Low doses of LPS and minimally oxidized LDL cooperatively activate macrophages via NF-kappaB and AP-1: Possible mechanism for acceleration of atherosclerosis by subclinical endotoxemia

Abstract

Rationale

Oxidized low-density lipoprotein (LDL) is an important determinant of inflammation in atherosclerotic lesions. It has also been documented that certain chronic infectious diseases, such as periodontitis and chlamydial infection, exacerbate clinical manifestations of atherosclerosis. In addition, low-level but persistent metabolic endotoxemia is often found in diabetic and obese subjects and is induced in mice fed a high-fat diet.

Objective

In this study, we examined cooperative macrophage activation by low levels of bacterial LPS and by minimally oxidized LDL (mmLDL), as a model for subclinical endotoxemia-complicated atherosclerosis.

Methods and Results

We found that both in vitro and in vivo, mmLDL and LPS (Kdo2-LipidA) cooperatively activated macrophages to express pro-inflammatory cytokines Cxcl2 (MIP-2), Ccl3 (MIP-1alpha), and Ccl4 (MIP-1beta). Importantly, the mmLDL and LPS cooperative effects were evident at a threshold LPS concentration (1 ng/ml) at which LPS alone induced only a limited macrophage response. Analyzing microarray data with a de novo motif discovery algorithm, we found that genes transcribed by promoters containing an AP-1 binding site were significantly upregulated by co-stimulation with mmLDL and LPS. In a nuclear factor-DNA binding assay, the cooperative effect of mmLDL and LPS co-stimulation on c-Jun and c-Fos DNA binding, but not on p65 or p50, was dependent on mmLDL-induced activation of ERK1/2. In addition, mmLDL induced JNK-dependent derepression of AP-1 by removing the corepressor NCoR from the chemokine promoters.

Conclusions

The cooperative engagement of AP-1 and NF-kappaB by mmLDL and LPS may constitute a mechanism of increased transcription of inflammatory cytokines within atherosclerotic lesions.

INTRODUCTION

Lipoprotein oxidation plays an in important role in initiation and progression of atherosclerosis, a vascular inflammatory disease, which leads to complications that cause myocardial infarction and stroke ¹. Oxidized low-density lipoprotein (LDL) accumulated in the vascular wall activates macrophages, endothelial and smooth muscle cells to produce pro-inflammatory cytokines and chemokines, thereby enhancing inflammation in the lesion. LDL oxidation is a gradual process, progressing from lipid hydroperoxides to degradation products, such as aldehydes and ketones, which can form covalent adducts with proteins and lipids. With the changing composition of lipid and protein oxidation products, different forms of oxidized LDL interact with different cellular and soluble receptors. Thus, we have demonstrated that hydroperoxide-rich minimally oxidized LDL (mmLDL) activates macrophages via toll-like receptor-4 (TLR4) ^2–4. More extensively oxidized LDL (OxLDL), containing aldehydes and modified/degraded apoB-100, binds to CD36 and other scavenger receptors ⁵.

Certain chronic infectious diseases, such as periodontitis and chlamydial infection, accelerate atherosclerosis and exacerbate its clinical manifestations, leading to acute cardiovascular events ^6–10. This is likely a consequence of immune responses to bacterial pathogens resulting in intensified vascular inflammation within atherosclerotic lesions. Moreover, recent studies have revealed that ingestion of high-fat meals leads to transient increases in plasma endotoxin levels ¹¹, likely via chylomicron-mediated transport of endotoxin derived from intestinal microflora ¹². Metabolic endotoxemia has been also observed in patients with type 2 diabetes mellitus ¹³ and in mice fed a high-fat diet ¹⁴. Thus, a combinatorial activation of vascular cells by oxidized LDL and by threshold levels of endotoxin is a likely in vivo event, particularly relevant to the development of atherosclerosis in overweight individuals consuming a Western-type diet.

Because we previously demonstrated that mmLDL activates macrophages via TLR4 ^2–4, which is also activated by bacterial lipopolysaccharide (LPS), a “classic” TLR4 ligand ¹⁵, we hypothesized that low levels of mmLDL and LPS would exert cooperative activation of macrophages. TLR4-mediated responses involve the recruitment of a number of adaptor proteins to the TLR4 cytoplasmic domain. At the plasma membrane, the LPS-induced recruitment of MyD88 to TLR4 results in rapid expression of NF-κB-dependent genes and activation of JNK and p38 kinases. In the endosomal compartment, the recruitment of TRIF to TLR4 induces IRF3-dependent gene expression. In contrast to the LPS-induced TLR4 responses, mmLDL induces moderate expression of proinflammatory genes, but profound cytoskeletal rearrangements and ROS generation in macrophages ^2,3,16. These MyD88-independent effects of mmLDL are mediated by the recruitment of spleen tyrosine kinase (Syk) to the C-terminal domain of TLR4 and subsequent activation of ERK1/2 and PLCγ ^4,16.

