Phospholipids impact the protective effects of HDL-mimetic nanodiscs against lipopolysaccharide-induced inflammation

Abstract

Aim:

The impacts of synthetic high-density lipoprotein (sHDL) phospholipid components on anti-sepsis effects were investigated.

Methods:

sHDL composed with ApoA-I mimetic peptide (22A) and different phosphatidylcholines were prepared and characterized. Anti-inflammatory effects were investigated in vitro and in vivo on lipopolysaccharide (LPS)-induced inflammation models.

Results:

sHDLs composed with 1,2-dimyristoyl-sn-glycero-3-phosphocholine (22A-DMPC) most effectively neutralizes LPS, inhibits toll-like receptor 4 recruitment into lipid rafts, suppresses nuclear factor κB signaling and promotes activating transcription factor 3 activating. The lethal endotoxemia animal model showed the protective effects of 22A-DMPC.

Conclusion:

Phospholipid components affect the stability and fluidity of nanodiscs, impacting the anti-septic efficacy of sHDLs. 22A-DMPC presents the strongest LPS binding and anti-inflammatory effects in vitro and in vivo, suggesting a potential sepsis treatment.

Plain language summary

Sepsis is triggered by endotoxins released by bacteria. These endotoxins trigger an exaggerated inflammatory response, leading to widespread inflammation and organ damage. Synthetic high-density lipoprotein (sHDL) is a potential treatment of sepsis by neutralizing endotoxins and regulating inflammatory responses. The phospholipid components of sHDL may affect the effectiveness of sHDL against sepsis. In this study, we prepared sHDLs with different phospholipids and compared their anti-septic effects on cells and in animal models. We found that sHDL made from DMPC presented the best anti-septic effects, possibly because DMPC-sHDL had the best fluidity at body temperature.

Sepsis, a life-threatening organ dysfunction caused by a dysregulated host response to infection, remains a major health challenge responsible for high morbidity, mortality and cost. Despite progress in anti-inflammatory and antimicrobial treatments, effective treatments for sepsis are still challenging due to the complex pathological mechanisms of sepsis. Lipopolysaccharides (LPS), an endotoxin component present on the outer membrane of Gram-negative bacteria, are one of the most potent inflammation initiators inducing sepsis [1]. TLR-4 is the major sensor for extracellular LPS [2–4]. Upon LPS recognition, TLR-4 activates several downstream factors, such as NF-κB, leading to the upregulation of proinflammatory mediators [5]. Such overactivation of TLR-4 and consequent overproduction of proinflammatory cytokines are considered key pathological mechanisms for sepsis.

High-density lipoproteins (HDL) are the smallest and densest plasma lipoproteins. HDL mainly consists of ApoA-I and phospholipids, and it plays pivotal roles in reverse cholesterol transport. Additionally, HDL exerts an array of anti-inflammatory and immune regulation effects through several mechanisms. HDL binds to and neutralizes circulating endotoxin, attenuating the host response to endotoxins [6,7]. Independent of endotoxin binding, HDL could modulate inflammatory responses by reducing TLR-4 recruitment, activating ATF3 expression and inhibiting NF-κB signaling pathways [8–11]. Unfortunately, HDL level is compromised in sepsis., which is associated with mortality and clinical outcomes [12]. Moreover, HDL isolated from septic patients presented impaired cholesterol transport capacity and compromised anti-inflammatory effects [12–14]. Thus, restoring HDL levels and enhancing HDL functionality would be a potential therapeutic strategy for sepsis [15–17].

Synthetic HDL (sHDL) nanodiscs, composed of phospholipids and ApoA-I or ApoA-I mimetics, have been developed as a promising strategy to restore the level and functionality of endogenous HDLs. Our previous study showed that infusion of sHDL nanodiscs significantly alleviates inflammatory cytokine levels and improves the survival of septic mice by neutralizing bacterial endotoxins and inhibiting endotoxin-induced inflammatory cytokine production [18]. To optimize the formulation of sHDL nanodiscs for sepsis treatment, we aim to further investigate the anti-sepsis mechanisms and structure–function relationship of sHDL nanodiscs in the present study. The phospholipid component greatly impacts the physiochemical properties of sHDL nanodisc, which in turn affects the efficacy and pharmacokinetics/pharmacodynamic (PK/PD) profiles of sHDL nanodiscs [19]. In this study, a series of sHDL nanodiscs consisting of phospholipids with different chain lengths and saturation were prepared. The particle properties, LPS neutralization capacity, as well as anti-inflammatory mechanisms, were investigated to understand the structure–function relationship of sHDL nanodiscs for sepsis treatment.

Materials & methods

Animals & reagents

7–9 week-old female C57BL/6 mice were purchased from Jackson Laboratory (ME, USA). All protocols were approved by the Institutional Animal Care & Use Committee (IACUC) at the University of Michigan, Ann Arbor.

