Relationship of lipoprotein-associated phospholipase A2 and oxidized low density lipoprotein in carotid atherosclerosis.

Plasma levels of lipoprotein-associated phospholipase A2 (Lp-PLA2) and oxidized low density lipoprotein (oxLDL) have been identified as risk factors for cardiovascular disease. Lp-PLA2 is the sole enzyme responsible for the hydrolysis of oxidized phospholipids on LDL particles in atherosclerotic plaques. We have studied the relationship between Lp-PLA2 and oxLDL in carotid endarterectomy (CEA) tissues and in matched plasmas. In extracts from CEA anatomical segments, the levels of oxLDL were significantly associated with the levels of Lp-PLA2 protein (r = 0.497) and activity (r = 0.615). OxLDL and Lp-PLA2 mass/activity were most abundant in the carotid bifurcation and internal segments where plaque was most abundant. In extracts from CEA atheroma, the levels of oxLDL and Lp-PLA2 were significantly correlated (r = 0.634). In matched plasma and atheroma extracts, the levels of Lp-PLA2 were negatively correlated (r = − 0.578). The ratio of Lp-PLA2 to oxLDL was higher in atheromatous tissue (277:1) than in normal tissue (135:1) and plasma (13:1). Immunohistochemical experiments indicated that in plaques, oxLDL and Lp-PLA2 existed in overlapping but distinctly different distribution. Fluorescence microscopy showed both oxLDL and Lp-PLA2 epitopes on the same LDL particle in plasma but not in plaque. These results suggest that the relationship between Lp-PLA2 and oxLDL in the atherosclerotic plaque is different from that in the plasma compartment.


Determination of Lp-PLA 2 activity
Lp-PLA 2 activity was measured by a colorimetric activity method (diaDexus). The assay was performed in a 96-well microtiter plate, using 1-myristoyl-2-(4-nitrophenylsuccinyl)-phosphatidylcholine as substrate. The level of Lp-PLA 2 activity in nmol/ min/mg was calculated from the slope of the kinetic absorption curve (410 nm), using a standard conversion factor derived from a p -nitrophenol calibration curve. The assay had a coeffi cient of variation of <4% for duplicate measurements and a sensitivity down to <1.0 nmol/min/mg.
Although there are at least 19 mammalian enzymes that possess PLA2 activity, the Lp-PLA 2 (platelet-activating factoracetylhydrolase) family of phospholipases contains only four enzymes that exhibit unique substrate specifi city toward plateletactivating factor and/or oxidized phospholipids ( 31 ). Calcium chelators in blood and tissue collection, procurement, and assay were used to assess only calcium-independent lipase activities.

Determination of OxLDL levels
The quantitation of oxLDL was performed with the Mercodia Oxidized LDL ELISA kit (Mercodia, Uppsala, Sweden). This microtiter plate capture ELISA utilizes the monoclonal antibody mAb-4E6, which has been previously described ( 32 ). Tissue extract samples were diluted 1:81 compared with 1:6561 used for plasma; otherwise, the protocol was performed as described ( 32 ). Copper oxidized LDL carefully prepared by the method of Lopes-Virella et al. ( 33 ) was used as standard.

Immunohistochemistry
Detailed protocols are found in the supplemental materials section, including the methods for immunohistochemical analysis of plasma oxLDL, oxLDL controls, and fl uorescent microbead standards.

Statistics
The sample set for all experiments reported in Fig. 1 consisted of 16 CEA tissue samples, each cut into four anatomical segments (n = 64). The experiments reported in Figs. 2 and 3 were performed using an additional sample set of 22 CEAs. These CEAs were independent of the fi rst sample set and were matched to homologous plasma from individual subjects. The 20 matched plasma/CEA pairs were obtained over a period of 16 months. Data are expressed as mean ± SD. When comparing two groups, a Mann-Whitney ( t -test) was used. Differences between groups (>2) were evaluated by one-way ANOVA, with Newman-Keuls posttest. A value of P < 0.05 was considered signifi cant. Normal probability plots were used to evaluate normality of data and determine appropriate statistical analyses. Correlation analysis (Pearson or Spearman) was followed by linear regression methods with 95% confi dence. Statistical analysis was performed using GraphPad Prism Software (San Diego, CA).