Because of the differences in the LPS- and mmLDL-induced activation of TLR4, we hypothesized that co-stimulation of macrophages by mmLDL and by low levels of bacterial LPS would result in cooperative effects that would result in greater activation than achieved by either stimulus alone. Herein, we demonstrate that mmLDL and low levels of Kdo2-LipidA (KLA), the active moiety of LPS, cooperatively upregulate expression of a number of proinflammatory genes, including chemokines Cxcl2 (MIP-2), Ccl3 (MIP-1α), and Ccl4 (MIP-1β), both in vivo and in vitro. A de novo motif analysis of promoters of cooperatively activated genes suggested involvement of AP-1 transcription factors. Indeed, ERK1/2-dependent activation of AP-1 was characteristic for mmLDL and LPS co-stimulation. In addition, mmLDL induced phosphorylation of c-Jun and the release of the corepressor NCoR from the promoter regions of Cxcl2, Ccl3 and Ccl4. These molecular mechanisms suggest that the combination of minimally oxidized LDL and low levels of LPS cooperatively activate inflammatory responses in macrophages.

METHODS

An expanded Methods section is available in the online-only Data Supplement.

Cells, LDL Preparations and Other Reagents

Murine macrophage-like cell lines RAW267.1 (abbreviated in the text as RAW) and J774A.1 (abbreviated as J774) were from ATCC. Mouse peritoneal macrophages were isolated as described below. LDL (density=1.019–1.063g/ml) was isolated from plasma of normolipidemic donors by sequential ultracentrifugation ¹⁷. Native and modified LDL preparations were tested for possible endotoxin contamination using a Limulus Amoebocyte Lysate kit (Cambrex, Walkersville, MD). LDL preparations with LPS content higher than 50 pg/mg protein, corresponding to 2.5 pg/ml in most cell culture experiments, were discarded. To produce mmLDL, we incubated 50 μg/ml of LDL in serum-free DMEM for 18 hours with murine fibroblast cells overexpressing human 15-lipoxygenase, as reported in detail ^2,18,19. Kdo₂-lipidA (KLA), obtained from Avanti Polar Lipids (Alabaster, AL), was used as a well-characterized active component of LPS and a highly specific TLR4 agonist ²⁰.

Mice, intraperitoneal injections and macrophage isolation

All animal experiments were performed according toNIH guidelines and were approved by the Animal Subjects Committee of the UCSan Diego. C57BL/6J mice were purchased from the Jackson Laboratory. Mice were anesthetized by a short exposure to isoflurane and injected intraperitoneally with 1 ml of 50 μg/ml mmLDL, 5 ng/ml KLA or their combination. One hour later, mice were sacrificed, peritoneum was lavaged and peritoneal cells were isolated and prepared for real time qPCR as described below. In some experiments, CD11b+ cells (predominantly macrophages) were rapidly separated at 4oC using magnetic bead reagents from Miltenyi (Germany).

Real-time PCR–based quantitative gene expression analysis

Total RNA was isolated using RNeasy columns (Qiagen, Valencia, CA), treated with DNase and reverse transcribed using oligo-dT and a First Strand Synthesis kit (Invitrogen, Carlsbad, CA). Real-time qPCR analysis was performed using reagents from Applied Biosystems (Foster City, CA) and a Rotor Gene Q (Qiagen, Valencia, CA).

Microarray and motif discovery algorithm

RNA concentration and integrity were verified with a NanoDrop ND-1000 (Thermo Scientific, Waltham, MA) and an Agilent 2100 BioAnalyzer respectively (Agilent, Santa Clara, CA). RNA was amplified, labeled and hybridized to the Sentrix BeadChip array, MouseRef-8 (Illumina, San Diego, CA). The slides were then processed according to the Illumina’s protocol, stained with Cy3-streptavidin and scanned with a BeadStation scanner (Illumina, San Diego, CA). Intensities were normalized using the modified Loess procedure ²¹. Sorting of genes and statistical analysis of pathways and gene ontology terms were performed as described ^21–23. De novo motif discovery was performed using HOMER (http://biowhat.ucsd.edu/homer/index.html), a comparative algorithm that searches for motifs that are specifically found in a set of regulated promoter sequences [−500, +100 bp] compared to an invariant background set of promoters.

