The association of C-reactive protein with an oxidative metabolite of LDL and its implication in atherosclerosis.

C-reactive protein (CRP) is one of the strongest independent predictors of cardiovascular disease. We have previously reported that oxidized LDL (oxLDL) interacts with β2-glycoprotein I (β2GPI), implicating oxLDL/β2GPI complexes as putative autoantigens in autoimmune-mediated atherosclerotic vascular disease. In this study, we investigated the interaction of CRP with oxLDL/β2GPI complexes and its association with atherosclerosis in patients with diabetes mellitus (DM). CRP/oxLDL/β2GPI complexes were predominantly found in sera of DM patients with atherosclerosis. In contrast, noncomplexed CRP isoforms were present in sera of patients with acute/chronic inflammation, i.e., various pyrogenic diseases, rheumatoid arthritis (RA), and DM. Immunohistochemistry staining colocalized CRP and β2GPI together with oxLDL in carotid artery plaques but not in synovial tissue from RA patients, strongly suggesting that complex formation occurs during the development of atherosclerosis. Serum levels of CRP correlated with soluble forms of intercellular adhesion molecule-1 and vascular cell adhesion molecule-1, and oxLDL/β2GPI complexes correlated with total cholesterol and hemoglobin A1c. Thus, the generation of CRP/oxLDL/β2GPI complexes seems to be associated with arterial inflammation, hyperglycemia, and hypercholesterolemia. CRP/oxLDL/β2GPI complexes can be distinguished from pyrogenic noncomplexed CRP isoforms and may represent a more specific and predictive marker for atherosclerosis.

C-reactive protein (CRP) is an acute-phase reactant that belongs to the highly conserved "pentraxin" family and plays a major role in innate immunity (18)(19)(20). CRP binds to a range of autologous and exogenous ligands, including phosphorylcholine originating from either the capsular polysaccharide component of microorganisms, native LDL, oxLDL, or apoptotic cells. This interaction moderates the clearance of opsonized CRP by macrophages (21)(22)(23)(24). Further, CRP binds to multiple ligands, activating the classical complement pathway, which in turn enhances phagocytosis via complement receptors. Epidemiologic studies have shown an association between CRP and the risk for CVD (25,26). Multiple clinical studies have demonstrated a predictive relationship between increased CRP levels and atherothrombotic events. It has also been suggested that CRP is not only a serologic marker for atherosclerosis but also a causal risk factor, stimulating adhesion molecule expression and chemokine production by endothelial cells (27,28). Thus, CRP aids in the first line of host defense against infection and is also involved in inflammatory processes that promote atherosclerosis (atherothrombosis) (29).
In the present study, circulating CRP/oxLDL/b2GPI complexes demonstrated by ELISA and immunoblot analysis were predominantly detected in DM patients with atherosclerosis. Because CRP/oxLDL/b2GPI complexes were not detected in patients with pyrogenic diseases and rheumatoid arthritis (RA), they may be a part of the CVD predictor, i.e., high-sensitivity CRP (hsCRP). Immunohistochemistry also revealed the possibility that the com-plexes are formed in atherosclerotic plaques and released into the circulation. Thus, CRP/oxLDL/b2GPI complexes, which can be distinguished from pyrogenic noncomplexed CRP isoforms, may be a more specific and predictive marker for atherosclerosis (atherothrombosis).