22A (PVLDLFRELLNELLEALKQKLK) peptide was synthesized by GenScript (NJ, USA) and purity was approximately 85% as determined by HPLC. 1-palmitoyl-2-oleoyl-glycero-3phosphocholine (POPC), 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC), 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) and 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC) were purchased from Nippon Oil and Fat (Osaka, Japan). LPS conjugated with Alexa Fluor™ 488 was purchased from Thermo Fisher Scientific (MA, USA). All LPS (from E. coli O111:B4) were purchased from Sigma Aldrich (MO, USA). LPS purified by ion-exchange chromatography (L3024) was used throughout the entire experiment, except for the survival study in which LPS purified by phenol extraction (L2630) was used. Alexa Fluor 488-conjugated anti-mouse TLR4 (Clone: UT41) and Alexa Fluor 488-conjugated cholera toxin subunit B (CT-B) were obtained from Thermo Fisher Scientific. ATF-3 Antibody (C-19) (sc-188, 1:800 dilution) was purchased from Santa Cruz Biotechnology (Dallas, TX). GAPDH (D16H11) XP^® Rabbit Monoclonal Antibody (5174, 1:4000 dilution) and Anti-rabbit IgG, HRP-linked Antibody (7074, 1:5000 dilution) were purchased from Cell Signaling Technologies (MA, USA).

Cell culture

RAW 264.7 and J774A.1 macrophages were cultured in Dulbecco's Modified Eagle Medium (DMEM) containing 10% fetal bovine serum (FBS), 1% penicillin–streptomycin (10,000 U/ml), and 100 μg/ml Normocin™. HEK-Blue™ hTLR4 cells, which stably express CD14, MD2, NF-κB reporter, and human TLR4, were purchased from InvivoGen (CA, USA) and grown in DMEM containing 10% FBS. HEK-Blue hTLR4 cells express the secreted embryonic alkaline phosphatase (SEAP) reporter gene under the control of the NF-κB promotor, which enables the quantification of cell activation by measuring SEAP activity in media containing specific enzyme substrates. All cell lines were cultured at 37°C in a humidified 5% CO₂ incubator.

Preparation of synthetic high-density lipoprotein nanodiscs

sHDL nanodiscs were prepared via a co-lyophilization procedure that we have previously developed in our lab [20]. Briefly, 22A (PVLDLFRELLNELLEALKQKLK) and phosphatidylcholines either POPC (T_m = -2°C), DMPC (T_m = 23°C), DPPC (T_m = 41°C), or DSPC (T_m = 55°C) were mixed at 1:2 weight ratio in acetic acid. The resulting solution was then flash-frozen in liquid nitrogen and placed on a freeze-dryer for at least 2 days to remove organic solvents. The lyophilized powder was rehydrated with phosphate-buffered saline (PBS) and thermal-cycled above and below the transition temperature of each phospholipid to facilitate peptide-lipid binding. Finally, the pH of sHDL solutions was adjusted to 7.4 using NaOH and filtered with a 0.2 μm sterile filter. All sHDL concentrations are expressed in terms of 22A peptide concentration.

Characterization of synthetic high-density lipoprotein nanodiscs

The quality of the resulting sHDL nanodiscs was analyzed by the following analytical techniques. The purity of sHDL nanodisc was determined by gel permeation chromatography (GPC), with UV detection at 220 nm, using a Tosoh TSKgel G3000SWxl column (King of Prussia, PA, USA). The particle size of sHDL was determined by dynamic light scattering (DLS) on Malvern Zetasizer Nano ZSP (MA, USA), and the volume intensity average values were reported. The morphology of sHDL was assessed by transmission electron microscopy (TEM). sHDL samples were loaded on a carbon film-coated 400 mesh copper grid from Electron Microscopy Sciences (Hatfield, PA, USA) that were negatively stained with 1% (w/v) uranyl formate and dried before TEM observation. All specimens were imaged with 100 kV Morgagni TEM equipped with a Gatan Orius CCD. The transition temperature (T_m) of sHDL was analyzed by two-state modeling using TA Nano Differential Scanning Calorimetry (DSC; DE, USA).

Analysis of fluorescent-LPS binding to synthetic high-density lipoprotein HDL nanodiscs

LPS conjugated with Alexa Fluor™ 488 (10 μg/ml) was preincubated for 1 h at 37°C, then mixed with different formulations of sHDL (1 mg/ml) and incubated for 1 h at 37°C. Samples were centrifugated at 15,000 r.p.m for 10 min. 25 μl of samples were injected into Shimazu Nexera-I LC 2040d Plus system connected with an RF-20A prominence fluorescence detector (Kyoto, Japan) and separated with a Tosoh Bioscience TSKgel G3000Wxl (7.8 mm × 30 cm, 5 μm). PBS (pH 7.4) was chosen for the mobile phase with a flow rate of 0.5 ml/min. The signal was detected at 220 nm and an excitation wavelength of 495 nm and an emission wavelength of 519 nm.

Analysis of LPS-induced NF-κB expression

The HEK-Blue cell system from InvivoGen was used to analyze the neutralization of the LPS-induced inflammatory response. HEK-Blue hTLR4 cells stably express reporter-linked human TLR4, CD14, MD2 and NF-κB that are designed for studying the stimulation of human TLR4. Briefly, HEK-Blue hTLR4 cells were cultured in DMEM containing 10% low endotoxin FBS and selective antibiotics according to the manufacturer's instructions. The growth medium was discarded, and cells were resuspended in the HEK-Blue Detection medium. Cells were seeded at 25,000 cells per well. Cells were treated with various formulations of sHDL and were added at a peptide concentration of 10, 30 or 100 μg/ml in the presence or absence of 2 ng/ml of LPS. The cells were then incubated for 18 h. LPS binding to TLR4 induces NF-κB reporter expression, causing the HEK-Blue detection medium to turn blue. The blue color was quantified by measuring absorption at 650 nm using a SpectraMax M3 plate reader from Molecular Devices (CA, USA).

