The regulation of tissue factor mRNA in human endothelial cells in response to endotoxin or phorbol ester.

Tissue factor (TF) is the membrane-bound glycoprotein whose cofactor activity with factor VIIa causes activation of the extrinsic pathway of coagulation. The transition of endothelium to a procoagulant state by agents such as bacterial lipopolysaccharide (LPS) is the result of TF expression by these cells. The mechanism of TF induction in human umbilical vein endothelial cells (HUVEC) was investigated in response to LPS and phorbol 12-myristate 13-O-acetate (PMA). Northern blot analysis of total RNA from HUVEC showed a rapid rise in TF mRNA levels which was maximal at 2 h and had fallen to low levels by 6 h following both LPS (10 micrograms/ml) and PMA (10 ng/ml) stimulation. Nuclear-run on experiments showed at most a 2-fold increase in transcription of the TF gene following LPS stimulation but a 10-fold increase following PMA stimulation. In addition 24-h pre-incubation with PMA desensitized HUVEC to further PMA exposure, but caused no alteration in the response to LPS. Cycloheximide (10 micrograms/ml) alone caused induction of TF mRNA. Treatment of cells previously exposed to LPS for 1 or 4 h with actinomycin D indicated a 12-fold difference in the TF mRNA half-life. Therefore the rapid accumulation of TF mRNA in HUVEC stimulated by LPS is largely a result of an increase in mRNA stability rather than an increased rate of transcription of the gene.

Tissue factor (TF) is the membrane-bound glycoprotein whose cofactor activity with factor VIIa causes activation of the extrinsic pathway of coagulation. The transition of endothelium to a procoagulant state by agents such as bacterial lipopolysaccharide (LPS) is the result of TF expression by these cells. The mechanism of TF induction in human umbilical vein endothelial cells (HUVEC) was investigated in response to LPS and phorboll2-myristate I3-O-acetate (PMA). Northern blot analysis of total RNA from HUVEC showed a rapid rise in TF mRNA levels which was maximal at 2 h and had fallen to low levels by 6 h following both LPS (10 @g/ml) and PMA (10 rig/ml) stimulation.
Nuclear-run on experiments showed at most a a-fold increase in transcription of the TF gene following LPS stimulation but a lo-fold increase following PMA stimulation.
In addition 24-h pre-incubation with PMA desensitized HUVEC to further PMA exposure, but caused no alteration in the response to LPS. Cycloheximide (10 pg/ml) alone caused induction of TF mRNA. Treatment of cells previously exposed to LPS for 1 or 4 h with actinomycin D indicated a 12-fold difference in the TF mRNA half-life.
Therefore the rapid accumulation of TF mRNA in HUVEC stimulated by LPS is largely a result of an increase in mRNA stability rather than an increased rate of transcription of the gene.
is released upon bacterial lysis (7). The effects of LPS upon endothelium, in addition to the induction of TF (5), include the synthesis and release of interleukin 1 (8) and interleukin 6 (9), increased neutrophil adherence by a CDlS-dependent mechanism (10) and increased plasminogen activator inhibitor 1 expression (11). TF expression in the monocyte, induced by LPS, has been shown to require increased transcription of the TF gene (12), and similar conclusions have been reached for the TF response to tumor necrosis factor (13). The TF response to phorbol esters is less clear; in endothelium there is an increase in TF activity (14), but in the monocyte there are conflicting results (15,16).
This paper reports the effects of LPS and the phorbol ester, phorbol 12-myristate 13-O-acetate (PMA) upon TF mRNA levels, rates of transcription of the TF gene, and TF activity in human umbilical vein endothelium (HUVEC).

Materials
Cell Cultures-Medium 199 (MlSS), L-glutamine, penicillin, and streptomycin were from Flow Laboratories, Irvine, United Kingdom. Fetal calr serum, newborn calf serum, and trypsin, were from Gibco, Paislev. UK. Endothelial cell growth supplement was prepared inhouse"from bovine brain by the method previously described (17). Heparin (sodium salt from porcine intestinal mucosaf was from Sigma, Poole, UK. Tissue culture flasks were from Falcon, Irvine, UK unless otherwise stated. Gelatin was from British Drug House (BDH) Poole. UK.
