Oxidized low density lipoproteins regulate synthesis of monkey aortic smooth muscle cell proteoglycans that have enhanced native low density lipoprotein binding properties.

Oxidized low density lipoproteins (Ox-LDL) affect several biological processes involved in atherogenesis. However, it is not known whether Ox-LDL can regulate proteoglycan expression and thus affect arterial wall lipoprotein retention. This study evaluated whether Ox-LDL, as compared with native LDL, regulates proteoglycan expression by monkey arterial smooth muscle cells in vitro and whether proteoglycans synthesized in the presence of Ox-LDL exhibit altered lipoprotein binding properties. Ox-LDL stimulated glycosaminoglycan synthesis, as measured by (35)SO(4) incorporation, by 30-50% over that of native LDL. The effect was maximal after 72 h of exposure to 5 microg/ml of Ox-LDL. The molecular sizes of versican, biglycan, and decorin increased in response to Ox-LDL, as indicated by size exclusion chromatography and SDS-polyacrylamide gel electrophoresis. These effects could be mimicked by the lipid extract of Ox-LDL. These size increases were largely due to chain elongation and not to alterations in the ratio of (35)SO(4) to [(3)H]glucosamine incorporation. Affinity chromatography indicated that Ox-LDL stimulated the synthesis of proteoglycans with high affinity for native LDL. Ox-LDL also specifically stimulated mRNA expression for biglycan (but not versican or decorin), which was correlated with increased expression of secreted biglycan. Thus, Ox-LDL may influence lipoprotein retention by regulating synthesis of biglycan and also by altering glycosaminoglycan synthesis of vascular proteoglycans so as to enhance lipoprotein binding properties.

The vascular extracellular matrix is a major site of lipid accumulation in the arterial wall. Although several matrix molecules have been reported to interact with low density lipoproteins (LDL) 1 (1)(2)(3)(4), proteoglycans are hypothesized to be the major lipoprotein binding site within atherosclerotic lesions (5)(6)(7)(8). According to the response-to-retention hypothesis of atherogenesis, two processes thought to be central to the development of atherosclerosis are the retention of lipoproteins by proteoglycans in the arterial wall and the subsequent mod-ification of these retained lipoproteins by processes such as oxidation (8). However, little is known about whether and how oxidative processes might influence lipoprotein retention within the arterial wall. Therefore, this study examined whether lipoprotein oxidation could influence proteoglycan synthesis by arterial smooth muscle cells in culture and whether these proteoglycans exhibited altered lipoprotein binding characteristics.
Vascular proteoglycans are a heterogeneous group of molecules that have the common structure of a core protein, to which glycosaminoglycan chains are covalently attached (9). Usually, one type of glycosaminoglycan is associated with a single core protein, giving rise to proteoglycans within the arterial wall extracellular matrix that contain chondroitin sulfate, dermatan sulfate, heparan sulfate, and keratan sulfate glycosaminoglycans (10). The major types of vascular proteoglycans identified in blood vessels are the large chondroitin sulfate proteoglycan versican, the small dermatan sulfate proteoglycans decorin and biglycan, the heparan sulfate proteoglycan perlecan, and the keratan sulfate proteoglycan lumican (10).
Although proteoglycans are only a minor component of the normal arterial wall, their accumulation in intimal lesions is characteristic of all phases of atherogenesis (11)(12)(13). As examples, versican is prominent in the diffuse intimal thickening of early atherosclerosis, whereas biglycan and decorin are localized to the fibrous cap of more advanced atherosclerotic lesions (13). Furthermore, the structural properties of proteoglycans isolated from atherosclerosis-prone regions of the aorta (14) and from atherosclerotic lesions (2,(15)(16)(17) are altered. These structural changes include increases in glycosaminoglycan chain length (14) and increases in dermatan sulfate content (15), both of which were found to be positively correlated with increased lipoprotein binding by LDL affinity chromatography (14,15). Another structural characteristic of glycosaminoglycan chains that has been associated with increased lipoprotein binding is an increased ratio of chondroitin 6-sulfate to chondroitin 4-sulfate (5,19). Thus, the amount and structure of extracellular matrix proteoglycans is altered in atherosclerosis in ways that potentially affect their interactions with lipoproteins within the arterial wall.
Although there is considerable information regarding the regulation of proteoglycan synthesis by cytokines (20 -22), there are little data as to whether lipoproteins affect proteoglycan synthesis. Hyperlipidemia and high concentrations of LDL, but not high density lipoproteins (23), have been shown to increase total glycosaminoglycan secretion by smooth muscle cells in culture (23)(24)(25). However, changes in specific proteoglycan composition or structure were not characterized in those studies.
