Atherosclerosis 186 (2006) 448–457

Modulation of angiogenic processes in cultured endothelialcells by low density lipoproteins subfractions from patients with

familial hypercholesterolemia

Ming-Hong Tai a,b, Shiao-Mei Kuo b, Hui-Ting Liang a, Kuan-Rau Chiou c,Hing-Chung Lam a, Ching-Mei Hsu b, Henry J. Pownall d, Hsin-Hung Chen d,

Max T. Huang d, Chao-Yuh Yang d,e,∗a Department of Medical Education and Research, Kaohsiung Veterans General Hospital, Kaohsiung 813, Taiwan

b Department of Biological Sciences, National Sun Yat-Sen University, Kaohsiung 804, Taiwanc Department of Vascular Medicine, Kaohsiung Veterans General Hospital, Kaohsiung 813, Taiwan

d Department of Medicine, Section of Atherosclerosis and Lipoprotein Research, Baylor College of Medicine,6565 Fannin St., Houston, TX 77030, USA

e Department of Biochemistry, Baylor College of Medicine, 6565 Fannin St., Houston, TX 77030, USA

Received 19 April 2005; received in revised form 10 August 2005; accepted 12 August 2005Available online 26 September 2005

A

OfcMavFimBCv©

K

1

tt

0d

bstract

bjective: Electronegative low density lipoprotein (LDL) subfractions are cytotoxic to endothelial cells. To continue our study of homozygoticamilial hypercholesterolemic (FH)-LDL, we report the effects of FH-LDL subfractions (FH-L1 to FH-L5) on the angiogenic processes inultured endothelial cells.ethods and results: Subconfluent bovine aortic endothelial cells (BAEC) were treated with LDL subfractions (20 �g/ml), and the effects on

ngiogenic functions, including cell proliferation, migration, apoptosis, tube formation, secretion of matrix metalloproteinases (MMPs), andascular endothelial growth factor (VEGF) were determined. The electronegative FH-L4 and FH-L5 inhibited cell proliferation while the otherH-LDL subfractions and LDL from normocholesterolemic subjects (N-LDL) had negligible effects. Like Cu2+ ox-LDL, FH-L5 strongly

nhibited endothelial cell viability and FH-L4 had a milder effects. Similarly, FH-L4 and FH-L5 but not the other subfractions retarded celligration, induced cell apoptosis, and perturbed tube formation by BAEC in matrigel. FH-L5 inhibited secretion of MMP-2 and MMP-9 byAEC without affecting their endogenous levels. In contrast, FH-L5 increased the VEGF expression in endothelial cells.onclusions: Our results show for the first time that FH-L5, a circulating LDL subfraction from hypercholesterolemic patients, modulatesarious angiogenic processes, thereby dysregulating endothelial function in a way that may be atherogenic.

2005 Elsevier Ireland Ltd. All rights reserved.

eywords: Electronegative LDL; Endothelial cell; Angiogenesis; Matrix metalloproteinase; Apoptosis

. Introduction

Oxidatively modified human plasma low density lipopro-eins (ox-LDL) induce endothelial cell apoptosis, an impor-ant early step in atherogenesis [1–3]. In vitro, electronegative

∗ Corresponding author. Tel.: +1 713 798 4210; fax: +1 713 798 4121.E-mail address: cyang@bcm.tmc.edu (C.-Y. Yang).

human LDL, LDL (−), is proinflammatory and cytotoxictoward endothelial cells [3]. Using fast protein liquid chro-matography (FPLC), we separated plasma LDL from hyperc-holesterolemic patients into five subfractions, L1–L5, one ofwhich, L5 was highly electronegative and suppressed DNAsynthesis in cultured bovine aortic endothelial cells (BAEC)and stimulated mononuclear cell adhesion to cultured ECsunder flow conditions in vitro [4]. Marked apoptosis was

021-9150/$ – see front matter © 2005 Elsevier Ireland Ltd. All rights reserved.oi:10.1016/j.atherosclerosis.2005.08.022

M.-H. Tai et al. / Atherosclerosis 186 (2006) 448–457 449

induced in cultured BAECs by L5, the most electronegativeLDL subfraction isolated from hypercholesterolemic plasma[5].

Dysregulation of endothelial cells, a key cell type inangiogenesis, is an important initiating factor in atheroge-nesis [6]. Angiogenesis comprises several distinct steps inendothelial biology that include secretion of matrix met-alloproteinases (MMPs), migration, proliferation, and tubeformation (interaction with extracellular matrix). ox-LDLcould impair each of these processes, thereby dysregulatingendothelial function. The mechanism by which LDL fromhypercholesterolemic patients induces vascular EC apopto-sis is not known. L5, by far the most proapoptotic subfraction,was also the most highly oxidized subfraction. This observa-tion supports the hypothesis that in vivo oxidation of LDL ispathogenic and is mechanistically linked to atherogenesis. Inthe present study, the modulation of various angiogenic pro-cesses of FH-LDL subfractions were investigated in endothe-lial cells. Our results indicated for the first time that FH-L5perturbed angiogenesis in vitro and thereby induced endothe-lial dysfunction.

2. Materials and methods

2.1. Reagents

pw

2

apoaLpgab

eUTw(Aoftf3t

with nitrogen gas and stored at 4 ◦C, prior to further analy-ses, which were conducted within 2 week.

