THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 266, No. 4, Issue of February 5, pp. 2459-2465,1991 Printed in U. S. A.

Lipoprotein (a) Regulates Plasminogen Activator Inhibitor- 1 Expression in Endothelial Cells A POTENTIAL MECHANISM IN THROMBOGENESIS*

(Received for publication, July 31, 1990)

Orli R. EtinginSQ, David P. Hajjarn (1, Katherine A. Hajj,,**$$, Peter C. Harpel$, and Ralph L. NachmanS From the Departments of $Medicine, llBiochemistry, Pathology, and **Pediatrics, and the National Institutes of Health Specialized Center for Thrombosis Research, Cornell University Medical College, New York, New York 10021

Lipoprotein (a) (Lp(a)) is a low density lipoprotein- like particle which contains the plasminogen-like apo- lipoprotein a. Lp(a) levels are elevated in patients with atherosclerotic coronary artery disease. Recent studies suggest that Lp(a) competitively inhibits plasminogen binding to the endothelial cell and interferes with sur- face-associated plasmin generation. In this study, we present evidence for the presence of Lp(a) in the mi- crovasculature of inflamed tissue. In addition, we dem- onstrate that Lp(a) regulates endothelial cell synthesis of a major fibrinolytic protein, plasminogen activator inhibitor-1 (PAI-1). In cultured human endothelial cells, Lp(a) enhanced PAI-1 antigen, activity, and steady-state mRNA levels without altering tissue plas- minogen activator activity or mRNA transcript levels. This effect was cell-specific. Although other lipopro- teins did not coordinately raise PAI- 1 mRNA levels in endothelial cells, low density lipoprotein treatment se- lectively raised the level of the 3.4-kilobase mRNA species of PAI-1 without a concomitant increase in PAI-1 activity or antigen. Endothelial cell exposure to Lp(a) did not cause generalized endothelial cell acti- vation since the functional activity and mRNA levels for tissue factor, platelet-derived growth factor and interleukin-6 were not elevated following Lp(a) expo- sure. These data suggest a molecular mechanism whereby Lp(a) may support a specific prothrombotic endothelial cell phenotype, namely by increasing PAI- 1 expression.

Vascular endothelial cells may play a critical role in throm- boregulation by virtue of a membrane-oriented fibrinolytic system (1-3). Endothelial cells synthesize and secrete tissue plasminogen activator (t-PA)’ which binds to a surface recep-

* This work was supported in part by National Institutes of Health Grants HL-01687, HL-18828, HL-39701, and HL-42493. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduer- tisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Q To whom correspondence should be addressed.

York Affiliate). (1 Established Investigator of the American Heart Association (New

$$ Established Investigator of the American Heart Association and a Syntex Scholar. ’ The abbreviations used are: t-PA, tissue plasminogen activator; apo(a), apolipoprotein a; GAPDH, glyceraldehyde-3-phosphate de- hydrogenase; IL, interleukin; LDL, low density lipoprotein; Lp(a), lipoprotein (a); PAI-1, plasminogen activator inhibitor 1; PDGF, platelet derived growth factor; SDS, sodium dodecyl sulfate; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid.

tor. This membrane-binding site appears to protect t-PA from its major physiologic inhibitor, PAI-1, and preserves the cat- alytic activity of t-PA (4). NH2-terminal glutamic acid plas- minogen, the circulating form of the fibrinolytic zymogen, interacts with the endothelial cell surface and is converted to the catalytically more favored NH2-terminal lysine plasmin- ogen (2). This modification of circulating plasminogen at the endothelial cell surface augments the fibrinolytic potential of the blood vessel wall. The macromolecular assembly of a plasmin generating system at the endothelial cell surface may be important in the physiologic maintenance of blood fluidity. PAI-1 inhibits the generation of plasmin by forming an in- active complex with t-PA, thereby preventing plasminogen activation (5, 6).

In cultured endothelial cells several factors have been shown to regulate PAI-1 activity. Thrombin and interleukin- 1 coordinately enhance synthesis of both PAI-1 activity and t-PA while endotoxin selectively enhances PAI-1 (7-11). Ac- tivated protein C decreases PAI-1 protein activity directly (12). Endothelial cell growth factor and heparin have also been shown to decrease PAI-1 at the RNA level without altering t-PA mRNA or protein levels (13). This is consistent with the profibrinolytic activity of heparin.

Recent data from several laboratories may link some of these endothelial cell-mediated events to atherogenesis (14- 17). Lp(a), an atherogenic molecule which contains apolipo- protein a, is a structural homolog of plasminogen (18-22). Lp(a) is thought to interfere with surface-oriented endothelial cell fibrinolysis by inhibiting plasminogen binding and down- regulating plasmin generation. Accumulation of Lp(a) in var- ious atherosclerotic lesions provides a potential link between impaired cell surface fibrinolysis, localized thrombosis, and progressive atherogenesis.

As an acute phase reactant, Lp(a) may accumulate in sites of inflammation independent of serum levels (23). In this study, we present evidence that Lp(a) is present in the micro- vasculature of inflamed tissue and that Lp(a) enhances PA1 expression and function in human vascular endothelial cells. These observations support the hypothesis that Lp(a) binding to endothelium participates in the regulation of a thrombo- genic phenotype.