Western blot analysis

SDS-PAGE and western blot were performed according to standard protocols as we previously described⁴.

Chemiluminescent assay for transcription factor-DNA binding

RAW cell nuclear extracts were isolated using a Nuclear Extraction kit (Active Motif, San Diego, CA). Transcription factor activation was measured using sensitive Active Motif’s TransAM NF-κB p65, NF-κB p50, c-Jun and c-Fos assays in an ELISA format.

Enzyme-linked immunosorbent assay

RAW cells or J774 cells were incubated for 5 hours with 50 μg/ml mmLDL, 1 ng/ml KLA or mmLDL + KLA in a serum-free DMEM. Supernatants were harvested and spun for 5 min at 10,000 rpm to remove floating cells. Levels of Cxcl2, Ccl3 and Ccl4 proteins were measured in ELISA assays using reagents from R&D Systems.

Chromatin Immunoprecipitation

The ChIP assay was performed as we previously reported ²⁴. qPCR with SYBR Green (Invitrogen, Carlsbad, CA) was carried out with a Rotor Gene Q (Qiagen). Primer sequences for the promoter regions were: Cxcl2 forward, gggctctgtgcttcctgat; Cxcl2 reverse, cagtctggggctctgaggt ²⁵; Ccl3 forward, cctcagtccctcactgtggt; Ccl3 reverse, catggaacggaaactctcgt; Ccl4 forward, ttgtggcaggtgtgaacatt; Ccl4 reverse, tgtcatggcatcgagaaaga. Results are presented as enrichment of the precipitated target sequence compared to input DNA.

Statistical analysis

Graphs represent means ± standard error. Significance of differences was calculated using one-way ANOVA, with Bonferroni correction, where indicated.

RESULTS

mmLDL and LPS cooperatively upregulate gene expression in RAW cells

To assess the effect of cooperative stimulation with mmLDL and LPS on a genome-wide scale, we conducted microarray analysis and focused on early (1 hour) gene expression responses. Throughout this study, we used Kdo2-LipidA (KLA; from Avanti Polar Lipids) as a well-characterized active component of bacterial LPS and a highly specific agonist of TLR4 ²⁰, at threshold levels of 1 ng/ml for in vitro experiments. We used 50μg/ml of mmLDL as previously reported ^3,4,16. RNA isolated from RAW macrophages was analyzed using an Illumina microarray platform.

The analysis of macrophages stimulated with mmLDL alone and KLA alone demonstrated activation of both common and distinct signaling and metabolic pathways (Online Table I). While both mmLDL and KLA activated TLR, MAP kinase, NF-κB, and IL-6 proinflammatory signaling pathways, the mmLDL stimulation was uniquely characterized by activation of PI3K/Akt, ERK1/2 and signaling pathways involved in cytoskeletal rearrangements, which agrees with the mmLDL-induced effects we previously documented with other techniques ^2–4,19,26. Downregulation of the sterol biosynthetic pathway by mmLDL was indicative of mmLDL uptake and intracellular cholesterol accumulation, as we have recently reported ⁴.

Although there was little overlap in expression of individual genes stimulated by mmLDL and KLA alone (Figure 1A), co-stimulation of macrophages with mmLDL and KLA resulted in additive and synergistic increases in expression of a number of genes, including transcription factors, kinases, phosphatases, and cytokines. In further studies, we focused on three proinflammatory chemokines important in the development of atherosclerosis, Cxcl2 (MIP-2), Ccl3 (MIP-1α), and Ccl4 (MIP-1β), which were additively/synergistically upregulated by mmLDL and KLA.

Figure 1. Gene expression changes in macrophages stimulated with mmLDL, KLA and their combination.

A, Microarray analysis. RAW macrophages were stimulated for 1 hour with 50 μg/ml mmLDL and/or 1 ng/ml KLA (Kdo2-LipidA, the active component of bacterial LPS). Gene expression in macrophages without any stimulation (media alone) was used as a reference.

