Elaboration of Type-1 Plasminogen Activator Inhibitor From Adipocytes

Obesity is an independent risk factor for atherosclerosis and cardiovascular disease.¹ It is accompanied often by hypertriglyceridemia, hyperinsulinemia, and non–insulin-dependent diabetes.²³ Fibrinolysis may be attenuated in part because of increased concentrations in plasma of PAI-1, the primary physiological inhibitor of endogenous fibrinolysis.⁴ One factor implicated in augmentation of PAI-1 in obese subjects is increased plasma insulin.⁵ Elucidation of other factors responsible and obviation of increases in PAI-1 and attenuation of fibrinolysis are needed because of the well-documented association between impaired fibrinolysis and coronary disease.

Fibrin deposition within atherothrombotic lesions contributes to obstruction in coronary arteries and to acute myocardial infarction.⁶ It also may lead to subtle changes, such as endothelial injury and dysfunction⁷ and potentiation of proliferation of neointimal vascular smooth muscle cells.⁸ Subjects at high risk with diabetes mellitus exhibit high PAI-1 activity in plasma whether or not overt ischemic heart disease is present.⁹

PAI-1 has been detected in human lipid-laden vascular smooth muscle and foam cells.¹⁰ However, a direct pathogenetic link between obesity involving excess adipose tissue and attenuated fibrinolysis has not yet been identified. This study was performed to determine whether adipocytes express PAI-1 and if so, whether elaboration of adipocyte PAI-1 can be modulated by growth factors and cytokines already implicated in the development of vascular disease.¹¹

Methods

Cell Culture Procedures

Murine 3T3-L1 preadipocytes (American Type Culture Collection No. CCL 92.1, Rockville, Md) were cultured to confluence in DMEM (Sigma), with 10% FBS (Hyclone), 100 U/mL penicillin, and 100 μg/mL streptomycin. Differentiation of preadipocytes to adipocytes was initiated as described by Weiss et al.¹² In brief, confluent preadipocytes were exposed to 0.5 mmol/L 1-methyl-3-isobutylxanthine and 0.25 μmol/L dexamethasone for 48 hours. The cells were then placed in standard medium in which they accumulated small lipid droplets that grew to occupy a large fraction of total cell volume within 5 days. Seven days after initiation of the differentiation procedure, 85% to 90% of the cells exhibited these characteristics.

To stimulate PAI-1 synthesis, the differentiated cells (7 days after initiation of drug treatment) were exposed to TGF-β₁ (TGF-β, Boehringer Mannheim), murine TNF-α (Life Technologies), or LPS (Sigma); media were harvested 24 hours later. All experiments were performed with media containing 10% FBS. Cell proliferation was assayed by measurement of the incorporation of ³H-thymidine (ICN Biomedicals), as previously described.¹³ The total amount of DNA per culture flask was determined as previously described.¹⁴

Procedures in Animals

Recombinant TGF-β (Boehringer Mannheim) was diluted in 300 μL of endotoxin-free saline and injected intraperitoneally (10 ng/g body wt) into adult mice (Charles River, Wilmington, Mass) weighing 25 to 30 g. Control mice were given equivalent amounts of saline. After 3 hours, blood samples were obtained by cardiac puncture, and the mice were killed. Abdominal adipose tissue was removed rapidly, minced, and fast-frozen in liquid nitrogen.

Assays

PAI-1 activity in conditioned media and plasma was assayed spectrophotometrically¹⁴ ; PAI-1 protein was assayed by Western blotting¹⁵ ; and PAI-1 synthesis was assayed by immunoprecipitation of ³⁵S-labeled PAI-1¹⁶ as previously described. Before Western blotting, the cells were incubated with TGF-β or TNF-α for 24 hours and conditioned media harvested. Samples were subjected to SDS-PAGE and transferred to nitrocellulose membranes (Schleicher and Schuell). Membranes were incubated with 1 μg/mL rabbit anti–PAI-1 IgG (American Diagnostica). Immune complexes were detected by autoradiography with 2×10⁵ cpm/mL of goat anti-rabbit IgG (Sigma) labeled with ¹²⁵I (Amersham). Bands of interest were quantified by laser densitometry (Ultrascan XL, Pharmacia LKB).

