Mildly Oxidized LDL Induces Activation of Platelet-Derived Growth Factor β-Receptor Pathway
Background Mildly oxidized LDL (moxLDL) is thought to play a role in atherogenesis. MoxLDL induces derivatization of cell proteins and triggers a variety of intracellular signaling. We aimed to investigate whether moxLDL-induced protein derivatization may influence the activity of platelet-derived growth factor receptor β (PDGFRβ), a tyrosine kinase receptor of major importance in vascular biology and atherogenesis.
Methods and Results In cultured rabbit arterial smooth muscle cells, moxLDL induces activation of the PDGFRβ signaling pathway, as shown by PDGFRβ tyrosine phosphorylation on Western blot and coimmunoprecipitation of SH2-containing proteins. The cellular events involved in the moxLDL-induced PDGFRβ activation can be summarized as follows. Oxidized lipids from moxLDL trigger two phases of PDGFRβ activation involving two separate mechanisms, as shown by experiments on cultured cells (in situ) and on immunopurified PDGFRβ (in vitro): (1) the first phase may be mediated by 4-hydroxynonenal, which induces PDGFRβ adduct formation and subsequent PDGFRβ activation (antioxidant-insensitive step); (2) the second phase involves ceramide-mediated generation of H2O2 (these steps being inhibited by tosylphenylalanylchloromethylketone, an inhibitor of ceramide formation, and by antioxidant BHT, exogenous catalase, or overexpressed human catalase). Because 4-hydroxynonenal–PDGFRβ adducts are also detected in atherosclerotic aortas, it is suggested that this novel mechanism of moxLDL-induced PDGFRβ activation may occur during atherogenesis.
Conclusions MoxLDL acts as a local autoparacrine mediator in the vascular wall, and PDGFRβ acts as a sensor for both oxidized lipids and oxidative stress. This constitutes a novel mechanism of PDGFRβ activation in atherosclerotic areas.
Received March 5, 2001; revision received June 22, 2001; accepted July 5, 2001.
Atherogenesis involves a complex sequence of events associating endothelial activation, transendothelial migration of mononuclear cells, lipid accumulation, and smooth muscle cell proliferation, leading finally to fibroatheroma plaque formation.1 LDL is thought to become atherogenic after undergoing oxidative modifications,2,3 characterized by oxidized lipid formation and structural alterations of apoB.3,4
Oxidized LDL (oxLDL) is present in atherosclerotic areas3 and elicits changes in (1) lipoprotein metabolism (foam cells), (2) gene expression (adhesion molecules, cytokines, growth factors, coagulation proteins), (3) cell migration, motility, and contractility, (4) cell proliferation; (5) cell viability/death, and (6) local immune response.3
OxLDL triggers or alters various signaling pathways,5 namely calcium,6 trimeric G proteins and cAMP,7 phospholipase D,8 ceramide,9 PPARγ,10 and epidermal growth factor receptor (EGFR activation being associated with 4-hydroxynonenal [4-HNE] derivatization).11 We have examined whether this concept may be extended to platelet-derived growth factor receptor β (PDGFRβ), a tyrosine kinase receptor of importance in atherogenesis.12
PDGFRα and PDGFRβ belong to the tyrosine kinase receptor family. PDGFRβ binds only PDGF B-chain, whereas PDGFRα binds A and B PDGF isoforms.13–15 PDGFRα is involved in smooth muscle cell (SMC) hypertrophy, and PDGFRβ is involved in the migration and proliferation of SMC,13–15 a major event in atherosclerotic plaque formation.12 PDGFRβ activation triggers tyrosine phosphorylation, which creates binding sites for SH2-containing proteins, including Src, phospholipase C-γ1, phosphatidylinositol-3′-kinase, SHP2 protein tyrosine phosphatase, GTPase-activating protein of p21ras (rasGAP), Nck, Grb2, and Shc.13–15
We found that oxidized lipids of mildly oxidized LDL (moxLDL) induce derivatization of PDGFRβ (among other cell proteins16) and trigger activation of the PDGFRβ signaling pathway. Two separate mechanisms are involved: a first antioxidant-independent early phase resulting possibly from PDGFRβ derivatization by 4-HNE and a second antioxidant-sensitive phase mediated (at least in part) by ceramide and reactive oxygen species (ROS).
