Plasma lipoproteins such as LDL (low-density lipoprotein) are important therapeutic targets as they play a crucial role in macrophage biology and metabolic disorders. The impact of lipoprotein profiles on host defense pathways against Gram-positive bacteria is poorly understood. In this report, we discovered that human serum lipoproteins bind to lipoteichoic acid (LTA) from Staphylococcus aureus and thereby alter the immune response to these bacteria. Size-exclusion chromatography and solid-phase-binding analysis of serum revealed the direct interaction of LTA with apolipoproteins (Apo) B100, ApoA1, and ApoA2. Only ApoB100 and the corresponding LDL exerted biological effects as this binding significantly inhibited LTA-induced cytokine releases from human and murine immune cells. Serum from hypercholesterolemic mice or humans significantly diminished cytokine induction in response to S. aureus or its LTA. Sera taken from the patients with familial hypercholesterolemia before and after ApoB100-directed immuno-apheresis confirmed that ApoB100 inhibited LTA-induced inflammation in humans. In addition, mice in which LDL secretion was pharmacologically inhibited, displayed significantly increased serum cytokine levels upon infection with S. aureus in vivo. The present study identifies ApoB100 as an important suppressor of innate immune activation in response to S. aureus and its LTA.
Disturbances in lipid profiles, especially hypercholesterolemia, are increasingly encountered conditions that reach endemic proportions in Western societies. Consequently, statins and other lipid-lowering medicines are blockbuster drugs these days. While the beneficial effects of these drugs through lipid lowering and prevention of cardiovascular events are apparent , the potential impact of altered plasma-lipid compositions on the inflammatory response to pathogens is poorly understood. This might be of importance since earlier studies already showed that lipoproteins, especially HDL (high-density lipoprotein), can neutralize LPS from Gram-negative bacteria, and prevent its toxic effects in vitro and in vivo .
Serum lipoproteins also contribute to the host defense against Gram-positive bacteria. Staphylococcus aureus is the most commonly isolated bacterial pathogen in humans and an important cause of severe invasive infections and sepsis . The pathogenesis of sepsis involves the excessive and harmful production of cytokines. The inflammatory response to S. aureus mainly depends on the recognition of LTA (lipoteichoic acid)  and bacterial lipoproteins  by TLR 2/6 heterodimers [6, 7] and different coreceptors such as CD36  or CD14 . Despite accumulating evidence that lipoproteins are part of innate immunity [2, 10, 11], their role in S. aureus-mediated inflammation has rarely been addressed. During infections with S. aureus, Apolipoprotein (Apo) B100, the main protein component of LDL (low-density lipoprotein), antagonizes the staphylococcal accessory gene regulator (agr) quorum-sensing system  as well as the α-toxin , thereby preventing bacterial invasion and cellular damage. In human as well as murine macrophages, the LTA-induced release of TNF was inhibited by HDL, LDL, and LPS-binding protein (LBP) [14, 15]. However, the absence of LBP in a murine LTA-induced lung inflammation model did not show any augmentation of LTA-associated inflammatory responses in vivo . Also, it remains to be defined whether HDL and LDL selectively inhibit LTA or rather bacterial lipoproteins that possibly contaminate LTA preparations . In this study, we compared LTA from wild-type (WT-LTA) and diacylglycerol-transferase deletion mutant S. aureus (lgt-LTA) that lack contaminating bacterial lipoproteins  to investigate the binding of serum (lipo)proteins to LTA and its consequence for cytokine induction. Furthermore, LDL receptor knockout (LDLR-KO) and drug-induced hypercholesterolemic mice were used to analyze the role of serum LDL in innate immune responses against S. aureus and its LTA in vivo.
