Lipopolysaccharide Induction of Tissue Factor in THP-1 Cells Involves Jun Protein Phosphorylation and Nuclear Factor κB Nuclear Translocation*

Tissue Factor (TF) gene expression is transiently induced in human monocytic THP-1 cells by lipopolysaccharide (LPS). We characterized the transcription factor complexes binding to the TF gene promoter LPS response element (LRE) (−220 to −172), which contains binding sites for nuclear factor κB (NFκB) and activator protein 1 (AP1) transcription factors, and examined the nature of the activation of these factors during a 24-h time course of LPS stimulation. We found proteolysis of the cytoplasmic inhibitory protein IκBα and nuclear translocation of the NFκB/Rel family proteins p65 and c-Rel, corresponding to the transient binding of a p65/c-Rel heterodimer to the κB-like site of the LRE. AP1 binding to the LRE was found to be constitutive, with the majority of the AP1 complexes being JunD/Fra-2 heterodimers. A change in the activation state of the AP1 complexes was, however, found to be transient, as determined by JunD phosphorylation of AP1 bound to the proximal binding site. This directly correlates to the transient activation of Jun N-terminal kinase (SAPK/JNK). These data indicate that LPS induction of TF gene expression in monocytic THP-1 cells is regulated by both the transient phosphorylation of Jun-family proteins and the nuclear translocation and transient binding of NFκB/Rel proteins.

Tissue factor (TF) 1 is the primary initiator of the serine protease cascade of the coagulation system (1). TF is constitutively expressed in a number of different cell types that do not normally come into contact with blood (2) but is of necessity not usually expressed within the vasculature. However, in various disease states, aberrant TF expression in vascular cells may lead to thrombosis, such as during sepsis, when bacterial endotoxin (lipopolysaccharide) induction of TF in monocytes can lead to disseminated intravascular coagulation (3).
Monocytes are the only circulating cells in which TF expression is subject to inducible regulation (4). Synthesis may be up-regulated by a number of different stimuli including phorbol esters (phorbol 12-myristate 13-acetate) (5), tumor necrosis factor-␣ (6), and bacterial lipopolysaccharide (LPS) (7,8). In LPS-induced monocytes, TF mRNA levels increase as a result of transcriptional activation (7). In the monocytic cell line THP-1, TF gene expression is induced by LPS in a similar manner with, however, some increase in mRNA stability as well (9).
Functional studies in THP-1 cells identified an enhancer in the TF gene promoter that mediates LPS induction. This 56base pair region (Ϫ227 to Ϫ172) is termed the LPS response element (LRE) (10). In addition, a second region (Ϫ85 to Ϫ52) containing Egr-1 binding sites has been identified that is also subject to inducible binding (11). Our study focused on the LRE that contains two AP-1 sites (a distal, low affinity site and a proximal, high affinity site) and an NFB-like site. Mutation of any of these sites compromises LPS inducibility (10,12), suggesting that all three are required for optimal LPS induction.
In this study we have analyzed the protein interactions with the LRE binding sites over a 24-h time course of LPS induction in THP-1 cells. We have determined that a number of regulatory mechanisms act to control TF gene transcription in response to LPS stimulation. These include the transient binding of a p65/c-Rel heterodimer from at least 30 min to 2 h, resulting from the proteolysis of IB␣ and nuclear translocation of p65 and c-Rel and transient phosphorylation of JunD in LRE-bound AP-1 complexes correlating to the activation of Jun N-terminal kinase (SAPK/JNK) from 10 min up to 1 h of LPS treatment. These data suggest multiple mechanisms acting co-operatively at the LRE enhancer element to direct a transient increase in TF mRNA levels in monocytic THP-1 cells.

EXPERIMENTAL PROCEDURES
Cell Lines-THP-1 cells (13) were grown in RPMI 1640 with Lglutamine and 25 mM HEPES buffer (Life Technologies, Inc.) supplemented with 10% fetal bovine serum and 100 units/ml penicillin, 100 g/ml streptomycin. Cells were routinely grown to a density of 1 ϫ 10 6 cells/ml and induced with 10 g/ml LPS from Salmonella typhimurium (Sigma) for the times indicated in the figures.
