HEMOSTASIS, THROMBOSIS, AND VASCULAR BIOLOGY
Oxidized phospholipids in oxidized low-density lipoprotein down-regulate
thrombomodulin transcription in vascular endothelial cells through a decrease
in the binding of RAR␤-RXR␣ heterodimers and Sp1 and Sp3 to
their binding sequences in the TM promoter
Hidemi Ishii, Tsuyoshi Tezuka, Hiroyuki Ishikawa, Kimihiko Takada, Koji Oida, and Shuichi Horie
The present work investigated the mechanism for down-regulation of thrombomodulin (TM), an anticoagulant glycoprotein, on cultured umbilical vein endothelial
cells (HUVECs) exposed to lipid extracts
from oxidized low-density lipoprotein (oxLDL). HUVECs exposed to phospholipid
extracts, but not to free cholesterol, triglyceride, or cholesterol ester, isolated
from ox-LDL reduced TM mRNA levels to
nearly the same extent as native ox-LDL.
Oxidized 1-palmitoyl-2-arachidonyl-snglycero-3-phosphocholine (ox-PAPC), but
not native PAPC or a reduced form of
ox-PAPC, markedly decreased TM mRNA
levels. The apparent half-life (t 1/2 ⴝ 2.7
hours) of TM mRNA in control cells was
not significantly different from that in
cells exposed to ox-LDL or ox-PAPC. TM
mRNA levels were regulated by transcriptional activation via a retinoid receptor ␤
(RAR␤). The binding activities of nuclear
proteins from HUVECs treated with oxLDL or ox-PAPC to the DR4 or stimulatory
protein 1 (Sp1) sequence in the TM promoter were significantly reduced with decreased expression of RAR␤, retinoid X
receptor ␣ (RXR␣), Sp1, and Sp3 in the
nuclei. The promoter activity in HUVECs
transfected with a reporter plasmid expressing the TM promoter with targeted
deletions in the DR4 and Sp1 binding
elements was decreased to about 20% of
that with the wild-type construct. Treatment of the cells with ox-PAPC had no
additional effect on the promoter activity.
These results suggest that oxidized phospholipids in ox-LDL inhibit transcription
of the TM gene in HUVECs by inhibiting
the binding of RAR␤-RXR␣ heterodimer
and Sp, including Sp1 and Sp3, to the
DR4 element and Sp1 binding element,
respectively, in the TM promoter with
reduced expression of RAR␤, RXR␣, and
Sp1 and Sp3 in the nuclei. (Blood. 2003;
101:4765-4774)
© 2003 by The American Society of Hematology
Introduction
Vascular endothelial cells actively participate in the regulation of
hemostasis. Under normal circumstances, these cells are thromboresistant, a state maintained by antithrombotic mechanisms, as well
as control of expression of procoagulant factors, such as tissue
factor. Thrombomodulin (TM) is a high-affinity receptor for
thrombin on the endothelial cells, which acts not only as a direct
inhibitor for the procoagulant activity of thrombin toward fibrin
formation and platelet activation, but also as a key cofactor for the
TM–thrombin–protein C pathway, which is a major physiologic
antithrombotic property of endothelial cells.1,2 Protein C, which
circulates in the plasma as a protease zymogen, is activated by the
TM-thrombin complex, and the activated protein C proteolytically
inactivates coagulation factors Va and VIIIa, which are indispensable to blood coagulation.1,2 Therefore, thrombin formation in the
coagulation cascade is reduced by activation of the TM–thrombin–
protein C pathway. The physiologic relevance of this pathway for
antithrombotic properties of endothelial cells is emphasized by the
thrombosis disorders frequently observed in patients with protein
C–resistant factor V3,4 or protein C deficiency,5 and it has been
postulated that TM is an important regulator in maintaining the
fluidity of circulating blood and lymphatic fluid. The cofactor
activity of TM isolated from human placenta is markedly enhanced
by reconstitution into vesicles consisting of phosphatidylcholine
and acidic phospholipids, such as phosphatidylserine and phosphatidylethanolamine, so that optimal antithrombotic TM activity
depends on the local cell membrane status.6,7
The cDNA sequence of human TM encodes for 575 amino
acids.8,9 The 5⬘-flanking region of the human TM gene contains a
TATA box, a CAAT box, and several conserved transcription factor
binding elements such as a retinoic acid receptor responsive
element (RARE) and stimulatory protein 1 (Sp1) binding element.10-12 The RARE in the TM gene consists of direct repeats of
RARE separated by 4 base (DR4) sequences, and it binds a
heterodimer of retinoic acid receptors (RAR␣, RAR␤, and RAR␥)
and retinoid X receptor ␣ (RXR␣).10-12 We previously reported that
TM expression on cultured human umbilical vein endothelial cells
(HUVECs) was up-regulated by treatment of the cells with retinol
or retinoids containing retinoic acid,13 increases mediated by the
interactions of the RAR␤-RXR␣ heterodimer to the DR4 sequence
and nuclear Sp1 to the Sp1 binding site.12 Up-regulation of the TM
cofactor activity and antigen levels in HUVECs was also observed
when the cells were stimulated with histamine14 and cAMP.15 In
contrast, down-regulation of TM antigen levels in HUVECs was
observed in response to bacterial endotoxin16 or inflammatory
From the Department of Public Health, Showa Pharmaceutical University,
Machida, Tokyo, Japan; the Third Department of Internal Medicine, Fukui
Medical University, Fukui, Japan; and the Department of Clinical Biochemistry,
Faculty of Pharmaceutical Sciences, Teikyo University, Suarashi, Tsukui,
Kanagawa, Japan.
Reprints: Hidemi Ishii, Department of Public Health, Showa Pharmaceutical
University, 3-3165 Higashi Tamagawa Gakuen, Machida, Tokyo 194-8543,
Japan; e-mail: [email protected].
Submitted August 9, 2002; accepted January 28, 2003. Prepublished online as
Blood First Edition Paper, February 6, 2003; DOI 10.1182/blood-2002-08-2428.
BLOOD, 15 JUNE 2003 䡠 VOLUME 101, NUMBER 12
The publication costs of this article were defrayed in part by page charge
payment. Therefore, and solely to indicate this fact, this article is hereby
marked ‘‘advertisement’’ in accordance with 18 U.S.C. section 1734.
© 2003 by The American Society of Hematology
4765
4766
ISHII et al
cytokines such as interleukin-1␤ (IL-1␤) and tumor necrosis
factor-␣ (TNF-␣).17-20 TM down-regulation in cells exposed to the
inflammatory cytokines appears to require internalization and
degradation of surface TM21 and/or inhibition of TM transcription.20,22,23 The mechanism of the down-regulation of TM transcription is, however, obscure.
Oxidized low-density lipoprotein (ox-LDL) plays a pivotal role
in the pathogenesis of atherosclerosis and alters various physiologic circumstances, including vascular tone, endothelial permeability, and coagulation and fibrinolysis controls within the microenvironments of atherosclerotic lesions.24 A study on the
mechanisms of altered coagulation and fibrinolysis control in
endothelial cells exposed to ox-LDL would contribute to understanding the pathophysiology of thrombus formation in atherosclerotic
lesions. We previously demonstrated that HUVECs exposed to
ox-LDL, but not to native or acetylated LDL, reduced TM
expression on the cells. The decreased TM mRNA levels were
mediated by the lipid components of ox-LDL through internalization of ox-LDL.25 Subsequently, down-regulation of TM expression on endothelial cells was found in lesions of coronary
atherosclerosis.26 The present work was undertaken to identify an
active lipid species from ox-LDL important for down-regulation of
TM mRNA levels and to elucidate the molecular basis for
down-regulation of TM transcription in endothelial cells treated
with ox-LDL.
Materials and methods
Materials
Reagents were purchased from Wako Pure Chemical Industries (Osaka,
Japan), unless otherwise indicated. Bovine serum albumin (BSA), human
thrombin (4000 NIH units/mg), 3-(4,5-dimethylthiazole-2-yl)-2,5-diphenyltetrazolium bromide (MTT), 1-palmitoyl-2-arachidonyl-sn-glycero-3phosphocholine (PAPC), lysophosphatidylcholine (lysoPC) from egg yolk,
7␤-hydroxyl-cholesterol, 25-hydroxycholesterol, and 7-oxocholesterol were
purchased from Sigma Chemical (St Louis, MO); 5,6-dichloro-1-␤-Dribofuranosylbenzimidazole (DRB), from Calbiochem (San Diego, CA);
and poly(dI-dC) and PD-10 column, from Pharmacia (Piscataway, NJ). A
kit for assay of TM antigen was donated from Mitsubishi Gas Company
(Tokyo, Japan). Rabbit anti–human RAR␣, RAR␤, RAR␥, RXR␣, Sp3, or
Sp4 immunoglobulin G (IgG) and mouse monoclonal anti–human Sp1 IgG
were purchased from Santa Cruz Biotechnology (Santa Cruz, CA).
