Tumor Necrosis Factor α Up-regulates in an Autocrine Manner the Synthesis of Plasminogen Activator Inhibitor Type-1 during Induction of Monocytic Differentiation of Human HL-60 Leukemia Cells

Tumor necrosis factor-α (TNFα) critically regulates several cellular functions during monocyte/macrophage differentiation. We therefore investigated during the phorbol ester (phorbol 12-myristate 13-acetate (PMA))-induced monocyte/macrophage differentiation of the human HL-60 leukemia cells, if TNFα contributed to plasminogen activator inhibitor type-1 (PAI-1) synthesis that is initiated by a protein kinase Cβ-extracellular signal-regulated kinase 2-dependent pathway (Lopez, S., Peiretti, F., Morange, P., Laouar, A., Fossat, C., Bonardo, B., Huberman, E., Juhan-Vague, I., and Nalbone, G. (1999) Thromb. Haemostasis 81, 415–422). Following PMA treatment, the level of TNFα mRNA strongly increased and appeared earlier than PAI-1 mRNA. An anti-TNFα antibody significantly inhibited the PMA-induced PAI-1 mRNA and protein levels. The recombinant human TNFα, which is inactive on native HL-60 cells in terms of PAI-1 synthesis, optimally potentiates it once HL-60 cells are committed into the differentiation process. The use of 1) the HL-525 cell line, a clone issued from HL-60 cells rendered resistant to PMA-induced differentiation, and 2) the transforming growth factorβ-1/vitamin D3 differentiative mixture confirmed the relationships between the induction of differentiation and the potency of TNFα to up-regulate PAI-1 synthesis. In conclusion, we showed that during the induction of monocyte/macrophage differentiation, TNFα and PAI-1 gene expressions are activated and that synthesized TNFα up-regulates and prolongs, in an autocrine manner, the synthesis of PAI-1.

thesize TNF␣ (30), which was recently demonstrated to upregulate, in an autocrine manner, the synthesis of 92-kDa gelatinase via ␣ 5 ␤ 1 integrin expression (31). Because a close relationship exists between PAI-1 and pericellular proteolytic activity, this drove us to use this model to examine if TNF␣ contributes to PAI-1 synthesis during induction of leukocyte differentiation and if so, how it exerts this regulation.
Results showed that, during the early steps of the monocytic differentiation program, TNF␣ and PAI-1 mRNA levels increased. The synthesized TNF␣ up-regulates and prolongs, in an autocrine manner, elevated PAI-1 mRNA levels resulting in enhanced protein antigen synthesis and secretion.

EXPERIMENTAL PROCEDURES
Materials-Phorbol esters (PMA), phorbol dibutyrate (PdBu), and PD 098059 were purchased from Alexis (Coger). PKC inhibitor Ro 31-8220 was provided by Dr. Bradschaw (Roche Laboratories). Monoclonal antibodies (12A4 and 15H12) specific for human PAI-1 were generously given by Dr. P. Declerck (Center for Thrombosis and Vascular Research, Leuven, Belgium). Monoclonal antibody against TNF␣ were from R&D Systems. Recombinant human TGF-␤1 (rhTGF-␤1) and TNF␣ (rhTNF␣) were from Amersham Pharmacia Biotech. The Moloney-murine leukemia virus reverse transcriptase and its appropriate buffer were purchased from Life Technologies, Inc. Taq polymerase and its appropriate buffer were from Bioprobe. Other molecular biological products (dNTP, random hexaprimers, RNasin ® , and appropriate buffers) were from Promega. Phorbol esters, PKC, and MAPKK inhibitors were stored at Ϫ20°C in Me 2 SO. All trans retinoic acid (RA) and 1␣,25-dihydroxyvitamin D3 (D3) were stored at Ϫ20°C in ethanol. Appropriate dilutions were made in warm culture medium in such a way that the final Me 2 SO or ethanol concentration in the presence of cells did not exceed 0.1% (v/v). Proper controls made with similar Me 2 SO or ethanol concentrations were without effects on the parameters studied.