TF Antigen Assay-THP-1 cells were recovered from suspension and washed in phosphate-buffered saline. Cell extracts were then prepared, and TF antigen levels were determined by enzyme-linked immunosorbent assay as described by Consonni and Bertina (14), using TF 4503 monoclonal antibody (American Diagnostica, Greenwich, CT) as the catching antibody and biotinylated TF 5 monoclonal antibody (Costar) as the tagging antibody. Recomboplastin S/Innovin (Baxter Diagnostica Inc., Deerfield, IL) was used as a standard after calibration against the standard of the Immubind tissue factor enzyme-linked immunosorbent assay kit (American Diagnostica Inc., Greenwich, CT). The final results were expressed as ng TF/10 6 cells.
Electrophoretic Mobility Shift Assays-Crude nuclear extracts were prepared for use in electrophoretic mobility shift assays (EMSAs) essentially as described previously (11). Modifications include cell shearing by repeated aspiration through a 27-gauge needle rather than a Dounce homogenizer and the addition of phosphatase inhibitors (0.25 mM orthovanadate and 25 mM ␤-glycerophosphate) to all solutions. Protein concentrations of all extracts were determined using the Bio-Rad protein assay reagent.
Western Blot Analysis-Nuclear and cytoplasmic extracts used in the Western blot analyses were prepared essentially as described for EM-SAs with the exception of Nonidet P-40 being added to the cells to a final concentration of 0.5% following shearing by aspiration, for extracts used to examine NFB compartmentalization. After centrifugation of the lysed cells, the supernatant was frozen in liquid nitrogen and stored at Ϫ70°C as the cytoplasmic extract. Again, the protein concentrations of all extracts were determined using the Bio-Rad protein assay (a modification of the Bradford protein assay (20)). 20 g of nuclear extracts or 60 g of cytoplasmic extracts were fractionated on 10% SDS-polyacrylamide gels and then electrotransferred to a polyvinylidene difluoride membrane (Millipore) in 48 mM Tris, 58.6 mM glycine, 0.1% SDS, 20% methanol at 0.8 V/cm 2 for 2.5 h using the Nova Blot system (Amersham Pharmacia Biotech). Western blots were blocked, and immunoreactive products were detected according to the protocol of the Boehringer Mannheim chemiluminescence Western blotting kit. Briefly, blots were blocked in 1% blocking solution overnight at 4°C then washed in TBST (50 mM Tris, pH 7.5, 150 mM NaCl, 0.1% Tween 20). The blots were then incubated with the primary antibodies (Santa Cruz, Transcruz antibodies diluted 1:10,000 in 0.5% (w/v) blocking solution and phospho-specific antibodies diluted 1:1,000 in TBST, 5% (w/v) bovine serum albumin). The secondary antibody incubation was in 0.5% blocking solution containing 1:10,000 goat antirabbit horseradish peroxidase-conjugated antibody (Bio-Rad) and 1:2,000 anti-biotin antibody (New England Biolabs) to detect the bioti-nylated protein standard used (New England Biolabs). Comparative blots of nuclear and cytoplasmic proteins were performed simultaneously and exposed to film for the same length of time.

LPS Stimulation of Monocytic THP-1 Cells Produces Transient Increases in the Levels of TF Antigen and mRNA-To
determine their response to LPS, monocytic THP-1 cells were incubated with 10 g/ml LPS for various times up to 24 h. TF mRNA, analyzed by Northern blot hybridization ( Fig. 1), increased by 30 min and reached a peak at 1 h. Levels dropped considerably by 2 h and had returned to preinduction levels at the following time points. Larger transcripts resulting from incomplete splicing (17) can be seen in addition to the mature 2.2-kilobase mRNA. A corresponding increase in the level of TF antigen was also observed, reaching a peak at 2 h then decreasing through to 24 h (data not shown).