Horseradish peroxidase (HRP)–conjugated sheep anti–rabbit IgG and
HRP-conjugated sheep anti–mouse IgG were purchased from Southern
Biotechnology Associate (Birmingham, AL). Dulbecco modified Eagle
medium (DMEM) and Dulbecco phosphate-buffered saline (PBS) were
purchased from Nissui Pharmaceuticals (Tokyo, Japan). [␣32P]–deoxycytosine triphosphosphate (dCTP) and [␥32P]–adenosine triphosphate (ATP)
were purchased from NEN Research Products (Tokyo, Japan). Oligonucleotides were synthesized by Amersham Pharmacia Biotech (Tokyo, Japan).
Cell culture
HUVECs were cultured on collagen-coated 96- or 6-well plates (Corning;
Iwaki Glass, Tokyo, Japan) as described previously.15 Confluent monolayer
cells were washed in serum-free medium and then incubated in DMEM
supplemented with 20% fetal calf serum (FCS) containing LDL, lipid
extracts from LDL, or PAPC and/or various compounds. After incubation
for the indicated periods, cell viability was determined by the MTT assay as
described previously.27
BLOOD, 15 JUNE 2003 䡠 VOLUME 101, NUMBER 12
ultraviolet (UV) irradiation (254 nm, 0.4 mW/cm2) of LDL for 12 hours at
4°C as described previously.25 Oxidation of LDL was estimated as the
formation of thiobarbituric acid reactive substances (TBARS) and expressed as malondialdehyde (MDA) equivalents per milligram of protein
using 1,1,3,3-tetraethoxypropane, a precursor of MDA, as a standard.28
Phospholipid and triglyceride content in the LDL were determined using
assay kits Phospholipid C-test Wako and Triglyceride G-test Wako,
respectively. Total cholesterol was determined using the Cholesterol
CII-test Wako kit with cholesterol esterase and cholesterol oxidase. Free
cholesterol was determined using the Cholesterol CII-test Wako with only
cholesterol oxidase. Cholesterol ester content was determined as the total
cholesterol content minus the free cholesterol content. Protein content was
determined by the method of Bradford29 using BSA as a standard.
Lipid extraction from LDL and fractionation of lipid extracts
Crude lipid extracts from LDL (⬇400 ␮g protein) in 0.2 mL PBS were
obtained by the method of Bligh and Dyer30 as described previously.25 The
crude lipid extract in the organic phase was evaporated and dispersed in
DMEM containing 20% FCS by sonication. The lipid suspension was used
in experiments as a crude lipid from native LDL or ox-LDL. For additional
fractionation of lipids, crude lipid extracts in the organic phase were dried
under a stream of nitrogen and then dissolved in 0.5 mL hexane–2-propanol
(9:1, vol/vol). The dissolved lipids (0.5 mL) were applied onto a 2.0-mL
Sep-Pak silica cartridge (Millipore Corporation, Milford, MA) equilibrated
with hexane–2-propanol (9:1, vol/vol), and each lipid was separated with a
modification of the method described by Hamilton and Comai.31 Free
cholesterol, triglyceride, and cholesterol ester were eluted isocratically with
5 mL of hexane–2-propanol (9:1, vol/vol) and dried under a stream of
nitrogen and then dissolved in hexane–ether–formic acid (90:10:0.2,
vol/vol/vol). Phospholipids eluted with 24.0 mL methyltertiarybutylether–
methanol–ammonium acetate (5:8:2, vol/vol/vol) were dried and dissolved
in chloroform. The fraction containing free cholesterol, triglyceride, and
cholesterol ester was further applied onto Sep-Pak silica cartridge column
equilibrated with hexane–ether–formic acid (90:10:0.2, vol/vol/vol). Triglyceride and cholesterol ester eluted isocratically with 4 mL of the same
solvent were dried and dissolved in chloroform. Free cholesterol was eluted
with 3 mL hexane–2-propanol (9:1, vol/vol) and dried and then dissolved in
chloroform. To identify each lipid, the separated lipids were developed in
parallel by silica gel thin-layer chromatography with hexane-isopropanol
(9:1). Each lipid was detected under iodine vapors and identified by
comparison with the mobility of standard lipids. Each lipid in chloroform
was transferred to a glass test tube and evaporated under a stream of
nitrogen. The lipid residue was dispersed in DMEM containing 20% FCS
by sonication for about 30 seconds. These lipid suspensions were used for
the experiments.
Oxidation and reduction of phospholipid
PAPC (1 mg) in chloroform was evaporated under a stream of nitrogen. The
residue dissolved in 1 mL methanol was poured in 1 well of a 6-well plate
and irradiated by UV light (254 nm, 0.4 mW/cm2) for 12 hours at 4°C. The
ox-PAPC was dried under a stream of nitrogen and dissolved in chloroform.
Reduction of ox-PAPC or oxidized phospholipids from ox-LDL was
performed as described by Watson et al.32 Ox-PAPC or oxidized phospholipids in chloroform were dried under a stream of nitrogen and resuspended in
1.0 mL of 0.1 M borate buffer (pH 8.0) by sonication for 30 seconds. The
suspension was instantly reduced by adding 500 ␮g sodium borohydride.
After 5 minutes, the suspension was mixed with 3 mL chloroform-methanol
(1:1) and the chloroform phase was extracted. The reduced form of
ox-PAPC or oxidized phospholipids from ox-LDL was dried under a stream
of nitrogen and the lipid residue was dispersed in DMEM containing 20%
FCS by sonication.
Assay of TM antigen and TM mRNA
Isolation and modification of LDL
LDL (density range, 1.019-1.063 g/mL) was isolated from fresh human
plasma of healthy subjects and ox-LDL (2 mg protein/mL) was prepared by
TM antigen levels in cultured HUVECs were determined by enzyme
immunoassay as previously described.33 TM mRNA levels were measured
by reverse transcription–polymerase chain reaction (RT-PCR) using TM
BLOOD, 15 JUNE 2003 䡠 VOLUME 101, NUMBER 12
primers and ␤-actin primers9 as described previously.34 Amplified cDNA
was detected with 0.1 mg/mL ethidium bromide staining after electrophoresis through a 2% agarose gel and photographed under UV light (254 nm).
The gel areas corresponding to amplified cDNA were cut out and counted
for radioactivity. There was a linear relationship between the Cerenkov
counts obtained for each PCR product and the amount of total RNA
obtained under the experimental conditions, in the range from 0 to 1.25 ␮g
RNA. The PCR product counts from the TM cDNA were normalized to
those from the ␤-actin cDNA. The half-life of mRNA was determined by
the method of Xie et al.35 In the experiments with DRB, a transcriptional
inhibitor, postconfluent HUVECs were incubated with 65 ␮M DRB and/or
ox-LDL or ox-PAPC in culture medium for 0, 2, 4, and 6 hours at 37°C.
Electrophoretic mobility shift assay (EMSA) and
supershift assay
Cultured HUVECs (5 ⫻ 105 cells/plate) were homogenized in a Potter
Elvejem–type homogenizer and centrifuged at 2000 rpm to obtain the
nuclear fraction. This fraction was suspended in 800 ␮L of 20 mM Tris
(tris(hydroxymethyl)aminomethane)–borate (TB) buffer (pH 8.0) containing 10 mM KCl, 0.1 mM EDTA [ethylenediaminetetraacetic acid], 0.1 mM
EGTA [ethylene glycol tetraacetic acid], 1 mM dithiothreitol (DTT), 2
␮g/mL leupeptin, 2 ␮g/mL aprotinin, 0.5 mg/mL benzamidine hydrochloride, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), and 0.6% nonidet
P-40 (NP-40) and stored at ⫺80°C as a nuclear extract. Double-stranded
oligonucleotides were prepared by annealing (heating at 95°C for 5 minutes
and chilling at room temperature for 60 minutes) single-strand oligonucleotides of 5⬘-CGTTTGGTCACTGCAGGTCAGTCCA-3⬘ (at position ⫺1535
to ⫺1511 containing underlined consensus DR4 sequences in the 5⬘flanking region of the human TM gene), 5⬘-CTGTGTTGCACGTGCAAGCTCCGTA-3⬘ (scrambled DR4 sequences), or 5⬘-GCGAGGGCGGGCGCAAGGGCGGCCA-3⬘ (at position ⫺145 to ⫺121 for
the tandem Sp1 sites containing second and first Sp1 sites in the 5⬘-flanking
region of the human TM gene). The purity of each double-stranded
oligonucleotide was confirmed as described previously.12 The EMSA
reaction mixture contained 0.5 ng labeled probe (1 ⫻ 105 cpm), 1 ␮g
poly(dI-dC), and 2 ␮g nuclear proteins in a final volume of 6 ␮L TB buffer
(pH 8.0) containing 50 mM KCl, 5 mM MgCl2, 1 mM ZnCl2, 0.5 mM DTT,
0.1 mM PMSF, 5 mM spermidine, 0.008% NP-40, and 15% glycerol. The
mixture was incubated for 30 minutes at 25°C and subjected to 4%
polyacrylamide gel electrophoresis (PAGE) using a buffer consisting of 10
mM Tris-HCl (pH 7.9) containing 1.5 mM EDTA and 5 mM sodium
acetate. After electrophoresis, the gels were dried on filter paper and
subjected to autoradiography. To identify specific RARs, RXR␣, Sp1, Sp3,
or Sp4 proteins involved in DNA binding (the supershift assay), 5 ␮g
nuclear proteins were incubated with 1 ␮g rabbit IgG directed against
human RAR␣, RAR␤, RAR␥, RXR␣, Sp3 or Sp4 protein, or monoclonal
anti–human Sp1 IgG for 30 minutes at 25°C. After incubation, labeled DR4
or Sp1 oligonucleotide was added and reaction was allowed to proceed for
30 minutes at 25°C before EMSA was performed.