B, Chemokine mRNA expression in RAW macrophages stimulated with mmLDL, KLA and their combination. RAW macrophages were incubated for 1 hour with 50 μg/ml mmLDL, 1 ng/ml KLA (Kdo2-LipidA), or 50 μg/ml mmLDL + 1 ng/ml KLA. The relative expression of Cxcl2, Ccl3 and Ccl4 mRNA was measured by qPCR and normalized to Gapdh, with subsequent normalization to the KLA sample in each experiment. Mean ± SEM from 4 independent experiments. *, p<0.05 vs. mmLDL; ^#, not statistically significant vs. mmLDL. ANOVA analysis with Bonferroni correction: p < 0.01 for all 3 chemokines.

To confirm the microarray data, we conducted independent experiments and measured the expression of Cxcl2, Ccl3, Ccl4 and Gapdh mRNA in RAW macrophages induced my mmLDL, KLA and their combination (Figure 1B). The co-stimulation with mmLDL and KLA resulted in 1.5–3 fold increases in chemokine expression (normalized to Gapdh), and ANOVA analysis demonstrated a significant trend toward the increase in gene expression in the mmLDL+KLA samples compared to mmLDL and KLA alone. These increases in RAW cells were moderately synergistic (Online Table II). The changes in the chemokine expression levels were not due to variations in the Gapdh mRNA levels (Online Figure I).

Cooperative stimulation of cytokine expression by mmLDL and LPS in vivo

In order to test the in vivo relevance of our findings with RAW macrophages, we used a model of sterile peritonitis. Mice were injected intraperitoneally with vehicle, mmLDL, KLA, or mmLDL+KLA. One hour later, peritoneal cells were isolated and mRNA levels of Cxcl2, Ccl3 and Ccl4 were determined using qPCR. Similarly to the results with RAW cells, expression of Cxcl2, Ccl3, and Ccl4 was higher in peritoneal cells co-stimulated with mmLDL and KLA, but the synergistic effect was more dramatic, reaching as much as 2.5–5 fold increases compared to the cells stimulated with KLA alone (Figure 2A, Online Table II). Peritoneal macrophages constitute ~40% of all leukocytes in the peritoneum; other cells include lymphocytes, neutrophils, and dendritic cells. Using a rapid positive selection procedure with CD11b antibodies, we enriched the population of CD11b-positive cells (predominantly macrophages) from 40% to nearly 80% (Online Figure II) and confirmed that intraperitoneal stimulation with mmLDL and KLA resulted in cooperative stimulation of peritoneal macrophages (Figure 2B, Online Table II). These results, however, do not exclude the possibility that other peritoneal leukocytes can also respond to the mmLDL/KLA stimulation in an additive/synergistic manner.

Figure 2. Chemokine mRNA expression in murine peritoneal cells stimulated in vivo with mmLDL, KLA and their combination.

One milliliter of either vehicle (media), 50 μg/ml mmLDL, 5 ng/ml KLA (Kdo2-LipidA) or 50 μg/ml mmLDL + 5 ng/ml KLA were injected intraperitoneally into C57BL6/J mice. Peritoneal cells were isolated 1 hour post injection, and mRNA expression of Cxcl2, Ccl3 and Ccl4 was measured by qPCR. All values were normalized to the KLA response.

A, Total peritoneal cell population. qPCR measurement from individual mouse samples. Four mice per group. Mean ± SEM.

B, In a separate experiment, CD11b-positive cells were isolated from the peritoneal lavages (pooled from 3 mice) by positive selection as described in Methods. Representative results from two independent experiments are shown. Mean ± SD from technical triplicates.

*, p<0.05 vs. mmLDL and vs. KLA; ^#, non-significant vs. KLA; ANOVA with Bonferroni correction: p < 0.01 for all 3 chemokines.

Co-stimulation with mmLDL and LPS upregulates AP-1 transcriptional activity

Further analysis of the microarray data, using a de novo motif discovery algorithm developed in our group ²⁷, suggested that promoter regions of many genes upregulated in macrophages co-stimulated with mmLDL and LPS, including Cxcl2, Ccl3 and Ccl4, contained the common motif presented in Figure 3. This motif matched with the known c-Jun/c-Fos consensus sequence, suggesting the involvement of AP-1 transcription factor regulation in macrophages co-stimulated with mmLDL and LPS. The most common AP-1 components, transcription factors c-Jun and c-Fos, are regulated by MAP kinases JNK and ERK1/2, respectively. Indeed, microarray analysis demonstrated highly significant activation of MAP kinase related pathways (Figure 3). These data agree with our earlier findings of strong ERK1/2 phosphorylation and Ccl5 (RANTES) expression in macrophages stimulated with mmLDL alone ^3,4,16.