2.3. Cell culture

BAEC (6–10 passages) were cultured with Dulbecco’sModified Eagle Medium (DMEM; Gibco BRL, Rockville,MD) containing 10% fetal calf serum (Gibco BRL, Rockville,MD), 2 mM glutamine, 100 U/ml penicillin, and 100 �g/mlstreptomycin (Gibco BRL, Rockville, MD) in 5% CO2 at37 ◦C.

2.4. Cell proliferation assay

The effect of LDL subfractions on the viability of variouscells was determined using crystal violet stain assay as pre-viously described [5]. Briefly, cells were cultured in 96-wellplate ((2–4) × 103 cells/well) and treated with DMEM mediacontaining various doses of FH subfractions. After 24 h, cellswere fixed with 2.5% glutaraldehyde at room temperaturefor 15 min, stained with 0.1% crystal violet solution (in 20%methanol; 20 �l/well) for 20 min, washed with distilled waterfor three times, solublized with solution containing 50%ethanol and 0.1% acetic acid. The dye in viable cells was mea-sured by reading the optical density at 590 nm using a scan-ning multi-well spectrophotometer (ELISA reader; DynatechL

2

[mespsP(sI

2

[ftmebfibwct

Recombinant basic fibroblast growth factor (bFGF) wasurchased from R&D System (Minneapolis, MN). Matrigelas from BD PharMingen (La Jolla, CA).

.2. LDL preparations and LDL subfractions

Plasma was obtained from three FH subjects (two femalesnd one male), between 6 and 42 years of age, who had totallasma cholesterol levels >500 mg/dL, as described previ-usly [5]. FH patients had been treated with lipid-loweringgents, such as Lovastatin or Atorvastatin, in addition toDL-apheresis. The treatments were not discontinued whenlasma samples were obtained. The diagnoses of homozy-ous FH in the HC subjects were made on the bases of DNAnalyses of leukocytes, using polymerase chain reaction-ased techniques.

LDL preparations were chromatographed with an ionxchange FPLC system (Pharmacia Biotech Co.) through anoQ12 column (BioRad) equilibrated with buffer A (20 mMris–HCl, pH 8.0, containing 0.5 mM EDTA). Subfractionsere eluted by use of a multi step gradient of buffer B

1 M NaCl in buffer A). Samples equilibrated with bufferwere eluted with a linear gradient program at a flow rate

f 2 ml/min. Effluent was monitored at 280 nm and protectedrom ex vivo oxidation with 2 mM EDTA. Protein concentra-ions were determined by the Lowry method. The respectiveractions were concentrated with Centriprep filters (YM-0, Millipore Co., Bedford, MA) and sterilized by passinghrough 0.20 �m filters. The isolated fractions were filled

aboratories, Chantilly, VA).

.5. Apoptosis assays

Apoptosis assay was performed as previously described5]. Briefly, subconfluent endothelial cultures were washed,aintained in DMEM containing 10% serum, and then

xposed to PBS or graded (1, 5, and 20 �g/ml) FH-LDLubfractions, N-LDL for 24 h. Cu2+ ox-LDL was used as aositive control for apoptosis induction. Treated cells weretained for 10 min with 1 �M Hoechst 33342 (Molecularrobes) to assess nuclear morphology. Fluorescence imaging500 cells/well) was performed using a Zeiss inverted micro-cope (Axiovert; ×400) with MetaView software (Universalmaging Corp.) in triplicate.

.6. Cell migration assay

Migration assay was performed as previously described7]. Briefly, endothelial cells were treated with FH-LDL sub-ractions of indicated doses for 6 h, harvested by trypsiniza-ion, collected by centrifugation, resuspended in DMEM

edia containing 0.1% BSA, and seeded in triplicate forach dose and controls in the upper compartment of the cham-er (1.2 × 105 cells in 400 �l). The lower compartments werelled with 200 �l of the DMEM media containing 100 ng/mlFGF (R&D, Minneapolis, MN) as the chemo attractant, orith DMEM media containing 0.1% BSA as the negative

ontrol (to evaluate random migration). A polycarbonate fil-er (8-�m pore size; Nucleopore, Costar, Chambridge, MA)

450 M.-H. Tai et al. / Atherosclerosis 186 (2006) 448–457

coated with 0.005% gelatin to allow cell adhesion separatedthe compartments. After incubation for 2–4 h in a humidified5% CO2 atmosphere at 37 ◦C, cells on the upper side of thefilter were removed, and those that had migrated to the lowerside were fixed in absolute ethanol, stained with 10% Giemsasolution (Merck, Germany), and counted as a mean ± S.E.M.per filter under five different high power fields.

2.7. Tube formation assay

Tube formation assay was performed as previouslydescribed [8]. Briefly, Matrigel (Becton Dickinson, Bedford,MA) was diluted with cold serum-free medium to 10 mg/ml.Two hundred microliters of the diluted solution were addedto each well in 24-well plate and allowed to form a gel at37 ◦C for 30 min. BAEC (1.5 × 105 cells/ml) were initiallyincubated for 15 min with various doses of FH subfractionsin DMEM medium. Two hundred microliters of the cell sus-pension (3 × 104 cells) were then subsequently added to eachwell and incubated for 6–8 h at 37 ◦C in 5% CO2. Under theseconditions, endothelial cells form delicate networks of tubesthat are detectable within 2–3 h and are fully developed after8–12 h. After incubation, the endothelial tubes were fixedwith 3% paraformaldehyde and counted three to four differ-ent high power fields.