Subjects and specimens
The present cross-sectional study was performed at Okayama University Hospital (Okayama, Japan). The protocols were approved by the Institutional Ethical Review Board of Okayama University Hospital. Informed consent was given by all participants.
Type 2 DM patients (n 5 125) fulfilling the diagnostic criteria of the World Health Organization (WHO) (30) were enrolled in the present study. The diagnosis of DM was based on the presence of chronic hyperglycemia and metabolic disturbances of lipid, carbohydrate, and protein metabolism due to defects in insulin production or activity. Impaired fasting glycemia and/or glucose tolerance test results were used according to WHO established criteria. DM patients with significant CVD events, pregnancy, inflammatory disease, and/or steroid use were excluded. DM patients had a male to female ratio of 73:52 and a mean age of 61.4 6 13.3 years (mean 6 SD; range 34-67 years). Agematched RA patients (n 5 48) and healthy subjects (n 5 48) were enrolled. All RA patients met the American College of Rheumatology revised criteria (31). RA patients with DM and/ or significant CVD events were excluded. RA patients had a male to female ratio of 7:41 and a mean age of 64.0 6 10.0 years (range 33-80 years). Thirty-five patients with acute pyrogenic diseases (i.e., acute enteritis, acute pancreatitis, acute prostatitis, acute pyelonephritis, acute tonsillitis, appendicitis, bronchitis, diverticulitis of colon, cystitis, epididymitis, femoral necrotizing fasciitis, gonitis, fever of unknown cause, infectious enteritis, mesenteric panniculitis, pharyngitis, or pneumonia) without DM, CVD, or collagen diseases were enrolled. All serum samples were stored at 280jC until use.
Tissue samples were prepared from human carotid endarterectomy specimens from 17 patients with transient ischemic attack or minor completed strokes before surgery. Synovial tissue samples were prepared from five RA and five osteoarthritis patients. All tissues were fixed with neutral buffered formalin, embedded in paraffin, and sectioned by standard procedures.
Isolation and oxidation of LDL LDL (1.019,d,1.063) was isolated by ultracentrifugation from fresh healthy human plasma, as described (10). LDL (100 mg/ml of apoB equivalent) in 10 mM Hepes and 150 mM NaCl, pH 7.4, (Hepes buffer) was oxidized by incubation with 5 mM CuSO 4 for 12 h at 37jC. Aliquots were taken to determine thiobarbituric acid-reactive substances and electrophoretic migration in agarose gels.

Reduction of endogenous and/or exogenous endotoxin (lipopolysaccharide) from reagents
Affinity chromatography using a Detoxi-Gel column (Pierce; Rockford, IL) was performed to remove exogenous endotoxin that might have contaminated the CRP and b2GPI preparations during isolation. In experiments to assess endotoxin involvement in complex formation, additional purification steps were performed. For example, endotoxin-free water and sterile materials, including glassware, which were either commercially apyrogenic or depyrogenated by steam autoclaving and/or dry heat, were used during LDL isolation. Endotoxin levels were determined by the Limulus amoebocyte lysate assay.

ELISA for oxLDL complexes
ELISA for oxLDL/b2GPI complexes. This procedure was performed as previously described (10). Briefly, anti-b2GPI MAb, WB-CAL-1, was adsorbed onto microtiter plates (Immulon 2HB; Thermo Labsystems, Franklin, MA) by incubating at 8 mg/ml (dissolved in Hepes buffer, 50 ml/well) at 4jC overnight. After blocking with Hepes buffer containing 1% skim milk, samples diluted 1:100 with Hepes buffer containing 0.5% skim milk were added to the wells (100 ml/well) to be incubated for 2 h. The wells were then incubated with HRP-labeled antihuman apoB-100 MAb (N2E10). Extensive washing between steps was performed with Hepes buffer containing 0.05% Tween 20. Color was developed with tetramethylbendizine and H 2 O 2 . The reaction was terminated, and optical density at 450 nm was measured.
ELISA for CRP/oxLDL/b2GPI complexes. The assay was similar to that for the ELISA for CRP/oxLDL complexes, with the exception that captured N2E10 MAb was replaced by anti-b2GPI MAb WB-CAL-1 on the plate.

Nondenaturing PAGE (native PAGE) and immunoblot analysis
Nondenaturing PAGE (native PAGE without SDS) was carried out on 2-15% polyacrylamide gradient gels according to the method of Krauss and Burke (35), with slight modifications. A pH of 8.3 was used for running the gels to facilitate the characterization of LDL subclasses. To avoid the influence of protein charge, protein markers having similar isoelectric points (pIs) are generally used in nondenaturing PAGE. A set of commercially available molecular mass markers (Daiichi Pure Chemicals; Tokyo, Japan) for nondenaturing PAGE composed of thyroblobulin (669 kDa, pI 4.6), ferritin (443 kDa, pI 4.5), lactate dehydrogenase (140 kDa, pI 4.0), BSA (66 kDa, pI 4.7), and soybean trypsin inhibitor (20.1 kDa, pI 4.50) was used for the study. Then the proteins were transferred to a polyvinylidene difluoride membrane, and immunoblot was performed with anti-CRP Abs anti-b2GPI MAb (Cof-23) or anti-apo-B100 MAb (N2E10).