Cell Culture-The human promyelocytic HL-60 leukemia cell line and the HL-525 cell line that is resistant to PMA-induced differentiation (32) were kindly provided by Pr. E. Huberman (Argonne National Laboratory, Argonne, IL). HL-60 and HL-525 cells were grown in RPMI containing 10% fetal calf serum as already described (22). Cells in 72-h-old culture medium were resuspended for 1 h in fresh medium and then treated with differentiative agents.
RNA Extraction and Semi-quantitative RT-PCR Analysis-Total RNA extraction (extraction kit for total RNA, RNeasy from Quiagen) and cDNA synthesis were performed as recently described (22). The amplified fragment for human PAI-1 (GenBank TM accession number X04744) is 284 bp, base position 10977-11260 and that for human TNF␣ (GenBank TM accession number M10988) is 296 bp, base position 761-1056. The amplified fragment of TNF-RI (p55) (GenBank TM accession number X55313) is 261 bp, base position 786 -1046 and that of TNF-RII (GenBank TM accession number M55994) is 391 bp, base position 594 -984. In PMA-treated cells, eukaryotic elongation factor (eEF1␣) was observed to be a stable house keeping gene, whereas in TGF-␤1/D3-treated HL-60 cells, ␤-actin was preferred. The amplified fragment for eEF1␣ (GenBank TM accession number X03558) is 289 bp, base position 382-670 and is 388 bp for ␤-actin (GenBank TM accession number X00351), base position 379 -766. PCR was performed on a Perkin-Elmer thermocycler (GeneAmp 2400). We selected a number of 25 cycles for TNF␣ and its receptors, PAI-1, and ␤-actin cDNAs and a number of 18 cycles for eEF1␣. Specificity of the amplified fragment was assessed by the demonstration that appropriate restriction enzymes generated the expected cleavage fragments. PCR started for 2 min at 95°C followed by cycles consisting of: 60 s at 58°C (for PAI-1, eEF1␣, and ␤-actin) or 57°C (for TNF␣ and its receptors), 90 s at 72°C, and 45 s at 97°C. Amplification was terminated after 5 min at 72°C. Products were visualized and photographed under UV radiation following gel-agarose (2%) electrophoresis.
Protein Assays-Total proteins of cell lysates were assayed according to specifications of the bicinchoninic acid protein assay kit from Sigma. PAI-1 antigen assay was performed on supernatants from conditioned culture medium and on cell lysates (Triton X-100 0.1% final) by enzymelinked immunosorbent assay as described by Declerck et al. (33). TNF␣ antigen assay was performed on culture supernatants according to the specifications provided with the enzyme-linked immunosorbent assay kit (Coulter-Immunotech).
Flow Cytometry Analysis-Expression of TNF␣ receptors p55 (RI) and p75 (RII) were analyzed by flow cytometry. The experimental protocol was comparable to that previously described for the urokinase receptor (27). Briefly, HL-60 cells (10 5 ) were first acid-treated (glycine buffer, pH 3.0, 1 min at 37°C) to remove bound TNF␣. Cells were then incubated with fluorescein isothiocyanate-conjugated monoclonal antibody against TNF RI or RII (R&D Systems) and washed. Fluorescein isothiocyanate-labeled cells were analyzed on a XL-cytofluorograph (Coulter Electronics Inc.) at 488 and 525 nm, corresponding to excitation and detection wavelengths, respectively.
Statistics-Each experiment was performed in duplicate or triplicate. Results are expressed as mean Ϯ S.D. Comparisons were analyzed by an analysis of variance (ANOVA) test, and significance was calculated at p Ͻ 0.05 or p Ͻ 0.01 using the Scheffe F-test.