EXPERIMENTAL PROCEDURES

Cell Culture-Human umbilical endothelial cells were isolated and grown to confluence (24). Cells were subpassaged one to three times prior to use and confirmed to be endothelial cells by immunofluores- cent staining with von Willebrand factor antiserum. Cells were grown in “199 containing 20% heat-inactivated fetal calf serum (GIBCO). 100 IU/ml penicillin and 50 pg/ml streptomycin (M. A. Bioproducts). Since endotoxin can cause altered levels of PAI-1, polymyxin B (20

2459

2460 Lipoprotein (a) Regulates Endothelial Cell PAI-1 Expression

pg/ml) was added to cell cultures. Limulus lysate assay was used to test serum-free medium for endotoxin. HT 1080 cells were purchased from American Type Culture Collection, Rockville, MD and grown in "199 medium containing 5-10% fetal calf serum. Human venous fibroblasts were grown from explants in "199 medium, 10% fetal calf serum and 2% penicillin-streptomycin. Cells were treated with Lp(a), LDL, or other lipoproteins after the cells were washed in phosphate-buffered saline, and re-fed with fresh medium. In some studies, HT 1080 cells or fibroblasts were exposed to M dexa- methasone or 5 ng/ml transforming growth factor p , respectively (25, 26).

Conditioned Media and Cell Lysates-In assays for PAI-1 func- tional activity, cells were grown to confluence as above and then re- fed with medium containing 2% ITS (insulin, transferrin, selenium) instead of serum with Lp(a) for 24 h. Media were removed and stored at -70 "C for up to 1 week until the PA1 assay was performed. Cell lysates were prepared by extracting cellular components with 0.5% Triton X-100 for 10 min at 37 "C (27).

PAI-1 Antigen Detection-PAI-1 antigen was measured using a double antibody enzyme-linked immunosorbant assay according to the manufacturer's instructions (American Diagnostica). Endothelial cell matrices were prepared as described (27). Cells grown to conflu- ence in 96-well Costar plates were removed with 0.5% Triton X-100 in phosphate-buffered saline, pH 7.4, incubated for 10 min with 25 mM NH,OH to remove cytoskeletal elements, and washed extensively with distilled water. No cells were detected on extracted matrix by light microscopy.

t-PA Antigen-t-PA antigen levels in endothelial cell-conditioned medium was determined by a double antibody sandwich enzyme- linked immunosorbant assay using the Immunobind TPA kit accord- ing to the manufacturer's instructions (American Diagnostica).

Preparation of Plasma Lipoproteins-LDL (density range 1.019- 1.063 g/ml) and high density lipoprotein (1.083-1.210 g/ml) were isolated by preparative ultracentrifugation followed by dialysis against HEPES-buffered saline, membrane-sterilized, and stored un- der nitrogen gas at 4 "C (28). For some studies, LDL was further purified by Sepharose S-400 column chromatography. All lipid prep- arations were routinely screened for peroxides by thiobarbituric acid assay (29). Lp(a) was purified from plasma as described (30). Concen- tration of purified Lp(a) was determined by the Lowry procedure using bovine serum albumin as the standard (31). Lp(a) depleted of apo(a) (Lp(a-)) was prepared by reduction with dithiothreitol fol- lowed by heparin-Sepharose affinity chromatography as detailed (32). The peak fractions did not contain Lp(a) as assessed by native acrylamide gel electrophoresis or by Western blot analysis using a monospecific rabbit anti-human Lp(a). Lp(a) was dialyzed against phosphate-buffered saline at 4 "C, membrane-sterilized, and filtered prior to its application on cells. Oxidized LDL was prepared by overnight dialysis against 5 PM CuSO,, membrane-sterilized, and filtered.

Mitogenesis Assay-The mitogenic capacity of post culture medium from Lp(a)-treated endothelial cells was estimated in a bioassay. 13H] Thymidine incorporation into confluent BALB/C 3T3 fibroblasts or bovine aortic smooth muscle cells was measured upon exposure to test media. Cells were growth arrested using serum-free media, then exposed to conditioned media from Lp(a) (25 pg/ml)-treated endo- thelial cells for 24 h. Cells were harvested onto glass fiber paper using an automated cell harvester and incorporated [3H]thymidine esti- mated by liquid scintillation counting.

P A I Functional Assay-Post culture medium (serum-free) contain- ing polymyxin B was assayed for PAI-1 in the presence of t-PA (5 units/ml), plasminogen, and the fluorogenic plasmin substrate (165 p ~ , D-Val-Leu-Lys-aminofluoromethylcoumarin). Substrate hydrol- ysis was measured from 0 to 120 min in a Perkin-Elmer model 650- 10s fluorescence spectrophotometer as previously described (33). Plasmin generation was estimated as the slope of the line defined by plotting relative fluorescence units uersus time. Standard curves were constructed using serial dilutions of PAI-1 (American Diagnostica). Anti-PA1 monoclonal antibody and an irrelevant antibody were added to some samples of post-culture media controls. One unit of activity was defined as the amount of PA1 capable of inhibiting one interna- tional unit (IU) of t-PA. In order to activate latent PAI-l, some samples of conditioned media were treated with SDS as previously described (34).

RNA Isolation and Northern Analysis-Total RNA was isolated and prepared according to Chirgwin (35). Cells were harvested in 4.0 M guanidinium isothiocyanate containing 25 mM sodium acetate and 0.1% 2-mercaptoethanol. The RNA extract was triturated through

an 19-gauge needle and applied to a cushion of 5.7 M CsCl containing 25 mM sodium acetate, pH 6.0. After overnight centrifugation at 35,000 rpm, the RNA pellet was solubilized in 0.3 M sodium acetate, precipitated in 70% ethanol, and stored at -70 "C. In some studies, messenger RNA was purified using the fast track mRNA isolation kit (In Vitrogen). RNA was quantitated by UV spectrophotometry.

