Low Levels of IgM Antibodies against an Advanced Glycation Endproduct–Modified Apolipoprotein B100 Peptide Predict Cardiovascular Events in Nondiabetic Subjects

Atherosclerosis is characterized by retention, aggregation, and oxidation of low-density lipoprotein (LDL) in the arterial wall (1); this activates an inflammatory response and development of atherosclerotic lesions. It has been shown that both innate and adaptive immune responses have important roles in the atherosclerotic disease process (2). Oxidized LDL is one of the major targets of these immune responses. Oxidation of LDL includes oxidation of cholesterol and phospholipids as well as aldehyde (malondialdehyde [MDA] and 4-hydroxynonenal) modifications and fragmentation of apolipoprotein B (apoB) 100 (3). Autoantibodies recognizing different parts of the oxidized LDL particle have been identified in healthy subjects and in patients with cardiovascular disease. Although the exact role of these autoantibodies remains to be fully understood, several studies have shown associations between Abs against oxidized LDL and cardiovascular disease (3).

In particular, IgM Abs against MDA-LDL and oxidized LDL have been associated with less severe atherosclerosis in several but not all studies (4–8). There is evidence from experimental studies that IgM against oxidized LDL has a protective role in atherosclerosis and that this is due to the ability of these Abs to facilitate uptake and removal of apoptotic cells and modified LDL particles (9).

Patients with diabetes have increased risk for cardiovascular disease. Diabetes is associated with increased glycation of many proteins resulting in generation of advanced glycation endproducts (AGEs) (10). Accordingly, glycated apoB100 was shown to be increased in serum of diabetic patients and positively associated to glycated hemoglobin and fasting glucose concentration (11). Similar to oxidized LDL, autoantibodies against glycated LDL have been identified in immune complexes from diabetic patients (12, 13). In addition, IgG immune complexes containing AGE-LDL were associated with increased carotid intima-media thickness (IMT) in patients with type 1 diabetes (14, 15).

Methylglyoxal (MGO) is a highly reactive α-oxoaldehyde forming AGEs. The α-oxoaldehyde metabolites arise from glucose, but can be generated during oxidation of fatty acids (16). We have previously shown that IgM Abs against MGO-modified apoB100 were associated with decreased coronary artery calcium score in patients with type 2 diabetes (17). In this study, we screened a library consisting of 302 peptides comprising the complete apoB100 molecule to identify the immune-dominant epitopes of MGO-modified apoB100. Next, we determined whether autoantibodies against these epitopes could predict cardiovascular events. We found that IgM against MGO–peptide 220 (p220) are inversely associated with plasma glucose and that low levels of IgM against MGO-p220 are associated with an increased risk for cardiovascular events. Interestingly, the association between MGO-p220 IgM and cardiovascular events remained when subjects with diabetes were excluded from the analysis. Furthermore, these Abs were shown to be produced by B1 cells.

Abs against modified apoB100 peptides were screened using a previously described peptide library (18). The library consisted of 302 peptides, each 20 aa long with 5 aa overlap.

Abs against modified or native peptides were measured using ELISA as described previously (17). The apoB100 peptides were coated to microtiter plates (Nunc MaxiSorp), washed, and subsequently MGO-modified by addition of 100 mM MGO in 0.2 M phosphate buffer (pH 7.4). After blocking with Superblock (Pierce), plasma (dilution 1:100) was added and incubated overnight. Bound Abs were detected by biotinylated anti-IgM (ICN Biomedicals) and anti-human IgG (Abcam), followed by alkaline phosphatase-conjugated streptavidin and absorbance at 405 nm was measured. Absorbance values were normalized against a plasma pool present on each plate. The values are presented as relative units (RU). Carboxyethyl-lysine (CEL) epitopes on MGO-modified peptides were measured with ELISA using anti-CEL (KNH-30; Cosmo Bio, Tokyo, Japan) as previously described (17).

Native and MGO-modified peptides were screened for IgM and IgG Ab responses. Two selection criteria were used for further evaluation of peptides: 1) at least a 2-fold increased absorbance value compared with a control peptide 8 and 2) an Ab response against modified-to-native peptide ratio >2.