Figure 3.

Promoter motif discovery and pathway analysis in macrophages co-stimulated with mmLDL and KLA.

Next, we compared time-dependent ERK1/2 activation in macrophages stimulated with either mmLDL or KLA. Interestingly, mmLDL induced ERK1/2 phosphorylation in RAW cells as early as at 5 minutes post stimulation, but the levels of phosphorylated ERK1/2 rapidly declined and by 60 min were back to the levels in unstimulated cells (Figure 4). In contrast, ERK1/2 phosphorylation by a low dose of KLA (1 ng/ml) was delayed by nearly 30 minutes, but was sustained for over 1 hour. Co-stimulation with mmLDL and KLA resulted in a pattern of ERK1/2 phosphorylation similar to that of mmLDL, except that the length of activation was more sustained, for the entire hour of the experiment. Thus, we hypothesized that early and sustained phosphorylation of ERK1/2 increases activation of the AP-1 transcription complex during combined mmLDL/KLA stimulation.

Figure 4. ERK1/2 phosphorylation in RAW macrophages stimulated with mmLDL, KLA and their combination.

RAW macrophages were incubated for indicated times with 50 μg/ml mmLDL, 1 ng/ml KLA (Kdo2-LipidA), or 50 μg/ml mmLDL + 1 ng/ml KLA. Cell lysates were separated on SDS-PAGE and probed with antibodies against phosphorylated ERK1/2 and GAPDH and pan-Actin. Representative blots from 4 independent experiments.

To test this hypothesis, we measured binding of c-Jun and c-Fos to consensus AP-1 DNA binding sites. mmLDL alone and KLA alone induced only moderate c-Jun and c-Fos DNA binding, which was synergistically increased in RAW macrophages co-stimulated with mmLDL and KLA (Figure 5). Importantly, inhibition of ERK1/2 phosphorylation completely abolished the synergistic effect of mmLDL and KLA. NF-κB transcription factors p50 and p65 showed the same DNA binding pattern as the AP-1 transcription factors, but, in contrast, inhibition of ERK1/2/did not have a significant effect on p50 and p65 DNA binding. These data agree with the role of ERK1/2 in regulation of AP-1, but not the NF-κB transcription program.

Figure 5. AP-1 and NF-κB DNA binding in RAW macrophages stimulated with mmLDL, KLA and their combination.

RAW macrophages were incubated for 1 hour, in the presence or absence of 5 μM U0126, with 50 μg/ml mmLDL, 1 ng/ml KLA (Kdo2-LipidA), or 50 μg/ml mmLDL + 1 ng/ml KLA; “no inhibitor” samples were supplemented with an equal volume of DMSO. Nuclear extracts were analyzed for transcription factor DNA binding using TransAM plate-based assays. Representative results of 3 independent experiments.

To confirm that ERK1/2 activation mediates cytokine expression in macrophages co-stimulated with mmLDL and LPS, we measured CXCL2, CCL3 and CCL4 protein secretion by RAW and by J774 cells. We found that co-stimulation of both macrophage cell lines with mmLDL and KLA led to additive-to-synergistic upregulation of cytokine secretion, which, importantly, was blocked with the inhibition of ERK1/2 signaling (Figure 6).

Figure 6. Chemokine secretion by RAW and J774 macrophages stimulated with mmLDL, KLA and their combination.

RAW (top) or J774 (bottom) macrophages were incubated for 5 hours, in the presence or absence of 5 μM U0126, with 50 μg/ml mmLDL, 1 ng/ml KLA (Kdo2-LipidA), or 50 μg/ml mmLDL + 1 ng/ml KLA; “no inhibitor” samples were supplemented with an equal volume of DMSO. Cell culture media were collected and CXCL2, CCL3 and CCL4 protein levels were measured in an ELISA assay. Results shown are technical duplicates of biological duplicates and are representative of 2 independent experiments.

