Oxidative stress activates the plasminogen activator inhibitor type 1 (PAI-1) promoter through an AP-1 response element and cooperates with insulin for additive effects on PAI-1 transcription

Oxidative stress is one of the characteristics of diabetes and is thought to be responsible for many of the pathophysiological changes caused by the disease. We previously identified an insulin response element in the promoter of plasminogen activator inhibitor 1 that was activated by an unidentified member of the forkhead/winged helix (Fox) family of transcription factors. This element mediated a 5- to 7-fold increase in PAI-1 transcription due to insulin. Here we report that oxidative stress also caused a 3-fold increase in PAI-1 transcription and that the effect was additive with that of insulin. Antioxidants prevent this response. Mutational analysis of the PAI-1 promoter revealed that oxidative stress acted at an AP-1 site at -60/52 of the promoter. Gel-mobility shift analysis demonstrated that binding to an AP-1 oligonucleotide was increased 4-fold by oxidative stress. Jun levels were increased by oxidants as assessed by RT-PCR. Western blotting demonstrated that a rapid and prolonged nuclear accumulation of phospho-c-Jun followed oxidant stimulation. The nuclear c-Jun phosphprylation was not observed in cells treated with reduced glutathione. Finally, JNK/SAPK activity was found to increase in response to oxidants and inhibition of JNK/SAP blocked TBHQ-increased PAI-1-Luciferase expression. Thus, oxidative stress stimulated AP-1 and activated the PAI-1 promoter.

We previously demonstrated that Insulin activated the PAI-1 promoter through a forkhead-related response element (TTATTT) at -52/-43 of the PAI-1 promoter (14). Insulin-increased PAI-1 gene expression was inhibited by the expression of the DNA binding domain of forkhead. Finally, LexA-FKHR increased the expression of a LexA-CAT reporter in insulin treated cells. Thus, it appeared that a FKHR related transcription factor mediated the effect of insulin on the PAI-1 gene. The FKHR-related factor was probably not FKHR (FoxO1) itself since FKHR was expressed in both insulin-responsive and non-responsive cell types.
Oxidative stress is a characteristic of diabetes. This is partially the result of increased blood glucose. The major cause of protein oxidation in diabetes is keto aldehydes and oxidizing intermediates that are formed under in vivo conditions by a reaction between glucose and oxygen (15). Additionally, excess 4 glucose is shunted through the aldose reductase pathway. This depletes NADPH reducing potential and leaves antioxidants in their oxidized state (16).
How cells sense oxidative stress arising from a multitude of different compounds was not completely determined. Oxidants activate the transcription of genes for antioxidant defense primarily through stimulation of either NF-kB or AP-1 (17)(18)(19)(20).
The additive activation of PAI-1 gene transcription by Insulin and oxidative stress could explain much of the increase in circulating PAI-1 in diabetics. These experiments sought to define that response. An AP-1 element in the PAI-1 proximal promoter mediated activation of PAI-1 gene transcription by oxidants.
Gel-mobility shift experiments demonstrated an oxidation responsive increase in a factor that binds the AP-1 element. RT-PCR showed that c-Jun is increased by oxidative stress. Western blot analysis demonstrated that oxidative stress increased the phosphorylation of JNK/SAP and nuclear localized c-Jun.

MATERIALS AND METHODS
Materials -[ 32 P]dGTP, 3000 Ci/mmol, was obtained from ICN Biochemicals Corporation. Oligonucleotides were from Operon and reagents for PCR were obtained from Roche. Medium components were obtained from Hyclone Laboratories. Antibodies to c-Jun, phospho-c-Jun, JNK/SAP and phospho-JNK/SAP were from Cell Signaling Technologies while horseradish peroxidase conjugated goat anti rabbit secondary antibody was from Upstate.
Inhibitors were from Calbiochem. All other reagents were of the highest purity 5 available and were obtained from Sigma, Calbiochem, Bio-Rad, Eastman, Fisher, or Roche.
Plasmids -The PAI-1 promoter reporter plasmid, p800neo-Luc, was the generous gift of Dr. D. Rifkin (NYU School of Medicine) (21). Chloramphenicol acetyltransferase (CAT) reporter plasmids were constructed from p800neo-Luc by polymerase chain reaction (PCR) as previously described (22). Deletion mutants of this plasmid were made by PCR and point mutations of these plasmids were also made by PCR using mutant primers as described (14,23).
The human insulin expression vector, pRT3HIR2, was the gift of Dr. J. Whittaker (Hagedorn Institute, Copenhagen, Denmark).