Oxidized LDL (Ox-LDL) has been shown to modulate a variety of functions in vascular wall cells in ways that can contribute to all stages of atherosclerosis development (26 -28). Although the effects of Ox-LDL on the regulation of proteogly-can synthesis have not been investigated to date, Ox-LDL has been shown to enhance the expression of other extracellular matrix molecules, including fibronectin (29,30), laminin (29), and collagen (31). These findings suggest that lipoprotein oxidation may play an important role in the extracellular matrix remodelling that occurs during atherogenesis.
Therefore, in this study, the role of Ox-LDL as a potential regulatory molecule of proteoglycan synthesis in arterial smooth muscle cells was examined. Ox-LDL was found (i) to enhance total proteoglycan synthesis and to preferentially stimulate synthesis of the core protein for biglycan, but (ii) to stimulate glycosaminoglycan chain elongation on versican, biglycan and decorin, all of which (iii) display stronger binding to an LDL affinity column than proteoglycans synthesized under the influence of native LDL (N-LDL). Thus, lipoprotein oxidation may play a role in the pathobiology of atherosclerosis by regulating the synthesis of proteoglycans with enhanced lipoprotein binding properties.
Stimulation and Metabolic Labeling-Quiescent cells were incubated for up to 72 h with 5-25 g of N-LDL or Ox-LDL/ml, or phosphate-buffered saline (control) in fresh medium containing 0.1% calf serum (see figure legends for specific details). Cells were metabolically labeled with 35 SO 4 (50 Ci/ml) for 6 h following the treatment period, in fresh medium containing 0.1% calf serum with no lipoproteins. Alternatively, cells were metabolically labeled with 35

Lipoprotein Preparation
Isolation-LDL were isolated from plasma of normal human volunteers as described previously (32). In brief, LDL (d ϭ 1.019 -1.063 g/ml) was separated from normal human plasma by preparative ultracentrifugation in a Beckman VTi 50 vertical rotor (Beckman Instruments, Palo Alto, CA) (33) and purified by sequential density gradient ultracentrifugation (32).
Oxidation-Ox-LDL was generated by incubation of N-LDL (300 g/ml) in the presence of 5 M copper sulfate for 18 h at 37°C in air (34). This procedure resulted in the formation of extensively oxidized LDL, as indicated by measurements of thiobarbituric acid reactive substances (35), lipid peroxides, conjugated dienes, free amino groups by trinitrobenzenesulfonic acid reactivity (36), and electrophoretic mobility on agarose gels in barbital buffer at pH 8.6.
Extraction-N-LDL and Ox-LDL were extracted with chloroform and methanol according the method of Folch (37). The lipid extracts were dried down under N 2 and then resolubilized in ethanol for treatment of cells. Cellular lipids were extracted with hexane and isopropanol for determination of cholesteryl ester and unesterified cholesterol, as described (38)

Proteoglycan Isolation and Characterization
Isolation-Media from metabolically labeled SMC cultures were harvested with the protease inhibitors benzamidine HCl, 6-aminocaproic acid, and phenylmethylsulfonyl chloride. Following thorough washing with phosphate-buffered saline, the cell layer was solubilized with 8 M urea buffer (8 M urea, 2 mM EDTA, 50 mM Tris-HCl, 0.5% Triton X-100, and 0.25 M NaCl) and harvested with protease inhibitors. The medium and cell layer were applied to separate DEAE-Sephacel mini-columns equilibrated in 8 M urea buffer, washed in 8 M urea buffer, and eluted with 8 M urea buffer containing 3 M NaCl. This ion exchange chromatography step served to remove free radiolabel and to concentrate proteoglycan samples. Radiolabeled proteoglycans then were applied to a preparative Sepharose CL-2B column (0.7 ϫ 110 cm) equilibrated in 8 M urea buffer. Fractions of 1 ml were collected, and an aliquot of each fraction was assayed for radioactivity by liquid scintillation counting. The radiolabeled proteoglycans eluting at a mean hydrodynamic size of K av Յ 0.3 were pooled for the versican peak (Peak 1) (21). Proteoglycans eluting at 0.55 Յ K av Յ 0.8 were pooled for the combined biglycan and decorin peak (Peak 2) (22). Versican, biglycan, and decorin then were reconcentrated over DEAE-Sephacel mini-columns.
Cetyl Pyridinium Chloride Precipitation-The incorporation of 35 SO 4 into proteoglycans was measured by cetyl pyridinium chloride (CPC) precipitation. Aliquots (3 ϫ 50 l) of the 35 SO 4 -labeled medium were spotted on filter paper and washed five times for 1 h in 1% CPC with 0.05 M NaCl (20,21,39). The amount of precipitate on the dried filter paper was determined by liquid scintillation counting. In some experiments, CPC precipitations were performed with and without the addition of carrier chondroitin sulfate to determine the recovery efficiency of radioactivity in the precipitate. The use of carrier chondroitin sulfate did not improve the recovery of radioactivity (data not shown).