Results and discussion
Serum apolipoproteins bind to soluble LTA
We previously demonstrated that the cytokine release from human PBMCs induced by soluble but not surface-immobilized LTA is abrogated in the presence of human serum  and we therefore hypothesized that inhibitory serum components might prevent the inflammatory response to soluble LTA. We now observed that preincubating soluble WT-LTA with human serum diminished the TNF and IL-1β release by human PBMCs in a concentration-dependent manner (Fig. 1A and Supporting Information Fig. 1). This inhibition by serum was specific for LTA since inhibition of lgt-LTA, which lacks bacterial lipoproteins, by serum was even more pronounced (Fig. 1 A and Supporting Information Fig. 1). To identify serum components that inhibit LTA-mediated cytokine induction, we applied an unbiased approach to detect serum-derived binding partners of soluble LTA. We incubated human serum with WT-LTA or PBS, subjected samples to size-exclusion chromatography and observed an increase in UV extinction in specific chromatography fractions when LTA-incubated serum was applied (Fig. 1B). SDS-PAGE analysis of these fractions revealed an enrichment of five distinct protein bands in the LTA sample. Subsequent MALDI-TOF mass spectrometry analysis of peptide fingerprints identified ApoA1, ApoA2, ApoA4, ApoB100, and complement factor C4 as potential LTA-binding partners (Fig. 1C). In order to verify the binding of these apolipoproteins to WT-LTA, we established a solid-phase-binding assay using purified Apo preparations. We immobilized ApoA1, ApoA2, and ApoB100 to 96-well plates, and tested their ability to bind biotinylated WT-LTA (Fig. 1D). We observed a dose-dependent increment in LTA binding to ApoA1, ApoA2, or ApoB100, and binding to the latter was most pronounced. We did not test complement factor C4 and ApoA4 as they were not commercially available. ApoB100 is the primary protein component of LDL whereas ApoA1 and ApoA2 represent major protein components of HDL. Apolipoproteins rarely exist in a free form in human serum, which suggests that LTA was bound to HDL and LDL, respectively, during the chromatography experiments. This is in line with an earlier report that found about 95% of plasma LTA bound to lipoproteins, mostly to HDL, followed by LDL and very low-density lipoprotein .
ApoB100 and LDL suppress LTA-induced cytokine release from human PBMCs
We next asked whether ApoA1, ApoA2, and ApoB100 or the respective lipoproteins HDL and LDL exert inhibitory effects on LTA-mediated cytokine induction and thus explain the anti-inflammatory effects we observed in the presence of human serum. Preincubation of WT-LTA or lgt-LTA with increasing concentrations of ApoB100 dramatically diminished the release of TNF (Fig. 2A) and IL-1β (Supporting Information Fig. 2) from human PBMCs. However, compared with TNF release, a higher concentration of ApoB100 was required to reduce IL-1β release, especially when WT-LTA was used for stimulation. The latter finding possibly relates to the presence of residual lipoproteins that specifically contributed to the induction IL-1β by WT-LTA but also rendered WT-LTA less sensitive ApoB100-mediated suppression. In contrast to ApoB100, cytokine production remained unaffected in the presence of ApoA2 (Fig. 2B and Supporting Information Fig. 3) or ApoA1 (data not shown). Supporting the potential biological relevance of our findings, we discovered that as little as 5 mg/dL of ApoB100 (i.e. 2–3% of the physiological serum concentration) was sufficient to substantially inhibit cytokine induction by LTA (Fig. 2A). In line with this, we also observed a significant reduction in LTA-induced cytokines when using 2.5% human serum (Fig. 1A).