Western blot analysis
After treatment with various reagents, nuclear proteins were extracted from
HUVECs (3 ⫻ 105 cells/well) according to the method described above.
Nuclear protein (5 ␮g) was subjected to sodium dodecyl sulfate (SDS)–
PAGE under nonreduced conditions as described previously.36 After
SDS-PAGE, protein was transferred to a nitrocellulose membrane and
incubated with 0.5 ␮g/mL polyclonal antibody against human RAR␤,
RXR␣, Sp3 or Sp4, or 0.5 ␮g/mL monoclonal antibody against human Sp1.
Detection of these proteins was carried out using HRP-conjugated sheep
anti–rabbit IgG or HRP-conjugated sheep anti–mouse IgG. A chemiluminescence detection system (Super Signal, Pierce Chemical, Rockford, IL)
was used.
Transient transfection, luciferase assay, and preparation
of cell lysates
The promoter activity of TM was determined using pGL3-pTM reporter
plasmid containing the 5⬘-flanking region of TM as described previously.12
ox-LDL DOWN-REGULATES THROMBOMODULIN
4767
The pGL3-pTM plasmid and deletion mutants were prepared by insertion of
an MluI-XhoI fragment derived from pTM-1562CAT into the pGL3 basic
plasmid (Promega). HUVECs were cultured in a 6-well plate with DMEM
medium supplemented with 20% FCS, human epidermal growth factor
(hEGF; 10 ng/mL), human fibroblast growth factor B (hFGF-B; 5 ng/mL),
heparin (10 ␮g/mL), and hydrocortisone (1 ␮g/mL) and were transfected
with pGL3-pTM plasmids (2 ␮g/well) in the presence of the pRL-TK
plasmid (0.1 ␮g/well) by the cationic liposome method at a cell density of
70% to 80% confluency in an OPTI-MEM medium.12 After 6 hours, the
culture medium was replaced. Cells were cultured for 12 hours and then
exposed to 50 ␮g/mL of either native PAPC (n-PAPC) or ox-PAPC for 12
hours. The promoter activity of the cell lysate was measured by a
dual-luciferase reporter assay system (Promega), and the transfection
efficiency was corrected with pRL-TK vector-dependent renilla luciferase activity.
Results
TM mRNA levels in HUVECs exposed to lipid extracts
from ox-LDL
The dose- and time-dependent effects of ox-LDL or crude lipid
extracts from ox-LDL on TM mRNA levels in HUVECs were
investigated (Figure 1). The TBARS value of ox-LDL prepared by
UV irradiation of LDL was 14.33 ⫾ 2.66 nmol MDA/mg protein,
while no TBARS was observed in native LDL. LDL containing 200
␮g protein contained 200 ⫾ 15 ␮g (mean ⫾ SD) phospholipid.
Cell viability as measured by the MTT assay was higher than 90%
in all these experiments (data not shown). A dose- (Figure 1A-B)
and time-dependent (Figure 1C-D) decrease in TM mRNA levels
were observed in the cells treated with ox-LDL, and the normalized
signal decreased by about 65% in the cells treated with 200 ␮g
protein/mL ox-LDL for 12 hours (Figure 1A,C) as previously
described.25 HUVECs treated with crude lipid extracts from
ox-LDL also decreased TM mRNA levels in a dose- (Figure 1E-F)
and time-dependent (Figure 1G-H) manner. The normalized TM
mRNA signal decreased by 60% in the cells treated with crude lipid
extracts containing 200 ␮g/mL phospholipids from ox-LDL for 12
hours (Figure 1E,G). No influence on TM mRNA levels was
observed with incubations of up to 12 hours with 200 ␮g
protein/mL native LDL or crude lipid extracts containing 200
␮g/mL phospholipids from native LDL (Figure 1A,E). The results
indicated that the lipid component of ox-LDL induced a decrease in
HUVEC TM mRNA levels to the same extent as the ox-LDL.
TM mRNA levels in HUVECs exposed to phospholipid
extracts from ox-LDL
To elucidate the active lipid species in LDL, the different lipid
components of ox-LDL were separated and analyzed for their effect
on TM mRNA levels in HUVECs (Figure 2A). Crude lipid extracts
from ox-LDL were separated into 3 fractions, which predominantly
contained phospholipid, free cholesterol, and triglyceride and
cholesterol ester. On average, LDL containing 200 ␮g protein
contained 200 ␮g phospholipid, 37 ␮g free cholesterol, 52.7 ␮g
triglyceride, and 252 ␮g cholesterol ester, and these concentrations
were used in the experiments. A significant decrease in TM mRNA
levels was observed in the cells treated with 200 ␮g protein/mL
ox-LDL, crude lipid extracts containing 200 ␮g/mL phospholipid,
or 200 ␮g/mL phospholipid. A decrease was not observed with the
cholesterol ester and triglyceride fraction. A slight decrease in TM
mRNA levels was observed in the cells treated with 37 ␮g/mL free
cholesterol isolated from the ox-LDL. However, commercially
4768
BLOOD, 15 JUNE 2003 䡠 VOLUME 101, NUMBER 12
ISHII et al
Figure 1. Dose- and time-dependent decrease in TM
mRNA levels in HUVECs treated with ox-LDL or crude
lipid extracts from ox-LDL. (A-B) Cells were treated for 12
hours with or without various concentrations of ox-LDL or 200
␮g protein/mL native LDL (n-LDL). (C-D) Cells were treated
with 200 ␮g protein/mL ox-LDL for various times. (E-F) Cells
were treated for 12 hours with or without crude lipids extracts
(lipids) from ox-LDL or from 200 ␮g protein/mL n-LDL. The
lipid concentrations are expressed as the equivalent concentration of phospholipids (PL) in various concentrations of oxor n-LDL. The crude lipid extracts containing 200 ␮g/mL PL
from LDL are of equal concentration to those in 200 ␮g
protein/mL LDL. (G-H) Cells were treated for various times
with the crude lipids containing 200 ␮g PL/mL extracted from
ox-LDL. Total RNA was extracted from the cells, and TM
mRNA levels in the cells were analyzed by quantitative
RT-PCR as described in “Materials and methods.” (B,D,F,H)
The autoradiograms of panels B, D, F, and H were analyzed
by densitometry and the results are expressed in panels A, C,
E, and G, respectively, as a percentage of control values after
normalization to the ␤-actin mRNA signal. Panels A, C, E, and
G represent the mean ⫾ SD from 4 independent experiments, and panels B, D, F, and H show representative results.
available cholesterol (40 ␮g/mL) irradiated with UV light for 12
hours failed to decrease in TM mRNA levels (data not shown).
About a 50% decrease in TM antigen levels was observed in
HUVECs treated for 12 hours with the ox-LDL phospholipid
fraction (200 ␮g/mL) and the extent of the decrease was comparable with that induced by 200 ␮g protein/mL ox-LDL (data not
shown). No significant decrease in TM antigen levels was observed
in the cells treated with either native LDL or the 3 lipid extracts of
native LDL (data not shown). A dose-dependent decrease in TM
mRNA levels was observed in HUVECs treated for 12 hours with
50 to 200 ␮g/mL phospholipid fraction of ox-LDL (Figure 2B).