2

bdewmcBjtgb0Camid

2

u41TBj

polyvinylidene difluoride membranes (Immobilon-P, Milli-pore) for Western detection. The filter was blocked with 5%non-fat dry milk in TBST buffer containing 10 mM Tris, pH7.5, 100 mM NaCl, and 0.1% Tween-20. Afterward, filterwas incubated with primary antibodies such as MMP-2 anti-bodies (1:500; Oncogene), MMP-9 antibodies (1:100; Onco-gene), or actin antibodies (1:1000; Oncogene) for 1 h, andthen incubated with goat-anti-rabbit antibodies conjugatedwith horseradish peroxidase (1:2500 dilution) for 30 minand detected with chemiluminescence ECL kit (AmershamPharmacia Biotech, UK). The intensity of protein band wasquantified by densitometer.

2.10. Quantitative reverse transcription-polymerasechain reaction (qRT-PCR)

RNA was isolated from endothelial cells using RNAzol(TEL-TEST, Inc., Friendswoods, TX). For reverse transcrip-tion, 5 �g of total RNA was used for reverse transcriptionwith Superscriptase II (Life Technologies, Rockville, MD)using olio-dT and random primers. One-twentieth of reverse-transcription products were used as template for real-timePCR in Lightcycler (Roche) using a SYBR green I assay. PCRreaction was performed in 20 �l SYBR Green PCR MasterMix (Roche) containing 10 �M forward primers and reverseprimers, and approximately 30 ng cDNA. Amplification andd415G1fGGmwaT

2

tamVtc

2

SeP

.8. Matrix metalloproteinases (MMPs) zymography

Secretion of MMPs by endothelial cells was assessedy 0.1% gelatin–SDS-PAGE zymography as previouslyescribed [9]. Briefly, endothelial cells near 80% conflu-nce were washed twice with serum-free media and treatedith various doses of LDL subfractions for 24 h. Conditionededia were collected and normalized with number of viable

ells using crystal violet assay or protein concentration byradford assay. Aliquots of conditioned media were sub-

ected to separation with 10% SDS-PAGE containing 0.1%ype A gelatin (Sigma, St. Louis, MO). After electrophoresis,el was washed twice with 2.5% Triton X-100, incubated inuffer containing 40 mM Tris–HCl, pH 8.0, 10 mM CaCl2,.01% sodium azide at 37 ◦C for 12–24 h, stained with 0.25%oomassie Blue R-250 in 50% methanol and 10% aceticcid for 1 h, and destained with 10% acetic acid, and 20%ethanol. The gelatinolytic regions by MMPs were visual-

zed as white bands in blue background and quantified byensitometer.

.9. Western blot analysis

Protein extracts were prepared from endothelial cellssing buffer containing 50 mM Tris–HCl, pH 7.4, 1% NP-0, 0.25% sodium deoxycholate, 1 mM EDTA, 1 mM PMSF,�g/ml aprotinin, 1 �g/ml leupeptin, and 1 �g/ml pepstatin.he concentration of protein extracts was determined byradford assay. Twenty micrograms of protein were sub-

ected to separation by 10% SDS-PAGE and transferred onto

etection were performed by: 1 cycle of 95 ◦C for 10 min,0 cycles of 95 ◦C for 5 s, and 62 ◦C for 5 s, and 72 ◦C for0 s. The primer sequences for MMP-2 were: forward primer′-CAACTACGATGATGACCGCAA-3′, reverse primer 5′-TGTAAATGGGTGCCATCAGG-3′, which amplified a40-bp MMP-2 cDNA fragment. The primer sequencesor VEGF were: forward primer 5′-GCCCACTGAGGA-T CCAACA-3′, reverse primer 5′-CTCTCCTATGTGCTG-CCTTG-3′, which amplified a 98-bp VEGF cDNA frag-ent. The �-actin mRNA level was determined using: for-ard primer 5′-TCACCCACA CTGTGCCTATCTACGA-3′

nd reverse primer 5′-CAGCGGAACCGCTCATTGCCAA-GG-3′, which amplified a 295-bp �-actin cDNA fragment.

.11. ELISA

After incubation of HUVEC under the indicated condi-ions, the conditioned medium was collected and centrifugedt 100 × g for 5 min, and the supernatant was used for theeasurement of VEGF protein with a Quantikine HumanEGF ELISA kit (R&D Systems, Inc.). The sensitivity of

he assay was 1 pg/ml. All samples were measured in tripli-ate.

.12. Statistical analysis

The significance of differences was assessed by a pairedtudent’s t-test with Bonferroni correction. Results arexpressed as mean ± S.E.M. values. Probability values of> 0.05 were considered significant.