To further corroborate our observations under physiological conditions where ApoB100 is associated with LDL , we examined the impact of purified LDL on the inflammatory activity of LTA. For this purpose, we added human PBMCs to lgt-LTA that was preincubated with LDL at physiological concentrations and determined the release of IL-6. Purified HDL devoid of ApoB100 but containing ApoA1 and ApoA2 was used as a control. Analogous to the results with purified apolipoproteins, the presence of LDL but not HDL significantly diminished lgt-LTA-mediated IL-6 release (Fig. 2C), confirming that LDL-mediated suppression of LTA-induced immune activation occurred solely through ApoB100. Assuming that enhanced serum LDL levels in humans would alter the cytokine response to LTA, we set out an experiment that enabled us to specifically test the inhibitory effects of ApoB100 in humans while simultaneously reducing the problems of interindividual variability in serum composition. For this purpose, we made use of serum from selected patients with familial hypercholesterolemia directly before and immediately after undergoing LDL-apheresis. Serum LDL levels of patients significantly decreased following immune apheresis employing anti-ApoB100 antibody-coated columns (Fig. 2D, left panel). This decrease in LDL was paralleled by an augmented cytokine response to lgt-LTA (Fig. 2D, right panel and Supporting Information Table 1). The level of serum LDL, and hence ApoB100 concentrations, conversely correlated with the amount of IL-6 elicited by lgt-LTA (Fig. 2E). Together, our supplementation and depletion experiments suggest that serum LDL dictates the inflammatory response to LTA of S. aureus. In support of our findings, low levels of serum LDL have been associated with increased fever and sepsis in patients requiring intensive care . We speculate that free LTA is cleared from the circulation after its incorporation into LDL and subsequently taken up and metabolized by recipient cells thereby preventing an inflammatory response.
Serum LDL determines pro-inflammatory responses to S. aureus in mice
Given that during S. aureus infection, large amounts of LTA are shed from the bacterial cell wall while undergoing lysis , we asked whether serum LDL can suppress cytokine releases following an S. aureus challenge in vivo. To address this question, we first analyzed the cytokine production induced by heat-killed (HK) S. aureus in murine PBMCs and macrophages supplemented with serum from WT or hypercholesterolemic LDLR-KO mice that display 7–9-fold higher serum LDL levels . Supplementation of WT PBMCs or peritoneal macrophages with serum from LDLR-KO mice significantly suppressed TNF release in response to HK S. aureus (Fig. 3A). Further, serum from LDLR-KO mice diminished the release of IL-6 from lgt-LTA-stimulated WT peritoneal macrophages (Fig. 3B) as well as the TNF release from S. aureus-stimulated LDLR-KO macrophages (Fig. 3C).
We next attempted to explore the in vivo contribution of LDL on the inflammatory response to S. aureus. An earlier study investigated the inflammatory response of LDLR-KO animals to LPS and discovered an increased cytokine release from LDLR-KO peritoneal macrophages independent of serum, suggesting an inflammatory phenotype of LDLR-deficient macrophages . Since we also observed an enhanced inflammatory response by macrophages from LDLR-KO mice upon LPS stimulation irrespective of the type of serum cells were supplemented with (Fig. 3D), we decided that the usage of LDLR-KO mice might cause biased results due to differences in the macrophage functions. Therefore, to address the precise role of LDL in S. aureus-mediated immune activation in vivo, without using the LDLR-KO mice, we made use of 4APP (4-aminopyrazolo[3,4-d]pyrimidine) that is known to decrease serum LDL levels in C57BL/6 mice . Intraperitoneal injection of 4APP reduced serum LDL by 80%, as compared with the vehicle control (Fig. 3E). Notably, compared with the WT mice, 4APP-treated mice showed a significant increase in the IL-6 release in response to the HK S. aureus (Fig. 3F, left panel). This effect was specific for S. aureus since the IL-6 induction by zymosan, a TLR ligand lacking fatty acids that could interact with ApoB100 in LDL, was unaffected by 4APP treatment (Fig. 3F, right panel). For mouse experiments, we administered the HK bacteria as live S. aureus can change its phenotype in dependence of host ApoB levels . Peterson et al. recently demonstrated that ApoB antagonizes the S. aureus agr quorum sensing system by sequestering autoinducing cyclic thiolactone peptide (AIP) 1 and thus represents and innate barrier against an invasive S. aureus infection in mice. Albeit distinct from each other, the quorum sensing antagonizing as well as the LTA-neutralizing activity represents two mechanisms of how ApoB contributes to the innate immune defense against S. aureus. Recently, a protective role for ApoA1 in LTA-induced acute lung injury and sepsis in mice has been proposed . However, in this study high concentrations of commercially available LTA were injected into mice making it difficult to distinguish whether the observed ApoA1-mediated reduction in cytokines and lung cell toxicity relates to an interaction with LTA or to the impurities found in these LTA preparations . Collectively, our data demonstrate for the first time an essential role for ApoB-containing lipoproteins in suppressing the proinflammatory cytokine release in response to S. aureus and its LTA.