Transient Gene Transfection Facilitated by Electroporation -
Electroporation experiments and reporter assays were performed as described (24). GH4 cells were harvested with an EDTA solution, and 20 to 40 x 10 6 cells were used for each electroporation. All electroporations contained 5 µg of the plasmid pHIR-RT3 that expresses high levels of the human insulin receptor (25). This is necessary to achieve the high levels of insulin stimulation seen in these studies and is consistent with numerous other systems where co-transfection of receptors has been necessary to achieve physiological regulation of transfected genes (25). Experiments with GH4 cells stably transfected with the human insulin receptor give similar results. Trypan blue exclusion before electroporation ranged from 95% to 99%. The voltage of the electroporation was 1550 volts.
This gives trypan blue exclusion of 70% to 80% after electroporation. The 6 transformed cells were then plated in multiwell dishes (Falcon Plastics) at 5 x 10 6 cells per 9-cm 2 tissue culture well in DMEM with 10% hormone-depleted serum (26,27). Cells were refed at 24 h with DMEM with 10% hormone depleted serum ± insulin (1µg/ml bovine insulin, Calbiochem). After 48 h, the flasks were washed three times with normal saline and frozen. The cells were harvested and reporter activity was assayed. Luciferase assays were performed on GH4 cell lysates using reagents and protocols from Promega. Luciferase activity was normalized for variability of transfections using b-galactosidase as described below. Control experiments demonstrated that stimulation of PAI-1-Luciferase by insulin and other hormones was identical to that seen with the corresponding CAT reporter (data not shown).
An RSV-b-galactosidase expression plasmid was included in the electroporations. This plasmid is not expressed and its inclusion has no effect on the overall results of the experiments, but it was included to control for minor

RESULTS
Insulin and oxidative stress activate the PAI-1 promoter additively -The high levels of circulating PAI-1 found in diabetics could result from insulin stimulation of PAI-1 production as suggested previously (14,28). High levels of oxidative stress associated with diabetes might also contribute to maintenance of high PAI-1 (29). Therefore, we determined the effect of oxidative stress caused by tertbutylhydroquinone (TBHQ) on PAI-1 mRNA levels. GH4 cells express measurable PAI-1 mRNA that was increased by TBHQ treatment with a maximal To determine whether the effect of oxidative stress was mediated by an increase in PAI-1 promoter activity, GH4 cells were electroporated with a PAI-1-CAT reporter plasmid and then treated with oxidants. Both TBHQ and H 2 O 2 induced a 3-fold increase in PAI-1-Luc expression ( Fig. 2A) in this assay. The 10 combined effect of insulin and oxidative stress on PAI-1 promoter function is shown in fig. 2B. Treatment of GH4 cells with both insulin and TBHQ increased PAI-1 transcription additively (Fig. 2B). These increases are fully prevented by incubation with antioxidants. Reduced glutathione blocked the effects of TBHQ but not the effect of insulin on PAI-1 gene transcription (Fig. 2C).
The AP-1 response element of the PAI-1 promoter mediated effects of oxidants -An analysis of the PAI-1 promoter was done to determine the response element that mediated the effect of oxidants. This would indicate the transcription factor family that mediated the response and allow us to examine signaling by the oxidants. Deletion of the promoter to -116 had no effect on oxidant-increased PAI-1 gene expression (Fig. 3A). This indicated that the oxidant response element was in the region -116/-6. This region contains elements for the Etsrelated transcription factors, for SP-1, for FKHR-related factors, and for AP-1.
Since AP-1 sites were shown to be oxidant responsive (30), a mutant of the AP-1 element of the PAI-1 promoter was tested and found to be unresponsive to oxidants while it retained insulin responsiveness (Fig. 3B). For comparison, a mutant of the insulin response element (14) remained oxidant sensitive while it lost insulin responsiveness as previously shown. Finally, a 3X-AP-1-Luciferase reporter plasmid was oxidant sensitive but unresponsive to insulin (Fig. 3B).
Oxidants increased binding to the AP-1 response element -To determine if oxidants increased the binding of transcription factors to the AP-1 element, nuclear extracts from untreated and oxidant treated cells were used in gel-11 mobility shift experiments. Two retarded complexes were observed using nuclear extracts from GH4 cells (Fig. 4). The fastest migrating complex was nonspecific since it was efficiently competed with either a 100-fold excess of specific or nonspecific competitor. The more slowly migrating complex was specific for the AP-1 response element since it was completely eliminated using a 100-fold excess of homologous competitor but it was only slightly affected by an oligonucleotide containing a mutation in the AP-1 response element. This was directly tested using the inhibitor SP600125 (JNKII, Calbiochem) (32).