M.-H. Tai et al. / Atherosclerosis 186 (2006) 448–457 451

3. Results

3.1. Electronegative FH-LDL subfractions inhibitedproliferation and induced apoptosis in endothelial cells

We investigated the effects of FH-LDL subfractions on theproliferation and apoptosis of BAEC. Subconfluent BAECcultures were exposed to PBS or graded concentrations (1, 5,and 20 �g/ml) of FH-LDL subfractions, Cu2+ ox-LDL, andN-LDL in DMEM containing 10% serum for 24 h. Effectsof FH-LDL subfractions on cell morphology and on theendothelial nuclei as visualized by Hoechst dye were shown(Fig. 1A). Like Cu2+ ox-LDL, FH-L5 inhibited endothelialcells viability and induced chromosome condensation or frag-mentation; an effect that was much lower with L4 while othersubfractions or N-LDL had no effect. Using the crystal vio-let assay, cell proliferation in response to the graded dosesof FH-LDL subfractions were quantified (Fig. 1B). Aftertreatment for 24 h, FH-L5 caused a dose-dependent inhibi-tion of endothelial proliferation (80.9 ± 2.4 and 58.5 ± 4.3%of control at 5 and 20 �g/ml; P < 0.05 and <0.001, respec-tively), which was similar to that of Cu2+ ox-LDL. Aninhibitory effect that was less profound with FH-L4 and wasnot observed with FH-L1-3 or N-LDL. As with proliferation,FH-L5 was the most potent subfraction to induce apoptosisin 19.2 ± 2.3% of BAEC (Fig. 1C), whereas FH-L4 was onlymn

3e

tttoLFtCrcFtcl

3f

tifv

FH-L1 to FH-L3 had no such effect (Fig. 3A). After fixationand measurement under light microscopy, quantification ofthe tubes formed in Matrigel revealed that like Cu2+ ox-LDL,FH-L4 and FH-L5 reduced BAEC tube formation (5.8 ± 0.2and 3.4 ± 0.4, respectively, versus 8.5 ± 1.5 tubes for control;Fig. 3B). In contrast, FH-L1 to FH-L3 subfractions or N-LDLdid not reveal notable affect. These results indicated that FH-L5 perturbed the interaction between endothelial cells andextracellular matrix.

3.4. Electronegative FH-LDL subfractions inhibitedMMP-2 and MMP-9 secretion by endothelial cells

MMPs, a family of zinc-containing endopeptidases, medi-ate selective degradation of extracellular matrix that isrequired for migration and invasion of endothelial cells at theonset of angiogenesis. To determine the effect of FH-LDLsubfractions on MMPs secretion, conditioned media fromendothelial cells treated with LDL subfractions were normal-ized with cell number then subjected to gelatin-zymographyanalysis. Like Cu2+ ox-LDL, FH-L5 treatment decreasedthe MMP-2 and MMP-9 activities in cultured media ofendothelial cells by approximately 40% of control (Fig. 4A),whereas N-LDL or FH-L1 had no such effect. This observa-tion was further supported by Western blot analysis (Fig. 4B),wdBLsaidcm

3V

hiweuLttBaeIVtV

ildly apoptotic (6.5 ± 1.1%), and the other subfractions didot reveal significant apoptotic effect.

.2. Electronegative FH-LDL subfractions reducedndothelial migration

The effects of various LDL subfractions on EC migra-ion were determined in vitro in a Boyden chamber in whichhe number of migrating cells in response to the chemoat-ractant bFGF was determined. The representative profilesf endothelial migration after treatment with various FH-DL subfractions were shown (Fig. 2A). Unlike FH-L1–L3,H-L4 and FH-L5 substantially reduced the migration of EC

oward bFGF. Additional experiments demonstrated that likeu2+ ox-LDL, FH-L4 or FH-L5 induced a dose-dependent

eduction in endothelial migration (Fig. 2B). Relative to aontrol value of 402 ± 34 migrating cells, in the presence ofH-L4 and FH-L5, respectively, cell migration was reduced

o 326.9 ± 19.4 and 215 ± 16 cells (P < 0.05 and <0.001). Inontrast, N-LDL and FH-L1 to FH-L3 did not affect endothe-ial migration.

.3. Electronegative FH-LDL subfractions reduced tubeormation by endothelial cells

The effects of various FH-LDL subfractions on the forma-ion of tube-like structure in extracellular matrix were alsonvestigated (Fig. 3A). Both Cu2+ ox-LDL and FH-L5 sub-ractions (20 �g/ml) profoundly perturbed the formation ofessel-like structures of BAEC in matrigel while N-LDL and

hich revealed that MMP-2 and MMP-9 protein levels wereecreased in cultured media from ox-LDL- or FH-L5-treatedAEC by about 60% of control, but not from N-LDL- or FH-1-treated ones. However, treatment with ox-LDL or FH-L5ubfractions did not influence the protein levels of MMP-2nd MMP-9 in BAEC (Fig. 4C). Similarly, qRT-PCR analysisndicated that treatment with ox-LDL or FH-L5 subfractionsid not influence the mRNA level of MMP-2 in endothelialells (Fig. 4D). These data suggested that FH-L5 subfractionsight perturb the release of MMPs in endothelial cells.

.5. Electronegative FH-LDL subfractions increased theEGF expression in endothelial cells

VEGF is a potent mitogen for endothelial cells. Besides, itas been shown that ox-LDL enhanced the VEGF expressionn endothelial cells and macrophages [10,11]. To investigatehether circulating LDL from FH patients altered VEGF

xpression, the VEGF mRNA level and secretion were eval-ated in cultured endothelial cells after treatment with FH-DL subfractions. By qRT-PCR analysis, it was found that

reatment with FH-L5, but not FH-L1, significantly elevatedhe VEGF mRNA level by nearly 2-fold of control (Fig. 5A).esides, similar to that in ox-LDL-treated cells, there wassignificantly higher VEGF content in cultured media from

ndothelial cells treated with FH-L5 subfraction (Fig. 5B).n contrast, N-LDL and FH-L1 to FH-L4 did not affect theEGF secretion. Together, these results indicate that elec-

ronegative LDL subfraction from FH patients increases theEGF expression at transcriptional level in endothelial cells.