Distribution of Lp-PLA 2 and oxLDL in anatomic segments of CEA tissue
CEA tissues (n = 16) were collected, imaged, and cut into common, bifurcation, internal, and external segments ( Fig. 1A ). Segments were halved for frozen sectioning and total protein isolation (n = 64). The internal segments had the greatest abundance of Lp-PLA 2 protein , followed by the bifurcation, common, and external segments ( Fig.  1B ). The bifurcation and internal branches possessed the Lp-PLA 2 activity, the prediction of CVD is increased ( 21 ). Lp-PLA 2 is secreted predominantly by macrophages ( 22,23 ). Its expression and secretion signifi cantly increase as human monocytes differentiate into macrophages and dramatically increase during activation of macrophages in the atherosclerotic lesion ( 24 ). Recently, Lp-PLA 2 has been localized to necrotic cores and infl ammatory areas of coronary vulnerable plaques ( 25 ). In plasma, Lp-PLA 2 is thought to circulate bound to LDL ( ‫ف‬ 80%) and HDL ( ‫ف‬ 20%), although hypercholesterolemia alters this dynamic ( 26 ). Lp-PLA 2 is reported to bind to the C-terminal segment of apoB-100 on LDL ( 27 ), but the relationship between Lp-PLA 2 and LDL or oxLDL in atherosclerotic lesions is not well understood.
It is currently unknown if levels of Lp-PLA 2 and oxLDL in plasma correlate to levels found in atherosclerotic lesions. Nevertheless, it is reasonable to postulate that the plaque burden in each atherosclerotic artery contributes to the circulating levels of each biomarker. Lavi et al. ( 28 ) have demonstrated a net increase of Lp-PLA 2 levels in blood that traverses a coronary vascular bed that contains signifi cant atherosclerotic plaque. This observation suggests that a primary source of Lp-PLA 2 in blood is macrophage-rich plaque (22)(23)(24). Measuring the distribution of Lp-PLA 2 and oxLDL in specifi c areas of the carotid atheroma should help to evaluate this possibility.
Accordingly, the objectives of this study were i ) to quantify Lp-PLA 2 (mass/activity) and oxLDL in carotid atherosclerotic tissue and plasma; ii ) to determine the association of plasma levels of Lp-PLA 2 and oxLDL with matched carotid tissue levels; and iii ) to determine the macroscopic distribution of Lp-PLA 2 and oxLDL as well as their microscopic association in atheroma and plasma.

Tissues
This study was approved by the Baylor College of Medicine institutional review board for human research, and all subjects participating in this study provided informed consent. Carotid endarterectomy (CEA) tissues were excised from >40 patients undergoing unilateral carotid endarterectomy procedures. Common, bifurcation, internal, and external carotid segments were cut from these CEA tissues. Atheromatous segments were dissected from these tissues based upon gross pathological features. Segments were digitally imaged (Leica DC300) and mounted in OCT blocks for frozen sectioning. The remaining tissue was recovered for total protein extraction. Blood samples were drawn from patients who had undergone CEA to obtain plasma, thereby providing tissue/blood matched pairs. Fasting bloods were generally collected during follow-up visits an average of 4.5 months after surgery.

Determination of Lp-PLA 2 protein mass
Plasma levels of Lp-PLA 2 protein were measured using the dual monoclonal ELISA diaDexus PLAC TM assay. This enzyme immunoassay utilizes two anti-human monoclonal Lp-PLA 2 specifi c antibodies, clones 2C10 and 4B4, and has been previously reported ( 29,30 ). Concentrations of Lp-PLA 2 (ng/ml) in tissue extracts and plasma were normalized to total protein concentration ( g/ l) and reported as g/ g total protein.