Analysis of Molecular Size-35 SO 4 -Labeled proteoglycans secreted into the medium were characterized by SDS-PAGE under reducing conditions according to the procedure of Laemmli (40) on 4 -12% gradient slab gels with a 3.5% stacking gel. Chondroitin and dermatan sulfate glycosaminoglycan chains of the 35 SO 4 -labeled proteoglycans were removed by digestion with chondroitin ABC lyase (0.03 U/ml) (ICN, Costa Mesa, CA) in 0.3 M Tris-HCl pH 8.0, 0.6 mg of bovine serum albumin/ml, and 18 mM sodium acetate with protease inhibitors for 4 h at 37°C (21). For estimation of apparent relative masses, prestained high molecular weight standards (Novex, San Diego, CA) were run in separate lanes. Gels were dried, treated with Enlightning enhancer (NEN Life Science Products) and visualized by fluorography after exposure to Kodak XAR-2 film at Ϫ70°C.
Analysis of Hydrodynamic Size-Hydrodynamic size of 35 SO 4 -labeled proteoglycans was analyzed on an analytical Sepharose CL-2B molecular sieve column (0.7 ϫ 110 cm) equilibrated in 8 M urea buffer with 0.5% Triton X-100 (22). Approximately 50,000 dpm of free 35 SO 4 radiolabel were added to each sample just prior to application on the column. Fractions of 1 ml were collected, and an aliquot of each fraction was assayed for radioactivity by liquid scintillation counting. The elution position of the free 35 SO 4 was used as a marker for the total volume (V t ). The void volume (V o ) of the column was determined separately from the elution position of 3 H-labeled DNA.
Glycosaminoglycan Analysis-Glycosaminoglycan chains were released from proteoglycans by reductive ␤-elimination with 1 M sodium borohydride in 50 mM NaOH for 4 h at 45°C (21). The reaction was terminated by neutralizing the sample with glacial acetic acid. Approximately 50,000 dpm of free 35 SO 4 radiolabel were added to each sample, and the liberated glycosaminoglycan chains then were applied to a Sepharose CL-6B column (0.7 ϫ 65 cm) in 0.2 M Tris, pH 7.0, with 0.2 M NaCl for analysis of chain length by size exclusion chromatography (22). The elution position of the free 35 SO 4 was used as a marker for V t . The void volume (V o ) of the column was determined separately from the elution position of 3 H-labeled DNA.
Northern Analysis-Total cellular RNA was isolated by lysis of cells in situ in guanidinium isothiocyanate, homogenized, and purified using guinidinium-acid-phenol (41). 15-g aliquots of RNA were electrophoresed overnight in a denaturing 1% agarose gel containing 6% formaldehyde and 1% MOPS running buffer. Agarose gels were stained with ethidium bromide to visualize 28 S and 18 S RNA bands prior to transfer of RNA to nylon membranes with 10ϫ standard saline citrate. Filters were prehybridized in a solution containing 0.5 M Na 2 HPO 4 , pH 7.2, with 1 g of bovine serum albumin/100 ml, 7 g of SDS/100 ml, and 10 mM EDTA and then hybridized overnight with random prime labeled cDNA for biglycan (cDNA for the full-length core protein of human bone PG-1) (42), decorin (cDNA for the full-length core protein of bovine bone decorin) (43), versican (cDNAs encoding the C-terminal and middle regions of fibroblast versican (corresponding to the sequences between base pairs 6412-8125 and 2607-6092, respectively) (44), and perlecan (45). Autoradiography of the filters was performed, and autoradiograms were scanned with a Hewlett-Packard Scan Jet IIcx using ImageQuant software. Autoradiograms were normalized to 28 S, as visualized by ethidium bromide staining.
Western Immunoblot Analysis-Core proteins of chondroitin ABC lyase-digested samples were separated on SDS-PAGE 10% gels and transferred to nitrocellulose membranes at 20 V for 35 min in 40 mM Tris, 50 mM glycine with 20 ml of methanol/100 ml, and 0.0375 g of SDS/100 ml in a semidry electrophoretic transfer apparatus (Bio-Rad). Membranes were blocked with 2 ml of calf serum/100 ml of 50 mM Tris-buffered saline, pH 7.4, with 0.05 ml of Tween-20/100 ml (TBST) and then incubated overnight at 4°C with rabbit antiserum prepared against a peptide sequence near the N terminus of human biglycan (1:1000) (kindly provided by Dr. L. Fisher, National Institutes of Health, Bethesda, MD) (46). Nitrocellulose membranes were washed with 0.1% calf serum in TBST before incubation with alkaline phosphatase-conjugated goat anti-rabbit antiserum (1:2000) dilution in TBST with 0.1% bovine serum albumin) (Roche Molecular Biochemicals) for 1 h at room temperature. After washing, membranes were developed by enhanced chemiluminescence (Pierce) and visualized by fluorography on Kodak XAR-2 film.