Binding of AP1 and NFB Complexes-The EMSA studies presented were predominantly carried out using the AP1 P B oligonucleotide (Ϫ213 to Ϫ172). This region of the TF LRE was chosen as AP1 consists of various dimers of Fos and Jun family proteins, and this oligonucleotide together with TFAP1 distal , allowed us to distinguish between AP1 complexes binding to the proximal and distal AP1 sites while maintaining possible effects on complex binding arising from the proximity of the adjacent high affinity proximal AP1 site and the B-like site. We found that variation in the binding reaction conditions had a noticeable effect on the relative intensities of the DNA-protein complexes that were observed (Fig. 2). Under conditions that seemed to favor AP1 binding (binding buffer 1) two complexes became apparent whose intensity appears to be relatively constant throughout the time course ( Fig. 2A). A very faint larger complex (I) and a smaller doublet complex (II), which constituted the majority of the binding to this oligonucleotide, are seen. Both complexes I and II could be competed with a 100-fold molar excess of AP1 P B and an AP1 consensus THP-1 cells were cultured in the absence or presence of 10 g/ml LPS for the times indicated. Total RNA was fractionated and transferred to a membrane; TF mRNA levels were determined using Northern hybridization with a TF cDNA probe. 15 g of total RNA/time point were loaded. Equivalence of loading was examined by rehybridization of the stripped blot with a cDNA probe for the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) house-keeping gene. kb, kilobases. competitor but not with an NFB consensus or a nonspecific competitor (Sp1), indicating that both complex I and II are AP1 complexes. Constitutive binding of an AP1 complex throughout the 24-h time course was also observed using an oligonucleotide containing the distal AP1 site (data not shown).
Under the conditions of binding buffer 2 (Fig. 2B), we see a much lower intensity of complex II, with complex I no longer visible, whereas complexes III and IV become more obvious. Complexes III and IV begin to appear at 30 min, with binding reaching a peak at 1-2 h. At 4 h and later, their presence is no longer detected. A 100-fold molar excess of AP1 P B or the NFB consensus oligonucleotide both compete complexes III and IV. The AP1 consensus competes complex II (as seen in Fig.  2A), whereas the nonspecific competitor (Sp1) appears to compete only complex IV. These data demonstrate the transient binding of two NFB complexes to the oligonucleotide between 30 min and 2 h, of which only complex III appears to have high affinity for the B-like site.
Under conditions where the ratio of nuclear proteins to oligonucleotide is much higher (200-fold higher, binding buffer 1, Fig. 2C) we see the appearance of a new larger complex (V). This complex is present from 30 min to 2 h, similar to complexes III and IV binding in Fig. 2B. The AP1 P B oligonucleotide and the NFB consensus oligonucleotide compete complex V, as does the AP1 consensus competitor (in addition to complex II). A nonspecific competitor (Sp1) has no effect. These results indicate that complex V is a complex of the oligonucleotide AP1 P B with both AP1 and NFB binding simultaneously.
Characterization of the NFB Complex-Supershift assays were carried out with the AP1 P B oligonucleotide, the TF Blike site oligonucleotide and the NFB consensus oligonucleotide, and antibodies against specific NFB/Rel proteins. With AP1 P B (Fig. 3, lanes 1-10), supershifts of complex III were observed with an antibody against p65 and to a lesser extent, anti-c-Rel, but not with antibodies specific for other members of the NFB/Rel family. The complex formed with the TFB oligonucleotide (Fig. 3, lanes 11-20) had the same mobility and transient binding characteristics as complex III and was also supershifted by antibodies against p65 and c-Rel. The affinity of protein complexes for the TFB site in isolation was much lower than with the adjacent AP1 proximal site present (lanes 11-20 of Fig. 3 were exposed four times longer than the rest of the assay). The complex formed with the NFB consensus oligonucleotide (Fig. 3, lanes 21-30) was supershifted by antibodies against p50 and p65 but not by antibodies specific for other members of the NFB/Rel family. These data show complex III to be a p65/c-Rel heterodimer.