The normalized TM mRNA signal decreased by 60% in the cells
treated with 200 ␮g/mL ox-LDL phospholipid (Figure 2C). In
contrast, the phospholipid (200 ␮g/mL) separated from native LDL
did not affect TM mRNA levels (Figure 2B-C). A time-dependent
Figure 2. Phospholipids separated from ox-LDL induce a decrease in TM mRNA
levels in HUVECs. The crude lipid extracts (Lipid) from ox-LDL were separated as
described in “Materials and methods” into the phospholipid (PL), cholesterol ester
and triglyceride (CE/TG), and free cholesterol (Cho) fractions. (A) HUVECs were
treated for 12 hours with each lipid fraction at a concentration equivalent to that found
in 200 ␮g protein/mL ox-LDL. TM mRNA levels were determined by quantitative
RT-PCR. The signal is expressed as the percentage of control values (CTL) after
normalization to the ␤-actin mRNA signal. (B) Phospholipids n-PL and ox-PL were
separated from crude lipid extracts of native LDL or ox-LDL, respectively, and
dispersed in DMEM containing 20% FCS. HUVECs were incubated with or without
increasing concentrations of the phospholipids for 12 hours and TM mRNA levels
were determined. (C) The autoradiogram of panel B was analyzed by densitometry,
and the data are expressed as the percentage of control values after normalization to
the ␤-actin mRNA signal. Data show the mean ⫾ SD from 4 independent experiments.
decrease in TM mRNA levels also was observed in the cells treated
with 200 ␮g/mL ox-LDL phospholipid extracts for up to 12 hours
(data not shown) with results similar to that observed in Figure
1G-H. Taken together, the results indicated that oxidized phospholipids may be the primary active species of ox-LDL that contributes
to reduced TM mRNA levels in HUVECs.
Ox-PAPC
Ox-PAPC is among the oxidized phospholipids generated during
oxidative modification of LDL. Ox-PAPC is a mixture of oxidized
arachidonic acid–containing phospholipids that is obtained by
auto-oxidation of 1-palmitoyl-2-arachidonyl-sn-glycero-3-phoshocholine and is reportedly responsible for some biologic activities of
ox-LDL relevant to atherogenesis.32,37,38 Therefore, the present
work investigated the effect of ox-PAPC on TM antigen and
mRNA levels in HUVECs (Figure 3). Treatment of HUVECs with
ox-PAPC (TBARS ⫽ 20.03 ⫾ 0.10 nmol MDA/mg phospholipid)
induced a dose-dependent decrease in TM antigen and mRNA
levels (Figure 3A-C). Treatment of the cells with 50 ␮g/mL
ox-PAPC for 12 hours induced a 54% decrease of TM antigen and
75% decrease of TM mRNA levels relative to control cells. No
significant decrease in TM antigen and mRNA levels was observed
in cells treated with 100 ␮g/mL nonoxidized native PAPC for 12
hours. A time-dependent decrease in TM mRNA levels in the cells
treated with 50 ␮g/mL ox-PAPC was observed during the 12-hour
incubation (Figure 3D). Ox-PAPC was reduced immediately by
incubation with sodium borohydride for 5 minutes and its effect on
TM mRNA levels was investigated (Figure 3E-F). Although
down-regulation of TM mRNA levels in the cells treated with 50
␮g/mL ox-PAPC for 12 hours were observed to the same extent as
the above results, no decrease in TM mRNA levels was observed in
the cells treated with the reduced form of ox-PAPC. These results
suggest that ox-PAPC, a representative component of oxidized
phospholipids in ox-LDL, may participate in the reduction of TM
antigen and mRNA levels in HUVECs.
Half-life of TM mRNA in HUVECs exposed to ox-LDL
or ox-PAPC
The half-life of TM mRNA in HUVECs treated with or without
ox-LDL or ox-PAPC was measured using DRB, a transcription
BLOOD, 15 JUNE 2003 䡠 VOLUME 101, NUMBER 12
ox-LDL DOWN-REGULATES THROMBOMODULIN
4769
ox-LDL or ox-PAPC decreased TM mRNA levels without an
apparent effect on the stability of the mRNA in HUVECs.
RAR, RXR, and Sp proteins in the nuclei
Figure 3. TM antigen and its mRNA levels in HUVECs treated with ox-PAPC,
native PAPC, or the reduced form of ox-PAPC. (A-B) HUVECs were incubated with
various concentrations of ox-PAPC or 100 ␮g/mL native PAPC for 12 hours. TM
antigen (A) and its mRNA (B) levels in the cells were assayed by the method as
described in “Materials and methods.” Data shown in panel A are the mean from 3
independent experiments. (C) The autoradiogram of panel B was analyzed by
densitometry and the data are expressed as the percentage of control values after
normalization to the ␤-actin mRNA signal. Data shown represent the mean ⫾ SD from
4 independent experiments. (D) HUVECs were treated with 50 ␮g/mL ox-PAPC for
various times, and TM mRNA levels were quantitated by the RT-PCR method. (E)
ox-PAPC was reduced by incubation with 500 ␮g sodium borohydride for 5 minutes,
and the reduced form of ox-PAPC was extracted and dispersed in DMEM as
described in “Materials and methods.” HUVECs were treated with 50 ␮g/mL ox-PAPC
or 50 ␮g/mL reduced form of ox-PAPC for 12 hours, and TM mRNA levels were
assayed. (F) The autoradiogram of panel E was analyzed by densitometry, and the
data are expressed as the percentage of control values after normalization to the
␤-actin mRNA signal. Data shown are the mean ⫾ SD from 4 independent
experiments.
inhibitor, to investigate effects on TM mRNA stability (Figure 4).
TM mRNA levels in the cells treated with 65 ␮M DRB alone
decreased in a time-dependent manner (Figure 4A,C), and the
apparent half-life of normalized TM mRNA was calculated as
2.7 ⫾ 0.3 hours (Figure 4B,D). The half-life of TM mRNA in cells
treated with ox-LDL plus DRB or ox-LDL alone was 3.1 ⫾ 0.3
hours and 3.0 ⫾ 0.3 hours, respectively, and did not significantly
differ from that in the cells treated with DRB alone. The half-life of
TM mRNA in cells treated with ox-PAPC plus DRB or ox-PAPC
alone was 2.8 ⫾ 0.2 hours and 3.1 ⫾ 0.2 hours, respectively. These
also did not differ significantly from that observed in the cells
treated with DRB or ox-LDL alone. The results indicated that
Figure 4. The half-life of TM mRNA in HUVECs treated
with DRB, ox-LDL, or ox-PAPC. HUVECs were treated
with 65 mM DRB alone (DRB), DRB plus 200 ␮g
protein/mL ox-LDL (DRB ⫹ ox-LDL), ox-LDL alone (oxLDL), DRB plus 50 ␮g/mL ox-PAPC (DRB ⫹ ox-PAPC),
or ox-PAPC alone (ox-PAPC) for various times. Total
RNA was isolated from the cells for RT-PCR analysis as
described in “Materials and methods” (A,C). The autoradiogram was analyzed by densitometry and the data are
expressed as the percentage of control after normalization to ␤-actin mRNA levels (B,D). Each symbol (f, F,
and Œ) represents 3 different experiments. The half-life (t
1/2) for each condition was calculated by linear regression analysis from semilog plots of the mRNA remaining
versus incubation times and is reported as the mean.
The 5⬘-flanking region of the TM gene contains a DR4 sequence
binding for the RARs-RXR␣ heterodimer and an Sp1 binding
sequence binding for Sp1 protein.10-12 The nuclear proteins from
untreated control HUVECs that bind to the DR4 oligonucleotide
sequence were investigated with an EMSA using rabbit antihuman
RAR␣, RAR␤, RAR␥, or RXR␣ IgG (Figure 5A). There were 2
unique bands observed in the control EMSA (lane 1). A slightly
reduced intensity of the lower band was observed by preincubation
of the nuclear proteins with anti-RAR␣ or RAR␥ IgG before assay
(lanes 2 and 4, respectively), and a marked decrease in the intensity
was observed by preincubation of the nuclear proteins with
anti-RAR␤ IgG (lane 3). This indicates that RAR␣, RAR␤, and
RAR␥ proteins were components in the lower band, and RAR␤
was a major component in the RARs. Preincubation of the nuclear
proteins with anti–human RXR␣ IgG before EMSA resulted in
disappearance of the lower band, a marked decrease in intensity of
the upper band and appearance of a new band at the gel top (lane 5),
indicating that RXR␣ was an essential component in the lower
band and a major component in the upper band.
The nuclear proteins from untreated control HUVECs that bind
to the Sp1 oligonucleotide sequence were investigated with an
EMSA using anti–human Sp1, Sp3, or Sp4 IgG (Figure 5B). There
were 3 bands observed in the control EMSA (lane 1). There was a
marked decrease in intensity of the complex 1 band and appearance
of a weak broad band just above the complex 1 band after
preincubation of the nuclear proteins with anti–Sp1 IgG (lane 3).