Immunohistochemistry
Details of the immunohistochemical staining procedure were previously described (36). Pretreatment of human carotid endarterectomy and synovial tissue sections in an oil bath (97jC, 20 min in 0.1 M Tris-HCl, pH 6.0) was performed after deparaffinization according to the modified method of Shi, Key, and Kalra (37), except for the CRP staying run. Endogenous peroxidase activity was blocked by treatment with 3% H 2 O 2 in methanol for 30 min, followed by blocking with 2% normal goat serum. Immunohistochemical staining of adjacent sections was carried out using Abs against CD68, SMCa, CRP, SR-PSOX, and apoB in either native LDL or oxLDL (N2E10), apoB in oxLDL (O1F9), noncomplexed and complexed forms of b2GPI (Cof-23), and complexed form of b2GPI (WB-CAL-1).

Other clinical and biological parameters
Intima media thickness (IMT) was measured by B-mode ultrasonography. Three IMT measurements of diastolic images on each side at 10 mm before or after the carotid bifurcation were obtained. Mean IMT was calculated for each point, and the highest value (maximum IMT) was recorded for each subject and was defined as the distance from the lumen-intima interface to the intima-adventitia interface. An IMT cutoff value of 0.75 mm was used to diagnose atherosclerosis. Serum hsCRP was measured by nephelometry, a latex particle-enhanced immunoassay (N-Latex CRPII; Dade Behring, Tokyo, Japan). Total cholesterol (T-chol), hemoglobin A1c (HbA1c), and CRP were measured by routine laboratory methods. Soluble forms of intracellular adhesion molecule-1 (sICAM-1) and vascular cell adhesion molecule-1 (sVCAM-1) were measured by commercially available ELISAs (R and D Systems, Inc.; Minneapolis, MN).

Statistical analyses
Statistical analysis was performed by StatView software (Abacus Concepts; Berkeley, CA). The Student t-test was used to assess possible correlation between levels of CRP or three types of oxLDL complexes and the occurrence of disease. The correlation between two variables was evaluated by the Pearson correlation test. Fisher's exact test was used to compare the appearance of biological markers, CRP and oxLDL complexes. Ninty-five percent confidence interval (95% CI) was calculated by Woolf's method. The level of P , 0.05 was considered statistically significant.