LPS Induces Proteolysis of IB␣ and Translocation of p65 and c-Rel from the Cytoplasm to the Nucleus-Nuclear and cytoplasmic extracts from THP-1 cells stimulated with LPS for the times indicated in the figure were examined by Western blot analyses (Fig. 4). Antibodies against p65 and c-Rel, the component proteins of the higher affinity complex III, and also the cytoplasmic inhibitory protein IB␣, were used. When nuclear extracts used in the EMSA studies were examined with an antibody against a cytoplasmic protein (␣-tubulin) in Western blot analysis a certain proportion of cytoplasmic protein was observed. Therefore, to attain a greater degree of nuclear/ cytoplasmic separation, an adapted technique of preparation was used (see "Experimental Procedures"). However, rather than single bands corresponding to the p65 and c-Rel proteins, we observe multiple bands after immunostaining of the Western blots, which we would suggest are because of partial degradation of the proteins during this preparation procedure.
However, because we are mainly concerned with the localization of these proteins, the degradation was not considered to impede the interpretation of the data. IB␣ and Fos/Jun family proteins show relatively little degradation on Western blots, indicating that p65 and c-Rel may be particularly susceptible to breakdown with this preparation method. We can see that very little p65 is evident in the nucleus in unstimulated cells. After 10 min of LPS induction, nuclear p65 begins to appear and peak at 1 h, declining again by 2 h. A concomitant decrease in cytoplasmic p65 corresponds to the observed increase in nuclear p65.
In the case of c-Rel, again very little protein is evident in the nuclei of unstimulated cells or at 10 min. However, by 30 min to 1 h, nuclear c-Rel has increased to a level that, unlike p65, appears to be maintained up to 24 h. The level of cytoplasmic c-Rel seems relatively constant with only a slight dip observed at 1 h, corresponding to the peak in nuclear c-Rel levels.
The amount of IB␣ seen in the cytoplasm drops sharply at 30 min, with levels beginning to increase again at 1 h, peaking at 2-4 h, and reaching preinduction levels by 24 h. This peak in IB␣ expression corresponds to a peak in IB␣ mRNA levels at 1 h (data not shown). Nuclear levels of IB␣ in Western analysis were found to be low and constant, confirming breakdown and re-synthesis of cytoplasmic IB␣ rather than translocation. These data are consistent with the proposal that IB␣ is proteolyzed after induction of the NFB system, releasing the NFB proteins, which then translocate to the nucleus (10,19,23,24). An NFB element in the IB␣ promoter (25) directs the increase in IB␣ levels at later time points.
Characterization of the AP1 Complexes-Nuclear extracts prepared from THP-1 cells stimulated with LPS through a 24-h time course were used to analyze the identity of complex II in supershift assays (Fig. 5). In stimulated and unstimulated cells, complex II is completely supershifted by an antibody against JunD (Fig. 5A). An unidentified complex that seems to have a slightly higher mobility than complex II becomes evident when complex II is supershifted by anti-JunD. Antibodies specific for c-Jun and ATF-2 have no observable effect on complex II. The effect of an antibody against JunB was also investigated with negative results (data not shown).
Further supershifts with antibodies against Fos proteins were also carried out. In the absence of LPS, complex II is completely supershifted by a broadly reactive Fos antibody. Anti-Fra-2, although not producing an observable supershift, does significantly decrease the intensity of complex II. Anti-c- Fos has no effect. This situation remains the same following LPS stimulation of up to 24 h. The effects of antibodies against FosB and Fra-1 were also analyzed with negative results (data not shown).
Our data show complex II to be a JunD/Fra-2 heterodimer both before and after LPS stimulation of up to 24 h. The residual complex II observed in the presence of anti-Fra-2 may either be because of the antibody having a relatively low affinity for Fra-2 or because of the presence of an as yet unidentified Fos-related protein that forms a dimer with JunD.
Supershift assays with the AP1 distal site oligonucleotide demonstrated the binding of both JunD/Fra-2 and JunD/c-Fos heterodimers with no change in the components of the complex through the 24-h time course (data not shown). These data suggest slight differences in binding at the two adjacent AP1 sites.