Rabbit arterial SMC and SMCcat+ (transduced with catalase cDNA and overexpressing catalase by 350%)17 and human ECV-304 EC (from ATCC) and LDL receptor–negative fibroblasts (GM-486A from NIGMD) were grown in RPMI-1640 containing 10% FCS. Subconfluent cells were starved for 24 hours before experiments in 0.5% FCS.11
New Zealand White male rabbits (2.5 kg) were randomly assigned to regular (n=3) and hypercholesterolemic (1% cholesterol for 8 weeks) (n=8) diets. After euthanasia (pentobarbital 50 mg/kg), thoracic aortas were removed and fatty streaks were stained with Sudan IV. Experiments were performed according to French Accreditation of Laboratory Animal Care.
LDL and Lipids
Human LDL was oxidized by UV-C irradiation (moxLDL)18 or by overnight incubation with ECV-304 EC (cell-moxLDL).9 MoxLDL contained 3.2 to 4.9 nmol TBARS and 6.5 to 8.7 nmol 4-HNE/mg apoB) and minor apoB alterations (TNBS-reactive amino groups were 89% to 95% of nonoxidized LDL; relative electrophoretic mobility ranged between 1.1 and 1.3).
[3H]4-HNE was prepared from [3H]4-HNE diethylacetal (a generous gift of J.P. Cravedi, Toulouse)19 and allowed to react with moxLDL for 1 hour. Cells (4 culture flasks of 80 cm2, containing 1.5 mg cell protein) were preincubated with inhibitors, incubated with [3H]4-HNE-moxLDL (250 000 dpm/100 μg apoB/mL for 1 hour), washed, harvested, and used for PDGFRβ immunoprecipitation and counting the PDGFRβ-associated radioactivity.
Sphingomyelinase (Smase) treatment of LDL was performed with B cereus Smase (Sigma) (40 mU/mL, 1 hour, 37°C in 5 mmol/L MgCl2, 0.1 mol/L tris/HCl, pH 7.5); Smase-treated LDL was separated from sphingomyelinase on a Sephadex G75 column (Pharmacia). Lipids were extracted in chloroform/methanol by the Folch procedure, as previously used.11
Western Blot Analysis
Cells or tissue samples were lysed, PDGFRβ was immunoprecipitated, and Western blots were performed as reported.11
Determination of PDGFRβ-Free Amino Groups
The free amino group content of PDGFRβ was evaluated after immunoprecipitation by [3H]NSP ([3H]N-succinimidyl propionate (Amersham) labeling.11
Intracellular ROS Determination
Intracellular ROS were determined fluorometrically with the use of dichlorodihydrofluorescein diacetate (H2DCFDA) or dihydrorhodamine (DHR) from Molecular Probes, incubated with cells 30 minutes before determination.20
Determination of Enzyme Activities
Data are given as mean±SEM. Estimates of statistical significance were performed by ANOVA (Student-Newman-Keuls multiple comparison test).
Chemicals and Biological Reagents
Antiphosphotyrosine (4G10) and anti–PDGF-AB antibodies were from UBI, monoclonal anti-PDGFRβ from Transduction Laboratories, human PDGF-BB, anti-SHP2, anti-Grb2, anti-ERK1/ERK2, and polyclonal anti-PDGFRβ from Santa Cruz, anti-Src, anti–P-serine from Calbiochem, anti–P-threonine from Chemicon, antiactivated MAPK from Promega, anti–4-HNE-protein polyclonal (K5 to 4412) from G. Jürgens Laboratory,22 4-HNE from Tebu-Biomol, H2DCFDA and DHR from Molecular Probes, cell culture reagents from Gibco, and other chemicals from Sigma.
MoxLDL Induces PDGFRβ Tyrosine Phosphorylation in Cultured SMC Independent of Any Autocrine Effect
MoxLDL (at mitogenic concentration) elicited tyrosine phosphorylation of PDGFRβ, identified by immunoprecipitation and Western blot (Figure 1). MoxLDL-induced PDGFRβ tyrosine phosphorylation is sustained for 5 hours (Figure 1A), is dose-dependent (maximal PDGFRβ tyrosine phosphorylation at 1 hour and 5 hours being obtained with 100 to 150 and 50 to 75 μg apoB/mL, respectively) (Figure 1B), and is associated with the lipid fraction (Figure 1C).