Distribution of Lp-PLA 2 and oxLDL in the plasma
Lp-PLA 2 (ng/ml) and oxLDL (mU/L) were measured in plasma from subjects from whom CEA tissues had been obtained. The lipid profi les and risk factors for 20 subjects are listed in Table 1 . One arbitrary unit of oxLDL immunoreactivity is equivalent to 300 ng ( 32 ). After unit conversion and normalization (µg/µl total protein), it was possible to compare the abundance of both factors in plasma to levels in atherosclerotic lesions. Matched plasma samples contained signifi cantly more Lp-PLA 2 than atherosclerotic ( P < 0.05) and normal ( P < 0.001) tissues contained ( Fig. 3A ). Additionally, oxLDL is more abundant in plasma than in either atherosclerotic or normal tissue ( P < 0.001; Fig. 3B ). During optimization of plasma ELISA assays to measure tissue extracts, it was necessary to modify dilutions of samples to fi t into the dynamic range of the assay standard curves. The normal dilution of plasma samples required for analysis with the Mercodia oxLDL kit was 1:6561; for tissue samples, a dilution of 1:81 was optimal.
greatest Lp-PLA 2 activities ( Fig. 1C ). The distribution pattern of oxLDL ( Fig. 1D ) was similar to that of Lp-PLA 2 protein and activity. The internal and bifurcation segments both possessed signifi cantly more oxLDL, as assessed by mAb-4E6, than the common and external branches of the CEA tissue ( P < 0.05). We observed that not only is the distribution of both factors similar in CEA tissue segments, but their mass abundances are also correlated ( r = 0.497, P < 0.0001, n = 64; Fig. 1E ). Consequently, it was not surprising that Lp-PLA 2 enzymatic activity was signifi cantly coupled to oxLDL abundance ( r = 0.6147, P < 0.0001; Fig. 1F ). Thus, the distribution of Lp-PLA 2 and oxLDL was greatest in the bifurcation and internal segments where atherosclerotic lesions were most prevalent.

Distribution of Lp-PLA 2 and oxLDL in normal and atherosclerotic regions of CEA tissues
CEA tissues (n = 22) were dissected into atherosclerotic and normal regions documented by digital imaging ( Fig.  2A ). These images were used to categorize these regions by their qualitative features, mainly plaque size and composition. Lp-PLA 2 was signifi cantly more abundant in atherosclerotic lesions than in normal tissue ( P < 0.05; Fig.  2B ). Similarly, oxLDL was also signifi cantly more abun- PLA 2 and oxLDL in plasma to that in matched CEA tissue extracts. Signifi cantly, in the case of Lp-PLA 2 , the plasma and tissue levels were negatively correlated ( r = Ϫ 0.578; Fig. 3D ). In the case of oxLDL, plasma and tissue levels were positively borderline correlated ( r = 0.482; Fig. 3E ).

Localization of Lp-PLA 2 and oxLDL in CEA atherosclerotic lesions shows overlapping but distinctly different distribution patterns
Deconvoluted microscopy and immunochemistry were used to localize Lp-PLA 2 and oxLDL in CEA tissues. Three types of localization were observed: distal, proximal, and colocalization. Distal localization is illustrated in Fig. 4B and C in which Lp-PLA 2 is highly localized to the lesion shoulder, whereas ox-LDL is broadly spread throughout the entire lesion area. Proximal localization is illustrated in Fig. 4F and G in which both species are localized to an underlying segment. Although Lp-PLA 2 and oxLDL had overlapping patterns, they were not coincident. Authentic colocalization is illustrated in Figure 4F-H in which green (oxLDL) combines with red (Lp-PLA 2 ) to give a yellow color (white arrow). This colocalization of these two species was limited to a relatively small area, and the majority of oxLDL and Lp-PLA 2 occupied similar areas but did not actually colocalize. The differences between Fig. 4D and H exemplify the variable distributions of oxLDL and Lp-PLA 2 in image sets acquired from CEA tissues.

Lp-PLA 2 colocalizes to individual oxLDL particles in plasma but not in atheroma
We sought to investigate the interaction between Lp-PLA 2 and oxLDL by employing immunohistochemistry. Using high magnifi cation (100×) and fl uorescence (de-These data were obtained to determine the stoichiometric relationship between oxLDL and Lp-PLA 2 in the plasma and tissue compartments. The average protein weight ratio of Lp-PLA 2 to oxLDL was ‫ف‬ 13:1 in plasma. This is signifi cantly less than the ratio of 204:1, which was observed in the total CEA tissue ( P < 0.01; Fig. 3C ). This ratio is substantially increased to 277:1 in atherosclerotic areas of the tissue ( P < 0.001; Fig. 3C ). The normal tissue areas possessed a ratio of 135:1. Hence, there is more Lp-PLA 2 (ng) per oxLDL (ng) in the atherosclerotic lesion than in the plasma. The ratio appears to be appreciably different in each compartment.