Analysis of proinflammatory mediators in vitro

RAW 264.7 cells were plated in a 96-well microplate at a density of 5 × 10⁴ cells/well and incubated until reaching 80% confluency. Cells were washed with PBS and different formulations of sHDL were added at peptide concentrations of 10, 30 or 100 μg/ml for 18 h followed by stimulation with LPS. To quantify the concentration of inflammatory cytokines including TNF-α, IL-6 and MCP-1, samples were prepared with BD Cytometric Bead Array Mouse Inflammation Kit (CA, USA) per manufacturer's instruction. Then, prepared samples were analyzed with flow cytometry, Beckman Coulter CytoFLEX (CA, USA).

Cellular cholesterol efflux analysis

J774.A1 cells were plated in a 24-well plate at a density of 1 × 10⁵ cells/well and incubated for 24 h. Cells were washed with PBS once and labeled with 1 μCi of [³H] cholesterol/ml for 24 h in DMEM containing 3% fatty acid-free bovine serum albumin (BSA) and 5 μg/ml ACAT inhibitor Sandoz 58-035. Cells were then washed with PBS and incubated overnight in DMEM containing 3% BSA and 5 μg/ml ACAT inhibitor Sandoz 58-035. Then cells were washed with PBS, and different formulations of sHDL were added at peptide concentrations of 100 μg/ml in DMEM containing 3% BSA. After 18 h of incubation, media were collected, and cells lysed in 0.5 ml of 0.1% SDS and 0.1 N NaOH. Radioactive counts in media and cell fractions were measured by liquid scintillation counting using Perkin Elmer Tri-Carb 2910TR (MA, USA) and percent cholesterol efflux was reported by dividing the media count by the sum of the media and cell counts.

Analysis of lipid raft & TLR4 recruitment

J774A.1 was plated in a 24-well microplate at a density of 5 × 10⁴ cells/well and incubated until reaching 80% confluency. Cells were treated with either PBS or different formulations of sHDL at a peptide concentration of 100 μg/ml for 18 h. The Control group was treated with 10 mM methyl-β-cyclodextrin for 30 min. Cells were washed with ice-cold PBS containing 2% FBS. Then, cells were incubated with 8 μg/ml Alexa Fluor 594-conjugated CT-B for 15 min and 2 μg/ml Alexa Fluor 488-conjugated anti-TLR4 for 30 min to label lipid raft and TLR4, respectively. Labeled cells were washed with ice-cold PBS containing 2% FBS and the percentage distribution of lipid raft and TLR4 were reported with the mean fluorescence intensity determined by MoFlo Astrios from Beckman Coulter.

RNA isolation & RT-PCR

RAW 264.7 cells were plated in a 6-well microplate at a density of 4 × 10⁵ cells/well and incubated until reaching 80% confluency. Cells were then washed with PBS and different formulations of sHDL were added at peptide concentrations of 100 μg/ml for 1, 2 or 4 h. Cells were lysed, and RNA was isolated using GeneJET RNA purification kit from Thermo Fisher Scientific. Approximately 1 μg of extracted RNA from each sample was transcribed to cDNA using SuperScript III First-Strand Synthesis System from Invitrogen. cDNA amplification was measured by quantitative real-time PCR on a StepOnePlus™ real-time PCR System from Applied Biosystems (MA, USA). TaqMan assays from Applied Biosystems were used to measure the following: Gapdh Mm99999915_g1; Atf3 Mm00476033_m1. Gene expression was determined using the DDCt method using Gapdh as the housekeeping control.

Cell lysis & immunoblotting

RAW 264.7 cells were plated in 6-well microplates at 5 × 10⁵ cells/well and incubated until reaching 80% confluency. Cells were then washed with PBS and different formulations of sHDL were added at peptide concentrations of 100 μg/ml for 18 h. Cells were washed with ice-cold 1× PBS twice and lysed on ice for 30 min with 1× RIPA buffer (25 mM Tris-HCl pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS) supplemented with cOmplete™ EDTA-free protease inhibitor cocktail and PhosSTOP from Roche (IN, USA). Lysates were clarified by centrifugation at 12,000 r.p.m. for 10 min at 4°C and measured protein concentration by BCA assay. An equal amount of protein per sample was loaded on 4–15% pre-casted SDS-PAGE gels from Bio-Rad (CA, USA) with Tris/glycine/SDS buffer and proteins were transferred onto PVDF membranes. Membranes were blocked in 5% (wt/vol) BSA in Tris-buffered saline with Tween-20 (TBS-T) for 1 h at room temperature and incubated 18 h at 4°C with specific primary antibodies (ATF-3 and GAPDH) diluted in BSA. Membranes were then washed with TBS-T, incubated with secondary antibodies for 1 h, and washed with TBS-T. Images were acquired on the Protein Simple FluorChem M imaging system (CA, USA).