For immunoprecipitation procedures, the adipocytes were incubated with 10 μCi/mL ³⁵S-methionine (ICN Biomedicals) for 24 hours. Labeled ³⁵S–PAI-1 was immunoprecipitated with the rabbit anti–PAI-1 IgG and goat anti-rabbit IgG coupled to Sepharose (Hyclone), and the complexes were subjected to SDS-PAGE followed by autoradiography and laser densitometry. In some experiments, labeled ³⁵S-TPA and ³⁵S-UPA were immunoprecipitated with the rabbit anti-TPA IgG or anti-UPA IgG (American Diagnostica). Expression of PAI-1 mRNA was determined by Northern blotting as previously described.¹⁴ Cells incubated with TGF-β for selected intervals and mouse tissue were harvested, and total RNA was isolated with RNAzol B (Tel-Test). Total cellular RNA (20 μg) was size-fractionated on 1.5% formaldehyde agarose gels and subjected to Northern blotting. Membranes were hybridized with a cDNA probe specific for murine PAI-1. A full-length mouse PAI-1 cDNA subcloned to pBS⁺ vector (Stratagene) was kindly provided by Dr M.D. Cole (Princeton University),¹⁷ and an EcoRI/Sph I fragment (nucleotides 1 through 1085) was used. Membranes were hybridized also with GAP cDNA (American Type Culture Collection No. 57091). Autoradiography was performed at −70°C, and the intensity of individual bands was quantified by laser densitometry. All reagents were acquired from Sigma unless otherwise specified.

Statistical Analysis

Data are presented as mean±SD. Statistical comparisons were made using ANOVA in testing the differences between groups. Significance was defined at the P<.05 level.

Results

PAI-1 Activity

TGF-β increased PAI-1 activity in the conditioned media of differentiated adipocytes in a concentration-dependent fashion, with a peak effect seen at 5 ng/mL (0.8±1.2 AU/mL at baseline, 1.7±1.0 AU/mL at 0.5 ng/mL, 9.2±6.0 AU/mL at 1 ng/mL, 11.4±8.6 AU/mL at 5 ng/mL, 10.7±9.4 AU/mL at 10 ng/mL, n=4) (Fig 1). TGF-β stimulated the adipocytes, resulting in significantly increased PAI-1 activity in the media. In contrast, TNF-α increased PAI-1 activity only marginally (1.5±0.8 AU/mL at 1 ng/mL, 1.6±1.0 AU/mL at 5 ng/mL, 2.1±1.4 AU/mL at 10 ng/mL, 2.2±1.2 AU/mL at 50 ng/mL, n=4 each, P=NS). LPS increased PAI-1 activity only modestly as well (1.3±1.0 AU/mL at 5 μg/mL, 1.3±0.6 AU/mL at 10 μg/mL, 2.0±1.4 AU/mL at 50 μg/mL, n=4 each, P=NS). These effects were observed without inducing detectable changes in ³H-thymidine incorporation or total amount of DNA per culture flask (data not shown).

PAI-1 Protein

To determine whether the TGF-β–induced increases in PAI-1 activity reflected accumulation of PAI-1 protein, Western blotting was performed. PAI-1 protein was increased 3.3±1.1-fold by 1 ng/mL TGF-β, 3.8±1.2-fold by 5 ng/mL, and 2.6±1.6-fold by 10 ng/mL (P<.05) in 24 hours (n=3, Fig 2). Thus, the increased PAI-1 activity was attributable, at least in part, to increased accumulation of PAI-1 protein in the media. In contrast, 5 ng/mL TNF-α increased PAI-1 protein only marginally (by 1.5±0.2-fold, n=3, P=NS).

To determine whether the increased PAI-1 protein reflected synthesis of PAI-1 by the adipocytes, immunoprecipitation of labeled PAI-1 protein released from cells that had been exposed to ³⁵S-methionine was performed. The increase in labeled PAI-1 protein was concentration dependent, with a peak increase of 6.0±1.9-fold (n=4, P<.05) at 5 ng/mL TGF-β (Fig 3). Thus, increased PAI-1 protein in the media resulted from increased PAI-1 synthesis by the adipocytes. TNF-α and LPS increased PAI-1 synthesis only modestly, by 1.5-fold to 2-fold. TGF-β (5 ng/mL) decreased the labeled TPA and UPA modestly by 20% to 30% (n=3). In contrast, TNF-α (5 ng/mL) and LPS (5 μg/mL) did not alter the labeled TPA or UPA levels in the media (data not shown, n=3 each).