Relation of levels of Lp-PLA 2 and oxLDL in plasma and matched atherosclerotic CEA tissue
It would be very desirable for the plasma levels of Lp-PLA 2 and oxLDL to refl ect their abundances in atherosclerotic lesions, thereby making them possible tools for noninvasive tracking of atherosclerotic plaque burden. We evaluated this possibility by comparing the levels of Lp-  validated by confi rming the positive staining of individual oxLDL particles using an oxLDL control ( Fig. 5G ) and a 20 nm fl uorescent microsphere standard ( Fig. 5H ). Discussion concerning the optical measurements can be found in the supplemental materials section.

Lp-PLA 2 colocalizes with large oxLDL-derived particles in carotid atherosclerotic lesions
In CEA tissue sections, immunohistochemistry was used to observe positively stained oxLDL particles of variable sizes. Individual particles ( ‫ف‬ 20 nm) and particles of much greater size ( ‫ف‬ 30 nm to 3 µm in diameter) were identifi ed. Whether these larger particles are oxLDL aggregates or oxLDL-derived liposomes is unclear. Positive staining for Lp-PLA 2, was observed on large DNA-free spheres ( Fig.  5I, J ). In Fig. 5J , the enlarged image shows the larger Lp-PLA 2 -localized spheres (red arrow). It was possible to rotate the image and positively identify Lp-PLA 2 on the surface of these spheres (data not shown). Signifi cantly, we were able to determine that these spheres were positive convolution) microscopy, we observed two distinct interactions between Lp-PLA 2 and oxLDL in plasma and in the atheroma ( Fig. 5 ). In plasma, we localized Lp-PLA 2 (red) to the surface of individual oxLDL (green) particles ( Fig.  5A, B ). Figure 5B is an enhanced image of a plasma individual oxLDL particle with Lp-PLA 2 colocalized to its surface. Conversely, in CEA tissue sections, actual colocalization of these two factors was not observed. Microscopy analysis revealed that often individual oxLDL particles were localized to lipid pools throughout the plaque as well as the shoulder and cap regions ( Fig. 5C-E ). In contrast, Lp-PLA 2 localized to the edges and adjacent matrix around these lipid pools and oxLDL deposits. We did not fi nd true colocalization of Lp-PLA 2 to the surface of individual ox-LDL particles within the atherosclerotic tissue sections. Figure 5F illustrates the lack of Lp-PLA 2 staining on the surface of a free-standing oxLDL particle in CEA tissue. To our knowledge, this is the fi rst report showing immunohistochemical (fl uorescence microscopy) identifi cation of individual oxLDL particles. These measurements were oxLDL levels were signifi cantly greater in diseased than nondiseased areas. Again, we observed signifi cant correlation between Lp-PLA 2 and oxLDL in this analysis ( r = 0.635; Fig. 2D ). It is noteworthy that the weight ratio of Lp-PLA 2 to oxLDL in plasma (13:1) was signifi cantly different from their ratio in the atheroma (277:1), a 20-fold difference.
Elevated plasma levels of Lp-PLA 2 and oxLDL are well documented as prognostic risk markers for cardiovascular disease ( 7-10, 29, 36, 37 ). This association suggested that the plasma levels of these analytes might be associated with their levels in atherosclerotic plaques. For example, we previously observed that the plasma level of the cardiovascular risk factor Lp[a] is associated with its levels in lesionbearing tissues, such as resected bypass vein grafts ( 38 ) and aortic aneurysms ( 39 ). In this study, however, the plasma levels of oxLDL are only borderline associated with carotid atheroma levels. This observation suggests that the carotid atheroma alone does not make a proportionate contribution of oxLDL to the total level in plasma. A moderate contribution of oxLDL from carotid plaques may be overshadowed by more abundant contributions from larger lesions, such as those in the aorta.
Plasma Lp-PLA 2 levels represent risk related to the total burden of vascular infl ammation, and not just the burden arising from a particular atheroma. Likewise, the plasma Lp-PLA 2 level refl ects contributions from not just a single lesion but from multiple atherosclerotic sites. Accordingly, we anticipated that the plasma levels of this biomarker might be greater than the level in a single specifi c tissue site, such as the carotid plaque. Indeed, we observed that for oxLDL, even though they were larger than authentic oxLDL. Our multiplexed images show the localization of Lp-PLA 2 to the surface of oxLDL-positive spheres ( Fig.  5K-O ). In Fig. 5L , the red arrow points to Lp-PLA 2 (red) on the surface of what appears to be a large oxLDL-derived particle (green). We observed that not all apoB-100 particles were positive for oxLDL (4E6), thus illustrating specifi city of the 4E6 antibody to recognize only a particular fraction (oxidized) of the LDL pool ( Fig. 5P ). The red arrow points to apoB-100 particles only, and the green arrow identifi es apoB-100-oxLDL positive particles. Taken together, these fi ndings indicate that the relationship between Lp-PLA 2 and oxLDL in the atheroma is very different from that in the plasma.