452 M.-H. Tai et al. / Atherosclerosis 186 (2006) 448–457

Fig. 1. The effect of LDL subfractions from FH patients on the proliferation of endothelial cells. (A) The morphologies of endothelial cells (upper lanes) andthe morphologies of nucleus endothelial cells (lower lanes) after treatment with LDL subfractions from FH patients. After treatment of LDL subfractions fromFH patients (at 20 �g/ml) in DMEM medium for 24 h, the morphologies of BAEC cells were monitored under phase contrast microscopy and the nucleusof BAEC cells were visualized by Hoechst 33258 and monitored under fluorescence microscopy. The pictures were taken at 200× magnification. (B) Thedose-dependent effect of FH-LDL subfractions on the proliferation of endothelial cells. After treatment with graded FH-LDL subfractions (1–20 �g/ml) or PBSfor 24 h, the viability of various cells was determined by crystal violet method in an ELISA reader. Each data is mean ± S.E.M. of quadruplicate experiments.Asterisks indicate statistic significance vs. normolipidemic LDL (N-LDL) (*P < 0.05). (C) The apoptotic effect of LDL subfractions in endothelial cells. Aftertreatment with FH-LDL subfractions (all at 20 �g/ml) or PBS for 24 h, apoptotic cells were quantified by employing the characteristic that apoptotic nucleihave higher integrated pixel intensity due to chromatin condensation. Counting of nuclei (500 cells/well) was performed by MetaMorph (Universal ImagingCorp.) in triplicates.

M.-H. Tai et al. / Atherosclerosis 186 (2006) 448–457 453

Fig. 2. The effect of FH-LDL subfractions on the migration of endothelial cells. (A) The cell image of migrated endothelial cells treated with FH-LDLsubfractions. BAEC treated with FH-LDL subfractions (20 �g/ml) in DMEM medium for 6 h were trypsinized and applied to top wells in Boyden chamber toinitiate migration toward chemoattractant bFGF (100 ng/ml) in the bottom wells for 6 h. The pictures were taken at 200× magnification. (B) The dose-dependenteffect of FH-LDL subfractions (1–20 �g/ml) on the migration of BAEC. After incubation in Boyden chamber for 6 h, the migrated cells on filter were stainedand counted. Each point represents mean ± S.E.M. of triplicate experiments. Asterisks indicated statistic significance vs. control (*P < 0.05). L1–L5 representedFH-L1 to FH-L5, respectively.

4. Discussion

Our studies show for the first time that electroneg-ative LDL fractions from familial hypercholesterolemicpatients, FH-L5, elicit global, and likely pathological effectwithin the context of endothelial cell biology. FH-L5 isanti-proliferative, pro-apoptotic, anti-chemotactic, and anti-angiogenic in endothelial cells. In addition, FH-L5 alters therelease of MMP-2, which is essential to the migration andinvasion of endothelial cells at the onset of angiogenesis. Inall our assays, the effects of FH-L5, a natural componentof plasma from familial hypercholesterolemic patients, werecomparable to those of Cu2+ ox-LDL, a chemically modifiedLDL frequently used in vitro model.

The pathophysiological effects of ox-LDL observed invitro have been implicated in the mechanisms of initiationand progression of atherosclerosis in vivo. Cu2+ ox-LDL arecytotoxic to cultured endothelial cells, inhibit endothelial cellproliferation and migration [12], and induce expression ofcellular adhesion molecules [2]. Cu2+ ox-LDL also impairsendothelium-dependent relaxation and organized (capillary-like) endothelial cells growth by down regulating endothelialnitric oxide synthase (eNOS) [12,13] and fibroblast growthfactor-2 (FGF-2) [14,15].

Endothelial cell migration is essential for healing of arte-rial injury and angioplasty sites. Iron or Cu2+ ox-LDL inhibitsendothelial migration in vitro and this inhibition can be atten-uated by antioxidants [16]. Migration of endothelial cells is

454 M.-H. Tai et al. / Atherosclerosis 186 (2006) 448–457

Fig. 3. The effect of FH-LDL subfractions on the tube formation of endothelial cells. (A) The profile of tube formation of endothelial cells treated with FH-LDLsubfractions in matrigel. Endothelial cells were applied to Matrigel-coated plate in the presence of FH-LDL subfractions (20 �g/ml) in DMEM medium for 8 h.The tubular structures of BAEC cells were monitored and recorded under light microscopy. The pictures were taken at 400× magnification. (B) Quantificationof tube formation by endothelial cells after treatment with various FH-LDL subfractions (20 �g/ml) for 8 h. Each data represent mean ± S.E.M. of tube numberfrom quadruplicate experiments. Asterisks indicated statistic significance vs. control (*P < 0.05).

a critical and initiating event in the formation of new bloodvessels and in the repair of injured vessels. Despite of con-vincing evidence that ox-LDL are formed in atheroscleroticlesions, the role of ox-LDL in lesion formation is not ascompelling. Our studies revealed that cell migration towardchemoattractant bFGF in FH-L5-treated endothelial cells wasinhibited in a dose-dependent manner by ox-LDL (Fig. 2).These results suggest that ox-LDL could impair wound heal-ing in response to endothelial injury. The mechanism bywhich ox-LDL attenuates cell motility could be attributedto the properties of ox-LDL in inhibiting cell proliferation orthe secretion of proteases that degrade extracellular matrixmolecules, which are essential for cell movement.