Affinity Chromatography
Preparation of LDL-Sepharose Columns-LDL (5 mg/ml) was bound to 1 ml of cyanogen bromide-activated Sepharose (Amersham Pharmacia Biotech) according to the manufacturer's procedure. 50 g of heparin/ml was included during the coupling step to protect the proteoglycan-binding arginine-and lysine-rich regions of apo B from being blocked during the reaction with activated gel (20). LDL-Sepharose was washed with 10 volumes of 2 M NaCl to dissociate the heparin from the covalently bound LDL and then equilibrated in 10 mM HEPES, 20 mM NaCl, 250 M butylated hydroxy toluene. A column containing no LDL but blocked with ethanolamine served as the control.
Analysis of Proteoglycan Binding to LDL-Sepharose Column-10 ml of 35 SO 4 -labeled medium was dialyzed extensively against 10 mM HEPES, 20 mM NaCl, 250 M butylated hydroxy toluene. The medium from control or N-LDL-or Ox-LDL-stimulated cells were passed over separate 1-ml LDL-Sepharose columns, and the column flow-through was collected. The columns were washed twice with 10 ml of 10 mM HEPES, 20 mM NaCl, 250 M butylated hydroxy toluene. The bound material then was eluted with a stepwise gradient consisting of 150, 200, and 400 mM NaCl. Five 1-ml fractions were collected at each concentration of NaCl, and an aliquot of each fraction was counted for 35 SO 4 radioactivity.

Statistical Analyses
The significance of differences in mean values between N-LDL or Ox-LDL and control were determined by Student's t test.

Other Methods
Cell protein was measured by the method of Lowry et al. (47). 35 SO 4 Incorporation into Proteoglycans-To characterize the effects of N-LDL and Ox-LDL on [ 35 SO 4 ]proteoglycan synthesis, SMC treated for up to 72 h with 0 -25 g/ml N-LDL or Ox-LDL were metabolically labeled with 35 SO 4 for the 6 h time period following lipoprotein treatment. After 72 h of treatment, an increase in CPC-precipitable counts was observed at all concentrations of Ox-LDL within this range. However, the response was maximal at 5 g of Ox-LDL/ ml. A 30 -50% (p Ͻ 0.005) increase in CPC-precipitable counts in both the medium (Fig. 1A) and cell layer extracts (Fig. 1B) of SMC treated with 5 g of Ox-LDL/ml for 72 h was observed. No significant effect of Ox-LDL was detected prior to 72 h. N-LDL had no effect at any time or at any concentration. Normalization of CPC-precipitable counts for cell number or cell protein did not alter these findings (data not shown), because neither N-LDL nor Ox-LDL had any effect on either cell numbers or cell protein (data not shown). Similar 35 SO 4 incorporation results were obtained with protocols for longer labeling periods up to 24 h (data not shown). Thus, Ox-LDL stimulates 35 SO 4 incorporation into proteoglycans.

Ox-LDL Stimulates
Ox-LDL Does Not Increase Intracellular Levels of Unesterified or Esterified Cholesterol-To investigate whether or not the effects of Ox-LDL on 35 SO 4 incorporation could be attributed to cholesterol loading of these monkey arterial SMC, unesterified and esterified cholesterol levels within the cells were determined. Ox-LDL (5 g/ml) did not increase either unesterified or esterified cholesterol levels after 72 h of treatment (Table I). N-LDL (5 g/ml) also did not increase either unesterified or esterified cholesterol levels after 72 h of treatment (Table I). Thus, the Ox-LDL effects on 35 SO 4 incorporation were not due to cholesterol loading of these cells.
Ox-LDL Increases the Size of Newly Synthesized Proteoglycans-To investigate whether or not Ox-LDL influences the size and proportion of proteoglycan molecules synthesized and secreted in its presence, both SDS-PAGE and size exclusion chromatography were performed. SMC synthesize a large proteoglycan that does not enter the resolving gel of a 4 -12% gradient SDS-PAGE. This proteoglycan previously has been identified as the chondroitin sulfate proteoglycan versican (21,44) and is referred to here as band 1. Band 2 has a molecular mass of Ͼ300 kDa and is a mixture of heparan sulfate and chondroitin sulfate proteoglycans that have not been characterized in detail. In addition, two smaller proteoglycans of apparent molecular masses ϳ200 and 75 kDa previously have been identified as biglycan (band 3) and decorin (band 4), respectively (48).
By SDS-PAGE, Ox-LDL increased the apparent molecular size of both bands 3 and 4, as compared with the size of these molecules secreted from control or N-LDL-treated cells ( Fig. 2A). Because band 1 does not enter the resolving gel and band 2 only just enters the resolving gel, no effect on the apparent molecular sizes of these proteoglycans could be determined by this method.