Coimmunoprecipitation experiments showed that SH2-containing relay proteins, such as Grb-2 (24 kDa), Src (60 kDa), and SHP2 (69 kDa), associated to PDGFRβ phosphotyrosines (Figure 1D), thus demonstrating that UV-moxLDL or cell-moxLDL triggers effective activation of the PDGFRβ signaling pathway. We prefered to use UV-moxLDL because cell-moxLDL preparations may contain potentially interfering bioactive compounds (released from ECV-304 cells).
A role for autocrine PDGF secretion was excluded because neutralizing anti-PDGF antibody (inhibiting PDGF-induced PDGFRβ autophosphorylation) did not block the moxLDL-induced PDGFRβ tyrosine phosphorylation (Figure 1E). Other autocrine mediators were also excluded by transfer of preconditioned medium on reporter cells (Figure 1F).
Mechanisms Implicated in moxLDL-Induced PDGFRβ Activation
Antioxidants Inhibit the Late Phase of moxLDL-Induced PDGFRβ Activation
The antioxidants BHT (100 μmol/L) or catalase (1500 U/mL, permitting effective loading of SMC with catalase23) were inactive in the first phase (1 hour) but strongly inhibited the second sustained phase (5 hours) of moxLDL-induced PDGFRβ activation (Figure 2, A and B). In SMCcat+ overexpressing catalase (by 350%), this second phase was abolished, in contrast to parental SMC (Figure 2, C and D). Altogether, the two phases are apparently mediated by two separate mechanisms, the first one involving an antioxidant-insensitive mechanism and the second one involving H2O2 (catalase sensitive).
Oxidized Lipids and 4-HNE Induce Derivatization and Activation of PDGFRβ
MoxLDL (but not natLDL or PDGF) induces 4-HNE derivatization of PDGFRβ, as shown by evaluating [3H]NSP-reactive free amino groups of immunoprecipitated PDGFRβ and 4-HNE–PDGFRβ adducts (Figure 3, A and B) and by transfer of [3H]4-HNE to PDGFRβ (and other cell proteins) in SMC incubated for 1 hour with [3H]4-HNE–labeled moxLDL. This transfer was reduced by chloroquine and concanamycin A (inhibitors of endolysosomal acidification and lipoprotein receptor recycling) and by cold (4°C) (Figure 3C). Chloroquine also inhibited moxLDL-induced PDGFRβ activation (Figure 3D): In these experiments, cold and concanamycin A cannot be used because they induced per se PDGFRβ autophosphorylation. Moreover, it may be noted that moxLDL elicited PDGFRβ activation and MAPK activation in normal fibroblasts but no PDGFRβ nor MAPK activation in LDL receptor–negative fibroblasts (data not shown). Altogether, these data suggest that a small part of moxLDL-associated 4-HNE derivatize PDGFRβ through a chloroquine-sensitive process.
Consistently, incubation of SMC with 4-HNE induced 4-HNE–PDGFRβ adduct formation, PDGFRβ tyrosine phosphorylation, and phosphotyrosine-binding SH2-protein recruitment (Figure 3, A, B, and E). 4-HNE–induced PDGFRβ activation was not associated with ROS generation and was antioxidant-insensitive (Figure 3, F and G), like the first phase of moxLDL-induced PDGFRβ activation.
To investigate whether 4-HNE acts directly on PDGFRβ, we designed in vitro experiments with the use of immunopurified PDGFRβ. 4-HNE or moxLDL lipid extracts induced PDGFRβ derivatization (decreased [3H]NSP-reactive amino group content) and PDGFRβ tyrosine phosphorylation, whereas lipid extracts of native LDL were inactive (Figure 3, H and I). This suggests that 4-HNE and oxidized lipids mimic and may participate in the first (antioxidant-insensitive) phase of moxLDL-induced PDGFRβ activation.
