Furthermore, hK1 levels were not associated with the cellular type and indexes of aggressiveness. An association was instead found with tumour location, with the highest values in GISTs of the small intestine. Our in vitro data show that expression levels in the GIST cell line are increased by hypoxia/starvation, pointing to the possibility that areas of the tumour far away from inhibitor Carfilzomib the vasculature could be stimulated to produce larger amounts of hK1. This is, at least in part, in line with the prevalent localization of hK1 in the central avascular core of GIST882 xenografts. In vitro studies on two different cell lines derived from Imatinib-resistant or Imatinib-sensitive tumours showed the common expression of hK1.

Cell biology results suggest two distinct molecular mechanisms by which hK1 could be implicated in GIST growth: (1) invasion through degradation of the ECM and (2) induction/stabilisation of host-derived tumour vasculature. A direct action of hK1 as an ECM-degrading enzyme could be complemented by the ability of hK1 to cleave and activate other proteases, such as pro-MMPs (Menashi et al, 1994). The proinvasive action was verified by a gene titration approach, using siRNA to decrease native hK1 expression and Ad.hK1 to force hK1 production. Silencing resulted in inhibition of GIST invasive capacity and overexpression enhanced it. The effect of silencing was further confirmed using inhibitors of hK1 activity. It is therefore likely that changes in hK1 production, release and clearance by endogenous inhibitors may confer a different invasive profile to the tumour.

Coculture experiments showed that GIST882 cells exert an attractive action on HUVEC, which is at least in part due to activation of kinin receptors, as verified by the use of receptor antagonists. Furthermore, GIST882-conditioned medium stimulates HUVEC to form more robust branches than those observed using unconditioned medium. The fact that kallistatin, which rapidly binds hK1 and inhibits its activity in vitro (Zhou et al, 1992), blocks the strengthening action of GIST882 cells on HUVEC networks argues in favour of hK1 as a promoter of cancer angiogenesis. Kallistatin itself was previously identified as an inhibitor of angiogenesis in gastric carcinomas (Zhu et al, 2007).

In the retrospective analysis of human GISTs, we could not find any correlation between hK1 expression and vascular density, suggesting Batimastat that other mechanisms may overwhelm or confound the proangiogenic effect of hK1 in vivo. On the other side, although HUVECs are widely used as models for tumoural angiogenesis, different mechanisms might drive cancer endothelial cells. In conclusion, results of our study show for the first time that hK1 is implicated in GIST invasion and angiogenesis. GIST882 xenografts express and release hK1 into the circulation, a result that calls for further validation of hK1 as a potential diagnostic biomarker.