Formation of CRP/oxLDL/b2GPI complexes
The nature of the interaction between oxLDL and b2GPI has been previously characterized and described (8)(9)(10). We first investigated the interaction among Cu 21 -oxLDL, b2GPI, and CRP. Figure 1A-C depicts the calcium dependency of the interaction of oxLDL with CRP and oxLDL with b2GPI. Figure 1A shows that the interaction between oxLDL and b2GPI was calcium independent from 0 to 5 mM of Ca 21 ion. In contrast, the interaction between oxLDL and CRP was calcium dependent (Fig. 1B). The complexes were not dissociated even if excess EDTA or Ca 21 ion was added after the completion of the reaction (data not shown). At physiological calcium concentrations (approximately 1.25 mM), both the calcium-independent and -dependent interactions took place ( Fig. 1A-C). CRP/ oxLDL/b2GPI complexes were formed proportionally when these three components were coincubated (Fig. 1C). No direct interaction between CRP and b2GPI was observed under any experimental conditions. Figure 1D-F reveals the time-dependent formation of oxLDL complexes with CRP and/or b2GPI in the presence of 1.25 mM of Ca 21 ion at 37jC. The interaction of oxLDL with CRP and oxLDL with b2GPI occurred in a time-dependent as well as dose-dependent manner. These two interactions progressed gradually and reached a plateau after ?12 h.
While the interaction between b2GPI and oxLDL was calcium independent, the interaction between CRP and oxLDL was calcium dependent. At physiological concentrations of calcium, long incubations (of several hours) were required for the in vitro formation of nondissociable CRP/oxLDL/b2GPI complexes or oxLDL/b2GPI complexes. The oxLDL-negative charges acquired during Fig. 1. Profiles of complex formation among Cu 21 -oxidized LDL (oxLDL), b2-glycoprotein I (b2GPI), and Creactive protein (CRP). For the experiment, oxLDL/b2GPI and CRP/oxLDL complexes were preformed by incubating Cu 21 -oxLDL [1 mg apolipoprotein B (apoB) equivalent/ml] and b2GPI (1 mg/ml) in the absence of CaCl 2 , and Cu 21 -oxLDL (1 mg apoB equivalent/ml) and CRP (1 mg/ml) in the presence of 2 mM CaCl 2 , respectively, at 37jC for 16 h. Subsequently, the complexes were purified by size exclusion column chromatography, as shown in Fig. 2B. A-C: CRP, oxLDL, b2GPI, preformed oxLDL/b2GPI complexes, and/or preformed CRP/oxLDL complexes were incubated at 37jC for 16 h in the presence of different concentrations of CaCl 2 . Concentration of each material was 50 mg/ml protein or 50 mg/ml apoB-100 equivalent. Reaction was terminated by immediate freeze, and an aliquot of each specimen was diluted up to 0.5 mg/ml of apoB equivalent (1:100 dilution) to apply to the ELISAs. Generation of oxLDL/b2GPI complexes (A), CRP/oxLDL complexes (B), or CRP/oxLDL/b2GPI complexes (C) was detected in ELISAs, as described in Materials and Methods. D-F: CRP, oxLDL, b2GPI, preformed oxLDL/b2GPI complexes, and/or preformed CRP/oxLDL complexes were incubated at 37jC for different periods in the presence of 1.25 mM CaCl 2 , and generated complexes (D-F) were measured by ELISA, as described above. All values in the in vitro experiments are expressed as the mean 6 SD. Circles, CRP 1 oxLDL 1 b2GPI; triangles, preformed oxLDL/b2GPI complexes 1 CRP; squares, preformed CRP/oxLDL complexes 1 b2GPI. Error bars represent mean 6 SD. CuSO 4 incubation were neutralized by the interaction with these proteins (Fig. 2A).
We also confirmed that significant in vitro production of CRP/oxLDL/b2GPI complexes occurs at reduced endotoxin concentrations (28 pg/ml). However, the potential influence of such small amounts of endotoxin contaminating the reaction mixture on complex formation should be investigated in further studies. Figure 2B shows the elution profile of size exclusion chromatography of the reaction mixture containing CRP and oxLDL/b2GPI complexes. CRP/oxLDL/b2GPI and oxLDL/b2GPI complexes were mainly eluted in fractions 17-30 (mainly 17-22) and 17-30 (mainly 20-30), respectively. CRP was detected in the first peak (corresponding to the size of CRP/oxLDL/b2GPI particles) and in a later fraction (38)(39)(40)(41)(42), where the commercially available pentameric CRP (120 kDa; derived from pleural fluid or serum) was eluted. The contamination of the modified CRP isoform having a molecular mass of 75 kDa (38) in the CRP protein(s) eluted in fractions larger than 50 kDa cannot be excluded. Thus, these results indicate that CRP/oxLDL/ b2GPI complexes can cross-react in the hsCRP assay. The Fig. 2. In vitro formation of CRP/oxLDL/b2GPI complexes. Cu 21 -OxLDL (1 mg/ml of apoB equivalent) was incubated with b2GPI (1 mg/ml) for 16 h at 37jC to form oxLDL/b2GPI complexes. The purified oxLDL/b2GPI (1 mg/ml of apoB equivalent) was further incubated with CRP (purified from human pleural fluid; 0.1 mg/ml) in the presence of 2 mM CaCl 2 for 16 h at 37jC, to form nondissociable CRP/ oxLDL/b2GPI complexes. The complexes were purified by size exclusion chromatography. A: Electrophoresis on an agarose gel was performed and proteins were visualized by amido black staining. The negative charge in oxLDL was significantly neutralized by the complex formation with CRP and b2GPI. B: Elution profile of the size exclusion column chromatography of CRP/oxLDL/b2GPI complexes is shown. The reaction mixture (100 ml aliquot) containing CRP and oxLDL/b2GPI complexes was applied on a Superose 6 HR 10/30 column (Amersham Pharmacia Biotech) equipped with fast-protein liquid chromatography and eluted with Tris buffer containing 1.25 mM CaCl 2 . CRP/oxLDL/b2GPI and oxLDL/b2GPI complexes in 100-fold diluted fractions were detected by ELISAs, and CRP in nondiluted fractions was detected by highsensitivity CRP nephelometry. total recovery of CRP from all eluates (including noncomplexed CRP) detected by hsCRP nephelometry varied in individual experiments, with ranges between 38% and 70%.