Second Antioxidant-Sensitive Phase of moxLDL-Induced PDGFRβ Activation Is Mediated by Ceramide and H2O2
The second phase (3 to 5 hours) of moxLDL-induced PDGFRβ activation is catalase-sensitive, thus possibly mediated by H2O2. This hypothesis was compatible with moxLDL-induced ROS generation (fluorescence of oxidized DHR) (Figure 4A) and the inhibitory effect of BHT and catalase (under loading conditions23). H2O2 also triggered a strong PDGFRβ autophosphorylation that was inhibited by BHT and catalase (Figure 4B).
Because moxLDL activates the sphingomyelin/ceramide pathway (ceramide peaking at 1 to 2 hours) in SMC9 and ceramide induces H2O2 generation,24 we investigated the possible role of ceramide in moxLDL-induced H2O2 generation and PDGFRβ activation. Theoretically, ceramide may be generated through sphingomyelin hydrolysis either in LDL (by secretory sphingomyelinase)25 or in cell membranes.9 Smase-treated LDL did not trigger PDGFRβ autophosphorylation (Figure 4C). In contrast, C2-ceramide and bacterial sphingomyelinase (generating ceramide at the plasma membrane9) trigger ROS generation (consistent with data from Reference 24)24 and PDGFRβ autophosphorylation (Figure 4D). Interestingly, TPCK (tosylphenylalanylchloromethylketone), which blocks moxLDL-induced sphingomyelin/ceramide pathway activation,9 also inhibited moxLDL-induced ROS generation and PDGFRβ autophosphorylation (Figure 4, A and E). Altogether, these data suggest that the second antioxidant-sensitive phase of moxLDL-induced PDGFRβ activation may be mediated (at least in part) by cellular ceramide and subsequent H2O2 generation. A sustained PDGFRβ tyrosine phosphorylation also may result from inhibition of regulatory mechanisms terminating PDGFR activation (namely PTPases and serine/threonine-phosphorylation of PDGFR).26,27
In moxLDL-treated SMC, a transient activation of PTPases was followed by inhibition (during the second phase of moxLDL-induced PDGFR activation), which was reversed by BHT. PDGF elicited activation of PTPases, which was inhibited by orthovanadate (a PTPase inhibitor) and H2O2, in agreement with data from Reference 26 (Figure 5, A and B).
Whereas PDGF triggered PDGFRβ serine and threonine phosphorylation (serine phosphorylation being involved in terminating PDGFRβ activation27), incubation of SMC with moxLDL for 5 hours induced no PDGFRβ serine phosphorylation (Figure 5C). Conversely, moxLDL incubated with cells for 1 hour and 5 hours did not downregulate PDGFRβ kinase, in contrast to PDGF (Figure 5D). All these data suggest that the regulatory mechanisms terminating PDGFRβ activation are inactive during the second phase of oxLDL-induced PDGFRβ activation.
4-HNE Derivativatization Occurs in Rabbit Aortic Atherosclerosis
To check the relevance of the data reported above to in vivo atherosclerosis, 4-HNE–PDGFRβ adduct formation was investigated in atherosclerotic rabbits. Cholesterol-fed rabbits exhibit hypercholesterolemia (25.10±3.46 g/L versus 1.04±0.10 g/L in normals) and large fatty streaks in thoracic aorta (Figure 6A). 4-HNE–PDGFRβ adducts were clearly detected in atherosclerotic aorta (exhibiting extensive fatty streaks) but not in normal aorta (Figure 6B).
The data suggest that (1) PDGFRβ is a target of moxLDL and is activated through two novel mechanisms involving oxidized lipids and ceramide/H2O2 (Figure 7); and (2) moxLDL may be considered as a local autoparacrine mediator (generated in vascular wall and acting on neighboring cells).