The Transient Phosphorylation of JunD-In the absence of a significant quantitative change in binding to the proximal AP1 site through the 24-h time course (see Fig. 2A), we examined whether there was a change in the phosphorylation state of JunD. Supershift assays (Fig. 6) were carried out using anti-JunD alone or in combination with an antibody specific for JunD phosphorylated at Ser-100 (c-Jun phosphorylated at Ser-73, a conserved site, is also recognized). In the absence of LPS, complex II is supershifted with anti-JunD, with the addition of the phospho-specific Jun antibody (␣-p-Jun) having little effect. At 30 min to 1 h of LPS stimulation, the anti-phospho-Jun antibody caused a further shift in the anti-JunD supershift (supershift I), producing a lower mobility complex (supershift II). The amount of observable "supershift II" is considerably reduced at periods of LPS incubation of 2 h and longer. These data indicate that the majority of the JunD bound to the AP1 P B oligonucleotide is not phosphorylated at its transactivation domain (Ser-100) in the absence of LPS, but with LPS induction of up to 1 h, transient phosphorylation occurs. The presence of phosphorylated JunD at 30 min to 1 h of LPS stimulation was confirmed by Western blot analyses using the phospho-specific Jun antibody (data not shown).
We also analyzed the activation state of SAPK/JNK, whose downstream targets include JunD. The data indicate no change in the levels of nuclear or cytoplasmic SAPK/JNK (Fig. 7, A and C) but clearly show activation by phosphorylation at the Thr-183/Tyr-185 residues beginning at 10 min and peaking at 30 min to 1 h (Fig. 7, B and D). The activation of SAPK/JNK directly reflects the appearance of phosphorylated JunD. DISCUSSION LPS activation of monocytes and monocytic THP-1 cells produces a transient increase in the levels of TF mRNA and TF antigen and activity (8,9). Our data charting these levels in THP-1 cells through a 24-h time course show a correlation to previous reports, with a rapid induction of TF mRNA from 30 min to 1 h of LPS stimulation and TF antigen levels peaking at 2 h.
When we examined the TF gene promoter LRE in DNA binding studies, we found the transient binding of an NFB complex and the constitutive binding of AP1 complexes. The NFB complex was observed to bind the LRE site from at least 30 min up to 2 h of LPS stimulation and was found to contain p65 and c-Rel in a heterodimeric complex. The TF LRE NFBlike sequence has been shown to be an optimal site for the binding of p65 and c-Rel but not p50 (26). A number of previous reports studying this NFB complex identified its component proteins as p65 and c-Rel, in agreement with our data (26,27).
A different study, however, suggested that an Ets transcription factor is binding to the core sequence of the B-like site upon LPS stimulation rather than an NFB complex. We found no evidence of Ets proteins binding to this site using an antibody against Ets 1/2 in supershift assays (data not shown).
To confirm the predicted model of NFB activation, we compared nuclear and cytoplasmic extracts for evidence of nuclear translocation. Our data show the proteolysis of the cytoplasmic inhibitory protein IB␣ and the translocation of both p65 and c-Rel from the cytoplasm to the nucleus, corresponding to the appearance of the NFB complex in gel shift studies. An earlier study charting the nuclear and cytoplasmic levels of the NFB proteins in THP-1 cells (19) found that p65 was present in the nucleus before LPS induction, with levels increasing on stimulation and continuing to be elevated for up to 24 h. This contrasts with our data, where p65 was not detected before stimulation and only translocated to the nucleus for a very short period (up to 2 h), corresponding to the appearance of NFB complexes in the binding studies and the transiently elevated levels of TF gene transcription. This same study, in agreement with our data, observed the rapid nuclear translocation of c-Rel with elevated levels persisting for up to 24 h. The fact that c-Rel is present at these longer periods of LPS stimulation yet does not appear to bind the TF B-like site suggests that either this site is specific for the binding of p65/c-Rel heterodimers and not c-Rel homodimers or that the DNA binding capacity of c-Rel is subject to regulation. A recent study demonstrated that IB␣ may also regulate the transcriptional activity of c-Rel in the nuclear compartment (28), and other studies have shown that the DNA binding capacity of certain NFB proteins may be controlled by redox mechanisms (29) or at the level of phosphorylation (30 -32).