Diagnostic accuracy of CRP/oxLDL/b2GPI complex assay
The principle of the ELISA for CRP/oxLDL/b2GPI complexes is schematically represented in Fig. 3A. The assay was specific for CRP/oxLDL/b2GPI complexes without any cross-reactivity to CRP/oxLDL complexes or oxLDL/b2GPI complexes (Fig. 3B). In addition, the CRP/oxLDL/b2GPI complex value (30 ng/ml of apoB equivalent) was not affected by addition of excess amounts of oxLDL/b2GPI complexes (300 ng/ml of apoB equivalent, expected as pathophysiological concentration) in the CRP/oxLDL/b2GPI ELISA. The ELISA for oxLDL/ b2GPI complexes was previously described elsewhere (10).
In the present study, the clinical significance of circulating CRP/oxLDL/b2GPI, oxLDL/b2GPI, and CRP/ oxLDL complexes was assessed by ELISA. Cutoff values for these ELISAs were analyzed by two methods: a receiver operating characteristic (ROC) curve using DM patients, and by calculating one to five standard deviations (SDs) above mean values using 48 healthy subjects. For the ROC analysis, atherosclerosis was diagnosed according to IMT measurements. A mean 1 3 SD of the healthy subjects was used as cutoff value. One U/ml was defined as the mean 1 3 SD of controls. The assay was reproducible, with intraand inter-assay coefficients of variation not exceeding 7.0%. The same procedure was used to calculate the CRP cutoff value (0.16 mg/dl).
Diagnostic accuracy of the CRP/oxLDL/b2GPI complex ELISA for atherosclerosis, as compared with CRP nephelometry and the other two complex ELISAs, was assessed by determining sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), odds ratio (OR), and 95% confidence interval (CI) ( Table 1). ROC curves for these three markers were similar (data not shown). In the present study, a subgroup of 69 DM patients (49 patients without atherosclerosis and 20 patients with atherosclerosis) was analyzed. Both CRP/oxLDL/b2GPI and CRP/oxLDL ELISAs showed lower sensitivity (25%) but 100% specificity and PPV, and similar NPV (76.6%) compared with the CRP assay. The measurement of circulating CRP/oxLDL/b2GPI complexes may reliably help to predict the development of atherosclerosis. Thus, a major benefit of these complex ELISAs is that they allow the exclusion of cross-reactive noncomplexed CRP isoforms frequently found in sera of patients with various unrelated acute/chronic infections.
Serum levels of oxLDL/b2GPI and CRP/oxLDL/ b2GPI complexes Figure 4 shows the serum levels of CRP, oxLDL/b2GPI, CRP/oxLDL, and CRP/oxLDL/b2GPI complexes in patients with various diseases and in healthy subjects. Elevated levels of CRP were observed in patients with DM, RA, and pyrogenic diseases [25.6% (32/125), 75.0% (36/48), and 100% (35/35), respectively] (Fig. 4A). Statistical significance was observed between DM and healthy subjects, RA and healthy subjects, and DM and RA patients. OxLDL/b2GPI complexes were found in 44.0% of DM patients (55/125) but not in healthy subjects, RA patients, or patients with pyrogenic diseases (Fig. 4B). Seven DM patients had high levels of CRP/oxLDL (5.6%) and nine had high levels of CRP oxLDL/b2GPI complexes (7.2%). Five DM patients had both (Fig. 4C, D). Although none of the patients in the other groups showed high levels of CRP/oxLDL/b2GPI   complexes, two RA patients had slightly elevated levels of CRP/oxLDL complexes ( just above the cutoff value). CRP/ oxLDL/b2GPI and CRP/oxLDL complex levels between DM and healthy subjects, DM and RA, and DM and patients with pyrogenic diseases were statistically significant.
Sixty-nine DM patients, a subpopulation of 125 DM patients described in Materials and Methods, were divided into two groups according to the presence (AT1) or absence (AT2) of atherosclerosis (AT1 with IMT .0.75 mm, n 5 20 and AT2 with IMT ,0.75 mm, n 5 49). Serum levels of CRP, CRP/oxLDL and CRP/oxLDL/b2GPI complexes but not oxLDL/b2GPI complexes were significantly higher in AT1 than those in AT2 (Fig. 4).
There was no correlation between serum levels of oxLDL/ b2GPI and CRP/oxLDL/b2GPI complexes (Fig. 5A). In contrast, strong correlation between CRP and CRP/ oxLDL/b2GPI complexes (Fig. 5B), CRP and CRP/oxLDL (Fig. 5C), and CRP/oxLDL and CRP/oxLDL/b2GPI (Fig. 5D) were observed in DM patients. Thus, CRP/ oxLDL and CRP/oxLDL/b2GPI found in DM patients may be a part of the hsCRP currently used as a predictive marker for CVD. Further, these correlations seem to indicate that CRP/oxLDL complexes in DM patients were mostly complexed with b2GPI.