The Matrigel, which has been used to assay tubuleformation for many years, is a mixture of basementmembrane components, including laminin, collagen typeIV, entactin/nidogen, and protoheparan sulfate [17], pre-pared from the murine Engelbrecht–Holm–Swarm tumorof C57BL/6 mice. Matrigel is often used in three differ-ent tests related to angiogenesis: the chemoinvasion assay,the morphology assay, and the in vivo sponge model. Whenseeded onto Matrigel, endothelial cells attach and differen-tiate as evidenced by both the morphological changes andreduced proliferation, and ultimately tubule formation. While

tubule formation in vitro does not simulate all aspects ofangiogenesis, it is a useful in vitro assay of at least twokey steps—migration and differentiation of endothelial cells.Like Cu2+ ox-LDL, FH-L5 effectively abolished the vessel-like structure of BAEC cells in Matrigel (Fig. 3), suggestingthat FH-L5 may be anti-angiogenic in vivo.

MMPs represent a family of endopeptidases named fortheir ability to degrade extracellular matrix components(ECM). These proteinases participate in tissue remodelingassociated with both development and disease [18,19]. Local-ized ECM breakdown plays a major role in the pathogenesisof atherosclerosis including the early migration of mono-cytes into the arterial wall and the mechanical strength ofthe plaque cap. MMPs expression is correlated with clin-ical manifestations of unstable angina and plaque rupture[20,21]. All vascular cells including endothelial cells andmacrophages secrete MMPs. MMP-1 (a collagenase), MMP-2 (a gelatinase), and MMP-3 (a stromelysin) are predom-inantly secreted by endothelial cells [22] while the majorMMP secreted from macrophages is the gelatinase, MMP-9.MMP activity is tightly regulated at several levels that requireactivation and inhibition by tissue inhibitors of metallopro-teinases (TIMPs), chiefly TIMP-1. TIMP-1 is co-expressedwith MMPs in atherosclerotic lesions. It has been reported

M.-H. Tai et al. / Atherosclerosis 186 (2006) 448–457 455

Fig. 4. The effect of FH-L5 on MMPs secretion and expression in endothelial cells. (A) The effect of FH-L5 on MMP-2 and MMP-9 gelatinase activities incultured media of endothelial cells. After treatment with N-LDL, Cu2+ ox-LDL, FH-L1, and FH-L5 (20 �g/ml) for 24 h, the cultured media of BAEC werenormalized by cell number for analysis of MMP-2 and MMP-9 activities by gelatin–SDS-PAGE zymography. (B) The effect of FH-L5 on MMP-2 and MMP-9protein level in cultured media of endothelial cells. After treatment with N-LDL, Cu2+ ox-LDL, FH-L1, and FH-L5 (20 �g/ml) for 24 h, the cultured media ofBAEC were normalized by cell number and analyzed for MMP-2 and MMP-9 protein levels by Western blot analysis. (C) The effect of FH-L5 on endogenousMMP-2 and MMP-9 protein level in endothelial cells. After treatment with N-LDL, Cu2+ ox-LDL, FH-L1, and FH-L5 (20 �g/ml) for 24 h, the MMP-2 andMMP-9 protein levels in cell extract of BAEC were determined by Western blot analysis. The actin level was analyzed as protein loading control. The numberbelow each lane represents the band intensity as mean ± S.E.M. fold of control from triplicate experiments. Asterisks indicated statistic significance vs. control(*P < 0.05; **P < 0.001). (D) The effect of FH-L5 on MMP-2 mRNA level in endothelial cells. After treatment with N-LDL, ox-LDL, FH-L1, and FH-L5(20 �g/ml) for 24 h, MMP-2 mRNA levels in endothelial cells were determined by qRT-PCR analysis. The MMP-2 mRNA level was expressed as ratio ofmelting temperature for MMP-2 vs. that of actin and each data represented mean ± S.D. of triplicate experiments.

that ox-LDL modulates the expression of MMPs and theirtissue inhibitors in different types of cells. ox-LDL regulatesMMP-9 and its tissue inhibitor in human monocyte-derivedmacrophages [23,24]. ox-LDL differentially regulates MMP-1 and TIMP-1 expression in vascular endothelial cells [25].Besides, ox-LDL regulates the production of MMP-1 andMMP-9 in activated monocytes [26]. Recent data further indi-cate that ox-LDL enhances the expression of MMP-1 andMMP-3 in carotid artery endothelial cells via activation of itsown receptor LOX-1 [27]. In arterial vasculature of diabeticpatients, the synthesis and activities of a matrix metallo-proteinase induction/activation system are decreased [28].Recently, Ague et al. reported that MMP-2 plays a pivotalrole in ox-LDL-induced activation of sphingomyelin/ceramidsignaling pathway and subsequent smooth muscle cell prolif-eration [29]. In the present study, we provide the first evidencethat MMP-2 and MMP-9 releases in endothelial cells weredisrupted by FH-L5 by unknown pathway (Fig. 4). It seemsplausible that, like Cu2+ ox-LDL, FH-L5 alters the secretory

activities of endothelium and causes it to become dysfunc-tional [30]. These data suggested that FH-L5 might affect therelease of MMPs in endothelial cells, but not the synthesisof MMPs in endothelial cells. Decreased MMP activity maycontribute to increase the collagen deposition and patholog-ical remodeling in associated with disease development.

