Response: Tissue factor de-encryption: the cell model system

In their correspondence, Liang and Hogg have suggested that MDA-MB231 cells, one of the 3 cell systems we used in our study of tissue factor (TF) de-encryption,¹  were not suitable for the study because they believe that TF is constitutively active in these cells. They base this conclusion on their observation that they were unable to find an increase in TF activity in these cells in response to HgCl₂ (100 μM for 30 seconds). We chose MDA-MB231 cells as a model system because we wanted to study the involvement of TF de-encryption in regulating the procoagulant as well as cell signaling activities of TF. We do not agree with Liang and Hogg in that all TF on MDA-MB231 cells is constitutively active. There are no specific assays that can quantify cryptic and active TF. However based on other parameters (ie, the discordance between the amount of FVIIa bound to TF and TF-FVIIa procoagulant activity, and many fold differences between TF activity on intact cells and lysed cells), a majority of TF on MDA-MB231 cells is encrypted. However, it is possible that the amount of active TF in MDA-MB231 cells might be higher than in monocytic cells. Liang and Hogg assert that the 4- to 7-fold increase in TF activity we observed with MDA-MB231 cells after the HgCl₂ treatment¹  is a relatively modest increase; we disagree with this assertion. More to the point, it is unclear to us why Liang and Hogg failed to observe any increase in TF activity in MDA-MB231 after treatment with HgCl_2. Perhaps the treatment conditions might explain this anomaly. In our experiments, we treated MDA-MB231 cells with various concentrations of HgCl₂ (1 to 100 μM) and chose a 15-minute treatment period because low concentrations of HgCl₂ may require a longer treatment period to show their effect. However, using 25 μM HgCl_2, we observed a 3-fold increase in TF activity after only a 1-minute treatment and a 5-fold increase after a 5-minute treatment (data not shown). It is unclear whether Liang and Hogg tested longer time periods of treatment of MDA-MB231 cells with HgCl₂ than 30 seconds. It is also conceivable that the source of MDA-MB231 cells, the culture conditions, and other experimental differences might be responsible for the discrepancy between our data and theirs.

Liang and Hogg also suggest that the increased TF activity we have observed with reduced glutathione (GSH) and dithiothreitol (DTT) in MDA-MB231, endothelial cells, and fibroblasts could be due to the cytotoxicity of these compounds. Indeed, not only GSH and DTT but also HgCl_2, when used in high concentrations for prolonged time periods, were cytotoxic and/or detach cells from the culture dish. However, this could be minimized by choosing the appropriate culture and experimental conditions (eg, low-passage cultures, 1- or 2-day-old cultures, removing the experimental compound promptly after the treatment period, and washing the cells gently, etc). We routinely observed the cells under microscope before, during, and after the experiment to monitor potential cytotoxic effects of the test compounds, and perform cytotoxic assays if warranted. Thus, it was unlikely that 30% to 35% of the cells were killed in our experiments in GSH and DTT treatments. Here it is important to point out that Liang and Hogg committed a mistake in stating that we treated MDA-MB231, endothelial, and fibroblast cells for 15 minutes with GSH and DTT. Endothelial cells and fibroblasts were treated only for 5 minutes with GSH and DTT (see Figure 2 legend in Pendurthi et al¹ ), and the report had no data on treating MDA-MB231 cells with DTT.

Finally our reservation on the validity of the disulfide bond switching in regulation of TF activity is not solely based upon the experiments discussed above but on a number of other facts as well.¹  In our view none of the recent publications on TF de-encryption by Cys186-Cys209 disulfide bond formation^2-4  are unequivocally convincing that the TF activation is independent of phosphatidylserine (PS). Although it had been concluded that TF de-encryption on HL60 could not be explained by exposure of PS on the plasma membrane,²  we reached the opposite conclusion based on the same set of primary data (see Figure 2 of Chen et al² ). Readers can examine these data for themselves and may reach their own conclusion. To provide convincing evidence that the disulfide bond formation plays a critical and important role in TF de-encryption, future studies should include a careful scrutiny of the data to rule out the possibility that previously proposed mechanisms (see Rao and Pendurthi⁵  and Bach⁶ ), particularly increased PS exposure, are not contributing to the observed effect. We are open and ready to change our opinion when such data become available.