Analysis of NF-κB expression & proinflammatory mediators from synthetic high-density lipoprotein pre-treatment

Depending on the experiments, either HEK-Blue cells or RAW 264.7 cells were plated in a 96-well microplate at a density of 25,000 or 5 × 10⁴ cells/well, respectively, and incubated until reaching 80% confluency. Cells were washed with PBS and different formulations of sHDL were added at peptide concentrations of 10, 30 or 100 μg/ml for 18 h. After 18 h incubation, sHDL were completely removed and cells were washed with PBS. Cells were then challenged with LPS (2 ng/ml) for 18 h again. The quantification of NF-κB expression and proinflammatory cytokines were obtained as described previously.

Effect of synthetic high-density lipoprotein nanodisc infusion on the endotoxemia mice

Female C57BL/6 mice were randomly assigned to six groups: vehicle, LPS, 22A-POPC, 22A-DMPC, 22A-DPPC and 22A-DSPC, containing ten mice each. Different formulations of sHDL were administered at a dose of 10 mg/kg via intravenous injection (iv.). Subsequently, LPS (0.05 mg/kg, ip.) was administered. The vehicle group was dosed with PBS (iv.) and then PBS (ip.). The LPS control group was dosed with PBS (iv.) followed by LPS (0.05 mg/kg, ip.). All blood samples were collected from the jugular vein in heparinized BD centrifuge tubes (NJ, USA) at 2 h post-LPS challenge. Plasma samples were separated immediately by centrifugation at 14,000 r.p.m. for 10 min at 4°C and stored at -80°C until further analysis.

In vivo analysis of proinflammatory mediators

The concentrations of inflammatory mediators, TNF-α, IL-6 and MCP-1 in the plasma of the LPS co-incubation study were quantified using eBioscience Ready-Set-Go ELISA (CA, USA) per manufacturer's instruction. The concentration of inflammatory mediators of TNF-α, IL-6 and MCP-1 in the plasma of the endotoxemia model was quantified using BD Cytometric Bead Array Mouse Inflammation Kit per manufacturer's instruction and analyzed on Beckman Coulter CytoFLEX Flow Cytometer.

Survival determination

Female C57BL/6 were randomly assigned into four groups, containing ten mice each: negative control vehicle group, positive control LPS alone group, 22A-DMPC treatment group, and 22A-DSPC treatment group. The vehicle group received PBS (ip. and iv.). The LPS group was first received LPS (10 mg/kg, ip.). Once the anal temperature increased 0.5°C from LPS (approximately 15 min), PBS (iv.) was administered. The 22A-DMPC and 22A-DSPC treatment groups were first received LPS (10 mg/kg, ip.). Again, once the anal temperature increased 0.5°C from LPS (approximately 15 min), either 22A-DMPC or 22A-DSPC (10 mg/kg, iv.) was administered to the corresponding treatment group. The mice were then observed for mortality every 6 h and survival rates were recorded. Their lungs and livers were isolated and collected for histological evaluation.

Tissue preparation

Tissues were fixed in 10% neutral buffered formalin for a minimum of 24 h. Histology preparation was performed by the Unit for Laboratory Animal Medicine In Vivo Animal Core at the University of Michigan. Briefly, tissues were cassetted and processed to paraffin on an automated processor, TissueTek VIP 6 from Sakura (CA, USA). Tissues were embedded in paraffin, sectioned at 4 μm thickness on a rotary microtome, and mounted on glass slides. Slides were stained with hematoxylin and eosin on an automated histostainer and coverslipped.

Histology evaluation & images

Histological sections were evaluated using light microscopy at magnifications ranging from 20× to 600× by a board-certified veterinary pathologist using an Olympus BX45 light microscope (Tokyo, Japan) Corporation. The evaluation was performed without the knowledge of the experimental groups. Representative images were taken after histology analysis using an Olympus DP73 microscope-mounted camera with associated software, Olympus cellSens v 1.18 (Tokyo, Japan). Images were processed into figures using Adobe Photoshop CC v 19.0. Image processing was confined to global adjustments of white balance, brightness, contrast and sharpness that did not affect image interpretation. Histology was assessed based on standardized nomenclature/criteria for rodent hepatobiliary lesions [21] and literature descriptions of relevant histology in LPS challenge experiments [22–25].

Statistical analysis

Statistical differences were compared with Student's t-test for comparing two groups or with one-way analysis of variance (ANOVA) with Tuckey's post-hoc test for comparing multiple groups. All samples were performed in triplicate unless noted otherwise. p < 0.05 was considered statistically significant. The Chi-square test was used to compare survival rates. Statistical analysis was performed using GraphPad Prism 7 (CA, USA). Measurements are presented as means ± standard error of the mean unless indicated otherwise.