In preparation for Northern hybridization analyses, RNA was electrophoresed in 1% denaturing agarose gels containing ethidium bromide. The RNA was transferred to a Nylon filter (Zetaprobe). The filter was washed in 10 X SSC buffer (1 X SSC = 0.15 M NaCl, 0.015 M sodium citrate), air-dried, and baked in a vacuum oven for 2 h at 80 "C. The filters were prewashed in 0.1 X SSC and 0.5% SDS for 1 h at 65 "C followed by prehybridization in 50% formamide, 0.25 M NaH PO4, pH 7.2, 0.2 M NaCI, 7% SDS, and 1.0 mM EDTA for 5 min at 43 "C. Filters were hybridized with cDNA probes labeled with [32P]dCTP for 16-24 h at 43 "C in a shaking water bath. The filters were washed under high stringency condition in 0.1 X SSC and 0.1% SDS at 65 "C. Filters were air-dried and exposed to Kodak XAR film with intensifying screens at -70 'C. Autoradiographs were examined by comparison to 8-actin or glyceraldehyde-3-phosphate dehydrogen- ase (GAPDH) mRNA levels, which are two stable genes that are constitutively expressed and unaltered by lipoproteins in human endothelial cells.

Preparation of Probes-cDNA probes were labeled with [32P]dCTP by random hexamer primer extension (36). The PAI-1 cDNA probe was a 3.0-kb EcoRI fragment of plasmid pGEM3 provided by Dr. D. Loskutoff (Scripps Research Institute) (37). The t-PA cDNA probe was a PstI fragment of plasmid pPAl provided by Dr. T. Ny (Uni- versity of Umea, Sweden) (38). The tissue factor cDNA probe was a 1.2-kb Hind111 fragment of plasmid pKS2B provided by Dr. W. Konigsberg (39). The IL-6 and PDGF A chain cDNA were provided by Dr. P. Sehgal of Rockefeller University and Drs. Christer Betsholtz and Robert Gallo of the National Institutes of Health, respectively (40, 41).

Materials-Human tumor necrosis factor (2 X lo7 units/mg pro- tein) and tissue plasminogen activator (t-PA) were kindly provided by Genentech Inc. IL-1 was purchased from Genzyme. PAI-1 was purchased from American Diagnostica.

Antibodies-Rabbits were immunized with purified human Lp(a). Monospecific polyclonal rabbit anti-human Lp(a) was prepared by extensive adsorption of the IgG fraction of rabbit anti-Lp(a) serum with insolubilized LDL, lys-plasminogen, and fibrinogen. The final preparation did not react with LDL, plasminogen, or fibrinogen by Western blot. Rabbit anti-plasminogen was prepared as previously described (15) and extensively adsorbed with insolubilized Lp(a). The final IgG fraction did not react with Lp(a) by Western blot. Mono- specific anti-apolipoprotein (B), anti-fibrinogen, and anti-albumin were obtained commercially (Organon Technica-Cappel).

Immunohistochemical Methods-Immunohistochemical staining was performed as previously described (15). Formalin-fixed paraffin- embedded sections were dewaxed and Pronase-treated. Endogenous peroxidase activity was blocked by treatment with hydrogen peroxide (3%, 30 min). The slides were preincubated with 10% normal rabbit serum in buffer for 1 h and then incubated (18 h, 4 "C) with primary antibody, rabbit anti-Lp(a) 1:2000 dilution. The slides were exposed to biotinylated goat anti-rabbit IgG (30 min, 21 "C, 1:250 dilution) and incubated with avidin-peroxidase complex. Peroxidase deposition was visualized using diaminobenzidine tetrahydrochloride. Sections were counterstained with hematoxylin.

Tissue Factor Actiuity-Tissue factor activity was measured as described by Carson and Archer (42).

RESULTS

Lp(a) Deposition in Inflammatory Microvasculature-Lp(a) was detected by immunohistochemical techniques in the mi- crovasculature associated with inflammatory lesions, includ- ing a gall bladder fistula, granulomatous lymph node, and pericarditis (Fig. 1). The Lp(a) appeared to be deposited in or on the endothelium as well as in the media of an arteriole (Fig. l a ) . Immunohistochemical analysis of adjacent sections showed colocalization of apolipoprotein (B) with Lp(a). Ad- jacent sections revealed no association deposition of fibrino- gen, albumin, or plasminogen (using a monospecific anti- plasminogen antibody adsorbed with Lp(a) (Fig. 1). M a ) was also detected in the small vessels of tonsils, pilonidal cysts,

Lipoprotein (a) Regulates Endothelial Cell PAI-1 Expression 2461

FIG. 1. Immunohistochemical identification of Lp(a) in mi- crovasculature of inflammatory tissues. Immunoperoxidase staining of small blood vessels is shown using monospecific rabbit anti-Lp(a) absorbed with and non-reactive to plasminogen. a, gall bladder fistula; b, granulomatous lymph node. (These tissues were freshly obtained following surgery). c, fungal pericarditis (obtained at post mortem examination). Adjacent sections were stained with mon- ospecific rabbit anti-plasminogen antibody adsorbed with Lp(a) (a’- c’) (X 425).

TABLE I Effect of Lp(a) on PAI-1 antigen in endothelial cells, matrix, and

conditioned media Endothelial cells (5 X lo5 cells/well) were incubated for 24 h at

37 “C in 2 ml of “199 medium with 2% ITS, polymyxin B (20 pg/ ml) with either Lp(a) (25 pg/ml), LDL (25 pglml), or thrombin (3 units/ml). Conditioned medium was removed. Cells were harvested by gentle scraping after washing in phosphate-buffered saline. Mat- rices were prepared as previously described (28, 33) and assayed for PAI-1 by enzyme-linked immunosorbant assay. Each point represents mean f S.D. for three separate determinations.