The EMSA data also suggest that the affinity of p65/c-Rel for the B-like site is considerably increased if the proximal AP1 site is present. Although this effect has not been quantitated in our study, comparison of the binding to the AP1 P B oligonucleotide and the isolated TFB oligonucleotide showed that much longer exposure times were required to observe complexes with TFB than with AP1 P B (Fig. 3, lanes 11-20 and  1-10, respectively, probes labeled to similar specific activities). Because under these conditions multiprotein-DNA complexes, which might involve protein-protein interactions, were not ob-served, the effect appears to lie in the structural influence of the DNA itself. Although this effect may be a result of the size of the oligonucleotides used and the position of the transcription factor binding sites relative to the ends of the oligonucleotides, it may also be a result of adjacent sequence influencing binding characteristics. A recent study demonstrated such a structural role for the AP1 sites in the TF promoter when their replacement by intrinsically bent DNA was able to partially restore LPS induction (12).
The AP1 sites of the LRE are essential for LPS induction, with mutation of either site compromising LPS inducibility (10,12). AP1 is a dimer of proteins of the Fos/Jun family. Jun proteins are able to bind to DNA as homodimers or as Fos/Jun heterodimers. Fos proteins, however, are obliged to form heterodimers to bind (33,34). Previous studies analyzing the composition of the LRE-bound AP1 complexes in THP-1 cells report differing binding profiles at these sites. Oeth et al. (12) looked at the two AP1 sites in isolation in THP-1 cells and found c-Jun/c-Fos heterodimers both pre-and post-LPS stimulation, with the amount of protein binding exhibiting some increase after stimulation; these results are consistent with the constitutive expression of c-Jun and c-Fos in human monocytes and their induction in response to LPS (35). A second study in THP-1 cells (11) used a larger oligonucleotide containing the entire LRE with an additional downstream Sp1 site and found LPS induction of c-Jun binding. c-Fos/JunD heterodimers were found preinduction, whereas following LPS stimulation, both c-Fos/JunD and c-Fos/c-Jun complexes were observed. Our data examining AP1 LRE binding through a 24-h time course of LPS induction shows a relatively constant degree of AP1 binding to AP1 P B throughout the 24-h time course, although we have consistently observed a decrease in the amount of complex II at 4 h and a subsequent increase at 24 h ( Fig. 2A). Consistent with the study of Groupp and Donovan-Peluso (11), we find the majority of AP1 binding activity to contain JunD. In our study, JunD/Fra-2 bound at the proximal AP1 site (complex II), and both JunD/Fra-2 and JunD/c-Fos bound at the distal AP1 site, both at pre-and post-induction time points.
The Jun proteins may be regulated in their DNA binding and transactivation activities by means of changes in their phosphorylation state. Phosphorylation of c-Jun at the C terminus by casein kinase II or glycogen synthase kinase 3 prevents DNA binding (36), whereas phosphorylation at the N-terminal region (Ser-63/Ser-73) by the mitogen-activated protein kinase homologue SAPK/JNK enhances transcriptional activation. Ser-100 of JunD is a conserved phosphorylation site corresponding to Ser-73 of c-Jun and is also a target for SAPK/JNK (37,38). Because our data showed no significant change in the degree of AP1 binding or in the composition of the complexes, we examined evidence for the activation states of the Jun proteins by assessing the phosphorylation of the amino acids that influence their transactivation domains. Our data clearly show the transient (30 min to 1 h) phosphorylation of JunD in the JunD/Fra-2 complex bound to the proximal AP1 site (Fig. 6,  lanes 6 and 9). The activation of SAPK/JNK, for which JunD is a target, directly correlates to the appearance of phosphorylated JunD. Although only a small proportion of the cellular JunD is phosphorylated when examined by Western blot analysis (data not shown), gel mobility shift assays indicate that a high proportion of the proximal AP1 site-bound JunD is phosphorylated (Fig. 6), raising the question as to whether there may be some form of coordinated regulation of DNA binding and transactivation enhancement. Although we did not assess the phosphorylation state of JunD bound to the distal AP-1 site, it seems probable that it, too, is phosphorylated at the transactivation domain upon LPS stimulation. Our evaluation of transcription factor binding to the TF gene promoter LRE throughout a 24-h time course of LPS induction demonstrates a number of regulatory features (Fig. 8). These include the transient, inducible binding of a p65/c-Rel heterodimer to the TF B-like site (30 min to 2 h) corresponding to the proteolysis of IB␣ and the nuclear translocation of p65 and c-Rel as predicted by previously proposed models of NFB activation (10,19,23)