The present study identifies Apo B100 as an important serum protein able to inhibit the cytokine induction by LTA from S. aureus. ApoB100 represents a major compound of the lipid metabolism known to play an important role in host defense by modulating effector functions of the innate immune system  In human as well as in mice, ApoB100 appears to act as a sink for free LTA in the blood, increasing the activation threshold of the inflammatory response. Furthermore, recent studies report that inflammatory cytokines upregulate ApoB100 production in hepatocytes [28, 29], indicating that ApoB100 may not only play a role in preventing but also in turning off an inflammatory reaction.
Material and methods
LTA from S. aureus 113 wild-type (WT-LTA) and mutant S. aureus strain 113 lgt::ermB (lgt-LTA) lacking the lipoprotein diacylglycerol transferase, was isolated as described previously . For biotinylation of LTA, 500 μg WT-LTA dissolved in 1 mL PBS and 500 μg PFP-biotin (pentafluorophenyl-ester biotin, Pierce) dissolved in 50-μL DMSO were mixed, incubated over night at 37°C on a shaker, and afterward submitted to ultrafiltration (Microsep 3K, Centricons, MI, USA). LTA concentration was determined as described previously .
Purified human ApoA1, A2, and B100 were obtained from Calbiochem, human LDL and HDL preparations from Millipore, zymosan (Saccharomyces cerevisiae; InvivoGen) and LPS (Salmonella enterica serovar abortus equi; Sigma). For stimulation of cells or iv injection into mice, S. aureus (ATCC 19059) were heat killed (HK) for 30 min at 65°C.
Identification of LTA-binding proteins
For chromatography, a sephacryl S-200-HR column (HiPrepTMSephacrylTM-S-200-HR-column, Amersham Pharmacia) with a void volume of 35 mL was used. First, the elution behavior of WT-LTA was determined. WT-LTA applied to the column already eluted in the void volume, starting at 30 mL, owing to the micelle formation. Next, IgG and albumin depleted serum (Albumin and IgG removal kit Amersham Biosciences) obtained from healthy donors was applied to the column and the macrocomplexes eluting in the void volume were collected and discarded. Afterward, plasma proteins derived from column fractions collected after the void volume were incubated either with PBS or with WT-LTA for 30 min at room temperature. Then, the samples were applied to the column, and the LTA-containing fractions of the void volume that showed an additional UV absorption peak (fraction 30–32 mL) were pooled and analyzed comparatively by SDS-PAGE and silver staining. The protein bands found only in the LTA-containing samples were identified using MALDI-TOF mass spectrometry at the core facility of the Biomedical Centre at the Ludwig-Maximillians-University (Munich, Germany). To confirm the binding between the identified apolipoproteins and WT-LTA, purified apolipoproteins were coated to 96-well plates (MaxiSorp, Nunc) in 0.1 M NaHCO3, pH 8.2, over night at 4°C, blocked with 3% BSA in PBS for 2 h at room temperature and subsequently washed twice with PBS containing 0.05% Tween-20. Biotinylated WT-LTA was added to the plate and incubated for 2 h. Bound LTA was quantified using streptavidin-conjugated peroxidase (Biosource) as described previously .