4-HNE may constitute a first mediator of moxLDL-induced PDGFRβ activation because 4-HNE evokes in vitro derivatization and tyrosine phosphorylation of immunopurified PDGFRβ (the fine molecular mechanism, eg, conformational changes, linking both events remains to be elucidated). In intact living cells, 4-HNE–induced PDGFRβ derivatization and activation is antioxidant-insensitive and independent of any ROS production, like the first phase of moxLDL-induced PDGFRβ activation. HNE exists in two forms in oxLDL: (1) the major part is 4-HNE covalently bound to amino groups of proteins or phospholipids; (2) a minor part of 4-HNE remains free, is poorly diffusible toward the water environment, and is probably dissolved in the phospholipid phase of moxLDL because of its hydrophobicity.4,28 Experiments with [3H]4-HNE–labeled moxLDL show that a small part of moxLDL-associated [3H]4-HNE is transferred to PDGFRβ (and other cell proteins). The mechanism of transfer of 4-HNE from moxLDL to PDGFRβ is yet unknown. Diffusion of free 4-HNE from moxLDL toward PDGFRβ through the culture medium is not probable because 4-HNE is hydrophobic (poorly diffusible)28 and because [3H]4-HNE trace amounts are rapidly scavenged by thiols (glutathione, cysteine) of culture medium (data not shown). Therefore, in moxLDL and during its transfer to PDGFRβ, free 4-HNE is probably prevented to react with thiols, hypothetically because of direct contact of moxLDL with PDGFRβ or diffusion through a hydrophobic environment, such as membrane lipids. Finally, our data suggest that 4-HNE (and possibly other oxidized lipids) may play a role in the first antioxidant-insensitive phase of the moxLDL-induced PDGFRβ activation.
The second antioxidant-sensitive phase (3 to 5 hours) of moxLDL-induced PDGFRβ activation is mediated by ceramide and ROS. Cellular moxLDL-induced ceramide does not originate in LDL-associated ceramide but is generated through hydrolysis of cellular sphingomyelin (step blocked by TPCK).9 The reported data suggest that moxLDL-induced ceramide9 triggers ROS generation, which evokes PDGFRβ activation: (1) TPCK, a nonantioxidant inhibitor of moxLDL-induced ceramide generation,9 blocks ROS formation and PDGFRβ activation; (2) C2-ceramide and ceramide generated by bacterial sphingomyelinase treatment triggers ROS generation, consistent with data from Reference 24, and PDGFRβ activation; (3) BHT and catalase block both moxLDL-induced ROS formation and PDGFRβ activation but not ceramide formation (data not shown); and (4) H2O2 triggers PDGFRβ activation (which is blocked by catalase and BHT).
It may be noted that although the fluorogenic oxidant–sensitive probes detect various ROS,20 H2O2 is very probably the prominent ROS because increased catalase level (induced by cell loading23 or overexpression) inhibited both moxLDL-induced ROS generation and the second phase of PDGFRβ activation. This is consistent with the requirement of H2O2 in PDGF signal transduction,23 ROS-induced PDGFRβ tyrosine-phosphorylation,29 and ROS-induced SMC proliferation.30
Moreover, moxLDL-induced inhibition of PTPases and serine phosphorylation of PDGFRβ (two regulatory mechanisms involved in terminating PDGFRβ activation) may be implicated in the second phase of moxLDL-induced PDGFRβ activation. In agreement with data from Reference 26, this PTPase inhibition may be mediated by H2O2 generated during the second phase of moxLDL-induced PDGFRβ activation. BHT (which blocks H2O2 generation) also prevented PTPase inhibition and the subsequent second phase of PDGFRβ activation.
The reported data suggest that in atherosclerotic areas, PDGFRβ can be activated by moxLDL, in addition to PDGF. However, in contrast to growth factors, moxLDL is a nonspecific mediator that triggers a broad spectrum of cellular signals.5–11 For instance, moxLDL-induced PDGFRβ activation is similar to that of EGFR,12 thus extending the concept that moxLDL activates various tyrosine kinase receptors and suggesting that besides their classic role of transducers of growth factors signaling, PDGFRβ and EGFR may act as cellular sensors for oxidative stress and oxidized lipids.