Detection of CRP/oxLDL/b2GPI complexes in serum samples by immunoblot analysis
To confirm the presence of oxLDL complexes containing CRP and b2GPI, nondenaturing PAGE (native PAGE without SDS)/immunoblot analysis was performed in serum samples from healthy subjects, CRP/oxLDL/b2GPI complex-positive DM patients with atherosclerosis, CRPpositive RA patients, and CRP-positive patients with pyrogenic diseases (seven samples from each group). Typical immunoblot patterns from two samples of each group are shown in Fig. 6. Two noncomplexed CRP isoforms were detected in DM and RA patients and those with pyrogenic diseases. In contrast, noncomplexed b2GPI was detected in all serum samples. The complex forms of LDL with CRP and of b2GPI were only detected in the DM sera (at the upper position of the LDL band). As described in Materials and Methods, the gels were run at pH 8.3, much closer to the reported pI of CRP (5.3 and 7.4) (39) and b2GPI (5.0 to 7.0) (40) than those of molecular markers (4.0 to 4.7) used in this experiment. Actually, mobility of CRP and b2GPI in agarose gel electrophoresis (run at pH 8.6) was relatively smaller than that of BSA (pI 4.7) (Fig. 2A). Therefore, molecular sizes of pentameric CRP (120 kDa) and b2GPI (50 kDa) determined by nondenaturing PAGE in this study were not consistent with previous reports.

Co-localization of b2GPI, CRP, and oxLDL in macrophages and smooth muscle cells from atherosclerotic lesions
We investigated the localization of b2GPI, CRP, and oxLDL in atherosclerotic carotid arteries and RA synovial tissues by immunohistochemistry. As shown in Fig. 7, clusters of foamy macrophages and layered smooth muscle cells with lipid droplets were commonly observed in cartoid endarterectomy specimens (Fig. 7A, B). These foamy macrophages and smooth muscle cells strongly expressed scavenger receptors for oxLDL (SR-PSOX) (Fig. 7C). Furthermore, CRP (Fig. 7D), apoB and oxLDL (Fig. 7E, F, respectively), noncomplexed/complexed b2GPI (Fig. 7G), and the complexed form of b2GPI (Fig. 7H) were always detected in these macrophages and smooth muscle cells. The presence of b2GPI in carotid atherosclerotic plaques is consistent with the previous report by George et al. (41).
Macrophage infiltration was also commonly observed in RA synovium (Fig. 8A) and colocalized with oxLDL (Fig. 8C), and some macrophages obviously showed CRP immunoreactivity (Fig. 8D). Interestingly, no expression  3 and 4), RA patients (lanes 5 and 6), or patients with pyrogenic disease (lanes 7 and 8) were run on the nondenaturing PAGE (native PAGE without SDS). The proteins were transferred onto the polyvinylidene difluoride membrane, and immunoblot analysis was performed with rabbit anti-CRP antiphospholipid antibodies, mouse anti-b2GPI monoclonal antibody (MAb), Cof-23, or mouse anti-apoB100 MAb, N2E10. The arrows indicate positions of marker proteins (on the left side) and of LDL, noncomplexed b2GPI or noncomplexed CRP isoforms (on the right side). of SR-PSOX (Fig. 8B) and no presence of b2GPI (Fig. 8E) was observed. In contrast, oxLDL, CRP, and b2GPI were not detected in osteoarthritis synovial tissues (data not shown).