PAI-1 mRNA

To determine whether the increases in PAI-1 synthesis reflected increased PAI-1 gene expression, Northern blotting was performed. Mouse adipocytes expressed only the 3.2-kb form of PAI-1 mRNA. The PAI-1 mRNA signals for untreated cells showed a transient increase at 3 hours after changing to fresh medium and a linear decrease thereafter and returned to baseline level at 24 hours, whereas the PAI-1 mRNA signals for cells treated with TGF-β showed strong induction at 3 hours and remained relatively constant over the same period (n=3, Fig 4). Thus, PAI-1 mRNA increased over control by 6.2±2.3-fold after exposure of the cells to 5 ng/mL TGF-β for 9 hours (P<.05). TGF-β exerted a concentration-dependent effect on PAI-1 mRNA levels that was evident in 6 hours and was specific, as indicated by the modest decline rather than an increase in GAP 1.3-kb mRNA used as a control.

Exposure in vivo to TGF-β increased the level of PAI-1 mRNA in mouse tissues. A strong response was seen in abdominal adipose tissue (Fig 5), with PAI-1 mRNA as determined by densitometric analysis being 3.9±0.8-fold greater in the TGF-β–treated mice than in controls given saline infusions (n=5 per group, P<.05). GAP 1.3-kb mRNA levels were not altered. At the time of tissue harvest, PAI-1 activity was 2.0-fold higher in TGF-β–treated mice than in those receiving saline (1.5±1.2 versus 3.0±0.8 AU/mL, n=5 per group, P<.05).

Discussion

Obesity is associated with cardiovascular disease.¹ Hypertriglyceridemia is associated with accelerated coronary disease, particularly in diabetic individuals,¹⁸ suggesting that hypertriglyceridemia and obesity may be linked. Resolution of hypertriglyceridemia in primary prevention trials, such as the Helsinki trial,¹⁹ led to prompt (within 1 year) reduction in the incidence of death associated with acute myocardial infarction. The rapidity of the reduction suggested to us that the cause could not have been simply regression of the coronary artery lesions mediated by cholesterol lowering. Instead, the decreased mortality may have reflected a favorable effect of the intervention on the balance between fibrinolysis and thrombosis and consequent diminution of the likelihood of thrombotic coronary occlusion.²⁰²¹ High concentrations of PAI-1 in blood tend to parallel those of triglycerides⁵²² and are associated with obesity.⁵ Hyperinsulinemia has been implicated in augmenting plasma PAI-1,¹⁶ a phenomenon present frequently in patients with insulin resistance manifest by hypertriglyceridemia along with obesity.

The unifying hypothesis that occurred to us predicated on these observations is that elevated PAI-1 is a common denominator linking accelerated coronary artery disease with diverse entities including obesity, hypertriglyceridemia, and diabetes, alone or in combination, and that the increase in PAI-1 may reflect diverse cellular contributions including those from liver cells,¹⁶ endothelial cells,¹⁴ and adipocytes, the focus of the present study.

Thus, we sought to explore the possibility that the association of coronary disease with obesity may be mediated in part by increased PAI-1 elaborated from adipocytes. Accordingly, in the present study we characterized PAI-1 gene expression in cultured adipocytes and in adipose tissue. The results indicate that adipocytes can express PAI-1 and respond to growth factors known to enhance PAI-1 synthesis in cells of other types such as TGF-β. Increased elaboration of PAI-1 protein and activity were detectable in conditioned media of adipocytes cultured in the presence of 10% FBS. These results are consistent with observations by Keeton et al,²³ who detected PAI-1 mRNA and PAI-1 antigen in perinephric fat in mice infused with LPS. They are consistent also with the observation that severe caloric restriction resulting in marked weight loss in markedly obese subjects consistently lowers plasma PAI activity.²⁴ Thus, it appears likely that increases in PAI-1 synthesis in adipose tissue can occur in vivo and contribute to the high concentrations of PAI-1 in plasma in obese subjects. In this context, the hypertriglyceridemia and high PAI-1 may be covariates of increased number, mass, or biochemical activity of adipose tissue.

