Tissue Factor Pathway Inhibitor Anticoagulant and Anti-PAR Signaling Activity Requires GPI-Anchor Attachment to the Cell Surface in Chinese Hamster Ovary (CHO) Cells.

Introduction: TFPI is an endothelial associated protein that prevents intravascular thrombosis. It produces active site directed inhibition of the TF-fVIIa catalytic complex. TF-fVIIa activates blood coagulation via cleavage of fIX and fX. TF-fVIIa also initiates inflammation via activation of protease activated receptors (PARs). TFPI is attached to the endothelial surface through a GPI-anchor that localizes it to lipid raft/caveolae microdomains. We examined how GPI-anchor mediated localization of TFPI to these microdomains alters its ability to inhibit TF-fVIIa mediated coagulation and signaling events in vitro and in vivo.

Methods: CHO cells were transfected with TF to create CHO-TF cells. CHO-TF cells were transfected with chimeric forms of TFPI containing either a GPI-anchor or a transmembrane (TM) attachment sequence at the C-terminus to create CHO-TF/TFPI-GPI and CHO-TF/TFPI-TM cells. Clones expressing equal amounts of TFPI-GPI and TFPI-TM were selected using flow cytometry. TF-fVIIa activity on the surface of these cells was compared to that of CHO-TF cells in three assays. TF-fVIIa coagulation and PAR signaling activity were examined in vitro using fXa activation assays and ERK1/2 phosphorylation assays, respectively. A TF-mediated tumor metastasis assay was used to assess TF activity in vivo.

Results: GPI-anchored TFPI displays significantly more anti-TF-fVIIa activity than TM-anchored TFPI in the in vitro assays. In the anticoagulant activity assays, when compared to CHO-TF cells, the amount of fXa generated on the cell surface for 30 min was reduced by 64% in the CHO-TF/TFPI-GPI cells but only 8% in the CHO-TF/TFPI-TM cells (CHO-TF cells generated fXa at 6.3 pmol/min; CHO-TF/TFPI-GPI at 2.3 pmol/min; CHO-TF/TFPI-TM at 5.8 pmol/min). The signaling assays were performed using flow cytometry. When compared to CHO-TF cells, ERK1/2 phosphorylation was reduced 73% in CHO-TF/TFPI-GPI cells while no effect was observed in CHO-TF/TFPI-TM cells (MFI for CHO-TF cells at 3.17±2.04; CHO-TF/TFPI-GPI at 0.86±1.9; CHO-TF/TFPI-TM at 3.5±0.95). To assess TFPI inhibitory activity in vivo, cell lines were injected into the tail vein of SCID mice and the lungs examined for tumors at 10 days. TF-fVIIa activity is required for tumor formation in this assay. Consistent with the in vitro results, the CHO-TF/TFPI-GPI cells had greatly decreased tumor burden while CHO-TF/TFPI-TM had much less effect. Interestingly, transfection of CHO-TF cells with wild type TFPI that is secreted by CHO cells had no effect on tumor burden indicating that cell surface association of TFPI is required for anti-TF-fVIIa activity in this assay. These CHO cell results contrast with our previously published data where we found no difference in the anticoagulant activity of TFPI-GPI and TFPI-TM on the surface of HEK293 cells (Thromb. Hemost. 2003, 89:65). Since CHO cells express abundant caveolin-1 while HEK293 cells express very little, we considered that caveolin-1 may be responsible for the differential activity observed. In support of this hypothesis we found that TFPI and caveolin-1 co-immunoprecipitate from human placental lysates.

Conclusions: A CHO cell model was used to examine the inhibition of TF-fVIIa by TFPI-GPI or TFPI-TM. GPI-anchor attachment of TFPI to the cell surface appears to be critical for its anti-TF activity. Localization to caveolae microdomains and/or a direct interaction with a protein complex containing caveolin-1 may be required for full TFPI activity in vivo.