PAI-1 Endothelial cells

Conditioned medium Matrix

nglml Control 38.0 ? 5.6 667.2 f 19.8 72.0 f 16.9 Lp(a) 69.9 f 16.9 1,341.4 f 115.9 139.6 f 55.8 LDL 26.7 f 1.4 562.5 f 73.5 80.4 f 16.4

and in the vasa vasora of coronary arteries associated with acute myocardial infarction (not shown). The microvascula- ture of normal tissues did not contain Lp(a) by immunohis- tochemical analysis.

Effect of Lp(a) on the Production of PAI-1 by Endothelial Cells-PAI-1 antigen was quantitated in conditioned medium, whole cell lysates and extracellular matrix from Lp(a)-treated, and untreated cells (Table I). Cells were maintained in media containing Lp(a) (25 pg/ml) for 24 h. Using a “sandwich” enzyme-linked immunosorbant assay, we found increased PAI-1 in all three compartments of Lp(a)-treated cells com- pared with control or LDL-treated cells.

As shown in Table 11, resting endothelial cells secreted abundant quantities of t-PA neutralizing activity into their conditioned medium. Lp(a) treatment (25 pg/ml) resulted in a 10-fold increase in this activity. The augmentation was observed as early as 4 h after Lp(a) treatment and lasted for a t least 24 h. Monospecific antibody to PAI-1 added to post- culture media completely eliminated detectable PAI-1 in this assay. Lp(a) doses as low as 5 pg/ml increased PAI-1 activity 4-6-fold in endothelial cell-conditioned media (data not

TABLE I1 Effect of Lp(a) on PAI-I production by endothelial cells

Human umbilical vein endothelial cells were incubated for 24 h at 37 “C in “199 medium with 2% ITS, polymyxin B 20 pg/ml, with Lp(a) (25 pglml), LDL (25 pglml), Lp(a-) (25 pg/ml) and monoclonal antibody to PAI-1 (1:500), or plasminogen (4.25 pg/ml). Conditioned medium was removed and assayed for PAI-1 activity as outlined under “Experimental Procedures.” Separate aliquots of conditioned medium were treated with guanidine HCI and subsequently assayed for PAI-1 activity as above. Each point represents mean k S.D. for four seDarate determinations.

PAI-1 activity

Control Lp(4 M a - ) LDL Lp(a) + PAI-1 antibody Plasminogen Guanidine-HC1-treated CM

Control LDL L d a )

IU/ml 3.3 f 0.3

36.0 f 3.4 3.3 f 0.4

Undetectable 4.9 f 1.0

2.5 f 0.6

29.0 f 3.1 37.2 k 5.9

277.8 f 31.8

3.4 kb-

2.4 kb -

P A I - I

FIG. 2. Lp(a) induction of PA1 &RNA in cultured endothe- lial cells. Endothelial cells were grown to confluence. Medium was replaced with either fresh medium alone (con), or medium supple- mented with tumor necrosis factor (100 ng/ml, 24 h), LDL (25 pg/ ml, 24 h), or Lp(a) (25 pg/ml, 24 h). Messenger RNA was isolated as outlined under “Experimental Procedures” and analyzed by Northern hybridization using the PAI-1 probe.

shown). At concentrations above 40 pg/ml, no further incre- ment in PAI-1 activity was observed. In these experiments, endothelial cells were cultured in medium containing 2% ITS in place of serum, and polymyxin B. Lp(a-), the reductively cleaved apo(a)-free particle, LDL, and plasminogen at ap- proximately equimolar concentrations did not stimulate PAI- 1 activity.

To test the hypothesis that Lp(a) might influence a con- version from latent to active PAI-1 in endothelial cell post- culture medium, we measured PAI-1 functional activity in medium from Lp(a)-treated endothelial cells that were sub- sequently treated with guanidine HCI and 0.1% SDS to acti- vate latent PAI-1. We observed a 7-10-fold increase in PAI- 1 activity in Lp(a)-treated post culture media as well as control post-culture media. The Lp(a) exposure did not appear to change the ratio of latent to active PA1 in the medium (Table 11).

Effect of Lp(a) on PAI mRNA Expression in Human Endo- thelial Cells-Northern blot analysis of messenger RNA from endothelial cells maintained for 18 h in medium containing Lp(a) (25 pg/ml) is shown in Fig. 2. A 2-4-fold increase in PAI-1 mRNA was seen in Lp(a)-treated cells. Both the 3.4- and 2.4-kilobase species of mRNA were increased. In contrast,

2462 Lipoprotein (a) Regulates Endothelial Cell PAI-1 Expression

mRNA isolated from endothelial cells maintained in medium containing LDL (25 pg/ml) demonstrated a selective 2-fold increase in the 3.4-kb species of PAI-1 mRNA. This increase in the 3.4-kb mRNA was not associated with an increase in PAI-1 antigen or functional activity, suggesting that the 3.4- kb species of PAI-1 mRNA may not be efficiently translated. Both species of PAI-1 mRNA were increased in cells treated with tumor necrosis factor (100 ng/ml, 24 h).

Superinduction of PAI-1 mRNA was demonstrated in endo- thelial cells treated for 2 h in medium containing cyclohexi- mide (10 pg/ml). We observed a 3-fold increase in both species of PAI-1 mRNA in endothelial cells treated with cyclohexi- mide for 2 h. Both species of PAI-1 mRNA were decreased by approximately 50% in cells exposed to actinomycin D for 2 h compared with control (data not shown).

