Involvement of Heat Shock Factor-1 in Glycated LDL–Induced Upregulation of Plasminogen Activator Inhibitor-1 in Vascular Endothelial Cells Free

LDL (density 1.019–1.063) was isolated from the plasma of healthy donors using sequential floatation density ultracentrifugation. LDL was glycated by incubation with 50 mmol/l glucose and 50 mmol/l sodium cyanoborohydride for 2 weeks at 37°C in the dark and overlaid with nitrogen as previously described (12). In a parallel preparation, LDL was glycated with the presence of 80 μmol/l butylated hydroxytulene (BHT) (BHT-gLDL) (16). Free glucose was removed from glycated LDL through dialysis. The extent of glycation in glycated LDL was estimated using trinitrobenzenesulfonic acid assay (24). Lysine residues (∼60%) were glycated in the preparations of glycated LDL used in the following experiments. Endotoxin level in lipoproteins was monitored using E-Toxate kit with a threshold of 0.05 ng/ml (Sigma, St. Louis, MO). LDL and its modified forms were stored in sealed tubes under a layer of nitrogen at 4°C in the dark to prevent auto-oxidation (25).

Seed human umbilical vein endothelial cells (HUVECs) were obtained from American Type Culture Collection (Manassas, VA). Cells were grown in F12K medium (Gibco, Burlington, ON, Canada) containing 10% FCS, 0.1 mg/ml heparin, and 30 μg/ml endothelial cell growth supplements (Sigma) (15). Seed human coronary arterial endothelial cells (HCAECs) and cultured supplements were obtained from Clonetics (San Diego, CA). Subconfluent endothelial cell cultures within eight passages from the seed cells were treated with or without an addition of lipoproteins or serum as indicated. No obvious cytotoxicity was detected in endothelial cells treated with LDL or glycated LDL in tested conditions using morphological observation or leucine incorporation assay (25).

Western blotting analysis was performed as previously described (25) using monoclonal antibodies against human HSF1, Hsp-70, PAI-1, β-actin, nonspecific mouse IgG, or anti-mouse IgG antibody conjugated with horseradish peroxidase obtained from Santa Cruz Biotechnology (Santa Cruz, CA) or Sigma. Enhanced chemiluminescence reagents (Amersham, Piscataway, NJ) were used for detecting targeted antigens on nitrocellulose membrane. The densities of the antigens were detected using Chemi-Doc system and Quantity One software (Bio-Rad, Hercules, CA). The abundance of targeted proteins was normalized with the level of β-actin in corresponding samples.

The levels of PAI-1 antigen and activity in experimental medium of endothelial cell cultures were analyzed using PAI-1 enzyme-linked immunosorbent assay or PAI activity assay kits (American Diagnostic, Stamford, CT) as previously described (26).

The levels of hydrogen peroxide (H₂O₂) in the conditioned media of endothelial cells were analyzed using PeroxiDetect kit (Sigma) as previously described (16).

The levels of HSF1 and PAI-1 mRNA were assessed using RT-PCR and justified with the level of β-actin mRNA in corresponding samples. Primers for HSF1 mRNA (sense, 5′-GACATAAAGATCCGCACGGA; and antisense, 5′-CTGCACCAGTGAGATCAGGA) were designed according to the sequence of HSF1 cDNA from GenBank (NM_005526). Primers for PAI-1 mRNA (sense, 5′-CAGACCAAGAGCCTCTCCAC; and antisense, 5′ATCACTTGGCCCATGAAAAG) and β-actin gene (sense, 5′-CGTGGGCCGCCCTAGGCACCA; and antisense, 5′-TTGGCCTTAGGGTTCAGGGGGG) were synthesized according to their reported cDNA sequences (27,28). PCR was performed at 95, 60, and 72°C for 1, 2, and 3 min for 35 cycles. HSF1 (210 bp), PAI-1 (202 bp), and β-actin (300 bp) mRNA fragments were visualized on ethidium bromide–stained 1% agarose gel and were semiquantified using the Chemi-Doc system.