Size exclusion chromatography on Sepharose CL-2B resolves the proteoglycan mixture into two peaks. Versican elutes in  Peak 1 at K av ϭ ϳ0.23 (21). The two small proteoglycans biglycan and decorin co-elute in Peak 2 at K av ϭ ϳ0.65 (22). Both radiolabeled peaks from cells stimulated with Ox-LDL were shifted to a lower K av relative to that of proteoglycans synthesized by control and N-LDL-treated cells (Fig. 2B). This shift toward a smaller K av for Peak 1 indicates an increase in hydrodynamic size for versican, which could not be detected by SDS-PAGE. Although this method does not resolve biglycan and decorin, the shift to a smaller K av for the combined small proteoglycans in Peak 2 is in agreement with the increase in size of the individual molecules as observed by SDS-PAGE. N-LDL did not result in an increase in the hydrodynamic size of either Peak 1 or Peak 2. Thus, these results indicate that Ox-LDL does not specifically affect the size of selected proteoglycan molecules but rather stimulates an increase in the overall sizes of versican, biglycan, and decorin secreted by SMC. The individual fractions from Peaks 1 and 2 were pooled and reconcentrated on DEAE-Sephacel for glycosaminoglycan chain length analysis (see below). Aliquots of each peak also were reanalyzed by SDS-PAGE to confirm the elution position of each peak on the basis of apparent molecular size (data not shown).
Ox-LDL Causes Glycosaminoglycan Chain Elongation on Biglycan, Decorin, and Versican-To evaluate whether the increase in the size of biglycan, decorin, and versican is correlated with glycosaminoglycan chain elongation, the chains from each peak were released from the core proteins of Peaks 1 and 2 by reductive alkaline ␤-elimination and analyzed by size exclusion chromatography. On Sepharose CL-6B, the glycosaminoglycan chains of versican from control or N-LDL-treated cells eluted with a K av of ϳ0.38, corresponding to a molecular weight of ϳ35,000, as determined according to Wasteson et al. (39). With Ox-LDL stimulation, the elution profile of versican glycosaminoglycan chains was shifted to a lower K av (ϳ0.32), indicating a larger molecular weight of ϳ55,000 (39) (Fig. 3A). Similarly, Ox-LDL caused an increase in chain length for the combined small proteoglycans biglycan and decorin isolated in Sepharose CL-2B Peak 2. Glycosaminoglycan chains for Peak 2 from control and N-LDL-treated cells eluted at a K av of ϳ0.43 (molecular weight, ϳ28,000) on CL-6B, whereas from Ox-LDLtreated cells, Peak 2 glycosaminoglycan chains eluted at a K av of ϳ0.38 (molecular weight, ϳ35,000) (Fig. 3B). Although the effects of Ox-LDL on chain length of the individual components of Peak 2 could not be resolved by this method, these results suggest that the Ox-LDL effect of glycosaminoglycan chain elongation is not selective for specific proteoglycan molecules but rather applies to all of the major proteoglycans secreted by SMC. 35 SO 4 incorporation is consistent with glycosaminoglycan chain elongation, the increase in 35 SO 4 incorporation also might reflect an increase in the degree of sulfation of the disaccharide residues. Therefore, the ratio of 35 SO 4 to [ 3 H]glucosamine incorporation into disaccharides was measured. The medium and cell layer from cells that were metabolically labeled with both 35 SO 4 and [ 3 H]glucosamine were harvested and analyzed by CPC precipitation. There was no significant change in the ratio of 35 SO 4 to [ 3 H]glucosamine with either N-LDL or Ox-LDL, as compared with control conditions (Fig. 4). Thus, the Ox-LDL-induced increase in [ 35 SO 4 ] incorporation does not suggest a change in the pattern of sulfation.

The Lipid Extract of Ox-LDL Mimics the Effects of Ox-LDL on Proteoglycan Synthesis-To understand what component of Ox-
LDL was responsible for these effects, lipid extracts of Ox-LDL were prepared and examined for their effects on proteoglycan synthesis. The lipid extract of Ox-LDL stimulated a 30% (p Ͻ 0.05) increase in medium 35 SO 4 incorporation over the control (Fig. 5A). This percentage of increase was similar to that found for Ox-LDL itself (Fig. 1A). Further, SDS-PAGE analysis showed that the lipid extract of Ox-LDL caused an increase in the apparent molecular weight of both bands 3 and 4, as compared with the size of these molecules secreted from the control cells (Fig.  5B), as did Ox-LDL itself. Lipid extracts of N-LDL had no effects on either 35 SO 4 incorporation (Fig. 5A) or apparent molecular weight (Fig. 5B). Thus, the oxidized lipid component was responsible for the effects of Ox-LDL on proteoglycan synthesis.