Similar results were obtained in dose-response and time- course studies. Messenger RNA was isolated from endothelial cells grown in media containing a range of Lp(a) concentra- tions (0-40 pg/ml). Levels of PAI-1 mRNA (represented as fold increase above control) are shown in Fig. 3. Similarly, levels of PAI-1 mRNA from endothelial cells exposed to Lp(a) (25 pg/ml) for varying time periods (0-48 h) were compared in Fig. 3. Maximal stimulation of PAI-1 mRNA levels by Lp(a) was seen a t 24 h and ranged from 2- to 4-fold for the 2.4-kb species and 2-3-fold for the 3.4-kb species. Messenger RNA levels of GAPDH, a stable “housekeeping gene” were compared with PAI-1 mRNA to normalize the data to the amount of RNA loaded/lane. GAPDH RNA levels were not

0.5 t 0

5.0 IO 20 30 40 Lpa (,ug/ml) ,&..

,/$/4kb ‘1

2 4 6 12 24 48

Lpa treatment (hours)

FIG. 3. Time course and dose response of Lp(a) induction of PAI-1 mRNA in endothelial cells. Endothelial cells grown to confluence were re-fed with fresh “199 medium containing 20% pooled human serum and Lp(a) (25 pg/ml). Cells were treated with guanidinium-isothiocyanate and mRNA was isolated and analyzed by Northern hybridization. The relative amounts of the mRNAs were estimated from the autoradiographs using scanning laser densitome- try. (Each point represents mean f S.D. from three separate autora- diographs.) In the top panel, mRNA was prepared from cells treated for 18 h with various doses of Lp(a) and PAI-1 mRNA compared densitometrically to mRNA prepared from untreated endothelial cells. In the bottom panel, cells were cultured in the presence of Lp(a) (25 pg/ml) for time periods up to 48 h. Northern blots were used to calculate relative amounts of PA1 mRNA in Lp(a)-treated cells com- pared with control cells a t each time point as described above.

altered in these cells following challenge with Lp(a) or LDL. Effect of Lp(a) on Tissue Factor, PDGF, and IL-6 mRNA

Expression-To determine whether the effect of Lp(a) on endothelial cell function was specific for PAI-1, we examined mRNA expression for tissue factor, PDGF, and IL-6. These proteins or polypeptides were selected because they are ex- pressed in “activated” endothelial cells under various patho- logical conditions. mRNA expression for these polypeptides is primarily under transcriptional regulation (40, 43-46). As shown in Fig. 4, Lp(a) treatment of endothelial cells did not increase the mRNA expression for tissue factor, PDGF, or IL-6. Functional assays for tissue factor and PDGF were also done to determine whether Lp(a)-treated endothelial cells induced protein synthesis or functional activity of these mol- ecules. We observed no difference in tissue factor or PDGF activity in these cells compared with untreated cells (data not shown). To determine whether Lp(a) might modulate expres- sion of endothelial cell adhesive molecules, we performed neutrophil adhesion studies on Lp(a)-treated endothelial cells according to the method of Bevilacqua (47). There was no difference in the extent of ”Cr-labeled human polymorpho- nuclear leukocyte adhesion to Lp(a)-treated cells compared with control cells (control: 156 & 33 polymorphonuclear leu- kocytes bound/mm2 uersus Lp(a)-treated cells: 276 f 130 polymorphonuclear leukocytes bound/mm2).

Taken together these data suggest that Lp(a) modulates endothelial cell procoagulant function by specifically increas- ing the expression of PAI-1 mRNA. Enhancement of PAI-1 synthesis and secretion was observed despite the lack of an effect on other markers of “activation” (PDGF, tissue factor, and endothelial cell adhesive molecules). The selectivity of this Lp(a) effect differs markedly from the typical pattern of cytokine-induced activation which is associated with coordi- nate expression of all of these activities (43-47).

Effect of Lp(a) on t-PA Antigen and mRNA-To determine whether the observed effect of Lp(a) on PAI-1 activity and mRNA was specific or involved other fibrinolytic proteins, t-

;“b w-

28s -

2,lkb-

18S-

T F

28S- 2.3kb- 1.8kb-

- P D G F

28s-

18s-

1.3kb-

I L - 6

FIG. 4. Effects of Lp(a) on mRNAs for tissue factor (TF), PDGF, and IL-6 in endothelial cells. Northern blots containing equivalent amounts of mRNA from endothelial cells in several treat- ment groups were separately hybridized to the tissue factor, PDGF, and IL-6 probes as outlined under “Experimental Procedures.” The four treatment groups were 1 , control (con), 2, TNF (100 ng/ml, 24 h), 3, LDL (25 pg/ml, 24 h), and 4, Lp(a) (25 pg/ml, 24 h). The migration of marker ribosomal RNA species is indicated.

Lipoprotein (a) Regulates Endothelial Cell PAI-1 Expression 2463

PA antigen, and mRNA transcript levels were measured in endothelial cells treated with Lp(a) (25 pg/ml) for 24 h. We observed no differences in t-PA antigen or mRNA levels in Lp(a)-treated endothelial cells compared with control cells (plasminogen activator antigen: LDL-treated cells 7.2 ng/well uersus Lp(a)-treated cells 7.0 ng/well) (Fig. 5).

Effect of Lp(a) on PAI-1 mRNA in HT 1080 cells and Human Fibroblasts-To determine whether the effect of Lp(a) on PAI-1 mRNA was specific to vascular endothelial cells, we examined H T 1080 fibrosarcoma cells and human skin fibro- blasts for PAI-1 mRNA levels in response to Lp(a) treatment. Cells were cultured in medium containing 10% fetal bovine serum with Lp(a) (25 pg/ml) for 24 h. Northern blot analyses revealed no increase in PAI-1 mRNA levels in Lp(a)-treated HT 1080 cells compared with non-treated H T 1080 cells. PAI- 1 mRNA level in human skin fibroblasts was decreased 50% by Lp(a) treatment (Fig. 6). In addition, PAI-1 activity was assayed in conditioned media from Lp(a)-treated H T 1080 cells and fibroblasts and demonstrated no differences in PAI- 1 activity compared with untreated control cells, while con- ditioned media from dexamethasone-treated H T 1080 and TGF-P-treated fibroblasts demonstrated markedly increased PAI-1 activity (data not shown). These data suggest that Lp(a) modulation of PAI-1 mRNA level and PAI-1 activity may be relatively specific for endothelial cells.