Small interference RNA (siRNA) targeting HSF1 mRNA (5′-GGAAAGUGACCAGUGUGUCtt) was obtained from Ambion (Autsin, TX). HSF1 siRNA was transfected to endothelial cells in serum-free medium using Silence siPort Lipid kit (Ambion). siRNA for β-actin or negative control siRNA (Ambion) was transfected in parallel cultures to verify the methodology.

The full-length HSF1 gene in pCMVHuHSFB plasmid (a kind gift from Dr. Carl Wu, National Cancer Institute, Bethesda, MD) (29) was excised using EcoRI and BglII. The insert was amplified via PCR, and the product was inserted into pcDNA3.1 vector/V5-His-TOPO expression vector (Invitrogen, San Diego, CA). The sequence of HSF1 gene was confirmed by DNA sequencing. HCAECs were transfected with HSF1/pcDNA3.1 or empty vector using CaCl₂ (125 mmol/l) as previously described (30).

PAI-1 promoter (−1,528/55)/luciferase reporter vector was constructed for transfection assay as previously described (13). Two additional PAI-1 promoter fragments (−1,197/55 and −1,105/55 bp) were generated through 5′ deletion. The products were inserted into the pXP1/luciferase vector to generate pPAI-1(−1,197/55)/luciferase and pPAI-1(−1,105/55)/luciferase vector. PAI-1 promoter/reporter constructs were precipitated with HEPES buffer (pH 7.05) containing 125 mmol/l CaCl₂ for 20 min. Cells were incubated with calcium phosphate–precipitated DNA as previously described (13). Chlormaphenicol acetyltransferase/pcDNA3 expression vector (31) was cotransfected as an internal control.

Point mutagenesis of the PAI-1 promoter within the −1,139/−1,126-bp region was generated using QuickChange Site-Directed Mutagenesis kits (Stratagene, La Jolla, CA) and the −1,528/55 PAI-1 promoter/luciferase reporter gene vector was used as the template. The sequences of mutants were verified through DNA sequencing.

Nuclear proteins were extracted from endothelial cells as previously described (32). A double-strand oligonucleotide corresponding to the −1,141/−1,126 bp (AATAGAAATAAAGCAC) of the PAI-1 promoter (GenBank no. J03764) was synthesized as a probe for electrophoretic mobility shift assay (EMSA). The probe was labeled with [³²P]dNTP at the single end. The labeled probe was incubated with nuclear extracts at 4°C for 15 min. DNA-protein complexes were analyzed using 5% nondenatured acrylamide gel electrophoresis and visualized using autoradiography. Monoclonal antibody against HSF1 (0.5 μg/μl; Santa Cruz Biotechnology) was used in supershift assay. Nonspecific mouse IgG was used as an antibody control for the supershift assay.

Student's t test was used for the determination of probabilities between two groups. One-way ANOVA analysis was performed for comparisons among multiple groups. The level of significance was defined as P < 0.05.

The effect of glycated LDL on cell-associated HSF1 was characterized using Western blotting in HUVECs treated with physiologically relevant concentrations of glycated LDL or LDL (50–150 μg protein/ml) for up to 24 h compared with vehicle control. The maximal effect of glycated LDL or LDL on HSF1 expression was detected in HUVECs treated with 100 μg/ml glycated LDL or LDL for 6 h (Fig. 1A and B). The stimulating effect of glycated LDL (100 μg/ml for 6 h) on HSF1 expression in HCAECs was similar to that in HUVECs (Fig. 1C and D). The abundance of Hsp-70 was examined in HUVECs and HCAECs exposed to 100 μg/ml glycated LDL or LDL for 6 h. Significant increase in cell-associated Hsp-70 was detected in HUVECs or HCAECs treated with glycated LDL compared with LDL (P < 0.05, Fig. 1E and F).