Tissue factor (TF)' is a 47-kDa membrane bound glycoprotein, which is the essential cofactor for activation of the extrinsic pathway of coagulation.
TF complexes with factors VII and VIIa permitting activation of factors IX and X, the substrates for VIIa (1). Factors VII and VIIa are devoid of activity in the absence of TF. Since factor VII circulates in the blood TF is not usually expressed by cells within the vasculature. However, two cells, the monocyte and the endothelium, having direct contact with the blood may be stimulated to express TF by a number of agents which include tumor necrosis factor (Z), interleukin 1 (3), and bacterial lipopolysaccharide (LPS) (4-6). LPS is the outermost part of a bacterial cell membrane and * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "adoertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
li To whom correspondence should be addressed. The cells were washed with ice-cold phosphate-buffered saline and removed by brief exposure to 1.5 ml of 0.1% trypsin, 0.02% EDTA.
The cells were resuspended in 10 ml of resuspension buffer (RSB) (10 mM Tris-HCl (pH 7.4), 10 mM NaCl, 5 mM MgCl, 1 mM phenylmethylsulfonyl fluoride, and 5 mM benzamidine) and then pelleted at 500 X g for 5 min and washed twice with 10 ml of RSB. The pellet was resuspended in 7 ml of lysis buffer (10 mM HEPES buffer (pH 7.9), 10 mM NaCl, 3 mM MgCl,, and 0.05% Nonidet P-40). Nuclei were released from the cells with 10 strokes of the B pestle of a Dounce homogenizer. The nuclei were pelleted at 500 x g for 50 min, washed twice with RSB, resuspended to a final volume of 210 ~1 in 50 mM HEPES buffer (pH 7.9), 40% glycerol, 5 mM MgCl,, and 0.1 mM EDTA, and frozen at -70 "C until required. Run on reactions were performed using the nuclei obtained from one 175~cm* flask for each time point z 3.0 x lo7 cells). The nuclei were added to 5 X run on buffer (25 mM (23), the a-actin and TF cDNAs described above, and control plasmids without inserts for the a-actin and TF cDNA were all applied to the membrane. Following prehybridization at 65 "C as above, the filters were hybridized for 36 h at 45 "C with the run-on RNAs and were washed twice for 15 min each in 2 X SSC, 0.1% SDS at room temperature and once at 55 "C in 0.1 X SSC, 0.1% SDS for 30 min. The filters were then exposed to Kodak XAR film at -70 "C with an image intensifying screen.
RNA Studies-One flask of cells was used for each time point with an agonist.
LPS ( Studies-HUVEC at passage 3 in 24well trays were exposed to 10 rig/ml PMA in Ml99 + 5% fetal calf serum for 24 h or medium alone. This was removed and either LPS (10 rig/ml) or PMA (10 rg/ml) was added to the cells for 6 h and the resulting TF activity measured as described above.

RESULTS
TF mRNA and Nuclear Run-on Studies-The 1.35-kb human antisense RNA probe for TF hybridized with a major transcript of 2.4 kb as well as a characteristic doublet of two minor transcripts at 3.5 and 3.1 kb (21). These species were in both polyadenylated placental RNA (data not shown) and total RNA from HUVEC following 2-h stimulation with LPS (10 pg/ml). Furthermore, the minor transcripts, as well as the major 2.4-kb species, were found to be resistant to RNase A digestion. TF mRNA was seldom detectable in unstimulated HUVEC and when observed was present at very low levels. Confluent cultures of cells at passage 3 were stimulated with maximally active concentrations of LPS (10 pg/ml) or PMA (10 rig/ml) (see below). Fig. 1 shows Northern blots of RNA from HU-VEC stimulated for 0, 1, 2, and 4 h with either PMA or LPS. TF mRNA shows rapid induction to achieve peak levels by 2 h following stimulation with either agent. Substantial depression of the TF mRNA has occurred by 4 h which appears to be more rapid in the case of PMA.