mmLDL-induced NCoR derepression

Nuclear receptor co-repressor (NCoR) is an important check point in the regulation of inflammatory responses to LPS via TLR4 signaling ²⁵. LPS induces IKKε-dependent phosphorylation of c-Jun, which results in detachment of NCoR from c-Jun and recruitment of c-Fos to complete the AP-1 assembly on a promoter binding site ²⁵. This ultimately leads to the transcription of inflammatory genes. Since mmLDL also signals via TLR4 and phosphorylates c-Jun ^3,4, we tested whether mmLDL has the ability to detach NCoR from the promoters of Cxcl2, Ccl3 and Ccl4, the mmLDL-induced proinflammatory genes. First, we measured time-dependent phosphorylation of c-Jun in macrophages stimulated with mmLDL, KLA and their combination. We found the same pattern as with phosphorylation of ERK1/2, characterized by rapid and transient c-Jun phosphorylation with mmLDL and delayed c-Jun phosphorylation with KLA (Figure 7). As with ERK1/2, the mmLDL and KLA combination induced early and sustained c-Jun phosphorylation.

Figure 7. c-Jun phosphorylation in RAW macrophages stimulated with mmLDL, KLA and their combination.

RAW macrophages were incubated for indicated times with 50 μg/ml mmLDL, 1 ng/ml KLA (Kdo2-LipidA), or 50 μg/ml mmLDL + 1 ng/ml KLA. Cell lysates were separated on SDS-PAGE and probed with antibodies against phosphorylated c-Jun and pan-Actin (pan-Actin loading control in this figure is identical to that in Fig. 4 as the same cell lysates were probed for p-ERK1/2 and p-c-Jun). Representative blots from 4 independent experiments.

In addition to ERK1/2, mmLDL activation of macrophages also results in JNK phosphorylation ³. Thus, we hypothesized that in contrast to LPS-induced c-Jun phosphorylation, which was dependent on IKKε²⁵, mmLDL stimulates c-Jun phosphorylation via JNK. Indeed, a specific JNK inhibitor SP600125 effectively inhibited mmLDL-induced c-Jun phosphorylation (Figure 8A). The specificity of SP600125 was validated by the lack of inhibition of ERK1/2 phosphorylation in the same macrophage lysates.

Figure 8. mmLDL-induced c-Jun phosphorylation and NCoR detachment from the Cxcl2 and Ccl3 promoters.

A, RAW macrophages were incubated for indicated times with 50 μg/ml mmLDL, in the presence or absence of 25 μM SP600125. Cell lysates were separated on SDS-PAGE and probed with antibodies against phosphorylated c-Jun, phosphorylated ERK1/2 and GAPDH.

Representative blots from three independent experiments.

B, RAW macrophages were incubated for 0, 15 or 30 minutes with 50 μg/ml mmLDL. NCoR ChIP and promoter qPCR were performed as described in Methods. Representative data from two independent experiments. Mean ± SEM of technical triplicates.

C, The NCoR ChiP assay was repeated for macrophages incubated with mmLDL for 15 minutes in the presence or absence of 25 μM SP600125; “no inhibitor” samples were supplemented with an equal volume of DMSO. The dashed line at 100% represents the NCoR occupancy levels measured in non-stimulated cells. Representative results of three independent experiments. Mean ± SEM of technical triplicates.

Next, in a chromatin immunoprecipitation experiment, we measured NCoR occupancy on the Cxcl2, Ccl3 and Ccl4 promoters. We found that as early as 15 minutes post stimulation mmLDL induced release of approximately 40–60% of NCoR molecules from the promoter regions of Cxcl2 and Ccl3 (Figure 8B), but the results with the Ccl4 promoter were inconsistent. Importantly, inhibition of JNK with SP600125 abrogated mmLDL-induced NCoR release from the promoter regions of Cxcl2 and Ccl3 (Figure 8C).

DISCUSSION

Inflammation plays a major role in the pathogenesis of atherosclerosis, and chronic infections are known to exacerbate its complications. In this study, we demonstrated that two likely pro-inflammatory agonists involved in endotoxemia-complicated atherogenesis, minimally oxidized LDL and bacterial LPS, induced cooperative upregulation of proinflammatory genes in macrophages when applied together. Specifically, we found that mmLDL and threshold levels of LPS (1 ng/ml KLA) cooperatively activated RAW and J774 macrophages to express pro-inflammatory genes, including chemoattractant cytokines Cxcl2, Ccl3, and Ccl4 (Figures 1 and 6, and Online Table II). This cooperative effect was even more pronounced in vivo, achieving synergistic levels in peritoneal macrophages and possibly in other peritoneal cells (Figure 2). We further found that AP-1 response elements were enriched in the promoter regions of the genes responsive to co-stimulation with mmLDL and KLA (Figure 3) and that c-Jun and c-Fos binding to the response element DNA sequence was higher in the nuclear lysates of macrophages co-stimulated with mmLDL and KLA compared to mmLDL or KLA alone (Figures 4 and 5). In addition, mmLDL rapidly induced AP-1 derepression by phosphorylating c-Jun and releasing corepressor NCoR from the chemokine promoters (Figures 7–8).