Previous studies in our laboratory demonstrated that glycated LDL increased the mRNA and protein release of PAI-1 from endothelial cells (12). The effect of glycated LDL on cell-associated PAI-1 in endothelial cells has not been characterized. The present study examined the effect of glycated LDL and LDL on abundance of PAI-1 in HUVECs exposed to 50–150 μg/ml lipoproteins for 12–72 h. The maximal increase of PAI-1 was detected in HUVECs treated with 100 μg/ml glycated LDL or LDL for 24 h (Fig. 2A and B). The effect of 100 μg/ml glycated LDL or LDL for 24 h on cell-associated PAI-1 in HCAECs was similar to that in HUVECs. Glycated LDL induced significantly greater expression of PAI-1 than LDL in HCAECs or HUVECs (P < 0.05; Fig. 2C and D). The levels of PAI-1 antigen and activity released from endothelial cells were examined in postcultural medium of HCAECs treated with 100 μg/ml glycated LDL or LDL for up to 72 h. The maximal increase of PAI-1 antigen or activity was detected in the media of HCAECs exposed to glycated LDL or LDL for 48 h. The level of PAI activity but not of PAI-1 antigen was attenuated after 72 h of incubation with glycated LDL, LDL, or vehicle (Fig. 2E and F).

Potential relationship between HSF1 and PAI-1 expression was determined in HCAECs transiently transfected with HSF1 gene. The abundance of HSF1 was apparently increased in endothelial cells transfected with HSF1 gene (Fig. 3A). The overexpression of HSF1 significantly increased the abundance of cell-associated Hsp-70 and PAI-1 in HCAECs compared with cells transfected with empty pcDNA3.1 vector (vector) or untransfected endothelial cells treated with vehicle control (Fig. 3B). The transfection of HSF1 did not substantially alter the abundances of β-actin in corresponding cells (Fig. 3A). The transfection of HSF1 gene also increased the expression of HSF1, Hsp-70, and PAI-1 in HUVECs (data not shown). The results indicate that the overexpression of HSF1 gene enhances the expression of PAI-1 in vascular endothelial cells.

We hypothesize that HSF1 mediates the increase of the expression of PAI-1 induced by glycated LDL in endothelial cells. The hypothesis was examined using siRNA against HSF1 mRNA in both venous and arterial endothelial cells. HSF1 siRNA efficiently blocked glycated LDL–or LDL-induced increases of HSF1 and PAI-1 protein or mRNA in HUVECs (Fig. 4). HSF1 siRNA also effectively prevented glycated LDL–or LDL-induced increase in cell-associated HSF1 or PAI-1 in HCAECs (Fig. 5). In endothelial cells transfected with HSF1 siRNA but without an addition of the lipoproteins, the expression of HSF1 and PAI-1 was partially inhibited. HSF1 siRNA did not evidently affect the abundance of β-actin protein or mRNA in endothelial cells with or without lipoprotein treatment (Figs. 4 and 5). Negative control siRNA or siRNA against β-actin did not noticeably alter the expression of HSF1 or PAI-1 protein or mRNA in endothelial cells (data not shown). The results suggest that the expression of HSF1 is required for the upregulation of PAI-1 expression in arterial or venous endothelial cells induced by glycated LDL.

We speculate that a responsive element within the PAI-1 promoter may indirectly mediate glycated LDL–induced activation of the PAI-1 promoter. The hypothesis was examined using transfection assay in HUVECs transiently infected with a battery of 5′-deletion PAI-1 promoter fragment/reporter gene vectors. Treatment with glycated LDL or LDL (100 μg/ml for 24 h) significantly increased the activity of −1,528/55- and −1,197/55-bp PAI-1 promoter but not that of −1,105/55-bp PAI-1 promoter in comparison with control (P < 0.05 or 0.01). The effects of glycated LDL on the activation of the PAI-1 promoter were significantly greater than LDL at comparable conditions (P < 0.05; Fig. 6A). The transfection of empty vector did not substantially alter in PAI-1 promoter activity compared with no-transfection cultures (data not shown). The results suggest that a responsive element activated by glycated LDL or LDL locates between −1,197 and −1,105 bp of the PAI-1 promoter. A homolog of heat shock–responsive element (HSE) was detected within the −1,140/−1,127-bp region of the PAI-1 promoter (TAGAAATAAAGCA) using a transcription factor searching tool (http://www.cbrc.jp/research/tfsearch.html). The results of point mutagenesis assay within the region confirmed that the −1,137/−1,128-bp region of the PAI-1 promoter was required for the activation of PAI-1 promoter induced by glycated LDL or LDL (Fig. 6B).