VEGF has been recognized as an angiogenic factor thatinduces endothelial proliferation and vascular permeability.Besides, VEGF can promote macrophage migration, whichis critical for atherosclerosis. VEGF is remarkably expressedin activated macrophages, endothelial cells, and smooth mus-cle cells within human coronary atherosclerotic lesions [10].Recent evidence indicates that ox-LDL upregulated VEGFsecretion and VEGF mRNA expression in human coro-nary artery endothelial cells [10,11], which is consistentwith our findings (Fig. 5). Since ox-LDL upregulates VEGFexpression in macrophages and endothelial cells through theperoxisome proliferator-activated receptor-gamma (PPAR-�) pathway [11], it remains to be elucidated whether FH-L5

456 M.-H. Tai et al. / Atherosclerosis 186 (2006) 448–457

Fig. 5. Effect of FH-LDL subfractions on VEGF expression in endothelialcells. (A) The effect of FH-L5 on VEGF mRNA level in endothelial cells.After treatment with N-LDL, ox-LDL, and FH-LDL subfractions (20 �g/ml)for 24 h, VEGF transcript levels in HUVEC were determined by qRT-PCRanalysis. The VEGF mRNA level was expressed as ratio of melting temper-ature for VEGF vs. that of actin and each data represented mean ± S.D. oftriplicate experiments. (B) Effect of FH-L5 on VEGF secretion in endothe-lial cells. After treatment with N-LDL, ox-LDL, and FH-LDL subfractions(20 �g/ml) for 24 h, conditioned media of HUVEC were collected and nor-malized with cell number for analysis of VEGF content by ELISA; *P < 0.05vs. control.

also activated PPAR-� signaling in endothelial cells. In thepresent study, we further demonstrate that, like ox-LDL, theelectronegative LDL subfraction from FH patients increasedVEGF expression in endothelial cells at transcriptional level.Together with our previous findings on FGF-2 downregu-lation by FH-L5 [5,14], we present evidence that FH-L5differentially regulates the expression of angiogenic factorsin endothelial cells.

The present study demonstrates for the first time thatL5 subfraction from FH patients is a potent inhibitor ofseveral angiogenic processes—migration, tube formation,MMP secretion and VEGF expression. Thus, the differ-ence in molecular composition between FH and normal mayoffer clues to their differential anti-angiogenic properties.Although FH-L5 occurs at a relatively low concentration,it could be physiologically important. Its effects on a smallregion of the endothelium that may be made more vul-nerable by local vascular anatomy, hypertension, flow and

hypercholesterolemia, all of which could conspire to pro-duce endothelial damage at a rate that exceeds the ratesof surveillance and repair that are associated with woundhealing.

Acknowledgments

This work was supported by in part by grantsfrom National Science Council, Taiwan (NSC-90-2320-B-075B-005), Kaohsiung Veterans General Hospital, Taiwan(VGHKS91-18 to M.H. Tai, VGHKS93-47 to K.R. Chiou,and VGHKS93-35 to H.C. Lam), American Diabetes Asso-ciation (ADA 7-03-RA-108 to C.Y. Yang), and NationalInstitute of Health (HL-63364 to C.Y. Yang).

References

[1] Choy JC, Granville DJ, Hunt DW, McManus BM. Endothelial cellapoptosis: biochemical characteristics and potential implications foratherosclerosis. J Mol Cell Cardiol 2001;33:1673–90.

[2] Takei A, Huang Y, Lopes-Virella MF. Expression of adhesionmolecules by human endothelial cells exposed to oxidized low den-sity lipoprotein. Influences of degree of oxidation and location ofoxidized LDL. Atherosclerosis 2001;154:79–86.

[3] Steinberg D, Witztum JL. Is the oxidative modification hypothesisrelevant to human atherosclerosis? Do the antioxidant trials con-

[

[

[

ducted to date refute the hypothesis? Circulation 2002;105:2107–11.

[4] Yang CY, Raya JL, Chen HH, et al. Isolation, characterization, andfunctional assessment of oxidatively modified subfractions of cir-culating low-density lipoproteins. Arterioscler Thromb Vasc Biol2003;23:1083–90.

[5] Chen CH, Jiang T, Yang JH, et al. Low-density lipoprotein inhypercholesterolemic human plasma induces vascular endothelial cellapoptosis by inhibiting fibroblast growth factor 2 transcription. Cir-culation 2003;107:2102–8.

[6] Moulton KS. Plaque angiogenesis: its functions and regulation. ColdSpring Harb Symp Quant Biol 2002;67:471–82.

[7] Tang MJ, Worley D, Sanicola M, Dressler GR. The RET-glial cell-derived neurotrophic factor (GDNF) pathway stimulatesmigration and chemoattraction of epithelial cells. J Cell Biol1998;142:1337–45.

[8] Kawasaki J, Hirano K, Hirano M, et al. Dissociation between theCa(2+) signal and tube formation induced by vascular endothelialgrowth factor in bovine aortic endothelial cells. Eur J Pharmacol2000;398:19–29.