PMA Increases TNF␣ mRNA Level Earlier than
That of PAI-1 In nonstimulated HL-60 cells, the level of TNF␣ mRNA was low (Fig. 1A). When HL-60 cells were treated with 20 nM PMA, the TNF␣ mRNA level rose rapidly. The optimum expression was attained 2 h after PMA addition. The increase in TNF␣ mRNA level was transient since 24 h after PMA addition, the level was below that of nonstimulated cells (Fig. 1A). TNF␣ antigen accumulation increased by a factor 3.6 in the culture medium of HL-60 cells treated with PMA (Fig. 1B). The PAI-1 mRNA level (Fig. 1A) was not detectable in basal conditions but was strongly induced 4 h after PMA addition and remained elevated over 24 h. PAI-1 antigen accumulation was dramatically enhanced at 24 h compared with nontreated HL-60 cells (Fig. 1B).

Endogenous TNF␣ Up-regulates PMA-induced PAI-1 Synthesis
In view of the above results and because PAI-1 is a TNF␣responsive gene in various differentiated cells, we then examined if the synthesis of PAI-1 was attributable only to the activation of the PKC␤-MAPKK-MAPKp42 pathway (22) or if the released TNF␣ contributed to the synthesis of PAI-1. HL-60 cells were simultaneously incubated with a neutralizing antibody against human TNF␣ and PMA. Preliminary experiments revealed that the optimal effect was obtained at 2.5 g/ml anti-TNF␣. We used 10 g/ml to ensure complete quenching of released TNF␣. As shown in Fig. 2A, anti-TNF␣ significantly lowered the level of PAI-1 mRNA 4 h after PMA addition indicating that endogenously released TNF␣ is already active at this time. This inhibition strongly persisted at 10 h. The intracellular level of PAI-1, measured 8 h after PMA addition, was reduced by 41% (from 6.9 Ϯ 2.1 down to 4.06 Ϯ 2.0 ng/mg proteins), whereas PAI-1 antigen accumulation in the culture decreased by 65% at 24 h (Fig. 2B). IgG1 (10 g/ml) used as a control isotype in anti-TNF␣ studies did not inhibit PMAinduced PAI-1 synthesis and even tended to increase it by 20% (not shown). This increase could be explained by the fact that in HL-60 cells, IgG immunoglobulins activate the production of TNF␣ (34). The IgG-induced TNF␣ release might in turn potentiate PAI-1 synthesis once differentiation is initiated by PMA. Morphological examination of HL-60 showed that the addition of anti-TNF␣ tended to reduce the spreading of cells, which appear more rounded when compared with PMA-treated cells (Fig. 3, A-C). Twenty four hours after PMA addition, the percentage of adherent cells was 95 Ϯ 3% and 62 Ϯ 10% when anti-TNF␣ was added. These results indicate that, in PMAstimulated HL-60 cells, endogenous TNF␣ up-regulates, in an autocrine manner, the PAI-1 synthesis.

The Up-regulating Effect of TNF␣ on PAI-1 Synthesis Is
Linked to the Induction of Differentiation rhTNF␣ Activates PAI-1 Synthesis in PMA-treated HL-60 Cells-rhTNF␣ did not induce any changes in the accumulation of PAI-1 antigen from native HL-60 cells, which did not exceed 1 ng/10 6 cells ( Fig. 2B) at 24 h. When rhTNF␣ was simultaneously added with PMA, it induced a modest increase in PAI-1 antigen level. This was characterized by an increase in the intracellular compartment of 18% after 8 h (from 6.9 Ϯ 2.1 to 8.3 Ϯ 2.4 ng/mg proteins) and 55% in the culture medium after 24 h (Fig. 2B). To rule out possible endotoxin contamination, rhTNF␣ was boiled for 5 min and added on PMA-treated cells. No enhancement of PAI-1 synthesis could be observed in these conditions. The addition of rhTNF␣ tended to increase the spreading of adherent cells ( Fig. 3D) but did not change the PMA-induced percentage of adherent cells. The increase in PAI-1 synthesis was not detected at the mRNA level ( Fig. 2A). The moderate effect of rhTNF␣ on PAI-1 synthesis could be due to the rapid down-regulation of the expression of TNF receptors I and II following PMA treatment as described previously in U937 cells (35) and HL-60 cells (36). To investigate this possibility, we examined TNF␣ receptor (RI and RII) expression by RT-PCR and flow cytometry. As shown in Fig. 4A, the level of RI mRNA progressively increased the first 14 -20 h then returned to the basal level at 24 h. The RII mRNA basal level, higher than RI, slightly decreased at 6 h and then was strongly  4B). The transient increase observed in untreated cells is likely the result of the renewal of the 72-h-old culture medium, because the basal expression of RI and RII was recovered around 36 h after culture medium renewal (not shown). However, this transient increase was less pronounced in PMA-treated than in untreated HL-60 cells. This is in line with previous data showing that PMA rapidly down-regulates RI and RII surface expression (35,36), but is not strong enough in our experimental conditions to induce an absolute decrease when compared with their levels at t ϭ 0.