Results

Preparation & characterization of synthetic high-density lipoprotein nanodiscs

sHDL nanodiscs were prepared by complexing ApoA-I mimetic peptide, 22A, with various PCs (POPC, DMPC, DPPC or DSPC) using a co-lyophilization procedure. Based on preliminary studies, the optimal weight ratio of peptide to phospholipid to result in homogenous preβ-HDL-like nanodisc is at 1:2 wt/wt peptide to phospholipid [26]. To validate the morphology and confirm preβ-like discoidal shape, each sHDL formulation was observed with TEM (Figure 1A). 22A-DMPC, 22A-DPPC and 22A-DSPC were observed with typical discoidal morphology and were uniform in size. In contrast, 22A-POPC displayed heterogeneity in both size distribution and morphology. This is plausibly owing to the presence of liposomal impurities. These characteristics were further confirmed with DLS. We observed that the average diameters for 22A-POPC (13.7 ± 0.2 nm), 22A-DMPC (9.7 ± 0.2 nm), 22A-DPPC (11.2 ± 0.3 nm) and 22A-DSPC (12.2 ± 0.3 nm) were all within range of previously reported sHDL nanodisc sizes [26,27] (Supplementary Table 1). Analysis via GPC verified the observed size differences and the purity of sHDL nanodiscs (Figure 1C). All sHDL nanodiscs resulted in a similar retention time of approximately 8 min (Supplementary Table 1), with 22A-POPC exhibiting a broader peak, indicating a more heterogeneous size distribution. The small peaks appearing at approximately 11 min represent free 22A peptide, accounting for less than 2% and considered negligible for all formulations.

Figure 1. . Characterization of synthetic high-density lipoprotein.

(A) Transmission electron microscopy image of different synthetic high-density lipoprotein (sHDL). (B) Size distribution profile of different sHDL analyzed by dynamic light scattering. (C) Size distribution and purity profile of different sHDL via gel permeation chromatography.

DMPC: 1,2-dimyristoyl-sn-glycero-3-phosphocholine; DPPC: 1,2-dipalmitoyl-sn-glycero-3-phosphocholine; DSPC: 1,2-distearoyl-sn-glycero-3-phosphocholine; POPC: 1-palmitoyl-2-oleoyl-glycero-3phosphocholine.

Next, the T_m of each sHDL nanodisc was evaluated, as shown in Supplementary Table 1. POPC is composed of 16:0/18:1 fatty acids, in which the unsaturated fatty acid causes a significantly low T_m (-3.3 ± 0.5°C) [28,29]. When complexed with 22A, 22A-POPC had an observed T_m value of 0.5 ± 0.5°C. Similarly, DMPC is composed of 14:0/14:0 (T_m: 24.5°C) [29], DPPC is composed of 16:0/16:0 (T_m: 41.6°C) [29], and DSPC is composed of 18:0/18:0 (T_m: 54.5°C) [29]. Once complexed with 22A to form sHDL, we observed sHDL T_m values of 27.0 ± 0.0°C, 45.4 ± 0.4°C and 57.8 ± 1.3°C, respectively. A slight temperature rise from PC to sHDL is observed, possibly due to the addition of 22A peptide adding rigidity to the phospholipids. Based on the T_m of each sHDL, 22A-POPC and 22A-DMPC are preferentially at fluid and mobile liquid crystalline phase, while 22A-DPPC and 22A-DSPC are at rigid and constrained gel phase at physiological temperature (37°C).

We further assessed the cytotoxicity of sHDL nanodiscs in multiple cell lines and none of the formulations exhibited cytotoxicity at concentrations up to 100 μg/ml (Supplementary Figure 1). Taken together, our results indicate the successful production of non-cytotoxic sHDL with preβ-like morphology and size.

Synthetic high-density lipoprotein nanodiscs binding & neutralization of LPS

HDL can directly neutralize the TLR4-mediated inflammatory cascade by sequestering LPS in its phospholipid layer [30–32]. To investigate whether sHDL nanodiscs made from different phospholipids could successfully neutralize LPS, a TLR4 ligand, by direct interaction, we analyzed the size-exclusion profile of sHDL nanodisc after incubation with fluorescent LPS. As expected, all formulations of sHDL promoted a shift of LPS-Alexa 488 to sHDL molecular weight fraction, demonstrating successful binding of LPS to sHDL (Figure 2).

Figure 2. . Absorbance profiles of fluorescent-lipopolysaccharide bound to synthetic high-density lipoprotein.

Fluorescent-LPS (10 μg/ml) and different formulations of sHDL (1 mg/ml) were mixed and incubated for 1 h at 37°C and the sHDL-LPS mixture was analyzed by HPLC.

DMPC: 1,2-dimyristoyl-sn-glycero-3-phosphocholine; DPPC: 1,2-dipalmitoyl-sn-glycero-3-phosphocholine; DSPC: 1,2-distearoyl-sn-glycero-3-phosphocholine; LPS: Lipopolysaccharide; POPC: 1-palmitoyl-2-oleoyl-glycero-3phosphocholine; sHDL: Synthetic high-density lipoprotein.