Disappearance of the complex 2 and 3 bands was observed after
preincubation of the proteins with anti–Sp3 IgG (lane 4). Preincubation of the nuclear proteins with both the anti–Sp1 and –Sp3
IgGs (lane 5) resulted in a marked decrease in the complex 1 band,
appearance of a broad supershift band above the complex 1 band,
and disappearance of the complex 2 and 3 bands. In contrast,
preincubation of the nuclear proteins with anti–Sp4 IgG did not
affect appearance of the original 3 bands (lane 6). Preincubation
with anti–Sp1, –Sp3, and –Sp4 IgGs (lane 7) resulted in the same
image as that observed in lane 5. These results indicated that Sp1
was a major component in the complex 1 band and Sp3 was an
essential component in the complex 2 and 3 bands. The binding
specificity of nuclear proteins to radiolabeled DR4 element or Sp1
element was confirmed by competition with excess unlabeled DR4
(Figure 5A, lane 6) or Sp1 (Figure 5B, lane 2) oligonucleotide.
LE540 is a specific antagonist of transcriptional activation of
RAR␤, but not RAR␣, RAR␥, and RXR␣.39 When HUVECs were
4770
ISHII et al
Figure 5. Supershift assay of nuclear proteins extracted from HUVECs with the
DR4 sequence and Sp1 binding sequence in the TM promoter. Nuclear proteins
were prepared from cultured control HUVECs as described in “Materials and
methods.” Nuclear proteins (5 ␮g) were incubated with or without 1 ␮g antibody
directed against human RAR␣, RAR␤, RAR␥, or RXR␣ protein (A) and Sp1, Sp3, or
Sp4 proteins (B) for 30 minutes at 25°C. After incubation, 0.5 ng 32P-labeled DR4 or
Sp1 oligonucleotide (1 ⫻ 105 cpm) was added, and the reaction was allowed to
proceed for 30 minutes at 25°C before EMSA was performed. EMSAs were
performed using 4% polyacrylamide gels, and complexes bound with the oligonucleotide were detected with a Bio-Imaging Analyzer (Fuji Film, Tokyo, Japan). Binding
specificity of nuclear proteins with the labeled oligonucleotides was confirmed by
addition of excess cold DR4 and a scrambled DR4 (A, lanes 6-7) or excess cold Sp1
oligonucleotide (B, lane 2). NS indicates nonspecific binding.
treated with 10 ␮M LE540 for 6 or 12 hours, TM mRNA levels
decreased by about 20% and 45%, respectively (Figure 6A), and
when the cells were treated with 10 ␮M LE540 for 12 or 24 hours,
TM antigen levels decreased by 30% and 55%, respectively (Figure
6B), relative to controls. These results indicated that TM mRNA
and antigen levels were controlled by transcriptional activation via
RAR␤. The effect of ox-LDL or ox-PAPC on RAR␤, RXR␣, Sp1,
and Sp3 expression levels in the nuclei of HUVECs was investigated (Figure 7). Sp3 formed 2 complexes, the complex 2 and 3,
with the Sp1 sequence as shown in Figure 5B. Therefore, it was
observed that 2 protein bands were recognized with anti-Sp3
antibody, a clear upper band (about 100 kDa) and a weak lower
band (about 60 kDa), in the Western blotting of the nuclear proteins
from untreated control HUVECs. Expression levels of RAR␤,
RXR␣, Sp1, and Sp3 in HUVECs treated with 200 ␮g protein/mL
native LDL or 50 ␮g/mL native PAPC for 9 hours did not differ
from those in the untreated control cells. However, a timedependent decrease in the expression levels of RAR␤, RXR␣, Sp1,
and Sp3 was observed in the nuclei of the cells treated with 200 ␮g
protein/mL ox-LDL (Figure 7A) or 50 ␮g/mL ox-PAPC (Figure
7B) for 9 hours.
Figure 6. The effect of LE540, a specific antagonist for RAR␤, on TM mRNA
levels in the nuclei of HUVECs. HUVECs were incubated with or without 10 ␮M
LE540 for 6, 12, or 24 hours. (A) TM mRNA levels in the cells were assayed by a
quantitative RT-PCR method described in “Materials and methods.” The normalized
signal is expressed as the percentage of control values (0 hours) after normalization
to the ␤-actin mRNA signal. (B) TM antigen levels in cell lysate were determined by
enzyme immunoassay. The data represent the mean ⫾ SD from 4 independent
experiments and are analyzed by the Student t test; *P ⬍ .05 when compared with
control without LE540 (0 hours).
BLOOD, 15 JUNE 2003 䡠 VOLUME 101, NUMBER 12
Figure 7. Western blots for RAR␤, RXR␣, Sp1, and Sp3 in nuclear proteins
extracted from HUVECs treated with native LDL, ox-LDL, native PAPC, or
ox-PAPC. HUVECs were incubated with or without 200 ␮g protein/mL native LDL
(n-LDL) or ox-LDL (A), or 50 ␮g/mL native PAPC (n-PAPC) or ox-PAPC (B) for various
times, and nuclear proteins were extracted from the cells. Nuclear proteins (5 ␮g)
were subjected to SDS-PAGE under nonreducing conditions and transferred to a
nitrocellulose membrane. The membrane was incubated with polyclonal antibody to
RAR␤, RXR␣, or Sp3, or with monoclonal antibody to Sp1. Detection of RAR␤, RXR␣
and Sp3, and Sp1 were carried out using HRP–conjugated sheep anti–rabbit IgG and
HRP-conjugated sheep anti–mouse IgG, respectively, with detection using chemiluminescence and exposure of X-ray film.
Binding activity of nuclear proteins from HUVECs treated with
oxidized phospholipids to the TM DR4 or Sp1 sequence
and deletion analysis of the DR4 and/or Sp1 sequences
within the TM promoter
Interaction of the DR4 sequence or Sp1 binding sequence with the
nuclear proteins from HUVECs treated with ox-PAPC or phospholipids from ox-LDL was investigated by the EMSA method (Figure
8). Binding activity of the nuclear proteins from HUVECs treated
with 50 ␮g/mL ox-PAPC for 9 hours to the DR4 sequence
Figure 8. EMSA of nuclear proteins from HUVECs treated with or without
oxidized lipids with the DR4 and Sp1 sequences from the TM promoter.
Ox-PAPC and ox-LDL were prepared by UV irradiation of PAPC and LDL, respectively, for 12 hours. Oxidized phospholipids (ox-PLs) and native phospholipids
(n-PLs) were separated from crude lipid extracts prepared from their respective
LDLs. Reduced forms of ox-PLs (Reduced ox-PLs) and ox-PAPC (Reduced oxPAPC) were prepared by incubation of the ox-PLs and ox-PAPC with sodium
borohydride as described in “Materials and methods.” (A-B) HUVECs were incubated
for 9 hours with or without 50 ␮g/mL ox-PAPC, reduced ox-PAPC, or n-PAPC. (C-D)
HUVECs were incubated for 9 hours with or without 200 ␮g/mL ox-PLs, reduced
ox-PLs, or n-PLs. Nuclear proteins (2 ␮g) isolated from the cells were incubated with
0.5 ng 32P-labeled oligonucleotide probes (1 ⫻ 105 cpm) for DR4 (A,C) or Sp1 (B,D)
for 30 minutes at 25°C. EMSAs were performed using 4% polyacrylamide gels and
complexes bound with oligonucleotide were detected with a Bio-Imaging Analyzer.
BLOOD, 15 JUNE 2003 䡠 VOLUME 101, NUMBER 12
Figure 9. Deletion analysis of the DR4 and/or Sp1 sequences within the TM
promoter. HUVECs were transfected with pGL3-pTM1562 plasmid or some deletion
mutants as well as pRL-TK plasmid as described in “Materials and methods.” The
culture medium was changed after 4 hours of catinoic liposome transfection, and 24
hours thereafter, the cells were treated with medium alone, or either 50 ␮g/mL of
n-PAPC or ox-PAPC for 12 hours. The activities of firefly and renilla luciferases in the
cell extracts were determined and the former activity normalized with respect to the
later activity was expressed as a percentage of that with pGL3-pTM1562 plasmid
treated without PAPC. 䡺 indicates medium alone; u, n-PAPC treated; and f,
ox-PAPC treated. Results were the average values of 3 independent experiments.
pGL3-basic indicates plasmid not containing any TM promoter region; pGL3pTM1562, pGL3-control plasmid containing ⫺1562 bp from transcription site;
pGL3-pTM(⫺DR4), pGL3-pTM1562 plasmid deleted from ⫺1524 to ⫺1521 in the
DR4 site; pGL3-pTM(⫺Sp1), pGL3-pTM1562 plasmid deleted from ⫺145 to ⫺121 in
the Sp1 site; and pGL3-pTM(⫺DR4/⫺Sp1), pGL3-pTM1562 plasmid deleted from
⫺1524 to ⫺1521 and from ⫺145 to ⫺121.
apparently decreased (Figure 8A, lane 2) compared with that of the
proteins from untreated control cells (lane 1). In contrast, no
significant changes were observed in the EMSA between the DR4
and the nuclear proteins from HUVECs exposed to the reduced
form of ox-PAPC (lane 3) or to native PAPC for 9 hours (lane 4).