Other experiments performed to examine the initial part (0, 10,30, and 60 min) and the late part of the time course (0, 3, 6, 12, and 24 h) of TF mRNA response to LPS and PMA confirm this rapid inductive effect. For both agents TF mRNA is easily detectable by 30 min following stimulation. In the late time-course studies TF mRNA levels were found to have fallen to very low levels by 12 h and fall to barely detectable amounts at 24 h. As is shown in Fig. 1, all three species of TF mRNA responded in a parallel manner.
Total RNA from HUVEC stimulated for 1 h with either LPS (10 pg/ml) or PMA (10 rig/ml) was serially diluted and applied to a dot blot filter for comparison of the change in steady state level of TF mRNA in each sample at that time point. Hybridization of the filter for TF mRNA showed that the stimulation produced by PMA caused a 30-fold greater increase in steady state mRNA levels compared to elevations achieved with LPS.
The effects of LPS and PMA upon the rate of transcription were examined in nuclear extracts from HUVEC. Fig. 2 shows the results from a nuclear run-on experiment where cells of the same batch had been exposed to either serum free medium alone, LPS (10 pg/ml) or PMA (10 rig/ml) for 1 h. A low level of transcription of the TF gene is seen in the control nuclei, and densitometric analysis, correcting to the actin signal, indicates only a 2-fold increase with LPS, in the experiment shown. A second run-on experiment (data not shown) confirmed an identifiable basal rate of transcription of the TF gene under control conditions, which was not altered by LPS. In contrast PMA caused a marked, lo-fold, increase in the FIG rather low level of transcription seen with control or LPStreated cells.
The level of transcription of the actin and von Willebrand's factor genes are also shown for comparison. SPARC mRNA hybridization is negative, indicating that there is no transcription of this gene by these cells under these conditions. SPARC is secreted by bovine aortic endothelial cells when injured in culture, or plated at nonconfluent density, and has been interpreted as a marker of culture induced damage (24). The absence of transcription of the SPARC gene suggests that the TF response seen was not a consequence of nonspecific cellular damage.
The effect of cycloheximide (10 pg/ml) was examined upon the basal and stimulated level of TF mRNA in HUVEC. Fig.  3 shows a Northern blot of RNA isolated from cells exposed to serum-free medium for 1 h, cells with cycloheximide (10 Kg/ml) for 1 h, cells with LPS (10 Kg/ml) for 1 h and cells exposed to LPS (10 pg/ml) and cycloheximide (10 pg/ml) together for 1 h. Substantial induction of the message for TF was seen with cycloheximide alone, to a level greater than achieved by LPS alone, and little further (super) induction was evident with the combination of the two agents. No changes in the actin mRNA levels were seen with cycloheximide.
In a separate nuclear run-on experiment the effect of exposure to cycloheximide (10 pg/ml) for 1 h upon TF gene transcription was examined. Correcting to the actin signal, as above, showed no change in the level of transcription of the TF gene in response to this dose of cycloheximide.

Tissue Factor mRNA Changes in Human
Endothelium 9785 The half-life of TF mRNA was examined by inhibiting transcription with actinomycin D (10 pg/ml) after stimulation of HUVEC with LPS (10 pg/ml) for either 1 h when the level of TF mRNA is increasing or 4 h when it is decreasing. Fig.  4 shows that TF mRNA was induced by LPS at both times compared with control, however, the message was substantially more stable following a l-h stimulation with LPS compared with 4 h. Densitometric analysis of autoradiographs exposed so that the zero time point for the l-and 4-h stimulated cells were quantitatively similar in density indicates half-lives for TF mRNA of 2 h following LPS stimulation for 1 h and 10 min following LPS stimulation for 4 h. Fig. 4 shows stability of the actin mRNA throughout both of these periods.