Effect of synthetic high-density lipoprotein lipids against LPS-induced inflammation in vitro

We next sought to evaluate how the ability of sHDL to sequester LPS could be translated to inhibition of inflammatory response, as TLR4-mediated recognition of LPS is thought to be one of the key triggers of the inflammatory response [3,4]. To understand which formulations of sHDL most effectively modulate TLR4-mediated signaling, the HEK-Blue cell system was used to quantify the activity of NF-κB. HEK-blue hTLR4 cells were incubated with different sHDL at various concentrations (10, 30 and 100 μg/ml) in the presence of LPS (2 ng/ml). 22A-POPC and 22A-DMPC displayed significant concentration-dependent inhibition of NF-κB (p < 0.001) and inhibited NF-κB at all tested concentrations, 22A-DPPC inhibited activity at concentrations of 30 μg/ml and greater, and 22A-DSPC had no effect (Figure 3A). When comparing the two most potent inhibitors, 22A-POPC and 22A-DMPC, 22A-DMPC showed enhanced inhibition at all concentrations (p < 0.001).

Figure 3. . Modulation of inflammatory response with synthetic high-density lipoprotein.

(A) Activation of NF-κB of HEK-BLUE hTLR4 cells with treatment of different formulations and concentrations of synthetic high-density lipoprotein (sHDL) in a presence of LPS (2 ng/ml). (B–D) Concentration of TNF-α (B), IL-6 (C) and MCP-1 (D) of macrophages with treatment of different formulations and concentrations of sHDL in a presence of LPS (2 ng/ml).

*p < 0.05; **p < 0.01; ***p < 0.001 compared with LPS group; ^#p < 0.05 compared with 22A-DSPC.

LPS: Lipopolysaccharide; sHDL: Synthetic high-density lipoprotein.

We further examined the downstream TLR4-mediated inflammatory response by evaluating proinflammatory cytokine production. To do this, macrophages were incubated with sHDL at various concentrations in the presence or absence of LPS (2 ng/ml). Concentrations of TNF-α, IL-6 and MCP-1 in the media were quantified. 22A-POPC, 22A-DMPC, and 22A-DPPC effectively reduced LPS-induced proinflammatory mediators compared with controls (p < 0.001) and to 22A-DSPC (^#p < 0.05) (Figure 3B–D). Here, we observed that fluid liquid crystalline phase 22A-DMPC resulted in the greatest inhibition of NF-κB activity and proinflammatory cytokine production. 22A-DPPC, although in the rigid gel phase, decreased the inflammatory response at the highest concentration as T_m is near physiological temperature, while 22-DSPC showed the least effective in LPS-induced inflammatory response modulation due to its limited fluidity making incorporating LPS into its phospholipid layer difficult.

Effect of synthetic high-density lipoprotein on TLR4 recruitment into lipid raft via cholesterol efflux

Lipid raft plays a significant role in LPS-induced cellular activation. HDL promotes cholesterol efflux from macrophages via reverse cholesterol transport, compromising the integrity of lipid rafts as cholesterol is depleted leading to reduced lipid raft and TLR4 recruitment into lipid raft [8]. First, to demonstrate whether sHDL could efflux cholesterol from macrophages, we incubated different sHDL (100 μg/ml) with [³H]-cholesterol-loaded macrophages. All sHDLs exhibited significant cholesterol efflux, with 22A-DMPC (61.8 ± 2.7%) exhibiting the greatest cholesterol efflux, followed by 22A-POPC (57.2 ± 0.9%), 22A-DPPC (38.1 ± 1.4%) and 22A-DSPC (37.5 ± 1.9%) (Figure 4A). Given these results, we then examined the impact of sHDL-mediated cholesterol efflux on the alteration of lipid raft and TLR4 surface expression. Intriguingly, despite the significant cholesterol efflux observed from all sHDLs, none markedly reduced the lipid raft content, where only 22A-DMPC non-significantly reduced the lipid raft (90.1 ± 3.1%) (Figure 4B). However, a minimal reduction of lipid rafts of 22A-DMPC was capable of a significant decrease of TLR4 recruitment on the cell surface (85.1 ± 4.5%) (Figure 4C). Overall, these results indicate that only 22A-DMPC was capable of inhibiting TLR4 recruitment into lipid raft via cellular membrane cholesterol depletion.

Figure 4. . Disruption of lipid raft content and TLR-4 with synthetic high-density lipoprotein.

(A) Relative quantification of cholesterol efflux from radio-labeled cholesterol-loaded macrophages with different formulations of synthetic high-density lipoprotein (sHDL; 100 μg/ml) to the non-treated control group. Percentage cholesterol efflux from radio-labeled cholesterol-loaded macrophages was reported by liquid scintillation counting (n = 4, mean ± SEM). (B–C) Relative measurement of lipid raft content of macrophages (B) and TLR4 expression (C) with different formulations of sHDL (100 μg/ml) to the non-treated control group. The percentage of lipid raft content and TLR4 expression was reported by the mean fluorescence intensity from flow cytometry.

ns: not significant; *p < 0.05; **p < 0.01; ***p < 0.001 compared with LPS group; ^#p < 0.05 compared with 22A-DSPC.