The Malmö Diet and Cancer (MDC) Study is a prospective cohort study examining the relationship between diet and cancer. Subjects born between 1926 and 1945 and living in Malmö were eligible for inclusion. Between October 1991 and February 1994, every other participant was invited to take part in a substudy focusing on the epidemiology of carotid artery disease (MDC, cardiovascular arm). In the current study, we randomly selected 700 participants, aged 63–68 y (mean age, 65 y), from the cardiovascular arm of MDC cardiovascular (n = 6103) (19). Participants underwent a medical history, physical examination and laboratory assessment as previously described in detail (20). The study was approved by the Regional Ethics Committee in Lund and was conducted in accordance with the Helsinki Declaration. All subjects gave written consent.

Analysis of common and bulb carotid IMT was performed using an Acuson 128 CT system with a 7-MHz transducer as described previously (19). Common carotid artery IMT area was calculated as described by Wendelhag et al. (21) as the difference between the total area inside the adventitia and the lumen area.

We examined the outcome for first cardiovascular events. The procedure for ascertaining outcome events has been described previously (22). Events were identified through linkage of the ten-digit personal identification number of each Swedish citizen with three registries: the Swedish Hospital Discharge Register, the Swedish Cause of Death Register, and the Stroke in Malmö Register. Ascertainment of cases and validity of the registries used have been shown to be high (23). A cardiovascular event was defined as a fatal or nonfatal myocardial infarction (i.e., International Classification of Diseases, 9th Revision [ICD-9]: 410), fatal or nonfatal ischemic stroke (ICD-9: 434), or death attributable to underlying ischemic heart disease (ICD-9: 412 or 414), whichever came first. Participants were followed from baseline examination until first cardiovascular event, emigration from Sweden, death, or until December 31, 2008.

Plasma cytokines were measured by Meso Scale Discovery multiplex technology. Th1/Th2 cells were determined as described previously (20).

Peripheral blood was obtained from healthy donors in heparinized tubes according to protocols approved by the North Shore-LIJ Health System Institutional Review Board. PBMCs were obtained by density gradient separation using lymphocyte separation medium (Cellgro). PBMCs were washed in wash buffer (1 mM EDTA in 1% PBS), and resuspended in cell sorting buffer (0.5% BSA in 1% PBS) to stain cells for cell sorting.

PBMCs were stained with FcR blocking reagent (normal mouse serum) and stained with an Ab mixture consisting of anti-CD19-APC-AF700, anti-CD27-APC, anti-CD43-FITC, anti-CD38-PerCP-Cy5.5, anti-CD20-Pacific Blue, anti-CD3-ECD, anti-CD4-ECD, anti-CD7-ECD, MitoTracker Green (MTG), and IgD-PE in cell sorting buffer, and then stained with Aqua LiveDead reagent to remove dead cells. After staining and washing, cells were resuspended in cell sorting buffer with 10 U/ml DNase. Stained cells were then subjected to sort-purification using a BD Influx cell sorter.

Surface phenotype of sorted cells:

• Transitional: CD3⁻CD4⁻CD7⁻CD19⁺CD20⁺CD27⁻MTG⁺CD38^hiIgD⁺

• Naive: CD3⁻CD4⁻CD7⁻CD19⁺CD20⁺CD27⁻MTG⁻IgD⁺

• Unswitched Memory: CD3⁻CD4⁻CD7⁻CD19⁺CD20⁺CD27⁺IgD⁺

• Switched Memory: CD3⁻CD4⁻CD7⁻CD19⁺CD20⁺CD27⁺IgD⁻

• Double Negative : CD3⁻CD4⁻CD7⁻CD19⁺CD20⁺CD27⁻IgD⁻

• B-1: CD3⁻CD4⁻CD7⁻CD19⁺CD20⁺CD27⁺CD43⁺

The gating strategy is shown in Supplemental Fig. 1.