DISCUSSION
The goal of this study was to characterize the association and relationship between Lp-PLA 2 and oxLDL in atherosclerotic lesions and plasma. The data presented offer new insights into the association between this centrally important enzyme and its substrate. In this study, we observed that Lp-PLA 2 (mass and activity) and oxLDL were distributed to the same anatomical segments of the carotid artery ( Fig. 1A-D ). The levels of these factors in these segments were signifi cantly correlated ( r = 0.497 and 0.615). Lp-PLA 2 and oxLDL were found in greatest abundance at the bifurcation and internal segments, regions known for having the most rupture-prone atherosclerotic lesions ( 34,35 ). Accordingly, we measured Lp-PLA 2 and oxLDL in atherosclerotic areas of the CEA tissues. Both Lp-PLA 2 and dized phospholipid epitopes. This antibody targets both oxLDL (in vitro) and/or malondialdehyde-modifi ed LDL (in vivo). These epitopes are reactive aldehyde-modifi ed lysines of apoB-100 that are products of lipid peroxidation. Nonetheless, the 4E6 antibody does not recognize all modifi ed forms of LDL (e.g., acetylated or oxPC). Furthermore, the 4E6 antibody recognizes only one epitope per LDL particle in an environment containing potentially numerous oxidized lipids. The localization studies and particle histology experiments lend support to the view that the atheroma in the vessel wall is a major site for production of Lp-PLA 2 and oxLDL, especially in advanced atherosclerosis where there are lesions of greater number and size. These studies show that Lp-PLA 2 and oxLDL are found in greatest abundance at the atheroma shoulder region and lesion cap, as has also been observed by Mannheim et al. ( 40 ). Both areas Lp-PLA 2 is more abundant in plasma than in CEA tissues when normalized to total protein ( Fig. 3A, B ). Lavi et al. ( 28 ) found that production of Lp-PLA 2 in a single vascular bed (i.e., coronary arteries) correlated with coronary atheroma volume determined by intravascular ultrasound by measuring the gradient in Lp-PLA 2 concentration from the coronary ostium to the coronary sinus. On the basis of this observation, we postulated that plasma and tissue levels of Lp-PLA 2 would be positively associated. Surprisingly, the levels were negatively correlated ( Fig. 3D ; r = Ϫ 0.578). Signifi cantly, Mannheim et al. ( 40 ) observed a negative correlation between plaque expression of Lp-PLA 2 and plasma HDL levels. Elucidation of how these sources of Lp-PLA 2 are inversely related will require further investigation.
We used the 4E6 antibody for detection of an oxidized form of LDL ( 10 ). The binding of this antibody to oxidized LDL is not necessarily a measure of binding to oxi- would likely be involved in lipid and protein infl ux and effl ux. However, the mechanism and/or rate of trans-intimal movement may vary among different permeating species. Studies directed toward measuring lipoprotein infl ux using optical coherence tomography are currently underway ( 41 ). Nonetheless, previous studies give added credence to the atheroma-epicenter model for production of Lp-PLA 2 and oxLDL. The LDL particle undergoes considerable modifi cation after becoming trapped in the subintimal space ( 5,6 ). It is unlikely that more than minimal amounts of LDL are oxidized in the plasma compartment due to its extensive reducing capacity (e.g., from reduced glutathione and paraoxonase). Additionally, the expression and secretion of Lp-PLA 2 dramatically increase during monocyte migration into the vessel wall and subsequent differentiation of monocytes to macrophages ( 24 ).
Signifi cant strengths of this study are i ) the use of anatomically and pathologically separate carotid tissue segments, ii ) tissue extraction and quantitation of oxLDL and Lp-PLA 2 , iii ) direct comparison of tissue and plasma levels of oxLDL and Lp-PLA 2 , and iv ) high-resolution imaging (colocalization) of Lp-PLA 2 and oxLDL in the carotid atheroma and plasma. Our results suggest that the relationship between Lp-PLA 2 and oxLDL in the carotid atheroma is distinctly different from what is observed in plasma.