Similarly, binding of the Sp1 sequence with the nuclear proteins
from HUVECs treated with ox-PAPC for 9 hours (Figure 8B, lane
2) also decreased in comparison with the proteins from untreated
control cells (lane 1). Binding of the Sp1 sequence with nuclear
proteins from the cells treated with the reduced form of ox-PAPC
(Figure 8B, lane 3) or with native PAPC (lane 4) was similar to
untreated control cells. Addition of excess unlabeled oligonucleotides abolished binding of the labeled oligonucleotide with the
nuclear proteins from the cells (Figure 8A-B, lane 5), demonstrating specificity of the binding of oligonucleotides and nuclear
proteins from the cells. Essentially identical results were obtained
using phospholipids (200 ␮g/mL) from ox-LDL or native LDL, or
the NaBH4-treated reduced form of phospholipid from ox-LDL
(Figure 8C-D). These results demonstrated that treatment of
HUVECs with oxidized phospholipids from ox-LDL or ox-PAPC
repress the binding activity of nuclear proteins to both the DR4 and
Sp1 binding elements of the 5⬘-flanking region of the TM gene.
The binding activity of nuclear proteins to both the DR4 and
Sp1 elements within the TM promoter with respect to oxidized
phospholipid-induced down-regulation of TM expression was
studied with TM promoter-luciferase fusion constructs. The wildtype construct pGL3-pTM1562, DR4 deletion mutant construct
pGL3-pTM (⫺DR4), Sp1 deletion mutant construct pGL3-pTM
(⫺Sp1), or DR4 and Sp1 deletion mutant construct pGL3-TM
(⫺DR4/⫺Sp1) was fused to a luciferase reporter plasmid and
cotransfected into HUVECs with a transfection efficiency reporter
construct. The transfected cells were treated with 50 ␮g/mL
n-PAPC or ox-PAPC for 12 hours (Figure 9). The promoter activity
was determined by firefly luciferase activity in the cell extracts and
expressed as a percentage of that with the wild-type construct not
treated with PAPC. All activities were normalized with renilla
luciferase activity. The DR4 deletion mutant (⫺DR4)–, Sp1
deletion mutant (⫺Sp1)–, and both deletion mutant (⫺DR4/⫺Sp1)–
ox-LDL DOWN-REGULATES THROMBOMODULIN
4771
transfected cells exhibited about 45%, 50%, and 75% decreases in
the promoter activity, respectively, compared with that of the cells
transfected with the wild-type construct. After treatment with
ox-PAPC, the cells transfected with the wild-type TM promoter
exhibited only 20% to 23% of the activity observed when they were
treated with medium alone or n-PAPC. The promoter activities of
⫺DR4-transfected and ⫺Sp1-transfected cells treated with oxPAPC exhibited 47% and 44%, respectively, of each control. In
contrast, the activity observed with ⫺DR4/⫺Sp1-transfected cells
treated with ox-PAPC did not decrease compared with treatment
with medium alone or n-PAPC. The results provided evidence that
the binding of nuclear proteins to both the DR4 element and Sp1
binding elements of the TM promoter is essential to promote TM
transcription in HUVECs. Furthermore, reduction of TM promoter
activity by ox-PAPC, but not n-PAPC, clearly relates to the
interaction of proteins with these binding elements.
Discussion
The present work demonstrated that endothelial cells treated with
phospholipid extracts, but not free cholesterol or triglyceride and
cholesterol ester, separated from ox-LDL significantly reduced TM
mRNA levels. Further, the degree of down-regulation was nearly
the same extent as that in cells treated with intact ox-LDL and with
extracts containing phospholipids at a concentration equivalent to
that in ox-LDL. This result suggests that the phospholipid component of ox-LDL is a major active species that contributes to
down-regulation of TM mRNA levels. This supports the previous
suggestion that down-regulation of TM antigen and its mRNA
levels in HUVECs exposed to ox-LDL could be mediated by the
lipid components in ox-LDL.25 Among the lipid components found
in ox-LDL, lysoPC (50 ␮g/mL), 7-oxocholesterol (40 ␮g/mL),
7␤-hydroxyl-cholesterol (40 ␮g/mL), and 25-hydroxyl-cholesterol
(40 ␮g/mL) failed to mimic the down-regulation of TM mRNA
levels in HUVECs treated with ox-LDL (data not shown). In
contrast, ox-PAPC produced by UV irradiation of PAPC induced a
decrease in TM mRNA levels in a dose- and time-dependent
manner. The oxidation was a requirement because treatment of the
cells with native PAPC or the NaBH4 reduced form of ox-PAPC
failed to affect TM mRNA levels. Ox-PAPC is a mixture of
products obtained by auto-oxidation of PAPC, some of which were
identified in mildly ox-LDL as 1-palmitoyl-2-oxovaleryl-sn-gylcero3-phosphocholine, 1-palmitoyl-2-glutaryl-sn-gylcero-3-phosphocholine, and 1-palmitoyl-2-epoxyisoprostane-sn-gylcero-3-phosphocholine.32,37,38,40 Treatment of ox-PAPC with sodium borohydride
reduces the hydroperoxides, epoxides, and carbonyls (aldehyde and
ketone) in ox-PAPC to hydroxides.32 Therefore, the present results
imply that a major active component of ox-LDL that influences TM
mRNA levels may be oxidized phospholipids, including ox-PAPC
with hydroperoxides, epoxides, and/or carbonyls moieties. OxPAPC is responsible for the biologic activity of mildly ox-LDL and
induces expression of adhesion proteins for monocytes in endothelial cells.32,41-43 The presence of these oxidized phospholipids in
fatty streak lesions from cholesterol-fed rabbits has been documented, suggesting that specific oxidized derivation of arachidonic
acid–containing phospholipids may be important initiators of
atherogenesis.32
When we treated HUVECs with cytochalasin B (final concentration 80 ng/mL), an endocytosis inhibitor, for 20 minutes before
exposure of the cells to ox-LDL or ox-PAPC, the ox-LDL induced
4772
ISHII et al
down-regulation of TM mRNA levels, but the effect of ox-PAPC
was abolished (data not shown). This suggests that downregulation of TM mRNA levels by ox-LDL was mediated by
endocytosis of the lipoprotein into the cells, whereas the ox-PAPC–
induced down-regulation was mediated by permeation of ox-PAPC
through the plasma membrane without endocytosis. Our previous
report suggested that down-regulation of TM expression by
ox-LDL could be mediated by degradation of the lipoprotein in
lysosomes after internalization of ox-LDL, because bafilomycin
A1, an inhibitor of lysosome proton pumps, prevented the loss of
TM antigen and mRNA by ox-LDL.25 Therefore, it is likely that
ox-LDL internalized into the cell cytosol may release oxidized
phospholipids through a lysosomal pathway, and the released
oxidized phospholipids may affect TM mRNA levels. However, the
precise mechanism for how this might occur is not known as yet.
Ox-LDL is internalized after binding several specific endothelial
receptors such as CD36,44 a lectinlike ox-LDL receptor-1 (LOX1),45 and a scavenger receptor (SREC).46 We pretreated HUVECs
with phosphatidylserine (200 ␮g/mL), a potent competitive binding inhibitor for CD36 and SREC; polyinosinic acid (200 ␮g/mL),
a potent competitive binding inhibitor for LOX-1 and SREC; or
acetyl LDL (400 ␮g protein/mL), a potent competitive binding
inhibitor for SREC, for 20 minutes before exposure to ox-LDL to
estimate the receptor involving the ox-LDL–induced downregulation of TM mRNA levels. The down-regulation of TM
mRNA levels, however, was not significantly influenced by pretreatment with these compounds (data not shown). Therefore, the extent
of participation of receptors for ox-LDL in the modulation of TM
expression is not apparent from the current study.
The present work was undertaken to evaluate the mechanism
for down-regulation of TM mRNA levels in HUVECs in response
to ox-LDL and ox-PAPC. The apparent half-life of TM mRNA in
the cells treated with ox-LDL plus DRB (3.1 hours), a transcription
inhibitor, and ox-LDL alone (3.0 hours) did not significantly differ
from that observed in the control cells treated with DRB alone (2.7
hours) as reported previously.22,25 Furthermore, the half-life of TM
mRNA in cells treated with ox-PAPC plus DRB or ox-PAPC alone
was 2.8 hours and 3.1 hours, respectively. These also did not
significantly differ from that observed in the cells treated with DRB
or ox-LDL alone. The results suggest that ox-LDL and ox-PAPC, a
representative oxidized phospholipid, could repress the transcriptional activity of the TM gene without affecting TM mRNA stability.