Effect of Other Lipoproteins on PAI-1 mRNA in Human Endothelial Cells-To investigate whether the Lp(a) effect on PAI-1 mRNA in endothelial cells was lipoprotein-specific, we examined PAI-1 mRNA in endothelial cells treated with high density lipoprotein, oxidized LDL, and very low density lipo- protein. Cells were cultured for 24 h in medium containing either high density lipoprotein (25 pg/ml) or very low density

TPA

FIG. 5. Effect of Lp(a) on t-PA mRNA in endothelial cells. Messenger RNA isolated from three groups of endothelial cells: I , controls (con) , 2, endothelial cells treated for 24 h in the presence of LDL (25 pglml) and, 3, endothelial cells treated for 24 h with Lp(a) (25 pg/ml). Messenger RNA was hybridized by Northern blot to the tPA probe and the autoradiograph developed on Kodak X-omat film for 3 days a t -70 "C. Scanning laser densitometry was performed. There was no effect in treated cell GAPDH mRNA levels.

HT 1080 Fibroblast

FIG. 6. Effect of Lp(a) on PAI-1 mRNA in HT 1080 and human skin fibroblasts. H T 1080 cells or human skin fibroblasts were cultured as described. Cells were re-fed with medium containing Lp(a) (25 pg/ml) and allowed to incubate a t 37 "C for 24 h. RNA was prepared as described and Northern analyses performed using the radlolabeled PAI-1 probe. Scanning densitometry revealed no differ- ence in the intensity of either of the two mRNA species for PAI-1 in H T 1080 cells treated with Lp(a) compared with untreated cells (con). In human fibroblasts, the intensity of the 3.4-kb mRNA for PAI-1 was diminished by 50% in Lp(a)-treated cells compared with control cells.

lipoprotein (25 pg/ml) or LDL (25 pg/ml) which was oxidized by overnight dialysis against 5 p~ CuS04. Comparison of Northern blot analysis revealed no significant increase in PAI-1 mRNA levels following exposure of endothelial cells to the other major lipoproteins other than LDL (Fig. 7).

DISCUSSION

These studies demonstrate that Lp(a) is present in the microvasculature of inflamed tissues and promotes a signifi- cant increase in PAI-1 protein, activity, and mRNA expres- sion in human endothelial cells, without altering endothelial cell t-PA antigen or mRNA levels. LDL (native or oxidized) and other circulating lipids had no effect on PAI-1 suggesting that Lp(a) acts specifically on the endothelial cell to increase the proportion of PAI-1 to t-PA at the endothelial cell surface as well as in the local microenvironment. The effect of Lp(a) mRNA appeared to be relatively specific for endothelial cells and was not seen in H T 1080 cells or fibroblasts, raising the possibility that these cells may not express a functional Lp(a) receptor. Our previous studies have suggested that Lp(a) may bind to the plasminogen receptor on the endothelial cell surface and thus compete for plasminogen binding (15). This molecular interaction appears to interfere with the generation of an active fibrinolytic complex. I t remains to be elucidated whether the Lp(a) effect on PAI-1 expression depends on direct occupancy of this receptor.

Previous reports have suggested that PAI-1 in endothelial cells is subject to transcriptional control by various growth factors notably epidermal growth factor and heparin and may also vary with successive cell passage (13, 48). Our experi- ments were performed using early passage endothelial cells (P3) which were exposed to Lp(a) in the absence of epidermal growth factor and heparin or other growth factors. As has been shown with other modulators of PAI-1 activity, actino- mycin D exposure decreased PAI-1 mRNA transcript levels by 2-3-fold in our study (26). I t is known that the ratio of the 3.4- to 2.4-kb mRNA for PAI-1 varies with cell of origin (49).

The two species of PAI-1 mRNA in endothelial cells were both increased in response to Lp(a) whereas LDL increased only the 3.4-kb species in these cells. This suggests a potential regulatory step in endothelial cell fibrinolytic control, namely that the larger mRNA species may not undergo efficient translation to PAI-1 protein to the same extent as the smaller 2.4-kb species. This observation is consistent with the reports of others that the 3.4-kb mRNA contains a 3' AU-rich region

5 1 T

Untreated Lpa LDL ox LDL HDL VLDL EC

FIG. 7. Effect of lipoproteins on PAI-1 mRNA in endothelial cells (EC). Endothelial cells were grown to confluence in "199 medium containing 10% fetal bovine serum, polymyxin B (20 pg/ml). Cells were re-fed with mprlinm mntninine Iipnpmtninc nt mnmntrx tions of 25 pg/ml and incubated at 37 "C for 24 h. RNA was prepared and Northern analyses were performed as described. Scanning den- sitometry was performed on autoradiographs and the intensity of the PAI-1 mRNA signal compared with that of GAPDH. Results repre- sent mean 2 S.D. from three separate experiments. VLDL, very low density lipoprotein.