The lack of effect of rhTNF␣ on PAI-1 synthesis observed on native undifferentiated HL-60 cells is not linked to some defect in TNF␣ receptors, because they are expressed in a constitutive manner as we demonstrated above and are operational in these cells (37,38). This led us to suggest the existence of a functional link between the differentiation state and the potency of TNF␣ to activate PAI-1 gene expression. We thus evaluated the relationship between the time-dependent monocyte/macrophage differentiation and the potency of TNF␣ to activate PAI-1 synthesis. To specifically evaluate the effect of rhTNF␣ from the global PMA effect, we treated HL-60 cells with a metabolizable phorbol ester (phorbol dibutyrate) for various times (i.e. 3, 6, 14, 24, 36, and 48 h after initial PdBu addition) at which conditioned medium was discarded. The cells were washed and then either stimulated or not (control) by rhTNF␣ for 24 h. As shown in Fig. 5, the enhancing effect of rhTNF␣ appears optimal 14 -24 h after the induction of monocytic differentiation and then declines progressively but remains higher than that of native undifferentiated HL-60 cells stimulated by rhTNF␣ alone. It should be mentioned that these results cannot be compared with those of the PMA-induced up-regulating effect described in Fig. 2. Indeed, once the PdBu-containing medium is removed, intracellular PdBu is rapidly degraded, which stops cellular differentiation related events. These results show that rhTNF␣ activates PAI-1 synthesis once HL-60 cells have been "primed" by a differentiative stimulus.

Relationships between TNF␣ and PAI-1 Synthesis in the HL-525 Cell Line Resistant to PMA-induced Differentiation-The
HL-525 cell line was cloned from HL-60 cells rendered resistant to PMA-induced monocytic differentiation after long term PMA treatment. This results in a down-regulation of PKC␤ expression (32,39). Restoration of PMA responsiveness in terms of monocyte/macrophage differentiation can be obtained by pretreatment of HL-525 cells with RA that re-induces PKC␤ gene expression (40). Fig. 6A shows that, in HL-525 cells, the level of TNF␣ mRNA was slightly enhanced by PMA alone, although it was lower and appeared later than in HL-60 cells. Treatment of HL-525 cells with RA alone had no effect on TNF␣ mRNA levels, except a slight transient increase at 6 -8 h. Clearly, RA pretreatment allows PMA to progressively and strongly enhance TNF␣ mRNA levels. It should be noticed that the optimum level of TNF␣ mRNA significantly increased much later than in HL-60 cells (10 versus 2 h). The amount of TNF␣ antigen secreted into the culture medium (Fig. 6B) showed that, in comparison with control HL-525 cells, PMA modestly enhanced TNF␣ antigen accumulation, whereas RA alone had no marked effect. The combination of RA ϩ PMA increased the antigen level of TNF␣ in the culture medium. This level was 60% higher than that from HL-60 cells (cf. Fig. 1B), and the increase continued for 48 h.