The magnitudes of the increase in PAI-1 protein in conditioned media of adipocytes induced by TGF-β as determined by Western blotting were more modest than the increases in PAI-1 activity in the conditioned media. This disparity is not surprising because of the concomitant modest decrease in TPA and UPA synthesis and the consequently altered balance between plasminogen activators and their inhibitors, which can affect detectable PAI-1 activity in the media.

Because PAI-1 mRNA signals increased transiently in control untreated cells subjected to a second exposure to fresh medium containing 10% FBS and decreased linearly over the following 24 hours, whereas signals remained relatively constant over the same interval in cells that had been exposed to TGF-β, it appears likely that FBS contains factor(s) that can regulate PAI-1 gene expression in cultured adipocytes and that TGF-β may act to maintain viability of adipose cells or to preserve metabolic balance that is lost in untreated cells.

TGF-β is a multifunctional agent present in α-granules of platelets, peripheral blood monocytes, and tissue macrophages that is released when those cells are activated.²⁵ It stimulates PAI-1 synthesis in human endothelial cells,²⁶ vascular smooth muscle cells,²⁷ and HepG2 cells, an immortal liver cell line.²⁸ LPS stimulates PAI-1 synthesis in endothelial cells in vitro²⁹ and in rabbits in vivo,³⁰ and TNF-α, a mediator of inflammation, stimulates PAI-1 synthesis in human endothelial cells in vitro³¹ and in rats in vivo.³² TNF-α is released from activated macrophages. Its concentration in plasma is increased in patients harboring infection.³³³⁴ In fact, many of the apparently toxic effects of LPS have been attributed to activity of TNF-α. Plasma PAI-1 increases markedly in blood when sepsis occurs,⁴ consistent with stimulation of PAI-1 synthesis by these and other mediators.

Sites of production of PAI-1 are diverse. Regulation may be tissue specific and may vary with disease. Thus, for example, in normal human liver, PAI-1 is expressed in endothelial cells.³⁵ In normal and atherosclerotic human arteries, it is expressed not only in endothelium but also in the vasa vasorum, vascular smooth muscle cells, and foam cells.¹⁰³⁵ As shown epidemiologically in the Atherosclerosis Risk in Communities (ARIC) study,³⁶ atherogenesis is linked intimately to thrombosis and fibrinolysis, particularly in its early stages. Plasma PAI-1 is increased in obesity.⁵³⁷³⁸ Because endogenous fibrinolytic activity constrains thrombosis and facilitates elimination of microthrombi with mitogens that may potentiate athrogenesis, the atherogenic risk accompanying obesity may be mediated in part by impaired endogenous fibrinolytic activity secondary to increased plasma PAI-1. It is likely that the increased plasma PAI-1 accompanying obesity depends in part on an increased mass of adipose tissue in view of our present observations that adipocytes can elaborate PAI-1 in response to growth factors and presumably diverse cytokines and mediators. The attenuation of fibrinolytic activity by increased PAI-1 in obese patients may predispose them to persistent or excessive formation of microthrombi or macrothrombi with prolonged or recurrent exposure of the luminal surfaces of vessel walls to high concentrations of clot-associated mitogens that can accelerate atherosclerosis.^39404142

The results of this study suggest a novel mechanism by which obesity leads to increased plasma PAI-1 activity in plasma and consequently, accelerated atherosclerosis. Both the elevated PAI-1 and accelerated atherosclerosis may be results, in part, of an increased mass of adipose tissue accounting for increased overall synthesis of PAI-1 in adipose tissue stimulated by mediators implicated in acceleration of atherogenesis in vivo. Accordingly, elucidation of mechanisms responsible for the increased plasma PAI-1 in obese subjects and ultimately attenuation of PAI-1 synthesis in vivo may ameliorate cardiovascular risk otherwise associated with obesity.

Selected Abbreviations and Acronyms

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

Footnotes