The premise for our study was that low concentrations of LPS and mmLDL, although having one common signaling receptor, TLR4, nevertheless, activate separate as well as common signaling pathways. mmLDL stimulates TLR-dependent, but MyD88-independent generation of ROS and cytoskeletal rearrangements in macrophages, the latter leading to macropinocytosis and lipoprotein uptake. These mmLDL effects are mediated by the recruitment of Syk to the cytoplasmic domain of TLR4 and subsequent Syk-dependent activation of ERK1/2 ^4,16. In this study, we compared the time courses of ERK1/2 phosphorylation and found that mmLDL activates ERK1/2 as early as in 5 min, but LPS (KLA) at a concentration of 1 ng/ml showed no significant ERK1/2 activation before 30 min. Interestingly, co-stimulation with mmLDL and KLA maintained the early ERK1/2 phosphorylation, characteristic for mmLDL only, which was sustained for a longer period of time due to the KLA component (Figure 4). It has been suggested that sustained ERK1/2-dependent phosphorylation of c-Fos leads to the AP-1 complex assembly and its DNA binding ²⁸.

In addition to the ERK1/2-mediated mechanism, mmLDL also activates a JNK pathway leading to c-Jun phosphorylation and AP-1 derepression via the detachment of corepressor NCoR (Figures 7–8). The dependence of NCoR detachment on c-Jun phosphorylation has been demonstrated in experiments with overexpressed phospho-mimic c-Jun mutant ²⁵. However, as we suggested earlier, the mechanism of LPS-induced c-Jun phosphorylation is via IKKε and is independent of JNK activity ²⁵. In contrast, experiments of this study demonstrate that mmLDL-induced c-Jun phosphorylation and NCoR detachment do depend on JNK activity (Figure 8). These findings of separate pathways by which mmLDL and LPS regulate NCoR derepression further explain cooperative upregulation of proinflammatory cytokines induced by these two stimuli. Another difference in the mmLDL- and LPS- induced mechanisms is that mmLDL induces translocation of p65 to the nucleus but not its phosphorylation nor binding to DNA, whereas LPS stimulates all these steps of NF-κB activation ³. Thus, taken together, mmLDL and LPS complement each other by strengthening AP-1 and NF-κB promoter activity of target proinflammatory genes.

Importantly, the mmLDL and LPS cooperative effects were evident at a threshold LPS concentration (1 ng/ml) at which LPS alone induced a very limited macrophage response. Such levels of bacterial endotoxin can be found in human plasma during low-grade but sustained chronic infections, like periodontitis or infections with Chlamydia, which have been shown to accelerate the progression of atherosclerosis ^6,7,9. In fact, the American Journal of Cardiology and the Journal of Periodontology jointly published an Editors’ Consensus, recommending the early treatment of periodontitis to slow the progression of atherosclerosis in affected subjects ²⁹.