Human PBMCs and human patient sera
PBMCs of healthy volunteers were prepared from heparinized whole blood using Lymphoprep (Axis-Shield) according to the manufacturer. For in vitro stimulation of human PBMC, 2.5 μg WT-LTA or lgt-LTA were preincubated with 20% (unless otherwise indicated) autologous or hypercholesterolemic serum, or with the indicated concentrations of ApoB100, ApoA2, LDL, or HDL. Preincubation was performed in polypropylene tubes in a total volume of 50 μL RPMI for 1 h at room temperature. Then, 5 × 105 PBMCs/tube in RPMI were added and to a total volume of 250 μL. PBMCs were incubated in the presence of 5% CO2 at 37°C and cell-free supernatants were collected after 22 h. Cytokines were quantified using ELISAs (R&D Systems) as described previously .
Hypercholesterolemic sera were collected from seven patients with hypercholesterolemia immediately before and after undergoing LDL-apheresis at the Department of Medicine III at the Vienna General Hospital. LDL levels in patient sera were measured as described previously . Enrolment of patients with hyper-cholesterolemia was approved by the Ethical Review Board of the Medical University Vienna and informed consent was obtained before blood was withdrawn.
LDLR-KO mice (kindly provided by Christoph J. Binder, Medical University Vienna, Austria) and age- and sex-matched C57BL/6 mice from Charles River were kept under SPF conditions. All animal studies were approved by the Animal Review Board of the Medical University Vienna and the Austrian Ministry of Science and comply with institutional guidelines (BMWF-66.009/0169-II/3b/2011). Peritoneal macrophages were harvested by flushing the peritoneal cavity. Murine PBMCs were obtained by Ficoll-Hypaque density-gradient centrifugation of whole blood according to the manufacturer's protocol (GE Heathcare Bio-Sciences AB).
For in vitro stimulation of murine PBMCs or peritoneal macrophages, 2.5 μg lgt-LTA, 25 ng LPS or 5 × 106 HK-S. aureus were preincubated with 20% serum obtained from either C57BL/6 WT or LDLR-KO mice. Preincubation was performed in 96-well cell culture plates in a total volume of 50 μL RPMI for 1 h at room temperature. Then, 1 × 105/well PBMCs or 5 × 104/well peritoneal macrophages in RPMI were added to a total volume of 250 μL. Cells were stimulated in the presence of 5% CO2 at 37°C and cell-free supernatants were collected after 22 h.
For in vivo infection experiments, female C57BL/6 mice (6 to 7 weeks old, Charles River) were pretreated with 4APP as described previously . For infection, WT mice either treated with 4APP or vehicle control or LDLR-KO mice were anesthetized by inhalation of isoflurane (Abbott, Vienna, Austria) and zymosan (10 μg/mouse) or HK S. aureus (3 × 109CFU/mouse) were administered intravenously via retro-orbital injection. Two hours post-infection blood was collected. Cytokines in supernatants from in vitro and in vivo experiments were measured using ELISA (R&D Systems).
To quantify circulating LDL, serum was obtained from mice treated with 4APP or vehicle control. Sera were applied to 6% SDS-PAGE and blotted to polyvinylidene fluoride membranes. ApoB100 was detected by immunoblotting using rabbit anti-ApoB antibody (Abcam) in a dilution of 1:1000. Detection and quantification of signals was performed using the infrared imaging system Odyssey (LI-COR Biosciences) after incubation with fluorophore-linked secondary antibodies (Rockland).
Paired t-test was used to compare data from two groups and one-way ANOVA followed by Tukey's multiple comparison was used to compare three or more groups.
We thank Andreas Peschel for providing the S. aureus mutant strain SA 113 lgt::ermB and are grateful to Thomas Hartung for helpful discussions. This work was supported by a grant from the German Research Council (International Research Training Group 1331) as well as the Austrian Science Fund (FWF): (I 289-B09).
Conflict of interest
The authors declare no financial or commercial conflict of interest.