After sort purification, 1–50 × 10³ cells of each fraction were cultured with CpG oligodeoxynucleotide 2006 (Oligos Etc; Wilsonville, OR; 1 μg/ml), anti-Ig F(ab′)₂ (2.5 μg/ml) and IL-2 (10 ng/ml). Cells were incubated in 5% CO₂ at 37°C for 36 h, after which cell-free supernatants were collected.

Different CpG/anti-Ig/IL-2–stimulated B cell supernatants were used in ELISA. Ninety-six–well microtiter plates were coated with 20 μg/ml of phosphorylcholine (PC)-BSA, p220, MGO-modified p220 (modified on the plate as described above) and incubated at 4°C overnight. All the wells were blocked with 2% fatty acid-free BSA (Sigma-Aldrich) in PBS for 2 h followed by addition and incubation of supernatants from different B cell populations (dilution 1:50) overnight at 4°C. Bound Abs in medium were detected by biotinylated goat anti-human IgM Abs (ICN Biomedicals) for 2 h, followed by streptavidin-alkaline phosphatase for 1 h, and by alkaline phosphatase substrate. The absorbance was measured at 405 nm. Washings between each step were performed with PBS with 0.05% Tween 20. The specificity for IgM against MGO-p220 was analyzed by preincubation of the supernatants to a plate previously coated with p220, MGO-BSA, or MGO-p220. After incubation for 24 h at 4°C the supernatants were transferred to another plate coated with MGO-p220 and Abs reacting with MGO-p220 were determined as described above (Supplemental Fig. 2).

The Student t test or nonparametric test for continuous variables and χ² test for categorical variables were used to examine the difference between groups. Spearman correlation coefficients were used to examine the relationship among continuous variables as appropriate. Kaplan-Meier analysis and Cox proportional hazards regression were used for time-to-event analysis to determine hazard ratios (HR) and 95% confidence intervals (CIs) between Ab levels (MGO-p220 IgM and MGO-p220 IgG) in tertiles and incident cardiovascular events during 15-y follow-up. The first model was unadjusted. In a second model, we adjusted for traditional cardiovascular risk factors (age, gender, diabetes, smoking, systolic blood pressure, LDL-to–high-density lipoprotein (HDL) ratio, C-reactive protein (CRP), fasting glucose, treatment with blood pressure-lowering, lipid-lowering or anti-diabetes medication and history of myocardial infarction or stroke at baseline). Plots of the hazard function in different groups over time did not indicate that the proportional-hazards assumption was violated. Statistical analyses were performed using IBM SPSS Statistics 20. A two-tailed p value < 0.05 was considered statistically significant.

To find key epitopes recognized by MGO autoantibodies, we screened an apoB100 peptide library, modified with MGO, for IgM and IgG Abs recognizing MGO-apoB100 peptides. We searched for MGO-apoB100-peptides targeted by relatively high levels of Abs that also exhibited at least twice the absorbance compared with Ab levels against the corresponding native peptide. Five peptides (p211, p214, p220, p272, and p279) fulfilled both criteria for IgM binding, whereas four peptides (p14, p210, p211, and p220) fulfilled the same criteria for IgG (data not shown). Thus, two peptides fulfilled the criteria for both IgM and IgG: p211 and p220. Peptide p211 shares 5 aa with p210, which has been extensively studied in immune responses against both the native and MDA-modified form (24, 25), prompting us to continue with p220.

Peptide p220 (amino acids 3286-3305; LKLSL PHFKE LCTIS HIFIP) contains two lysine residues (position 2 and 9), which may be modified by MGO generating CEL epitopes (26). Accordingly, CEL epitopes were detected on MGO-p220 by ELISA (Supplemental Fig. 3A). Anti–MGO-p220 Abs were found to be specific, and the binding could not be competed with another MGO-modified apoB100 peptide (Supplemental Fig. 3B, 3C).

Next, we measured IgG and IgM recognizing MGO-p220 in 700 subjects enrolled in the MDC Cohort. Baseline characteristics of the study population are presented in Table I. During follow-up, 139 subjects had an incident cardiovascular disease event of whom 84 had a coronary event and 55 had an ischemic stroke.