The present work further investigated the molecular basis for
transcriptional regulation of the TM gene in HUVECs in response
to ox-LDL or ox-PAPC. Treatment of HUVECs with retinol
(vitamin A) or its derivatives (retinoids) such as retinal and retinoic
acid increases transcriptional activity of the TM gene.13 Retinol and
retinoids exert profound effects on the transcriptional regulation of
various genes, mainly through 2 families of nuclear receptors, the
RARs and the RXRs. These receptors are ligand-dependent transcription factors that bind to cis-acting DNA sequences, RAREs
and RXR-responsive elements, located in the promoter region of
their target genes.47 RARs usually must form a heterodimer with
RXRs under physiologic conditions, and the heterodimer binds to a
RARE sequence and promotes transcription.47 The 5⬘-flanking
region of the human TM gene has the DR4 sequence at ⫺1531 to
⫺1516 bp and the 4 binding sites for the transcription factor Sp1 at
⫺276 to ⫺121 bp from the transcription site.10-12 Interactions
between RAR-RXR␣ heterodimers and the DR4 sequence and
between Sp1 and the second Sp1 binding sequence (⫺145 to ⫺121
bp) are observed with nuclear proteins from untreated HUVECs,
BLOOD, 15 JUNE 2003 䡠 VOLUME 101, NUMBER 12
and up-regulation of TM mRNA levels in cells exposed to retinoids
is accompanied by increased interaction between these nuclear
receptors and their cognate binding sites in the TM promoter.12 It is
further known that HUVECs contain predominantly RAR␤48 and
treatment of HUVECs with retinoic acid markedly increases
nuclear expression of RAR␤ and Sp1.12 The present results suggest
that the lower band observed in EMSA between the DR4 and
nuclear proteins from untreated control HUVECs was predominantly composed of RAR␤-RXR␣ heterodimer (Figure 5) and
results from the experiments with LE540, a RAR␤ selective
antagonist,39 indicate that TM mRNA levels in HUVECs are
controlled by transcriptional activation via RAR␤ (Figure 6). The
Sp family includes Sp1, Sp2, Sp3, and Sp4. In most promoters, Sp1
and Sp3 recognize the classical Sp1 consensus element with
comparable affinity and specificity.49 The present results indicated
that the complex 1 band observed in EMSA between the Sp1
binding element and nuclear proteins from untreated control
HUVECs predominantly bind with Sp1 protein and the complex 2
and 3 bands can bind with Sp3 protein. In addition, Western blot
analysis indicated that expression levels of RAR␤, RXR␣, Sp1,
and Sp3 in the nuclei were significantly decreased by treatment of
the cells with ox-LDL and ox-PAPC (Figure 7). The EMSA results
indicated that treatment of HUVECs with ox-PAPC and oxidized
phospholipid isolated from ox-LDL significantly decreased the
binding of the nuclear proteins to both the DR4 sequence and the
Sp1 binding sequence in the TM promoter (Figure 8). Furthermore,
the transfection experiments with deletion mutants of the DR4
element and/or Sp1 binding element in the TM promoter demonstrated that both elements are essential to promote TM gene
transcription and that ox-PAPC, but not n-PAPC, repress the
promoter activity (Figure 9). Overall, the data indicate that the
decreased TM transcription in HUVECs exposed to oxidized
phospholipids is mediated, at least in part, by decreased binding of
RXR␣-RAR␤ heterodimers and Sp proteins including Sp1 and Sp3
to the their DNA recognition sequences in the TM promoter
through suppression of nuclear expression levels of RAR␤, RXR␣,
Sp1, and Sp3.
Transcriptional down-regulation of the TM gene is found also in
HUVECs treated with lipopolysaccharide (LPS), TNF-␣, or IL1␤50 as well as ox-LDL through unknown mechanisms. RXR␣
expression in rodent liver is repressed during an acute-phase
response to inflammation after injection with LPS, TNF-␣, or
IL-1␤. This occurs through decreased RXR␣ transcription and
increased degradation of its mRNA, suggesting that the reduction
in RXR␣ levels alone or in association with other nuclear receptors
could be a mechanism for coordinately inhibiting the expression of
multiple genes during the acute-phase response.51 The present work
did not address whether ox-LDL–induced repression of RAR␤ and
RXR␣ levels is derived from decreased transcriptional activities
and/or increased degradation of their mRNA. The current study
observed decreased binding of Sp1 and Sp3 to the Sp1 binding
sequence and decreased nuclear contents of Sp1 and Sp3 in
HUVECS treated with ox-LDL. Pan et al52 recently showed that
transcription of the calmodulin gene in hepatoma cells is intimately
related to synthesis of Sp1, and the cells treated with insulin
showed a marked increase in Sp1 expression in the nuclei, although
the insulin-dependent up-regulation of Sp1 synthesis was antagonized by TNF-␣. The mechanism that results in a decrease in Sp1
expression by TNF-␣ as well as by ox-LDL is not known.
The present paper presents the novel observation that decreased
expression of TM mRNA levels in endothelial cells exposed to
BLOOD, 15 JUNE 2003 䡠 VOLUME 101, NUMBER 12
ox-LDL DOWN-REGULATES THROMBOMODULIN
ox-LDL is induced by transcriptional down-regulation of the TM
gene mediated by ox-LDL phospholipids. The proposed mechanism
involves a decrease in TM gene transcription mediated by reduced
binding of RXR␣-RAR heterodimers containing RXR␣-RAR␤ to
the DR4 sequence and Sp1 and Sp3 to the first and second Sp1
binding sites in promoter due to decreased content of the nuclear
transcription factors RAR␤, RXR␣, Sp1, and Sp3. The down-
4773
regulation of TM, a physiologic antithrombotic factor, on the
surface of endothelial cells exposed to ox-LDL may contribute
significantly to thrombotic properties of the cells in an atherosclerotic lesion. The present study contributes to an understanding of
the mechanisms of thrombus formation in atherosclerotic lesions
and transcriptional down-regulation of proteins containing RAREs
in the promoter site of genes.
References
1. Esmon CT. Regulation of blood coagulation. Biochim Biophys Acta. 2000;1477:349-360.
2. Dittman WA, Majerus PW. Structure and function
of thrombomodulin: a natural anticoagulant.
Blood. 1990;75:329-336.
3. Dahlbäck B, Hildebrand B. Inherited resistance to
activated protein C is corrected by anticoagulant
cofactor activity found to be a property of factor V.
Proc Natl Acad Sci U S A. 1994;91:1396-1400.
4. Bertina R, Koeleman BPC, Koster T, et al. Maturation in blood coagulation factor V associated
with resistance to activated protein C. Nature.
1994;369:64-67.
5. Griffin JH, Evatt B, Zimmerman TS, Kleiss AJ,
Wideman C. Deficiency of protein C in congenital
thrombotic disease. J Clin Invest. 1981;68:13701373.
6. Horie S, Ishii H, Hara H, Kazama M. Enhancement of thrombin-thrombomodulin-catalyzed protein C activation by phosphatidylethanolamine
containing unsaturated fatty acids: possible physiological significance of phosphatidylethanolamine in anticoagulant activity of thrombomodulin. Biochem J. 1994;301:683-691.
7. Galvin JB, Kurosawa S, Moore K, Esmon CT, Esmon NL. Reconstitution of rat TM into phospholipid vesicles. J Biol Chem. 1987;262:2199-2205.
8. Suzuki K, Kusumoto H, Deyashiki Y, et al. Structure and expression of human thrombomodulin, a
thrombin receptor on endothelium acting as a cofactor for protein C activation. EMBO J. 1987;6:
1891-1897.
9. Wen D, Dittman WA, Ye RD, Deaven LL, Majerus
PW, Sadler JE. Human thrombomodulin: complete cDNA sequence and chromosome localization of the gene. Biochemistry. 1987;26:43504357.
Bang NU. Endotoxin enhances tissue factor and
suppresses thrombomodulin expression in human vascular endothelium in vitro. J Clin Invest.
1987;79:124-130.
17. Nowroth PP, Handley DA, Esmon CT, Stern DM.
Interleukin 1 induces endothelial cell procoagulant while suppressing cell-surface anticoagulant
activity. Proc Natl Acad Sci U S A. 1986;83:34603464.
18. Scarpati EM, Sadler JE. Regulation of endothelial
cell coagulant properties: modulation of tissue
factor, plasminogen activator inhibitor, and thrombomodulin by phorbol 12-myristate 13-acetate
and tumor necrosis factor. J Biol Chem. 1989;
264:20705-20713.
19. Archipoff G, Beretz A, Freyssinet JM, Klein-Soyer
C, Brisson C, Cazenave JP. Heterogeneous regulation of constitutive thrombomodulin or inducible
tissue factor activities on the surface of human
saphenous-vein endothelial cells in culture following stimulation by interleukin-1, tumor necrosis
factor, thrombin or phorbol ester. Biochem J.
1991;273:679-684.