HSF1 activates the transcription of multiple stress-response–related genes through its binding to HSE in the promoters of target genes (16). The effect of glycated LDL on the binding of nuclear proteins to the targeted region of PAI-1 promoter was investigated using EMSA. Treatment with glycated LDL or LDL visibly enhanced the binding of a nuclear protein to the labeled −1,141/−1,126-bp PAI-1 promoter fragment compared with control. Glycated LDL induced considerably greater binding of the nuclear protein to the PAI-1 promoter fragment compared with LDL. The addition of 50- or 200-fold of the unlabeled −1,141/−1,126-bp PAI-1 promoter fragment blocked the binding of the nuclear protein to undetectable level (Fig. 7A). Antibody against HSF1 induced a visible upward shift in the migration of the DNA-protein complex induced by glycated LDL or LDL or at basal condition (Fig. 7B). Nonspecific mouse IgG did not affect the migration of the nuclear protein (data not shown). The results were reproduced in four experiments. The findings suggest that glycated LDL increased the binding of HSF1 to a putative HSE in the PAI-1 promoter.

Previous studies in our group demonstrated that 80 μmol/l BHT, a potent antioxidant, prevented glycated LDL–induced PAI-1 release from HUVECs or HCAECs (15). The effect of BHT on glycated LDL–induced HSF1 and PAI-1 expression in HCAECs was examined in the present study. BHT-gLDL significantly reduced glycated LDL–induced HSF1 and PAI-1 expression in HCAECs (P < 0.05 or 0.01; Fig. 8A). The levels of PAI-1 antigen or H₂O₂ in the conditioned media of HUVECs treated with BHT-gLDL were significantly lower than those treated with glycated LDL without BHT treatment (P < 0.05 or 0.01; Fig. 8B and C).

The major findings of the present study include that 1) glycated LDL stimulated the expression of HSF1 and Hsp-70 in cultured arterial or venous endothelial cells, 2) overexpression of the HSF1 gene upregulated the expression of PAI-1 in endothelial cells, 3) HSF1 siRNA blocked glycated LDL–induced PAI-1 expression in endothelial cells, and 4) glycated LDL enhanced the binding of HSF1 to a PAI-1 promoter fragment containing a HSE homolog. The findings suggest that HSF1 is involved in the upregulation of PAI-1 induced by glycated LDL in vascular endothelial cells.

Previous studies reported that proliferating endothelial cells responding to oxidized LDL on the expression of Hsp-70 were more active than nonproliferating endothelial cells (33). The results provided additional evidence that glycated LDL increased the expression of HSF1 in subconfluent human arterial or venous endothelial cells. The stimulating effect of glycated LDL on HSF1 was weaker in confluent endothelial cells compared with that in subconfluent endothelial cells (data not shown). The collection of these findings suggests that growing vascular endothelial cells may actively respond to hyperglycemia or oxidative stress–associated hyperbetalipoproteinemia on the expression of stress-response mediators.

Previous studies demonstrated that H₂O₂ activated HSF1 (34). H₂O₂ is derived from superoxide under the influence of superoxide dismutase (SOD) in cells. H₂O₂ is more stable than other ROS; therefore, it may function as an intracellular signal of oxidative stress (35,36). Our recent studies demonstrate that glycated LDL is a potent agonist for the generation of superoxide and H₂O₂ from endothelial cells compared with LDL, which implies that glycated LDL, a diabetes-associated lipoprotein, may lead to endothelial oxidative stress. The effect of glycated LDL on H₂O₂ generation from endothelial cells reached a peak after 2 h of incubation, which was associated with an elevated activity of SOD in endothelial cells (16). The results of the present study demonstrate that the expression of HSF1 in endothelial cells is significantly increased by glycated LDL after 4–6 h of incubation. The presence of antioxidant during the glycation suppressed the generation of H₂O₂ and the expression of HSF1 or PAI-1 in endothelial cells induced by glycated LDL. The combination of findings suggests that glyco-oxidation in glycated LDL and endothelial cell–derived ROS may contribute to glycated LDL–induced expression of HSF1 and PAI-1 in endothelial cells.