Subjects with incident cardiovascular event during follow-up had lower levels of anti-MGO-p220 IgM Abs, whereas no association was observed for IgG against MGO-p220 (Fig. 1, Table I). Cardiovascular risk factors and their relation to Ab tertiles of IgM against MGO-p220 are shown in Supplemental Table I. In a Cox regression model, subjects in the lowest compared with the highest tertile of anti-MGO-p220 IgM Ab level had an increased risk for incident cardiovascular events after taking age, gender, diabetes, smoking, systolic blood pressure, LDL/HDL ratio, CRP, fasting glucose, treatment with blood pressure-lowering, lipid-lowering or anti-diabetic medication, and history of cardiovascular event at baseline into account (HR 2.07, 95% CI 1.22–3.50, p = 0.004 for trend; Table II). In a sensitivity analysis this risk increase also remained after excluding subjects with prevalent diabetes (HR 2.51, 95% CI 1.37–4.61, p = 0.002 for trend). There was no difference in levels of IgM MGO-p220 between diabetic and nondiabetic individuals (99 [72–136] RU versus 109 [75–153] RU; p = not significant). There were no significant associations between baseline carotid IMT and anti-MGO-p220 IgM or IgG (data not shown).

Methylglyoxal, and downstream adducts like CEL, are believed to be formed in vivo by both glycoxidation and lipoxidation (16). Strikingly, IgM Abs against MGO-p220 were inversely related to blood glucose but not to LDL, cholesterol, or oxidized LDL (Table III). In accordance with our previous study focusing on MGO-modified apoB100 IgM, IgM against MGO-p220 was higher in women than in men (113 [77–153] versus 100 [70–142]; p < 0.05), whereas no gender difference was seen for anti-MGO-p220 IgG.

Previously, Sämpi et al. demonstrated that IgM Abs against MDA-LDL correlate with the plasma level of IL-5 (27). These Abs were considered to be natural Abs because IL-5 administration in mice increases IgM against MDA-LDL and PC (28). In the current study, IgM Abs against MGO-p220 were strongly associated to Th2 cytokines in plasma (IL-4, IL-5, IL-10, and IL-13), IL-2, and IL-12p70, whereas weaker correlations with markers of inflammation (IL-1β and TNF-α) and IFN-γ were observed (Table IV). No associations between IgM against MGO-p220 and Th1 (CD3⁺CD4⁺IFN-γ⁺) or Th2 cells (CD3⁺CD4⁺IL-4⁺) were observed (Table IV). In contrast to IgM, anti-MGO-p220 IgG correlated with inflammatory cytokines TNF-α, IL-1β, and IL-8 as well as with IL-10. In addition, IgG, but not IgM, against MGO-p220 correlated with blood counts of Th2 cells (Table IV).

To test whether IgM against MGO-p220 are natural Abs secreted by B1 cells, we purified different B cell populations from four healthy donors and measured binding of secreted Abs to MGO-p220. B1 cells from three of the four donors produced high amounts of IgM binding to MGO-p220 (Fig. 2), whereas transitional, naive, unswitched, switched, or double-negative memory B cells produced no or lesser amounts of these Abs. Specificity of the Abs secreted from B1 cells were established by inhibition assays (Supplemental Fig. 2). IgM from B1 cells reacted with PC-BSA, which was used as a positive control, but only to a minor extent with native p220. The fourth donor exhibited low titers of both MGO-p220 and PC-BSA IgM (data not shown).

Accumulation and oxidation of LDL in the arterial intima is considered to have an important role in the development of atherosclerosis. Short-chain aldehydes such as MDA and MGO, formed by lipoxidation and glycoxidation, respectively, modify the apoB100 protein of the LDL particle and generate neoepitopes recognized by the immune system. In this study, we investigated the relationship between autoantibodies recognizing an MGO-modified apoB100 peptide and cardiovascular events in 700 middle-aged individuals followed for a mean of 15 y. We show that low levels of IgM, but not IgG, recognizing an epitope of MGO-modified apoB100 are associated with increased risk for incident cardiovascular events. This association remained significant after adjusting for cardiovascular risk factors.