20. Ishii H, Horie S, Kizaki K, Kazama M. Retinoic
acid counteracts both the down-regulation of
thrombomodulin and the induction of tissue factor
in cultured human endothelial cells exposed to
tumor necrosis factor. Blood. 1992;80:2556-2562.
21. Moore KL, Esmon CT, Esmon NL. Tumor necrosis factor leads to the internalization and degradation of thrombomodulin from the surface of bovine aortic endothelial cells in culture. Blood.
1989;73:159-165.
22. Lentz SR, Tsiang M, Sadler JE. Regulation of
thrombomodulin by tumor necrosis factor-alpha:
comparison of transcriptional and posttranscriptional mechanisms. Blood. 1991;77:542-550.
10. Shirai T, Shiojiri S, Ito H, et al. Gene structure of
human thrombomodulin, a cofactor for thrombincatalyzed activation of protein C. J Biochem.
1988;103:281-285.
23. Conway EM, Rosenberg RD. Tumor necrosis factor suppresses transcription of the thrombomodulin gene in endothelial cells. Mol Cell Biol. 1988;8:
5588-5592.
11. Dittman WA, Nelson SC, Greer PK, Horton ET,
Palomba ML, McCachren SS. Characterization of
thrombomodulin expression in response to retinoic acid and identification of a retinoic acid response element in human thrombomodulin gene.
J Biol Chem. 1994;269:16925-16932.
24. Chait A, Heinecke JW. Lipoprotein modification:
cellular mechanisms. Curr Opin Lipidol. 1994;5:
365-370.
12. Horie S, Ishii H, Matsumoto F, et al. Acceleration
of thrombomodulin gene transcription by retinoic
acid: retinoic acid receptors and Sp1 regulate the
promoter activity through interactions with two
different sequences in the 5⬘-flanking region of
human gene. J Biol Chem. 2001;276:2440-2450.
13. Horie S, Ishii H, Kizaki K, Kazama M. Retinoic
acid stimulates expression of thrombomodulin, a
cell surface anticoagulant glycoprotein, on human
endothelial cells. Biochem J. 1992;281:149-154.
14. Hirokawa K, Aoki N. Up-regulation of thrombomodulin by activation of histamine H1-receptors
in human umbilical-vein endothelial cells in vitro.
Biochem J. 1991;276:739-743.
15. Ishii H, Kizaki K, Uchiyama H, Horie S, Kazama
M. Cyclic AMP increases thrombomodulin expression on membrane surface of cultured human umbilical vein endothelial cells. Thromb Res.
1990;59:841-850.
16. Moore KL, Andreoli SP, Esmon NL, Esmon CT,
25. Ishii H, Kizaki K, Horie S, Kazama M. Oxidized
low density lipoprotein reduces thrombomodulin
transcription in cultured human endothelial cells
through degradation of the lipoprotein in lysosomes. J Biol Chem. 1996;271:8458-8465.
26. Laszuj ZG, Zhou XJ, Ferrell GL, Silva FG, Esmon
CT. Down-regulation of endothelial expression of
endothelial cell protein C receptor and thrombomodulin in coronary atherosclerosis. Am J Pathol.
2001;159:797-802.
27. Ishii H, Takada K, Higuchi T, Sugiyama J. Verotoxin-1 induces tissue factor expression in human
umbilical vein endothelial cells through activation
of NF-␬B/Rel and AP-1. Thromb Haemost. 2000;
84:712-721.
28. Buege JA, Aust SD. Microsomal lipid peroxidation. Methods Enzymol. 1978;52:302-310.
extraction and purification. Can J Biochem
Physiol. 1959;37:911-917.
31. Hamilton JG, Comai K. Rapid separation of neutral lipids, free fatty acids and polar lipids using
prepacked silica Sep-Pak column. Lipids. 1988;
23:1146-1149.
32. Watson AD, Leitinger N, Navab M, et al. Structural identification by mass spectrometry of oxidized phospholipids in minimally oxidized low
density lipoprotein that induce monocyte/endothelial interactions and evidence for their presence in vivo. J Biol Chem. 1997;272:1359713607.
33. Ishii H, Nakano M, Tsubouchi J, et al. Establishment of enzyme immunoassay of human thrombomodulin in plasma and urine using monoclonal
antibodies. Thromb Haemost. 1990;63:157-162.
34. Ikeda T, Ishii H, Higuchi T, et al. Localization of
thrombomodulin in the anterior segment of the
human eye. Invest Ophthamol Vis Sci. 2000;41:
3383-3390.
35. Xie JF, Strouma J, Graves DT. IL-1 downregulates platelet-derived growth factor-alpha
receptor gene expression at the transcriptional
level in human osteoblastic cells. J Immunol.
1994;153:378-383.
36. Hamada H, Ishii H, Sakyo K, Horie S, Nishiki K,
Kazama M. The epidermal growth factor-like domain of recombinant human thrombomodulin exhibits mitogenic activity for Swiss 3T3 cells.
Blood. 1996;86:225-233.
37. Watson AD, Navab M, Hama SY, et al. Effect of
platelet activating factor-acetylhydrolase on the
formation and action of minimally oxidized low
density lipoprotein. J Clin Invest. 1995;95:774782.
38. Watson AD, Berliner JA, Hama SY, et al. Protective effect of high density lipoprotein associated
paraoxonase: inhibition of the biological activity of
minimally oxidized low density lipoprotein. J Clin
Invest. 1995;96:2882-2891.
39. Li Y, Hashimoto Y, Agadir A, Kagechika H, Zhang
X. Identification of a novel class of retinoic acid
receptor b-selective retinoid antagonists and their
inhibitory effects on AP-1 activity and retinoic
acid-induced apoptosis in human breast cancer
cells. J Biol Chem. 1999;274:15360-15366.
40. Watson AD, Subbanagounder G, Welsbie DS, et
al. Structural identification of a novel pro-inflammatory epoxyisoprostane phospholipid in mildly
oxidized low density lipoprotein. J Biol Chem.
1999;274:24787-24798.
41. Huber J, Vales A, Mitulovic G, et al. Oxidized
membrane vesicles and blebs from apoptotic
cells contain biologically active oxidized phospholipids that induce monocyte-endothelial interactions. Arterioscler Thromb Vasc Biol. 2002;22:
101-107.
29. Bradford MM. A rapid and sensitive method for
the quantitation of microgram quantities of protein
utilizing the principle of protein-dye binding. Anal
Biochem. 1976;72:248-254.
42. Subbanagounder G, Wong JW, Lee H, et al. Epoxyisoprostane and epoxycyclopentenone phospholipids regulate monocyte chemotactic protein-1 and interleukin-8 synthesis: formation of
these oxidized phospholipids in response to interleukin-1beta. J Biol Chem. 2002;277:7271-7281.
30. Bligh EG, Dyer WJ. A rapid method of total lipid
43. Leitinger N, Tyner TR, Oslund L, et al. Structurally
4774
BLOOD, 15 JUNE 2003 䡠 VOLUME 101, NUMBER 12
ISHII et al
similar oxidized phospholipids differentially regulate endothelial binding of monocytes and neutrophils. Proc Natl Acad Sci U S A. 1999;96:1201012015.
44. Uittenbogaard A, Shaul PW, Yuhanna IS, Blair A,
Smart EJ. High density lipoprotein prevents oxidized low density lipoprotein-induced inhibition of
endothelial nitric-oxide synthase localization and
activation in caveolae. J Biol Chem. 2000;275:
11278-11283.
45. Sawamura T, Kume N, Aoyama T, et al. An endothelial receptor for oxidized low-density lipoprotein. Nature. 1997;386:73-77.
46. Adachi H, Tsujimoto M, Arai H, Inoue K. Expres-
sion cloning of a novel scavenger from human
endothelial cells. J Biol Chem. 1997;272:3121731220.
47. Chambon P. A decade of molecular biology of
retinoic acid receptors. FASEB J. 1996;10:940954.
48. Fesus L, Nagy L, Basilion JP, Davies PJ. Retinoic
acid receptor transcripts in human umbilical vein
endothelial cells. Biochem Biophys Res Commun. 1991;179:32-38.
49. Suske G. The Sp-family of transcription factors.
Gene. 1999;238:291-300.
50. Kapiotis S, Besemer J, Bevec D, et al. Interleu-
kin-4 counteracts pyrogen-induced down regulation of thrombomodulin in cultured human vascular endothelial cells. Blood. 1991;78:410-415.
51. Beigneux AP, Moser AH, Shigenaga JK, Grunfeld
C, Feingold KR. The acute phase response is
associated with retinoid X receptor repression in
rodent liver. J Biol Chem. 2000;275:1639016399.
52. Pan X, Solomon SS, Borromeo DM, MartinezHernandez A, Raghow R. Insulin deprivation
leads to deficiency of Sp1 transcription factor in
H-411E hepatoma cells and in streptozotocininduced diabetic ketoacidosis in the rat. Endocrinology. 2001;142:1635-1642.