We then evaluated the synthesis of PAI-1 in HL-525 cells under various conditions. The PAI-1 mRNA level was hardly detectable in control HL-525 cells as well as in cells treated by RA alone for 72 h or longer (not shown). As shown in Fig. 7A, PMA alone has also no significant effect on the PAI-1 mRNA level, whereas the RA ϩ PMA treatment resulted in a modest increase, which appeared 3 h after PMA addition and contin-ued over 9 h to slightly decrease at 24 h. The accumulation of PAI-1 antigen that was not increased by either PMA or RA alone significantly increased in RA ϩ PMA-treated HL-525 cells confirming previous data (22). Anti-TNF␣ reduced the level of PAI-1 mRNA in RA ϩ PMA-treated HL-525 cells but, the inhibitory effect could not be firmly detected before 9 h (Fig.  7A). In HL-525 cells treated with PMA alone, rhTNF␣ did not significantly alter PAI-1 synthesis, both at mRNA and protein levels (not shown). However, when rhTNF␣ was added to RA ϩ PMA-treated HL-525 cells, a rapid and strong enhancement of the PAI-1 mRNA level was observed at 3 h and continued to increase over 24 h. This enhancement resulted in an increase in PAI-1 antigen accumulation by a factor of 3 at 24 h and of 6.5 at 48 h (Fig. 7B) when compared with RA ϩ PMA-treated HL-525 cells. Unlike native rhTNF␣, boiled rhTNF␣ was ineffective. Fig. 3, E-H shows that HL-525 cells treated by the combination of RA ϩ PMA develop a marked adhesion with some spreading and cell extensions. These morphological patterns, particularly cell extensions, were strongly intensified when rhTNF␣ was added.
Relationships between TNF␣ and PAI-1 Synthesis in the Presence of Other Inducers of Differentiation-We evaluated if the autocrine and stimulating effect of TNF␣ was specific to PMA-induced differentiation or also occurs with other agents known to trigger the monocytic differentiation program. The treatment of the promonocytic cell line U937 with rhTGF-␤1/D3 for 24 h has been described to induce monocytic differentiation (41). We treated HL-60 cells for 24 h with this mixture. The cells were then either stimulated or not by rhTNF␣ (t ϭ 0). PAI-1 and TNF␣ mRNA levels were analyzed at 0, 5, and 8 h after stimulation (Fig. 8A). The combination of rhTGF-␤1/D3 alone increased the level of PAI-1 mRNA but not that of TNF␣ mRNA. The level of TNF␣ secreted in the culture medium was not different from that of nontreated cells (not shown). Interestingly, in rhTGF-␤1/D3-treated HL-60 cells, rhTNF␣ enhanced the level of PAI-1 mRNA (Fig. 8A). Consistent with PAI-1 mRNA levels, PAI-1 antigen accumulation was increased by rhTGF-␤1/D3 and was drastically potentiated by a factor of 4 after the addition of rhTNF␣ (Fig. 8B). In cells treated with rhTGF-␤1/D3, the percentage of adherent cells did not exceed 10 -15% at 24 h, but increased up to 50% when rhTNF␣ was added. However, adherence is less firm than with PMA, because cells can be detached by shaking. We then investigated if PKC and MAPKp42/p44 were involved in the TGF-␤1/D3-induced PAI-1 synthesis, as we described with PMA (22). Neither Ro 31-8220 (0.5 M), a PKC inhibitor, nor PD 098059 (10 M), a MAPKK inhibitor, altered the level of PAI-1 antigen in the culture medium of rhTGF-␤1/D3-treated cells (not shown). This result indicates that the induction of differentiation by rhTGF-␤1/D3 involves a signaling pathway different from that induced by PMA. DISCUSSION TNF␣ is produced and released by activated macrophages in the atherosclerotic lesion and is a major inducer of PAI-1 synthesis in differentiated cells. However, its effect on PAI-1 in leukocytes appears linked to their differentiated stage because 1) TNF␣ is inactive on monocytic progenitors (22,29) and 2) mature macrophages produce more PAI-1 than monocytes isolated from circulation (7,8). To gain insight on this aspect, we used the undifferentiated human HL-60 leukemia cells that, when treated with differentiative agents, acquire the monocyte/macrophage phenotype (19) and synthesize PAI-1. During monocytic differentiation, PAI-1 gene expression depends on the early activation (5 min) of the PKC␤/MAPKK/MAPKp42 pathway (22). Herein, we show that an excess of anti-TNF␣ dramatically reduced the PMA-induced synthesis of PAI-1, observed at the mRNA level, 4 h after PMA stimulation. This result strongly suggests that PMA-induced PAI-1 synthesis is in large part dependent on the autocrine action of synthesized TNF␣. The modest up-modulating effect of rhTNF␣ on PAI-1 synthesis does not seem to result from a down-regulation of TNF␣ receptor surface expression, because it modestly increased after PMA addition, although lesser than in control cells. Therefore, the occupation of these receptors by endogenously released TNF␣ likely prevents rhTNF␣ to strongly up-regulate PAI-1 synthesis. Because the increase in the TNF␣ mRNA level preceded that of PAI-1, one may address the question if TNF␣ is absolutely indispensable to initiate PAI-1 synthesis. In the differentiation-resistant HL-525 cells cloned from HL-60 cells (32,39), RA ϩ PMA treatment, which restores the PKC␤-dependent Raf-1-MAPKK-MAPK pathway (42), increased PAI-1 mRNA levels before those of TNF␣ which plateaued at 10 h. Consequently, anti-TNF␣ is effective much later than in HL-60 cells (9 versus 4 h) to inhibit PAI-1 synthesis. Thus, the increase in PAI-1 mRNA level observed during the first 9 h in RA ϩ PMA-treated HL-525 cells is mainly the result of the restoration of the PKC-␤-dependent pathway. Also, in HL-60 cells a large excess of anti-TNF␣ did not fully inhibit the PMA-induced PAI-1 synthesis, suggesting that TNF␣ is not an obligatory preliminary step to initiate PAI-1 synthesis. As these data supported the idea that the critical role of TNF␣ is not to initiate PAI-1 synthesis, we therefore addressed the question as to whether TNF␣ up-regulates it once the differentiation process is committed. Our results indicate that this is the case.
First, in PMA-treated HL-525 cells, rhTNF␣ that did not activate PAI-1 synthesis strongly potentiates it when re-induction of differentiation by RA was allowed. However, it should be noted that in RA ϩ PMA-treated HL-525 cells PAI-1 synthesis is lower than in HL-60 cells (22), although TNF␣ is produced at levels comparable to those produced by HL-60 cells. The reasons for this low efficiency are not clear at present. A functional alteration of TNF␣ receptors appears unlikely, because the addition of rhTNF␣ recovered the elevated rate of PAI-1 synthesis. It is however, noteworthy that the optimum level of TNF␣ mRNA appeared much later than in HL-60 cells (10 versus 2 h) and obviously after that of PAI-1 mRNA. Therefore, it is possible that in HL-525 cells, a time-dependent uncoupling between early PKC␤-MAPKK-MAPKp42-induced PAI-1 gene activation and delayed action of TNF␣ prevents optimal upregulation of PAI-1 gene expression. This defect was corrected by the early addition of rhTNF␣, which can bind to TNF␣ receptors. A poor recovery in HL-525 cells treated by RA ϩ PMA was observed for other phenotypes than that of PAI-1, such as adherence and matrix metalloproteinase-9 production (31,39). This partial deficit can now at least be explained by a noncomplete autocrine action of synthesized TNF␣. Second, in HL-60 cells, the mixture of rhTGF-␤1/D3 was described to differentiate immature progenitors along the monocyte pathway (41). With this combination, the level of PAI-1 synthesis at 24 h is much lower than with PMA alone. As shown, this is because endogenous synthesis of TNF␣ is not activated by the combination of rhTGF-␤1/D3. Accordingly, the amplitude of up-regulation of PAI-1 synthesis provoked by rhTNF␣ was much higher than with PMA. The absence of significant TNF␣ synthesis is probably related to the signaling pathway triggered by TGF-␤1/D3, which is different from that induced by PMA. As proof, we demonstrated herein that PKC and MAPK p42/p44 were not involved in rhTGF-␤1/D3-induced PAI-1 synthesis unlike we previously described for the early steps of PMA-induced PAI-1 synthesis (22). Recently, TFE3 and Smad proteins were shown to be involved in TGF-␤1induced PAI-1 gene transcription (43,44). Whether these proteins can be up-regulated by TNF␣-mediated pathway remains to be verified.