Several studies have inferred a relationship between IL-5 and natural IgM Abs. IL-5 has been shown to promote release of natural Abs in mice (28) and is viewed as a cytokine capable of promoting release of natural Abs. A recent study demonstrated a relationship between IL-5 and anti-MDA-LDL IgM Abs, strengthening this claim. In addition, IL-5 levels were associated with less atherosclerotic disease (27). In our study, MGO-p220 IgM levels were significantly associated with plasma IL-5 levels, supporting the notion that these Abs could be natural Abs recognizing epitopes in modified LDL. Indeed, IgM from purified B1 cells reacted against MGO-p220. Moreover, the highest levels of IgM against MGO-p220 were produced by B1 cells, whereas only lesser amounts were produced by B2 cells.

Transfer of B1a cells to splenectomized mice reduced atherosclerosis, whereas no effect was seen for transfer of B1a cells without IgM secreting capability, demonstrating a protective role for natural Abs (29). This result is in line with our finding that lower levels of MGO-p220 IgM are associated with increased risk for cardiovascular events. Although the current study does not clarify whether IgM against MGO-p220 plays a protective role in cardiovascular disease or is merely a marker, there are several possibilities of how these autoantibodies could influence the risk for developing cardiovascular events. Vulnerable lesions are characterized by large lipid cores with foam cells in combination with thin fibrous caps. MGO-modified LDL has been shown to induce foam cell formation in vitro (30, 31). It has also been shown that Abs recognizing oxidized LDL inhibit uptake of oxidized LDL by the macrophage scavenger receptor in vitro and reduces atherosclerosis in mice (32, 33). Thus, autoantibodies against MGO-apoB100 could be protective by blocking the uptake of MGO-LDL in macrophages, thereby reducing plaque inflammation. Approximately 20% of cardiovascular events are associated with plaque erosion instead of plaque rupture. Because glycated LDL has been shown to both induce apoptosis in endothelial cells (34) and to increase platelet aggregation (35) and plasminogen activator inhibitor levels (36), Abs against MGO-apoB100 may protect against plaque erosions and thrombus formation. IgM recognizing MGO-p220 could possibly also mediate atheroprotection by inhibiting binding of MGO-modified LDL to the receptor for AGE (RAGE) present on vascular endothelial cells, smooth muscle cells and macrophages. Interaction of AGE-modified proteins to RAGE results in activation of the proinflammatory transcription factor NF-kβ and increased atherosclerosis. Interestingly, RAGE is proatherogenic also in nondiabetic LDLr^−/− mice, and oxidized LDL was proposed as ligand for RAGE (37).

Competition assays show that IgM binding to MGO-p220 could not be inhibited by native p220 or by another MGO-modified apoB100 peptide (MGO-p214). Similarly, binding of IgM from B1 cells to MGO-p220 was competed with MGO-p220, but not with native p220 and only to a lesser extent with MGO-BSA. These results indicate that the epitope recognized by the Abs includes both peptide and MGO-modification. However, we cannot exclude that these Abs can cross-react with p220 with other oxidative modifications induced during LDL oxidation or LDL glycation. For example, MDA is formed during lipid oxidation and is another reactive aldehyde reacting with amino groups on, for example, lysine residues. Interestingly, the levels of anti-MGO-p220 IgM Abs were inversely associated with blood glucose. A possible explanation for this finding is that glucose levels are related to the amount of glycemic stress and generation of MGO. Excess MGO in lesions reacting with LDL would generate MGO-p220 epitopes that bind Abs, thus reducing anti-MGO-p220 levels in plasma. In addition, the lack of relationship between anti-MGO-p220 IgM and oxidized LDL suggests that MGO-p220 IgM is primarily binding to structures formed by glycoxidation, rather than lipoxidation.

In conclusion, we show that B1 cells produce IgM that recognize an MGO-modified peptide sequence in apoB. Subjects with low levels of these Abs have an increased risk of developing cardiovascular events. This association is present also in nondiabetic subjects.

The authors have no financial conflicts of interest.