Association with biological markers related to atherosclerosis
To further evaluate the implication of CRP and these oxLDL complexes in atherosclerosis, the association with several biological serum markers was analyzed. As shown in Table 2, serum CRP levels were positively associated with those of sICAM-1 and sVCAM-1. In contrast, oxLDL/b2GPI complexes were associated with T-chol and HbA1c. However, there was no association between CRP/oxLDL/b2GPI complexes and any markers determined in this study. The results support the previous observations that CRP activates endothelial cells (29) and that oxidative stress (i.e., oxidation of LDL and complex formation) in the intima is related to hypercholesterolemia and hyperglycemia.

DISCUSSION
We have recently demonstrated that oxLDL interacts with b2GPI and the presence of oxLDL/b2GPI complexes circulating in patients with atherosclerotic and inflammatory diseases, such as systemic lupus erythematosus, APS, DM, and chronic nephritis (4,6,10). High levels of oxLDL/ b2GPI complexes were also present in atherosclerosis-prone mice with apoe 2/2 and ldlr 2/2 genotypes, especially those fed a high-cholesterol diet (unpublished observations). These observations strongly suggested that increased serum levels of oxLDL/b2GPI complexes could represent a novel and clinically useful serologic marker for the assessment of atherosclerosis. In this study, we found CRP/oxLDL/b2GPI (and CRP/oxLDL) as well as oxLDL/b2GPI complexes in sera of patients with DM. CRP/oxLDL/b2GPI (and CRP/ oxLDL) complexes were particularly present in patients with IMT-diagnosed atherosclerosis, and their levels strongly correlated with hsCRP but not with oxLDL/b2GPI complexes. Interestingly, LDL complexes containing b2GPI were not present in sera of patients with RA or pyrogenic diseases. Routine CRP/hsCRP nephelometry for diagnosing CVD may provide false-positive results. In contrast, in the present study, the CRP/oxLDL/b2GPI complex ELISA showed 100% specificity and positive predictive value due to the lack of reactivity of samples from pyrogenic diseases and RA.
We postulate that CRP/oxLDL/b2GPI complexes are mainly, or possibly, only formed in atherosclerotic lesions based on the following observations: Immunohistochemistry of carotid artery plaques showed colocalization of oxLDL, b2GPI, and CRP with SR-PSOX-positive foamy macrophages and activated/transformed smooth muscle cells. Our previous (10) and present in vitro studies actually demonstrated a stoichiometric interaction between Cu 21 -oxLDL and b2GPI and that stable/non-dissociable complexes of oxLDL/b2GPI and CRP/oxLDL/b2GPI were gradually generated up to 24 h of incubation at 37jC (Fig. 1D-F). OxLDL injected intravenously into experimental animals was quickly removed from the circulation by the liver (half life ?10 min), due to its negative charge (17). Even though high concentrations of b2GPI (about 200 mg/ml) are present in the circulation, it seems impossible that oxLDL could form complexes with circulating b2GPI in such a short period of time. In contrast, a 6-month follow-up study of DM patients indicated that serum levels of these complexes were very steady (data not shown). Therefore, CRP/oxLDL/b2GPI and oxLDL/ b2GPI complexes found in the circulation were probably generated in atherosclerotic lesions, not in the blood stream.
It is now widely accepted that CRP contributes to arterial inflammation and pro-atheroscelerotic phenotypes by upregulating adhesion molecules of endothelial cells, such as ICAM-1 and VCAM-1 (42). Dyslipoproteinemia and hyperglycemia contribute to oxidation of LDL, which leads to the progression of atherosclerosis via different mecha-nisms by activating or damaging endothelial cells with the participation of CRP (43). Two positive associations were observed in the present study: one between serum CRP and sICAM-1/sVCAM-1 levels, and another between oxLDL/ b2GPI complex and T-chol/HbA1c levels, both supporting the mechanisms mentioned above. As shown in Table 2, there were no significant associations of serum CRP/ oxLDL/b2GPI complexes with these pro-atherosclerotic parameters, because CRP/oxLDL/b2GPI complexes may represent a unique combination and a novel marker for the assessment of atherosclerosis severity. These complexes are formed by two distinct mechanisms: CRP generated from vascular inflammation, and oxLDL (oxLDL/b2GPI) generated from oxidative stress within the intima.