13
GH4 cells that were electroporated with the PAI-1 reporter plasmid were treated for two hours with inhibitors of MAP kinases and then exposed to TBHQ for 24 h.
The PD98059 that inhibits Erk1/2 and SB203580 that inhibits p38 stress activated kinase were without effect ( fig. 8A). The AKT inhibitor SH5, the PI 3kinase inhibitor LY294002, the mTOR inhibitor rapamycin, and the PKA inhibitors contradicted by a previous study that localized glucose activated PAI-1 transcription to the SP-1 sites in the promoter (38). It would be interesting, however, to determine if high glucose activation of PAI-1 in these cells could be inhibited by antioxidants. This would suggest that glucose could increase the oxidative state of the cells to activate PAI-1 transcription through AP-1 and establish a direct mechanism for activation of PAI-1 in diabetes.
Many protein kinase-signaling pathways were shown to activate AP-1. Oxidants have been reported to stimulate apoptosis in a number of cell types (18,54). This was attributed to activation of c-Jun/AP-1 in at least one case (18) although others found both NF-kB and AP-1 to be necessary (54). We did not see evidence for increased apoptosis. This was probably not due to the length of the incubation with oxidants since apoptosis was observed within 30 min. It might depend on the cell-type or dose that was used since the agents PAI-1 has been implicated in many of the complications of diabetes. It plays a role in atherosclerosis by its inhibition of fibrinolysis (61). It plays a role in wound healing through its effects on keratinocyte cell migration (62). It is a factor promoting glomerulosclerosis and tubulointerstitial fibrosis making it an important candidate in diabetic nephropathy (63). It is necessary for peripheral nerve regeneration establishing a link to peripheral neuropathy (64). It is overexpressed in diabetic retina and may play an important role in excess vascularization (65). Thus, it is important to limit PAI-1 production in diabetes.
We demonstrated that hyperinsulinemia and oxidative stress act additively to activate transcription of this gene by stimulating tandem elements in the promoter. Cytokines are also likely act through this stress element. Thus, the three factors that are most likely responsible for the inappropriate regulation of PAI-1 in diabetes act at this composite element. A concerted effort to block this element would likely yield important improvements in PAI-1 levels and diabetic complications. But selective blockade of this element using pharmacological inhibitors or transcription factor decoys could prove difficult. The realization that oxidative stress is responsible for a major proportion of the increase is significant since this could be prevented by antioxidant therapies. Which of these would be 19 effective is difficult to predict, but the response of the PAI-1 promoter could be used to screen for agents that would be most effective for this purpose.