From a pathophysiological point of view, the presence of 4-HNE–PDGFRβ adducts in atherosclerotic areas is direct evidence that oxidized lipids (originating from moxLDL or generated inside the cell during oxidative stress) may react in vivo with PDGFRβ, as in vitro and in cultured cells. Although tyrosine phosphorylation cannot be detected in aorta extracts (probably because of a rapid postmortem loss of phosphotyrosine in tissues), it may be speculated that oxidized lipids may interact with membrane tyrosine kinase receptors of vascular cells, thereby disturbing cellular responses such as migration, proliferation, and gene expression13–15 and playing a role in intimal changes occurring during atherogenesis.
This work was supported by grants from INSERM, Université Paul Sabatier-Toulouse III, Fondation pour la Recherche Médicale, European Community (Biomed-2 BMH4-CT983181) to U-466, the Austrian Research Council special research center “Biomembranes” project F (11) 710, and Jubiläumsfonds der Österreichischen Nationalbank project 6941 to Dr Jürgens. Dr Escargueil-Blanc was the recipient of a fellowship from SFA and VML. The authors wish to thank Dr J.P. Cravedi (INRA, Toulouse) for the generous gift of [3H]4-HNE diethylacetal.
Stary HC. The sequence of cell and matrix changes in atherosclerotic lesions of coronary arteries in the first forty years of life. Eur Heart J. 1990; 11 (suppl E): 3–19.
Hajjar DP, Haberland ME. Lipoprotein trafficking in vascular cells. J Biol Chem. 1997; 272: 22975–22978.
Parhami F, Fang ZT, Yang B, et al. Stimulation of Gs and inhibition of Gi proteins by minimally oxidized LDL. Arterioscler Thromb Vasc Biol. 1995; 15: 2019–2024.
Natarajan V, Scribner WM, Hart CM, et al. Oxidized LDL-mediated activation of phospholipase D in smooth muscle cells. J Lipid Res. 1995; 36: 2005–2016.
Auge N, Escargueil-Blanc I, Lajoie-Mazenc I, et al. Potential role for ceramide in MAPK activation and proliferation of vascular smooth muscle cells induced by oxidized LDL. J Biol Chem. 1998; 273: 12893–12900.
Suc I, Meilhac O, Lajoie-Mazenc I, et al. Activation of EGF-Receptor by oxidized LDL. FASEB J. 1998; 12: 665–671.
Claesson-Welsh L. Platelet-derived growth factor signals. J Biol Chem. 1994; 268: 32023–32026.
Vieira O, Escargueil-Blanc I, Jürgens G, et al. Oxidized LDL alter the activity of the ubiquitin-proteasome pathway: potential role in oxidized LDL-induced apoptosis. FASEB J. 2000; 14: 532–542.
Santanam N, Auge N, Zhou M, et al. Overexpression of human catalase gene decreases oxidized lipid-induced cytotoxicity in vascular smooth muscle cells. Arterioscler Thromb Vasc Biol. 1999; 19: 1912–1917.
Buscail L, Delesque N, Esteve JP, et al. Stimulation of tyrosine phosphatase and inhibition of cell proliferation by somatostatin analogues. Proc Natl Acad Sci U S A. 1994; 91: 2315–2319.
Jürgens G, Chen Q, Esterbauer H, et al. Immunostaining of human autopsy aortas with antibodies to modified apolipoprotein B and apoprotein (a). Arterioscler Thromb. 1993; 13: 1689–1699.
Sundaresan M, Yu ZX, Ferrans VJ, et al. Requirement for generation of H2O2 for platelet-derived growth factor signal transduction. Science. 1995; 270: 296–299.
Schissel SL, Jiang X, Tweedie-Hardman J, et al. Secretory sphingomyelinase, a product of the acid sphingomyelinase gene, can hydrolyze atherogenic lipoproteins at neutral pH: implications for atherosclerotic lesion development. J Biol Chem. 1998; 273: 2738–2746.
Bioukar EB, Marricco NC, Zuo D, et al. Serine phosphorylation of the ligand-activated β-PDGF receptor by casein kinase I-gamma2 inhibits the receptor autophosphorylating activity. J Biol Chem. 1999; 274: 21457–21463.
Stiko A, Regnstrom J, Shah PK, et al. Active oxygen species and lysophosphatidylcholine are involved in oxidized LDL activation of smooth muscle cell DNA synthesis. Arterioscler Thromb Vasc Biol. 1996; 16: 194–200.