A Large Proportion of the Proteoglycans Synthesized in the Presence of Ox-LDL Have Enhanced LDL Binding Properties-
Increases in glycosaminoglycan chain length have been correlated with increased LDL binding (14,20). Therefore, LDL affinity chromatography was performed to assess the binding properties of proteoglycans synthesized in the presence of Ox-LDL. 35 SO 4 -Labeled media from control cells or cells stimulated with N-LDL or Ox-LDL were applied to separate LDL-Sepharose columns and eluted with a stepwise NaCl gradient. In 20 mM NaCl, 75-90% of the 35 SO 4 radioactivity applied to the columns was bound for each sample. The majority of [ 35 SO 4 ]proteoglycans synthesized under control, N-LDL, or Ox-LDL conditions eluted at 150 mM NaCl, whereas a minority of [ 35 SO 4 ]proteoglycans were eluted at 400 mM NaCl (Table II). The proportion of [ 35 SO 4 ]proteoglycans that bound tightly and required 400 mM NaCl to be eluted was greatest under Ox-LDL conditions, as compared with control or N-LDL conditions. Thus, a larger proportion of proteoglycans synthesized in the presence of Ox-LDL have enhanced LDL binding properties, as compared with those from control or N-LDL-treated cells.
To examine whether specific proteoglycan molecules were represented among those with enhanced LDL binding properties, [ 35 SO 4 ]proteoglycans eluted from the LDL-Sepharose columns were analyzed by SDS-PAGE (Fig. 6A) and by size ex- clusion chromatography (Fig. 6B). The results for the control (data not shown), N-LDL (data not shown), or Ox-LDL treatment conditions were qualitatively the same. By SDS-PAGE (Fig. 6A), proteoglycan bands 1, 2, 3, and 4 all were represented in the affinity column fractions that eluted with 150 mM NaCl (lane 1). The fractions that eluted with 200 mM NaCl (lane 2) were diminished in band 4 (decorin). The fractions that eluted with 400 mM NaCl (lane 3) were further diminished in bands 2 and 3 (biglycan). Band 1 (versican) was present in the fractions that eluted at each NaCl concentration but was the predominant proteoglycan species remaining in the fractions that eluted with 400 mM NaCl. The fractions that eluted at 150 mM NaCl were resolved into two peaks by size exclusion chromatography on Sepharose CL-2B (Fig. 6B), as described in Fig. 2B. However, the fractions that eluted at 400 mM NaCl consisted only of Peak 1. Furthermore, the versican in Peak 1 from the 400 mM NaCl fractions had a higher K av than the versican in Peak 1 from the 150 mM NaCl fractions. This indicates that the versican that binds most tightly to native LDL is bigger on the basis of hydrodynamic size than the versican that binds less tightly to native LDL. Thus, proteoglycans could be selectively recovered from a native LDL affinity column on the basis of the concentration of salt at which they eluted. The tightest binding proteoglycans were found to be versican that was characterized by large hydrodynamic size. Taken together, these results indicate that there was no qualitative difference in these affinity column results between treatment conditions. However, quantitatively more versican with tight LDL binding properties was synthesized with Ox-LDL treatment than with control or N-LDL treatment.
Ox-LDL Increases Biglycan Expression-To determine whether N-LDL or Ox-LDL differentially regulates expression mRNA for versican, decorin, or biglycan, SMC were treated with N-LDL or Ox-LDL for up to 72 h. Northern blot analysis indicated that treatment with Ox-LDL for 72 h stimulated a 4 -8-fold increase (p Ͻ 0.005) in biglycan mRNA levels over that of control unstimulated or N-LDL-treated cells (Fig. 7A). This effect was time-dependent, because no increase in biglycan mRNA levels was observed after 24 or 48 h of treatment with Ox-LDL. This effect on biglycan mRNA also was dose-dependent and was maximal at 5-10 g of Ox-LDL/ml (Fig. 7B), concentrations that did not cause either proliferation or cytotoxicity (data not shown). Neither versican nor decorin mRNA levels were changed by Ox-LDL at any time point studied (Fig. 7A).