The increased expression of HSF1 was detected in human atherosclerotic lesions (37). The levels of circulating Hsp-72, an important member of inducible 70-kDa Hsp family, were significantly higher in type 1 diabetic patients with ketoacidosis compared with age-matching type 1 diabetic patients without ketoacidosis (22). The results of the present study indirectly support the hypothesis that diabetes-associated metabolic disorders may enhance stress responses in vascular endothelial cells. The upregulation of HSF1 in endothelial cells by glycated LDL may contribute to the upregulation of Hsp detected in endothelial cells or in the circulation of uncontrolled diabetic patients.

PAI-1 is an acute-phase reactant (38). Increased levels of PAI-1 have been observed in patients with sepsis (39) or recurrent myocardial infarction (40). During certain stresses, including wounding or bleeding, the increased generation of PAI-1 from endothelial cells may play a protective role through maintaining fibrin clots or the integrity of the extracellular matrix. Rücker et al. (41) demonstrated that heat shock increased the expression of PAI-1 in vascular tissue of rat muscle. Uchiyama et al. (42) reported that adenovirus vector-mediated overexpression of HSF1 gene reduced PAI-1 expression in cultured arterial endothelial cells, but the finding was not verified using any gene knockdown approach. The present study provided multiple lines of evidence for the involvement of HSF1 in glycated LDL–induced upregulation of PAI-1 in endothelial cells, including the inhibition of PAI-1 expression in arterial or venous endothelial cells by HSF1 siRNA and the suppression of PAI-1 promoter activity via mutagenesis of a HSE homolog in the PAI-1 promoter. The results suggest that HSF1 upregulates PAI-1 production through the binding of HSF1 to the PAI-1 promoter. The sequence of the −1,137/−1,128 bp of the PAI-1 promoter partially, but not completely, matches the authentic HSE. Considerable variations of HSE homologs have been described (43). HSF1 may interact with a homolog of HSE, the −1,137/−1,128 bp of PAI-1 promoter, which further enhances the transcription of the PAI-1 gene in endothelial cells.

The results of the present study demonstrated that LDL moderately increased the expression of HSF1 and cell-associated PAI-1 in endothelial cells. LDL may be oxidatively modified by a prolonged incubation with endothelial cells (44). A previous study in our group demonstrated that antioxidants reduced the abundance of lipid peroxides in LDL and prevented LDL-induced release of PAI-1 from endothelial cells (15). The effects of LDL on HSF1 and PAI-1 after a prolonged incubation with endothelial cells may result, at least in part, from cell-mediated oxidative modification on LDL.

In conclusion, glycated LDL stimulates the expression of HSF1 in vascular endothelial cells. HSF1 mediates glycated LDL–induced expression of PAI-1 in endothelial cells through enhancing the binding of HSF1 to the PAI-1 promoter. Glyco-oxidation of LDL and endothelial cell–derived ROS may play crucial roles in the upregulation of HSF1 and PAI-1 induced by glycated LDL. The findings from the present study provide a potential linkage between glycated LDL, stress response, and hypofibrinolysis. The results of the present study are generated from cultured endothelial cells. Subsequent studies in animal models potentially provide additional useful information on interrelationships between diabetes-associated lipoproteins and stress-response–or fibrinolysis-related proteins.

G.X.S. has received support from the Canadian Diabetes Association, the Canadian Institute of Health Research, Manitoba Health Research Council, the Innovation Fund through the province of Manitoba, and the Health Sciences Centre Foundation.