Tissue Factor Activity-TF activity was not detectable in unstimulated HUVEC. PMA and LPS each caused dosedependent increases in TF activity in HUVEC that were maximal with 10 rig/ml PMA and 10 pg/ml LPS. The maximal TF activity achieved with PMA was IO-fold greater than that with LPS. TF activity was not detectable until 2 h after stimulation with either agent, peaked at 6 h, and declined to low levels at 24 h (Fig. 5). Agonist-induced changes were consistently greater for both agents when incubation was performed in 5% serum but were still detectable when serumfree medium was used (data not shown). Exposure to serum- Half-life of TF mRNA following l-and 4-h exposure to LPS. HUVEC were exposed to 10 pg/ml LPS for either 1 or 4 h. 10 pg/ml actinomycin D was then added and total RNA extracted at 0, 0.25, 0.5, 1, 2, 4, and 6 h later. 5 pg of total RNA were added to each lane along with 5 pg of total RNA from a flasks of unstimulated cells (Basal). The filter was hybridized with antisense RNA probes for TF and cu-actin. The actin bands represent hybridization with the 2100base pair species. PMA. HUVEC were exposed of 1 wg/ml LPS (open circles) or 0.1 ng/ ml PMA (closed circles) for 0, 1, 2, 4, 6, and 24 h. TF activity was measured in cell lysates using a modified clotting assay (see "Methods"). Bars represent fl S. E. free medium alone for up to 6 h was without effect upon tissue factor expression.
When either cycloheximide (10 pg/ml) or actinomycin D (10 pg/ml) was co-incubated with either PMA (10 rig/ml) or LPS (1 pg/ml) there was complete abolition of the TF response to these agents.

Protein Kinase C Desensitization
Studies-Long-term exposure of endothelial cells to PMA down-regulates protein kinase C activity to almost undetectable levels (25). In other cell types this phenomenon has been shown to have a halflife of about 6 h and leads to near complete abolition of protein kinase C activity by 24 h (26). HUVEC when treated for 24 h with PMA (10 rig/ml) expressed little TF activity: medium alone 24 h, 5.2 f 0.9 milliunits of TF/105 cells; medium + PMA (10 rig/ml) for 24 h, 39.7 f 2.9 milliunits of TF/105 cells (mean f S.E., n = 12, two experiments). When cells were then further challenged for 6 h with PMA (10 ng/ ml) TF activity did not rise in those previously exposed to PMA (49.0 f 3.2 milliunits of TF/10" cells), whereas the expected response was obtained from cells preincubated for 24 h with medium alone (495 f 31.0 milliunits of TF/10" cells (mean f S.E., n = 12, two experiments)). In contrast, LPS (1 pg/ml) for 6 h induced equal amounts of TF activity in cells pretreated for 24 h with medium alone (111.0 + 7.7 milliunits of TF/105 cells), or with PMA (10 rig/ml) (120.8 f 26.7 milliunits of TF/105 cells (mean f S.E., n = 12, two experiments)).

DISCUSSION
The TF mRNA found in HUVEC upon stimulation with either LPS or PMA is composed of one major transcript (2.4 kb) and two larger, less abundant transcripts (3.1 and 3.5 kb). This is in accordance with the original description of this cDNA clone (21). The origin of these minor transcripts is not clear, but induction of these species parallels the change in level of the major transcript upon stimulation, and furthermore, the hybridization is resistant to RNase A digestion and is, therefore, specific. The presence of the larger transcripts in the polyadenylated placental RNA, and the demonstration by other groups of the presence of only one polyadenylation (28) signal in the genomic DNA sequence, suggest that these polyadenylated species are incomplete or alternatively spliced transcripts arising from the TF gene as suggested by Scarpati et al. (27). We do not see these additional transcripts in fibroblasts (data not shown) in agreement with others. The time courses of induction of TF mRNA by PMA and LPS are almost identical. A very rapid increase in mRNA level is detectable by 30 min with peak levels of mRNA occurring by 2 h. Considerably reduced, but detectable levels were seen at 6, 12, and 24 h following both agents. These changes are mirrored in the alterations of TF activity in HUVEC which, as expected, were slower in onset, being detectable at 2 h and reaching a maximum at 6 h in agreement with the results from other groups (29,30). The time course of TF mRNA in response to LPS in HUVEC demonstrated here contrasts with the results of similar experiments conducted in human monocytes, where peak TF mRNA was found at 4 h following LPS exposure (12). It is also at variance with the response of plasminogen activator inhibitor 1 mRNA to LPS in endothelium, which peaks at I2 h following stimulation (31). The TF mRNA response of HUVEC to tumor necrosis factor has been reported to show maximal levels at 4 h following stimulation, but in this study earlier time points were not included (13). The rapid TF mRNA response to LPS described here is similar to the mRNA time courses in HU-VEC of the early response gene KC induced by LPS, which was reported to be mediated by protein kinase C activation (32) and c-fos mRNA induced by interleukin 1 (33).