Third, the differentiation-dependent up-regulating effect of TNF␣ on PAI-1 synthesis is further demonstrated in PdButreated HL-60 cells. We showed that 14 -20 h of the differentiation process is necessary for TNF␣ to exert its optimal potentiating effect.
Our present results can be compared with those we obtained in promonocytic U937 cells (27) in which TNF␣ potentiated PAI-1 synthesis initially activated by thapsigargin, a non-PMA-type differentiative agent. Taken as a whole, these results indicate that whatever the type of differentiative agents inducing expression of factors involved in PAI-1 gene transcription, these factors are further activated by endogenous TNF␣ which in turn up-regulates PAI-1 synthesis. It is clear that the amplitude of the autocrine TNF␣ effect is dependent on the potency of differentiative agents to induce TNF␣ synthesis.
The up-regulation of PMA-induced PAI-1 synthesis by TNF␣ is consistent with the fact that TNF␣ stimulates AP-1 activity through a prolonged activation of c-Jun NH 2 -terminal kinase (45). In PMA-or vitamin D3-differentiated HL-60 cells, an increase in c-jun mRNA expression is observed (46). Interestingly, homodimers of c-Jun were shown to bind to the tetradecanoyl phorbol acetate-response element at position Ϫ58 to Ϫ50 of the PAI-1 promoter in HepG2 stimulated by PMA (23). Also, in human fibroblasts, the immediate-early gene egr-1 was recently demonstrated to be induced by TNF␣ (47) via the c-Jun NH 2 -terminal kinase pathway (48), and in the HT-1080 fibrosarcoma cell line, egr-1 enhanced synthesis of PAI-1 (18). We reported that SB203580 (22), a potent specific inhibitor of the stress-activated kinase p38 (49), and Emodin, 2 a potent inhibitor of NF-B activation (50), have no effect on PMA-induced PAI-1 synthesis. Collectively, these data are in favor of the involvement of c-Jun NH 2 -terminal kinase in the TNF␣ signaling pathway leading to PAI-1 up-regulation, which is presently under investigation.
The functional importance of TNF␣ and likely also PAI-1 is underlined here with the morphological aspect of adherent differentiated cells treated by rhTNF␣. Recently, it was shown that HL-60 cells treated with PMA secrete fibronectin and express its respective receptor, the integrin ␣ 5 ␤ 1 (31,51). This receptor is induced by TNF␣ in an autocrine manner and leads to the secretion of the 92-kDa gelatinase involved in tissue remodeling. The comparable autocrine regulation by TNF␣ of PAI-1 synthesis poses the question of the role of this inhibitor in participating to pericellular proteolysis during leukocyte differentiation. Interestingly, it was recently shown in the HT-1080 cell line that the induction of PAI-1 driven by TGF-␤1 via Egr-1 stimulation is coordinated with the secretion of fibronectin and its respective receptors, including a role for PAI-1 to stabilize cell attachment (18). Recent experimental data support the contention that PAI-1 is specifically directed at sites where pericellular proteolysis must be controlled (52). In cultured vascular smooth muscle cells, PAI-1 limits plasmin-mediated matrix metalloproteinase activation and consequently may prevent an excessive matrix proteolysis (53).
In conclusion, our results indicate that during induction of monocyte/macrophage differentiation, synthesized TNF␣ upregulates in an autocrine manner the PAI-1 synthesis, provided the PAI-1 gene has been primed by the differentiative stimulus.