Effect of synthetic high-density lipoprotein on ATF3 expression

ATF3 is a negative regulator of macrophage activation, acting as a negative feedback system upon TLR4 activation to limit excess production of proinflammatory cytokines [33,34]. A few studies have shown that HDL can regulate the expression of TLR-induced proinflammatory cytokines on the transcriptional level via the transcriptional repressor ATF3 in an LPS-independent manner [9–11]. To examine the ability of sHDL to promote ATF3 expression, we incubated macrophages with sHDL (100 μg/ml) and determined the mRNA and protein expression of ATF3. Notably, only 22A-DMPC induced Atf3 mRNA expression significantly, increasing fivefold in the first 1 h up to 21-fold after 4 h of incubation (p < 0.001), while 22A-POPC, 22A-DPPC or 22A-DSPC had no effect (Figure 5A). ATF3 protein expression was also examined by incubating macrophages with different sHDL for 18 h. Likewise, only 22A-DMPC induced prominent protein expression of ATF3, whereas expression was undetectable for the control group and other sHDL formulations (Figure 5B).

Figure 5. . The expression of ATF3 mRNA and protein with synthetic high-density lipoprotein.

(A) Kinetics of ATF3 mRNA expression in macrophages with different formulations of sHDL (100 μg/ml) (n = 9 ± SEM). (B) ATF3 protein expression in macrophages by immunoblot with different formulations of sHDL (100 μg/ml) followed by overnight incubation.

***p < 0.001.

Pre-treatment of synthetic high-density lipoprotein against LPS-induced inflammation in vitro

Previous assessments revealed that among the different formulations of sHDL, only 22A-DMPC reduced TLR4 recruitment and promoted ATF3 expression. Here, we further demonstrated how these mechanisms can be elucidated to modulate the inflammatory response. Macrophages were incubated with different sHDL at various concentrations (10, 30 and 100 μg/ml). After 18 h incubation, sHDL were completely removed to exclude the anti-inflammatory effects of sHDL sequestering LPS. Then cells were challenged with LPS (2 ng/ml) for 18 h again. When NF-κB expression was quantified from HEK-Blue hTLR4 cells, only 22A-DMPC showed significant inhibition in NF-κB expression in a dosage-dependent manner (p < 0.001) (Figure 6A). Similarly, when proinflammatory cytokines were measured from macrophages, again, only 22A-DMPC significantly altered TNF-α and IL-6 levels (p < 0.001) (Figure 6B–C). This study confirms that in addition to LPS sequestering, 22A-DMPC can promote anti-inflammatory activities by regulating cellular inflammatory responses.

Figure 6. . Attenuation of inflammatory response with pretreatment of synthetic high-density lipoprotein.

(A) Activation of NF-κB of HEK-BLUE hTLR4 cells. Cells were stimulated with 2 ng/ml of LPS after washing out the overnight pre-treatment of different formulations and concentrations. (B–D) Concentration of TNF-α (B), IL-6 (C), and MCP-1 (D) of macrophages. Macrophages were initially incubated with different formulations and concentrations of sHDL for overnight, then sHDL were completely removed and stimulated with LPS (2 ng/ml).

*p < 0.05; **p < 0.01; ***p < 0.001; compared with LPS group.

LPS: Lipopolysaccharide; sHDL: Synthetic high-density lipoprotein.

Effect of synthetic high-density lipoprotein on LPS-induced endotoxemia mice

The ability of sHDL to elicit an anti-inflammatory effect in vivo was examined in a murine endotoxemia model. Mice were initially administered with LPS (0.05 mg/kg) followed by different formulations of sHDL (10 mg/kg). 2 h post-LPS challenge, 22A-DMPC and 22A-DSPC caused a significant inhibition of TNF-α and IL-6 (p < 0.01 and p < 0.001, respectively), while 22A-DPPC resulted in a slight reduction of IL-6 (p < 0.05) and 22A-POPC had no effect (Figure 7A–B). In addition, 22A-DMPC and 22A-DSPC attenuated levels of MCP-1, although they were not statistically significant (Figure 7C). The good safety profiles of sHDL were shown in previous studies [35,36], as well as several preclinical and clinical studies [37,38]. The results indicated that 22A-DMPC and 22A-DSPC were the only formulations able to effectively attenuate the inflammatory response in endotoxemia mice.

Figure 7. . Serum cytokine levels postsynthetic high-density lipoprotein administration from endotoxemia model.

(A–C) Different formulations of synthetic high-density lipoprotein (sHDL) were administered at 10 mg/kg (iv.). Subsequently, LPS was challenged at 0.05 mg/kg (ip.). Blood was collected at 2 h post-administration and the levels of TNF-α (A), IL-6 (B), and MCP-1 (C) were measured (n = 10).

ns: not significant; *p < 0.05; **p < 0.01; ***p < 0.001; compared with LPS group.

LPS: Lipopolysaccharide.