The response to PMA was similarly brisk and decay of TF mRNA was, if anything, quicker. The latter may be the result of desensitization of protein kinase C upon prolonged exposure to PMA (see below). The steady state levels of TF mRNA achieved with PMA were some 30-fold greater than LPS when compared at 1 h of exposure, and this is reflected in the observation that PMA was consistently able to produce greater maximal TF activity in these cells.
Exposure of the cells to cycloheximide, at a concentration that blocked protein synthesis, induced TF mRNA to a level greater than that achieved with LPS alone. It is clear, therefore, that the induction of TF mRNA under these circumstances does not require protein synthesis, and this is in accordance with the rapidity of the LPS response.
The nuclear run-on experiments indicate that HUVEC in culture had a low basal rate of transcription which was only slightly increased or unaltered, by LPS, but substantially stimulated by PMA. The latter finding is in keeping with the presence of a concensus sequence for a phorbol response element in the 5'-flanking region of the TF gene (28). The absence of a clear stimulatory effect of LPS upon transcription is, however, at variance with the increase in TF transcription produced by LPS in the human monocyte (12). This may represent a significant cell type-specific difference in the regulation of the TF gene.
The predicted presence of AU sequences (known to be important in rendering mRNA unstable) (34) in the 3'-untranslated region of the TF mRNA, and the rapid induction of TF mRNA by cycloheximide without effect on the rate of transcription suggests that message stability is a potential control point regulating the level of TF mRNA. In addition, the lack of clear effects of LPS upon transcription, despite the demonstration of increases in transcription with PMA, suggested that LPS is acting by increasing stability of TF mRNA. The actinomycin D experiments reported here show that the half-life of TF mRNA at 1 h after the addition of LPS (when levels are increasing) is 12 times longer than at 4 h (when levels are decreasing). If control is entirely at the level of transcription such a difference should not occur, unless the rate of mRNA breakdown had become saturated. These findings give circumstantial support to the hypothesis that LPS stimulation of TF mRNA in HUVEC is substantially controlled by the rate of breakdown, i.e. message stability.
In contrast to LPS, PMA clearly stimulated the rate of transcription of TF mRNA. PMA acts as a diacylglycerol mimetic producing translocation and activation of protein kinase C. The phosphorylated substrates produced presumably include those which interact either directly or indirectly with the phorbol ester response elements of eukaryotic genes. The transduction system for and mechanism of action of LPS is not clear. However, the rapidity of its action, and absence of a requirement for protein synthesis, suggest that this is a direct effect of LPS and not mediated by the induction of other cytokines (such as interleukin l), which might feedback upon the endothelial cell in an autocrine manner. Interaction of LPS with substrates for protein kinase C has been shown (35). The long term PMA exposure experiments reported here have shown, however, that there was desensitization to PMA without loss of response to LPS. This suggests separate mechanisms of action for these two agonists. Differences between LPS and PMA have been demonstrated in other systems where it has been observed that LPS causes the phosphorylation of a different, additional protein kinase C substrate protein to those phosphorylated upon exposure to PMA in the murine macrophage (36).
In conclusion the effect of LPS and PMA upon TF mRNA in HUVEC have been demonstrated. The time course of TF mRNA in response to LPS is different from that previously reported for the human monocyte. In addition the clear stimulatory effect of LPS upon TF transcription seen in the mono&e (12), and by PMA in HUVEC shown here, is not seen when HUVEC are exposed to LPS, and may result from different, cell type-specific regulation of the TF gene. There are additional differences between the effect of LPS and PMA upon HUVEC, as evidenced by PMA-induced desensitization having no effect upon LPS induced TF activity. Therefore, while TF expression is absent in unstimulated HUVEC, it is a characteristic of these cells that they are able to produce an extremely rapid induction of TF upon stimulation, which relies upon a basal level of transcription and an increase in message stability, as well as the capacity to increase transcription in response to stimuli which activate PMA-sensitive protein kinases. These responses appear to facilitate the very rapid switch from an anticoagulant to a procoagulant status in these cells. situated at the luminal surface of blood vessels.