Recent clinical and experimental evidence suggests that low-level endotoxemia may be considerably more prevalent than previously realized ¹¹, underscoring the importance of our findings. The gastrointestinal tract constitutes an enormous reservoir of biologically active bacterial products, including LPS, and a small proportion of gut-derived LPS can translocate into the circulation even in relatively healthy subjects ¹¹. In the Bruneck (Italy) Study, plasma levels of endotoxin in the general population (516 men and women aged 50 to 79 years old) ranged from 6 to 209 pg/ml, with a 3-fold higher risk of carotid atherosclerosis in individuals with endotoxin levels 50 pg/ml and higher (90th percentile) ³⁰. A more recent study of a smaller population in England reported endotoxin levels of 3.1 EU/ml in non-diabetic and 5.5 EU/ml in patients with type 2 diabetes, which correspond to 310 pg/ml and 550 pg/ml LPS, respectively (according to the FDA’s EC-6 standard of 10 EU/ng) ¹³. Miller et al. analyzed ethnic differences and found a graded increase in endotoxin levels from black Africans to whites to South Asians, reporting significantly higher values than in other studies, as much as 14.4 EU/ml, or 1.44 ng/ml in South Asian men, and a strong correlation between plasma endotoxin and triglyceride concentrations ³¹. In addition, emerging evidence suggests that dietary habits alone may alter systemic exposure to endotoxin. For example, Erridge et al. have demonstrated that a single high-fat meal markedly increases endotoxin concentrations postprandially in healthy human subjects, sufficient to activate vascular cells in vitro ¹¹. These findings were supported by studies by Amar et al., correlating energy intake with endotoxemia in healthy subjects ³². In mice, a 4-week high-fat (72%) diet increased plasma endotoxin concentrations 2.7-fold to 4.9 EU/ml, or 490 pg/ml ¹⁴. Although absolute values vary considerably in different studies, the presence of sub-nanogram to nanogram per milliliter levels of endotoxin in plasma of seemingly normal subjects is well documented. To put the above numbers in perspective, the fatality rates for meningococcal patients with endotoxin plasma levels of 10–50 EU/ml, 50–100 EU/ml, 100–150 EU/ml, and > 150 EU/ml were 30%, 100%, 89% and 100%, respectively ³³. Because a large portion of LPS is carried in plasma by lipoproteins ^34,35, it is also plausible that LPS is present in atherosclerotic lesions, but this is yet to be determined.

In summary, our data demonstrate that cooperative engagement of AP-1 and NF-κB transcription factors by mmLDL and LPS results in additive/synergistic upregulation of pro-inflammatory genes in macrophages. This may constitute a mechanism of enhanced inflammatory activation within atherosclerotic lesions leading to the disease progression and increased risk of acute cardiovascular events, as suggested by human epidemiologic studies of atherosclerosis complicated by chronic infections and other conditions associated with subclinical endotoxemia in apparently healthy subjects.

Novelty and Significance.

What is known?

• Oxidation of low-density lipoprotein (LDL) significantly contributes to the progression of atherosclerosis, and oxidized LDL induces inflammatory responses in vascular cells

• Certain chronic infectious diseases, such as periodontitis and chlamydial infection, exacerbate clinical manifestations of atherosclerosis

• Low-level but persistent metabolic endotoxemia is often found in diabetic and obese subjects, who are at risk of cardiovascular disease, and is induced in mice fed a high-fat diet

What new information does this article contribute?

• Minimally oxidized LDL (mmLDL) and low doses of bacterial lipopolysaccharide (LPS) cooperatively activate macrophages to express higher levels of pro-inflammatory cytokines Cxcl2, Ccl3, and Ccl4

• The mechanism of this synergistic stimulation involves early and sustained activation of ERK1/2 and JNK, release of corepressor NCoR from the gene promoter regions, and derepression of the AP-1 transcription program

Chronic inflammation in atherosclerotic lesions in coronary and cerebral arteries leads to formation of plaques vulnerable to rupture. Ruptured plaques cause intravascular thrombosis, which often results in a myocardial infarction or a stroke. In this study, we tested the hypothesis that two factors involved in the pathogenesis and complications of atherosclerosis – mmLDL and LPS – may synergize in inducing pro-inflammatory responses in macrophages. We demonstrated that mmLDL and LPS have different modes of activation of intracellular signaling pathways and that combined mmLDL and LPS stimulation of macrophages leads to enhanced transcriptional activity and elevated expression of proinflammatory cytokines. These mechanistic findings may explain why chronic infectious diseases, such as periodontitis and chlamydial infection, characterized by low-level but systemic elevations in LPS levels, accelerate atherosclerosis and exacerbate its clinical manifestations, leading to acute cardiovascular events. Our results are also relevant to the development of atherosclerosis in diabetic and overweight individuals consuming a Western-type diet, since recent studies have demonstrated that high-fat diets result in subclinical metabolic endotoxemia. Because lipoprotein oxidation is a well-documented event in the development of atherosclerosis, our data showing cooperative activation of macrophages with mmLDL and LPS, suggest a possible mechanism of accelerated atherosclerosis in patients with subclinical endotoxemia.

Non-standard Abbreviations and Acronyms

LDL

low-density lipoprotein

mmLDL

minimally oxidized LDL

OxLDL

extensively oxidized LDL

LPS

lipopolysaccharide

KLA

Kdo2-lipidA

TLR

toll-like receptor

NCoR

nuclear receptor co-repressor