[9] Kim YM, Jang JW, Lee OH, et al. Endostatin inhibits endothelialand tumor cellular invasion by blocking the activation and catalyticactivity of matrix metalloproteinase. Cancer Res 2000;60:5410–3.

10] Ramos MA, Kuzuya M, Esaki T, et al. Induction of macrophageVEGF in response to oxidized LDL and VEGF accumulationin human atherosclerotic lesions. Arterioscler Thromb Vasc Biol1998;18:1188–96.

11] Inoue M, Itoh H, Tanaka T, et al. Oxidized LDL regu-lates vascular endothelial growth factor expression in humanmacrophages and endothelial cells through activation of perox-isome proliferator-activated receptor-gamma. Arterioscler ThrombVasc Biol 2001;21:560–6.

12] Chavakis E, Dernbach E, Hermann C, et al. Oxidized LDL inhibitsvascular endothelial growth factor-induced endothelial cell migrationby an inhibitory effect on the Akt/endothelial nitric oxide synthasepathway. Circulation 2001;103:2102–7.

M.-H. Tai et al. / Atherosclerosis 186 (2006) 448–457 457

[13] Blair A, Shaul PW, Yuhanna IS, Conrad PA, Smart EJ. Oxidizedlow density lipoprotein displaces endothelial nitric-oxide synthase(eNOS) from plasmalemmal caveolae and impairs eNOS activation.J Biol Chem 1999;274:32512–9.

[14] Chen CH, Jiang W, Via DP, et al. Oxidized low-density lipoproteinsinhibit endothelial cell proliferation by suppressing basic fibroblastgrowth factor expression. Circulation 2000;101:171–7.

[15] Chang PY, Luo S, Jiang T, et al. Oxidized low-density lipoproteindownregulates endothelial basic fibroblast growth factor through apertussis toxin-sensitive G-protein pathway: mediator role of platelet-activating factor-like phospholipids. Circulation 2001;104:588–93.

[16] van Aalst JA, Burmeister W, Fox PL, Graham LM. Alpha-tocopherolpreserves endothelial cell migration in the presence of cell-oxidizedlow-density lipoprotein by inhibiting changes in cell membrane flu-idity. J Vasc Surg 2004;39:229–37.

[17] Baatout S. Endothelial differentiation using Matrigel (review). Anti-cancer Res 1997;17:451–5.

[18] McFarlane S. Metalloproteases: carving out a role in axon guidance.Neuron 2003;37:559–62.

[19] Visse R, Nagase H. Matrix metalloproteinases and tissue inhibitorsof metalloproteinases: structure, function, and biochemistry. Circ Res2003;92:827–39.

[20] Galis ZS, Muszynski M, Sukhova GK, Simon-Morrissey E, Libby P.Enhanced expression of vascular matrix metalloproteinases inducedin vitro by cytokines and in regions of human atherosclerotic lesions.Ann N Y Acad Sci 1995;748:501–7.

[21] Shah PK, Falk E, Badimon JJ, et al. Human monocyte-derivedmacrophages induce collagen breakdown in fibrous caps ofatherosclerotic plaques. Potential role of matrix-degrading met-alloproteinases and implications for plaque rupture. Circulation1995;92:1565–9.

[22] Nguyen M, Arkell J, Jackson CJ. Activated protein Cdirectly activates human endothelial gelatinase A. J Biol Chem2000;275:9095–8.

[23] Uzui H, Harpf A, Liu M, et al. Increased expression of membranetype 3-matrix metalloproteinase in human atherosclerotic plaque: roleof activated macrophages and inflammatory cytokines. Circulation2002;106:3024–30.

[24] Kalela A, Koivu TA, Hoyhtya M, et al. Association of serumMMP-9 with autoantibodies against oxidized LDL. Atherosclerosis2002;160:161–5.

[25] Huang Y, Song L, Wu S, Fan F, Lopes-Virella MF. Oxidized LDLdifferentially regulates MMP-1 and TIMP-1 expression in vascularendothelial cells. Atherosclerosis 2001;156:119–25.

[26] Ardans JA, Economou AP, Martinson Jr JM, Zhou M, Wahl LM.Oxidized low-density and high-density lipoproteins regulate the pro-duction of matrix metalloproteinase-1 and -9 by activated monocytes.J Leukoc Biol 2002;71:1012–8.

[27] Li D, Liu L, Chen H, et al. LOX-1 mediates oxidized low-densitylipoprotein-induced expression of matrix metalloproteinases inhuman coronary artery endothelial cells. Circulation 2003;107:612–7.

[28] Portik-Dobos V, Anstadt MP, Hutchinson J, Bannan M, Ergul A. Evi-dence for a matrix metalloproteinase induction/activation system inarterial vasculature and decreased synthesis and activity in diabetes.Diabetes 2002;51:3063–8.

[29] Auge N, Maupas-Schwalm F, Elbaz M, et al. Role for matrixmetalloproteinase-2 in oxidized low-density lipoprotein-induced acti-vation of the sphingomyelin/ceramide pathway and smooth musclecell proliferation. Circulation 2004;110:571–8.

[30] Erl W, Weber PC, Weber C. Monocytic cell adhesion to endothelialcells stimulated by oxidized low density lipoprotein is mediated bydistinct endothelial ligands. Atherosclerosis 1998;136:297–303.