2464 Lipoprotein (a) Regulates Endothelial Cell PAI-1 Expression

that may confer mRNA instability or inability to be translated similar to that seen in mRNA encoding cytokines such as TNF, IL-1, and granulocyte macrophage colony stimulating factor (50-53). Our data further suggest that the ratio of 3.4- 2.4-kb mRNA species may be modulated by a specific agonist, Lp(a), and that this may influence both the amount and the activity of PAI-1 synthesized in these cells.

This effect of Lp(a) appears to be specific to that lipoprotein particle since other circulating lipoproteins in equivalent con- centrations such as LDL and very low density lipoprotein did not alter the PAI-1 mRNA profile of vascular endothelial cells. Interestingly, the effect of Lp(a) in raising PAI-1 mRNA in endothelial cells lasted for 24-48 h unlike the effects of other agonists such as thrombin whose effect lasted only 12 h (7). Others have reported increases in PAI-1 mRNA of similar duration in stimulated H T 1080 and rhabdo- myosarcoma cells (26, 54).

Unlike other agents that activate vascular endothelial cells (e.g. thrombin, IL-1) (7, 11, 45, 46), Lp(a) initiates a specific increase in PAI-1 activity without causing generalized cell activation that is characterized in part by tissue factor pro- duction, endothelial cell adhesive molecules expression, PDGF, and IL-6 production. This is unique for endothelial cell agonists and suggests a specialized role for circulating Lp(a) in vascular cell homeostasis.

The non-thrombogenic nature of the vascular endothelial cell surface is maintained by various means. The thrombo- modulin-protein C system and the heparin-antithrombin 111 system are both membrane oriented, and contribute to normal blood fluidity (55). The generation of plasmin at the endothe- lial cell surface similarly prevents local thrombus formation. Circulating lipids such as LDL have been postulated to be thrombogenic by interfering with eicosanoid production and promoting mononuclear cell adhesion to vascular endothelial cells (56, 57). Our data suggest that a specific atherogenic lipoprotein, Lp(a), promotes a thrombogenic state by increas- ing PAI-1 activity, thus interfering with endothelial cell and circulating plasmin generation.

The nature of Lp(a) as a vascular cell agonist appears to be unique in that it is specific for cell type and lipoprotein molecule. In addition, it does not appear to cause a generalized “activation” of the endothelial cell but is restricted to the fibrinolytic system. The role of cytokines and other inflam- matory mediators as regulators of the deposition and function of Lp(a) remains to be elucidated.

Acknowledgments-We wish to acknowledge the excellent techni- cal assistance of Barbara D. Summers, Barbara Ferris, and Emmanuel Cha.

1.

2.

3. 4.

5.

6. 7. 8.

9.

10.

11.

REFERENCES Hajjar, K. A., Harpel, P. C., Jaffer, E. A., and Nachman, R. L.

Hajjar, K. A., and Nachman, R. L. (1988) J . Clin. Inuest. 82,

Miles, L. A,, and Plow, E. F. (1988) Fibrinolysis 2, 61-71 Hajjar, K. A., Hamel, N. M., Harpel, P. C., and Nachman, R. L.

Erickson, L. A., Schleef, R. R., Ny, T., and Loskutoff, D. J. (1985)

Sprengers, E. D., and Kluft, C. (1987) Blood 69, 381-7 Dichek, D., and Quertermous, T. (1989) Blood 74 , 222-228 Nachman, R. L., Hajjar, K. A., Silverstein, R. L., and Dinarello,

Gelehrter, R. D., and Sznycer-Laszuk, R. (1986) J. Clin. Inuest.

Emeis, J. J., and Kooistra, T. (1986) J. Exp. Med. 163 , 1260-

Bevilacqua, M. P., Schleef, R. R., Gimbrone, M. A., Jr., and

(1986) J. Biol. Chem. 2 6 1 , 11656-11662

1769-1778

(1987) J. Clin. Inuest. 80, 1712-1719

Clin. Hematol. 14 , 513-529

C. A. (1986) J. Exp. Med. 163 , 1595-1600

77,165-169

1266

Loskutoff, D. J. (1986) J . Clin. Inuest. 78,587-591

12. Van Hinsbergh, V. W. M., Bertina, R. M., van Wijngaarden, A., van Tilburg, N. H., Emeis, J. J., and Haverkate, F. (1985) Blood

13. Konkle, B. A., and Ginsburg, D. (1988) J. Clin. Znuest. 82, 579- 585

14. Miles, L. A., Fless, G. M., Levin, E. G., Scanu, A. M., and Plow, E. F. (1989) Nature 339 , 303-305

15. Hajjar, K. A., Gavish, D., Breslow, J. L., and Nachman, R. L. (1989) Nature 3 3 9 , 303-305

16. Scott, J. (1989) Nature 341, 22-23 17. Gonzalez-Gronow, M., Edelberg, J. M., and Pizzo, S. V. (1989)

Biochemistry 28,2374-2377 18. Hoff, H. F., Beck, G. J., Skibinski, M. S., Jurgens, G., O’Nail, K.

