There are conflicting laboratory data regarding the mechanisms of procoagulant activity and alterations in vascular permeability in endothelial tissue following intravascular administration of TNF in isolation perfusion. These 2 actions (procoaguant activity and permeability) have been thought to represent distinct cellular responses to TNF. Procoagulant activity is induced via endothelial cell (EC) surface expression of tissue factor following activation of nuclear factor kappa beta and the AP-1 transcription factors and results in rapid and potent activation of the extrinsic clotting cascade. Increased vascular permeability has been ascribed to down regulation of integrin expression on ECs following exposure to TNF that results in disruption of the endothelial barrier. However, the experimental conditions that are known to alter integrin expression by TNF are not relevant or comparable to those conditions used in isolation perfusion. We established an in vitro model of vascular permeability using a 2-compartment system in which human umbilical vein ECs are plated in inserts that have a polycarbonate membrane with 1 micron pores that are placed in 6-well culture plates. Evans Blue bound albumin is added to the upper chamber to which the EC monolayer is impermeable and then the ECs are subjected to various experimental conditions that alter permeability and flux of albumin from the upper to the lower compartment. The amount of flux can be quantified by measuring absorbance of aliquots from the lower chamber with a spectrophotometer at 620 nanometers. Tissue factor production and procoagulant activity were quantified by ELISA and by a 1-stage coagulation assay in which EC cell lysates are mixed with factor VIII deficient plasma and time to clot formation was measured in seconds using a coagulation analyzer after adding calcium chloride as a catalyst. We confirmed that TNF induced tissue factor on ECs at doses and time intervals relevant to those used in vascular isolation and perfusion. At doses that maximally stimulated tissue factor production there was no effect on EC permeability under basal culture conditions. However, when factor VIII deficient plasma was added with TNF the barrier function of ECs changed rapidly and significantly. The addition of recombinant tissue factor pathway inhibitor (TFPI) or neutralizing anti-tissue antibody completely abrogated the permeability effects. It is noteworthy that the effects of TNF on EC monolayer permeability were rapid; significant changes in permeability were observed after only a 30 minute treatment with TNF. The site of administration (luminal or upper chamber versus abluminal or lower chamber) also influenced permeability in this model; EC monolayers treated with abluminal TNF had only a modest increase in permeability compared to control EC monolayers but there was a significant effect when the same dose of TNF was administered to the luminal EC surface. To characterize the EC morphological changes that occur after exposure to TNF and plasma, the effects on VE-cadherin, the main complex responsible for maintaining intracellular tight junctions in endothelial tissue was evaluated. As seen in figure 10, quiescent confluent HUVEC monolayers had homogeneous staining of VE-cadherin along cellular junctions and staining for F-actin cytoskeletal elements showed a dense circumferential band of colocalization of VE-cadherin resulting in diffuse yellow staining in these areas. TNF alone and factor VIII deficient plasma alone had no effect on VE-cadherin and F- actin staining but in combination caused a marked loss of cell to cell contact with decreased VE-cadherin staining in the areas of intercellular gap formation. The morphologic effects were completely prevented in the presence of anti-thrombin III indicating that the alterations in permeability in this model were due to cell surface induction of tissue factor, were rapid and significant, and a physiologic response to the TNF rather than a non-specific cellular injury. Additional data indicate that nuclear localization of the transcription factor, EGR 1, is likely responsible for the induction of tissue factor in ECs. These data demonstrated an essential role for tissue factor in mediating not only the procoagulant but the permeability effects with TNF in endothelium and suggested that modulation of tissue factor activity in tumor neovasculature may represent the central mechanism for mediating TNF's anti-vascular effects during isolation perfusion. In a broader perspective, the data also provide supporting evidence for an important role for tissue factor in the initiation of tumor associated angiogenesis because alterations in vascular permeability represent one of the early sentinel events in tumor neovessel formation by allowing extravasation of fibrin and other plasma proteins into the interstitium which then serve as a provisional matrix for subsequent EC migration in early angiogenesis. In this context it has been shown that tumor lines that have a high cell surface expression of tissue factor have increased growth and metastatic potential in vivo compared to low tissue factor expressing cell lines. Because TNF is always administered in isolation perfusion under hyperthermic conditions, we conducted experiments to characterize the effects of hyperthermia and TNF in the 2 compartment model of EC permeability. There was a progressive increase in plasma independent EC permeability after hyperthermia alone and EC integrity was restored 24 hours after hyperthermia. It is noteworthy that cell surface staining for VE-cadherin and F-actin were similar to the pattern observed with TNF induced permeability. These current data indicated that hyperthermia works through an alternate mechanism independent of tissue factor induction as it occurs in the absence of factor VIII plasma but results in the same physiologic cellular response. Additional studies were undertaken to characterize the effects of hyperthermia on TNF mediated endothelial cell coaguability in vitro under conditions relevant to those used clinically in isolated organ perfusion. The ability of TNF to induce a procoagulant phenotype in ECs at 37 and 41 degrees C was assessed using the one stage coagulation assay. Untreated ECs or cells treated with heat alone showed no increased coaguability. ECs treated with TNF at 370 C demonstrated rapid clot formation immediately following TNF treatment (time zero) which was sustained during the 31/2 hour duration of the experiment. In contrast, TNF treatment at 410 C resulted in delay of coagulant activity until 31/2 hours after treatment. We verified by ELISA that large quantities of tissue factor were produced by ECs in response to TNF at 37 degrees C whereas cells treated with TNF at 41 degrees C only produced the same amount of tissue factor protein as untreated control cells. In order to ensure that the results were not secondary to degradation of TNF under hyperthermic conditions we confirmed that there was no identifiable degradation of the protein under the experimental conditions being used. Similarly, hyperthermia did not affect viability of ECs. Tissue factor and tissue factor pathway inhibitor-2 mRNA expression was measured in ECs by quantitative RT-PCR following treatment with TNF at 41 or 37 degrees C. The experiments showed that tissue factor mRNA was barely detectable in untreated ECs and cells treated with hyperthermia alone whereas TNF at 37 degrees C caused a marked tissue factor mRNA up-regulation while the effect was markedly attenuated after TNF at 41 degrees C. A similar pattern of gene expression change was noted with TFPI-2. These data demonstrated that hyperthermia significantly but transiently decreased TNF induced procoagulant activity in HUVECs compared to TNF administered under normothermic conditions.