The interaction of CRP with oxLDL is considered a key event linked to atherosclerosis. It has been recently reported that monomeric or structurally modified (not pentameric) CRP isoform(s) can bind to apoB and lipid ligands either on native LDL or oxLDL in a calciumdependent manner (24). Interestingly, our native PAGE/ immunoblot analysis also revealed the presence of two noncomplexed CRP isoforms in all CRP/oxLDL/b2GPI complexes-positive DM sera, and one of them may be a spontaneously dissociated mononeric form and/or a structurally modified form (Fig. 6). Another possibility is that the doublet bands come from the pentraxin form of CRP and have different pIs because of their different conformation, as previously reported (39). In contrast, as we previously described elsewhere, the v-carboxyl function at the acyl chain of cholesteryl esters in oxLDL was responsible for the initial b2GPI binding (4,(8)(9)(10). The carboxyl function is a late oxidative product that appears after severe lipid peroxidation of LDL, but not in minimally modified LDL. Taken together, these results indicate that circulating CRP/oxLDL/b2GPI complexes are most probably generated from highly oxidized LDL in atherosclerotic lesions but not from native or minimally modified LDL.
As shown in the present study, significant amounts of noncomplexed isoforms of CRP were detected in sera of DM patients with atherosclerosis, as well as CRP/oxLDL/ b2GPI complexes (immunoblot analysis in Fig. 6). It is generally thought that circulating CRP comes from the liver and that the hepatic production of CRP is mainly upregulated by inflammatory cytokines, such as interleukin-6. However, it has been recently reported that CRP can also be produced locally in atherosclerotic lesions by endothelial cells, smooth muscle cells, and monocytic cells (44)(45)(46). It is still under discussion, but CRP as an autocrine and a paracrine factor probably promotes atherosclerosis by interfering with endothelial cell regulation, altering vascular smooth muscle cell and/or monocyte/macrophage functions (29). As described herein, b2GPI complexes with oxLDL or CRP/oxLDL are most probably generated only in atherosclerotic lesions; however, it is still unclear whether oxLDL is complexed with liver or locally generated CRP.
Finally, atherosclerotic complications have been described in RA patients, but b2GPI complexes with oxLDL or CRP/oxLDL complexes were not detected in the sera from these patients. We demonstrated that oxLDL, b2GPI, and CRP were colocalized with SR-PSOX-positive foamy macrophages and activated or transformed smooth muscle cells in carotid artery plaques, but not with b2GPI in RA synovial tissues. In addition, synovial macrophages did not express scavenger receptor SR-PSOX, nor did they develop into foam cells. This can be interpreted as indicating that synovial macrophages are not sufficiently activated. LDL in RA synovial fluid was slightly more electronegative than LDL from matched plasma samples, possibly representing minimally modified LDL (47,48). Because changes of minimally modified LDL are much less pronounced than those produced by Cu 21 treatment, the finding of minimal cholesterol or cholesteryl ester loading in synovial macrophages was expected. The degree of LDL oxidation and macrophage activation in carotid atherosclerotic plaques, unlike that in synovial tissue, is strong enough to cause the binding of b2GPI. Thus, the development of atherosclerosis in DM patients may occur as a result of a mechanism somewhat different from that in RA patients.
In conclusion, oxLDL complexes containing both b2GPI and CRP are formed under inflammatory and oxidative stress conditions in atherosclerotic lesions and are released into the circulation as electrostatically neutral and stable complexes. Determination of circulating CRP/ oxLDL/b2GPI complexes may be useful in assessing the development of atherosclerosis and/or in diagnosing atherosclerosis. However, the clinical significance of these novel markers (CRP/oxLDL/b2GPI and oxLDL/b2GPI) as well as CRP/oxLDL should be further elucidated in larger clinical studies.