This suggested increase in biglycan expression was supported by Western immunoblot analysis using an antibody specific for the core protein of biglycan. Newly synthesized proteoglycans were 35 SO 4 -labeled during the 6-h period following treatment with N-LDL or Ox-LDL. 35 SO 4 -Labeled proteoglycans were treated with chondroitin ABC lyase, separated on SDS-PAGE, and transferred to nitrocellulose for immunodetection. An increase in the amount of biglycan synthesized and secreted into the medium was found after 72 h of treatment with Ox-LDL (Fig. 7C). In some experiments N-LDL also appeared to increase the amount of biglycan core protein, although never as much as did Ox-LDL (Fig. 7C). Thus, Ox-LDL selectively increases biglycan mRNA and also increases biglycan expression in monkey arterial SMC. DISCUSSION Ox-LDLs have been found to have a variety of biological effects that may be important to atherogenesis (49 -51). This study demonstrates that Ox-LDL also regulates the expression of proteoglycans, molecules that are thought to be critical to the retention of lipoproteins within the arterial wall (5-7). The Ox-LDL effects are severalfold. First, Ox-LDL nonselectively increases the overall molecular sizes of versican, biglycan, and  SO 4 ]proteoglycans eluted from native LDL affinity columns by NaCl step gradient 35 SO 4 -labeled media from control or N-LDL-or Ox-LDL-stimulated cells was passed over LDL-Sepharose columns, as described under "Experimental Procedures." The bound material was eluted with a step-wise gradient consisting of 150, 200, and 400 mM NaCl. Five 1-ml fractions were collected at each concentration of NaCl, and an aliquot of each fraction was counted for 35 SO 4 radioactivity. Values are expressed as the percentages of radiolabeled material recovered at each concentration of NaCl. decorin synthesized and secreted by monkey arterial SMC in culture. Second, these effects were not due simply to cholesterol loading of the cells by Ox-LDL but could be isolated to the oxidized lipid fraction of Ox-LDL. Third, the increase in molecular size is due, at least in part, to glycosaminoglycan chain elongation on each of these molecules. Fourth, a greater proportion of proteoglycan molecules synthesized in the presence of Ox-LDL exhibit tight binding properties to native LDL. Fifth, Ox-LDL stimulates the expression of a large versican molecule that has high affinity for native LDL. Sixth, Ox-LDL specifically enhances the expression of mRNA for biglycan but not versican nor decorin. Finally, this enhanced expression of biglycan mRNA is correlated with increased expression of the biglycan molecule associated with the medium and cell layer in these smooth muscle cell cultures.
Ox-LDL regulates post-translational modifications by universally causing glycosaminoglycan chain elongation on versican, biglycan, and decorin. This lengthening of the glycosaminoglycan chains contribute at least partially to the increase in size of each of these proteoglycans. Such glycosaminoglycan chain elongation is believed to be functionally significant for binding to plasma proteins and growth factors. Proteoglycanlipoprotein binding involves an interaction between negative charges on the glycosaminoglycan chains of proteoglycans and positively charged lysine and arginine residues on LDL apo B (5,8). These interactions can be modulated by changes to either component of the interaction. For example, LDL forms complexes with proteoglycans, whereas apo E-free high density lipoproteins do not (52), and Lp(a) binds proteoglycans more avidly than LDL (53). A characteristic of proteoglycan that can affect its binding to LDL is the length of its glycosaminoglycan chai(ns); longer chains have been correlated with increased LDL binding (2,14,20). This is supported in vivo by the finding that LDL is selectively retained by proteoglycans with elongated side chains isolated from vascular sites that are prone to developing atherosclerotic lesions (14). Thus, the finding in the present study that Ox-LDL, as compared with N-LDL, stimulates synthesis of proteoglycans that bound more strongly to an LDL affinity column is consistent with the finding that Ox-LDL also causes glycosaminoglycan chain elongation.
The regulation of glycosaminoglycan chain elongation is likely to be under separate control from that of core protein synthesis. For example, PDGF regulation of versican core protein synthesis is protein kinase C-and mitogen-activated protein kinase kinase-dependent, whereas the effect of PDGF on chondroitin sulfate chain elongation is protein kinase C-dependent and mitogen-activated protein kinase kinase-independent 2 and appears not to be dependent on the core protein substrate for glycosaminoglycan attachment. The increase in chain length may be due to increases in the amount and/or activity of the enzymes responsible for chain synthesis, as has been shown for glycosaminoglycan chains synthesized by proliferating versus nonproliferating bovine arterial SMC (54). In these studies, the synthesis of longer chains by proliferating cells was accompanied by an increase in the activity of several glycosaminoglycan synthesizing enzymes, such as xylosyl transferase, N-acetylgalactosaminyl transferase I, and chondroitin sulfotransferases (54).
Whereas Ox-LDL has a nonselective impact on post-translational glycosaminoglycan chain elongation of versican, biglycan, and decorin, Ox-LDL appears to differentially regulate proteoglycan gene expression at the mRNA level by selectively enhancing biglycan mRNA. This increased expression of biglycan mRNA translates into increased expression of the biglycan molecule. Such differential regulatory effects over SMC proteoglycan genes have been demonstrated for other molecular mediators of atherogenesis as well (21,22). PDGF selectively increases versican mRNA, with no effects on biglycan or decorin mRNA, and transforming growth factor-␤ stimulates mRNA for versican and biglycan but not decorin. Mechanisms that might play a role in the specific effects of Ox-LDL on the biglycan gene involve regulation of intermediary cytokines or transcriptional factors by Ox-LDL.