J., and Lytle, B. (1988) Circulation 77, 1238-1244 19. Rhoads, G. G., Dahlen, G., Berg, K., Morton, N. E., and Dannen-

berg, A. L. (1986) J. Am. Med. Assoc. 256,2540-2544 20. Armstrong, V. W., Cremer, P., Eberle, E., Manke, A., Schulze,

F., Wieland, H., Kreuzer, H., and Seidel, D. (1986) Atheroscle- rosis 62,249-257

65,444-451

21. Brown, M. S., and Goldstein, J. L. (1987) Nature 330 , 113-114 22. Eaton, D. L., Fless, G. M., Kohr, W. J., McLean, J. W., Xu, Q.,

Miller, C. G., Lawn, R. M., and Scanu, A. (1987) Proc. Natl. Acad. Sci. U. S. A. 8 4 , 3224-3228

23. Maeda, S., Abe, A., Seishima, M., Makino, K., Noma, A,, and Kawade, M. (1989) Atherosclerosis 78 , 145-150

24. Jaffe, E. A., Nachman, R. L., Becker, C. G., and Minick, C. R. (1973) J. Clin. Inuest. 53,2745-2756

25. Sawdey, M., Podor, T. J., and Loskutoff, D. J. (1989) J . Biol. Chem. 264,10396-10401

26. Medcalf, R. L., Van den Berg, E., and Schleuning, W. D. (1988) J. Cell Biol. 106, 971-978

27. Schleef, R. R., Podor, T. J., Dunne, E., Mimuro, J., and Loskutoff, D. J. (1990) J. Cell Biol. 110 , 155-163

28. Havel, R., Eder, H., and Bragdon, J . (1955) J. Clin. Inuest. 34, 1345-1353

29. Marshall, P., Warso, M., and Lands, W. (1985) Anal. Biochem.

30. Harpel, P. C., Gordon, B. R., and Parker, T. S. (1989) Proc. Natl.

31. Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J.

32. Armstrong, V. W., Walli, A. K., and Seidel, D. (1985) J. Lipid

33. Knudson, B. S., Silverstein, R.L., Leung, L. L. K., Harpel, P. C.,

34. Hekman, C. M., and Loskutoff, D. J. (1985) J. Biol. Chem. 260,

35. Chirgwin, J. M., Przybyla, A. E., MacDonald, R., and Rutler, W. J. (1979) Biochemistry 18,5894-5899

36. Fienberg, A., and Vogelstein, B. (1984) Anal. Brockm. 137,266- 267

37. Ny, T., Sawdey, M., Laurence, D., Millan, J. L., and Loskutoff, D. (1986) Proc. Natl. Acad. Sci. U. S. A. 83,6776-6780

38. Edlund, T., Ny, T., Ranby, M., Heden, L., Palm, G., Holmgren, E., and Josephson, J. (1983) Proc. Natl. Acad. Sci. U. S. A. 80,

39. Spicer, E., Horton, R., Bloem, L., Bach, R., Williams, K. R., Guha, A,, Kraus, J., Lin, T., Nemerson, V., and Konigsberg, W. (1987) Proc. Natl. Acad. Sci. U. S. A. 84,8957-8961

40. May, L. T., Helfgott, D. C., and Sehgal, P. B. (1986) Proc. Natl. Acad. Sci. U. S. A. 83,8957-8961

41. Betsholtz, C., Johnson, A., Helden, C-H., Westermark, B., Lind,

A. L., Knott, T. J., and Scott, J. (1986) Nature 320 , 695-699 P., Ureda, M. S., Eddy, R., Shows, T. B., Philpott, K., Mellor,

42. Carson, S . D., and Archer, P. G. (1986) Thrornb. Res. 4 1 , 185- 195

43. Majesky, M. W., Benditt, E. P., and Schwartz, S. M. (1988) Proc. Natl. Acad. Sci. U. S. A. 85, 1525-1528

44. Seheal. P. B.. May. L. T., Tamm, I., and Vilcek, J . (1987) Science

145, 192-199

Acad. Sci. U. S. A. 86, 3847-3851

(1951) J. Biol. Chem. 193, 265-275

Res. 26, 1314-1323

and Nachman, R. L. (1986) J. Biol. Chem. 261,10765-10771

11581-11587

349-352

” ,

235; 731-732 45. Nawroth, P. P., and Stern, D. M. (1986) J. Exp. Med. 163 , 740-

745 46. Bevilacqua, M. P., Pober, J. S., Majeau, G. R., Fiers, W., Cotran,

R. S., and Gimbrone, M. A. (1986) Proc. Natl. Acad. Scz. U. S.

47. Nawroth, P. P., Handley, D. A., Esmon, C. T., and Stern, D. A. 84,9238-9242

(1986) Proc. Natl. Acad. Sci. U. S. A. 83, 3460-3464

Lipoprotein (a) Regulates Endothelial Cell PAI-1 Expression 2465

48. Dichek, D., and Quertermous, T. (1989) In Vitro Cell. 25, 289- 53. Kruys, V., Marinx, O., Shaw, G., Deschamps, J., and Huez, G.

49. Lucore, C. L., Fujii, S., Wun, T.-C., Sobel, B. E., and Billadello, 54. Mayer, M., Lund, L. R., Riccio, A., Skouv, J., Nielson, L. S.,

50. Kruys, V., Wathelet, M., Poupart, P., Contreras, R., Fiers, W., Chem. 263,15688-15693

292 (1989) Science 245, 852-854

J. J. (1988) J. Biol. Chem. 263, 15845-15848 Stacey, S. N., Dano, K., and Andreasen, P. A. (1988) J. Biol.

Content, J., and Huez, G. (1987) Proc. Natl. Acad. Sci. U. S. A. 55. Esmon, N. L., and Esmon, C . T. (1988) Sem. Thromb. Hemostasis

51. Shaw, G., and Kamen, R. (1986) Cell 46,659-667 56. Frostegard, J., Nilsson, J., Haegestrand, A., Hamsten, A,, Wigzell, 52. Caput, D., Beutler, B., Hartog, K., Thayer, R., Brown-Shimer, H., and Gidlund, M. (1990) Proc. Natl. Acad. Sci. U. S. A. 87,

84,6030-6034 1 4 , 210-215

S., and Cerami, A. (1986) Proc. Natl. Acad. Sci. U. S. A. 8 3 , 904-908 1670-1674 57. Pomerantz, K., and Hajjar, D. (1989) J. Lipid Res. 30,1385-1395