally, since glycosylation of PAR2 may alter its trafficking and function, the patterns observed in Fig.
2 may hold further clues to the "TF-release potential" property of the cell lines and need further investigation.