The effects of Ox-LDL were not found to be due to cholesterol loading of the SMC but could be isolated to the oxidized lipid fraction of Ox-LDL. The late time course of these effects on proteoglycan synthesis after exposure to Ox-LDL or the lipid fraction of Ox-LDL suggests that other intermediary molecules may be involved. Ox-LDL, or its components, has been shown to modulate the expression of some of the known regulatory molecules of proteoglycan synthesis, including PDGF-A chain (55) and transforming growth factor-␤ (30). However, the effects of Ox-LDL on regulation of proteoglycan synthesis are not 2 L. Cardoso and T. N. Wight, unpublished observations. FIG. 7. Ox-LDL increases biglycan expression. Northern blot analysis of quiescent SMC treated with phosphate-buffered saline (control), 5 g of N-LDL/ml, or 5 g of Ox-LDL/ml for 0, 24, 48, and 72 h. Total cellular RNA was harvested by the guanidinium isothiocyanate method and run on denaturing agarose gels prior to transfer to nitrocellulose filters. Filters were probed with 32 P-labeled cDNA for versican (n ϭ 3), biglycan (n ϭ 5), and decorin (n ϭ 3). Methylene blue staining of 28 S rRNA on the filter was used for normalization. Autoradiograms of filters probed with 32 P-labeled cDNA for biglycan, versican, and decorin were scanned and quantified with normalization to methylene blue-stained 28 S rRNA. Values are the means Ϯ S.D. B, Northern blot analysis of quiescent SMC treated with 0, 1, 5, 10, or 25 g of Ox-LDL/ml for 72 h. Filters were probed with 32 P-labeled cDNA for biglycan. C, Western immunoblot analysis of SMC treated with phosphatebuffered saline (control), 5 g of N-LDL/ml, or 5 g of Ox-LDL/ml for 72 h. [ 35 SO 4 ]Proteoglycans were isolated and treated with chondroitin ABC lyase, run on SDS-PAGE (10% resolving gel), and transferred to nitrocellulose for detection with an antibody specific for the biglycan core protein. Gels were loaded on the basis of equal percentages of the CPC precipitable radioactivity and were visualized by fluorography. The arrow indicates a core protein band with a molecular mass of 45-50 kDa, corresponding to the size of the biglycan core protein. The chondroitin ABC lyase (CABC) reaction buffer also is shown. entirely consistent with the effects of either one of these growth factors. For example, Ox-LDL selectively stimulates biglycan mRNA, whereas PDGF selectively stimulates versican mRNA, but transforming growth factor-␤ stimulates both biglycan and versican mRNA. Ox-LDL regulates a number of other molecules, such as monocyte chemotactic protein-1 (49) and tumor necrosis factor-␣ (56), whose effects on proteoglycan synthesis remain to be examined. Thus, other molecules also may play an intermediary role in the effects of Ox-LDL on proteoglycan synthesis.
Ox-LDL also is known to affect promoter elements found within the biglycan gene. Minimally oxidized LDL has been shown to induce NFB-binding promoter elements (57), whereas extensively oxidized LDL has been shown to induce AP-1 binding promoter elements (58). Among others, both NFB and AP-1 binding sites are present in the promoter sequence of the biglycan gene (59), raising the possibility that regulation of biglycan by Ox-LDL may involve these, or other, transcriptional factors. However, NFB and AP-1 binding sites are present in the promoter sequence of the decorin gene (60), which does not appear to be regulated by Ox-LDL.
Although biglycan is only a minor component of the normal arterial wall, it has been demonstrated to accumulate during atherogenesis. By Northern blot analysis, biglycan mRNA has been shown to be increased in the rat carotid artery after balloon injury (61). By immunohistochemistry, biglycan has been demonstrated in primary atherosclerotic and restenotic lesions of human coronary arteries (18) and in atherosclerotic lesions of human coronary arteries from explanted hearts (12). However, the specific function of biglycan in vascular tissue has not been determined. The colocalization of biglycan with apolipoproteins B and E in human coronary arteries suggests that biglycan may be important in the trapping of lipoproteins in atherosclerosis (12).
In the response-to-retention hypothesis of atherogenesis, two processes that are believed to be central to atherosclerotic development are the initial retention of lipoproteins and the subsequent modification of these retained lipoproteins by processes such as oxidation (8). The factors responsible for the initial retention of lipoproteins are not clear but may involve localized alterations in matrix molecules, including proteoglycans, which have enhanced lipoprotein binding capability (14). Such localized differences in proteoglycans have been described in prelesional and lesional sites (15) of atherosclerosis. Cellular oxidation of the lipoproteins may then occur as a normal consequence of lipoprotein retention. Our results thus suggest that lipoprotein oxidation might aggravate this process not only by regulating the synthesis of biglycan but also by enhancing the lipoprotein binding properties of all of the major proteoglycan molecules synthesized and secreted by SMC.