Effect of synthetic high-density lipoprotein on lethal endotoxemia & organ injury

We further hypothesized that sHDL, especially 22A-DMPC, could improve the survival rate and protect organ injury from lethal endotoxemia, as it was observed to promote exceptional anti-inflammatory activities in vitro and in mild endotoxemia mice. Once we determined the appropriate concentration of LPS for lethal endotoxemia (Supplementary Figure 2), mice were administered with a lethal endotoxin concentration of LPS (10 mg/kg). Later, mice were administered with either 22A-DMPC or 22A-DSPC (10 mg/kg), and their survival was monitored for 4 days. As shown in Figure 8 & Supplementary Table 2, the survival in the 22A-DMPC treatment group drastically improved from 30% to 90% compared with the LPS-only group (hazard ratio, 0.12 [95% CI, 0.03–0.46]; p < 0.01). Additionally, the mean survival time in the 22A-DMPC treatment group was prolonged dramatically from 53.4 ± 9.3 h to 93.6 ± 2.4 h relative to LPS-only treated animals (p < 0.001). There were no statistical differences observed between the 22A-DMPC group and the vehicle group (data not shown). In contrast, the 22A-DSPC treatment group failed to improve survival exhibiting a 10% survival rate and survival time of 49.8 ± 8.5 h which is similar to the LPS group.

Figure 8. . Effect of synthetic high-density lipoprotein on lethal endotoxemia & organ injury.

(A) The effect of 22A-DMPC and 22A-DSPC treatment on LPS-induced lethality in mice. Mice were first challenged with LPS at 10 mg/kg (ip.), then administered with either 22A-DMPC or 22A-DSPC at 10 mg/kg (iv.). Survival was then monitored every 6 h for 96 h. Data show survival proportions (%; n = 10). **p < 0.01. (B) Representative histologic images of liver. (B1) Normal levels of glycogen storage (evident as irregularly vacuolated hepatocytes) and no inflammatory infiltration. (B2) LPS + 22A-DMPC treatment group with normal levels of glycogen storage and no inflammation, similar to vehicle control. (B3) LPS + 22A-DSPC treatment group with widespread glycogen depletion, evidenced as loss of vacuolated hepatocyte appearance (asterisk) and moderate inflammation, evidenced as infrequent neutrophil and macrophage infiltrates (arrow). (B4) LPS-treated positive control with glycogen depletion (asterisk) and low level of inflammatory cell infiltration evidenced as individual infiltrating neutrophils and macrophages within sinusoids (arrow). (C). Representative histologic images of lung. (C1) Normal lung showing no inflammation. (C2). No inflammation, similar to normal lung in LPS + 22A-DMPC treatment group. (C3) Mild perivascular inflammation and neutrophil emigration from vasculature (arrow) with mild interstitial inflammation (asterisk) in LPS + 22A-DSPC treatment group. (C4) Extensive neutrophil emigration (arrow) and interstitial inflammation (asterisk) following LPS treatment.

LPS: Lipopolysaccharide.

We further compared the relative severity of LPS-induced pulmonary and liver pathology in the sHDL treatment group and LPS group. As shown in Figure 8 & Supplementary Table 3, the normal liver shows normal levels of glycogen storage and no inflammatory infiltration (Figure 8B) and the normal lung exhibits as open alveolar spaces separated by thin, delicate alveolar septae without interstitial expansion (Figure 8C). The 22A-DMPC treatment group showed no signs of alterations including normal levels of glycogen storage in the liver and no inflammatory infiltrations in both liver and lung (Figure 8) displaying a similar pathology to the negative control group. In contrast, positive LPS controls had the greatest histological changes in both liver and lung, consisting of inflammatory cell infiltration by neutrophils and macrophages was present within the vasculature and perivascularly in the liver and lung, hepatocyte degeneration, hepatic glycogen depletion and hepatic extramedullary hematopoiesis of the myeloid lineage (Figure 8). 22A-DSPC treatment group resulted in similar observations to the LPS control group in histological alterations with mild inflammatory cell infiltration in the liver and lung but prominent glycogen depletion in the liver as a loss of the typical indistinctly vacuolated appearance of the hepatocyte cytoplasm (Figure 8). These findings evidently indicate that sHDL, particularly 22A-DMPC, could protect endotoxemia mice from death and organ injury.

Discussion

Nanomedicine has been proposed as a promising strategy for inflammatory diseases [39–42]. Accumulating research has suggested the therapeutic potential of HDL replacement therapy for sepsis treatment [43–45]. However, the structure–activity relationship of sHDLs in the context of sepsis treatment has not been fully revealed. We have previously explored the different phospholipid compositions that can alter the physicochemical properties, anti-inflammatory capacities, and pharmacokinetics of sHDL in vitro [20]. In the present study, we further investigated how phospholipids in sHDL impact the anti-inflammatory activities in LPS-induced inflammation. Four different phospholipids with different T_m were selected to prepare sHDLs with different membrane fluidity [31]. All sHDLs presented uniform particle distribution except for 22A-POPC. We believe that heterogeneous size distribution in 22A-POPC is possibly due to the presence of liposomal impurities attributed to fluidity and instability of phospholipid membrane at room temperature which is above POPC T_m [46].

The anti-septic mechanisms of sHDLs involve direct neutralization of endotoxin, as well as regulating inflammatory pathways on immune cells such as macrophages. The LPS neutralization capacity of different sHDLs was compared in vitro. 22A-POPC exhibits the greatest fluidity and is expected to result in the greatest LPS neutralization. Nevertheless, 22A-DMPC led to the greatest LPS neutralization, which could be potentially explained by the better particle stability of 22A-DMPC sHDL. In addition, the presence of an unsaturated chain could result in phospholipid oxidation and oxidized HDL has been reported with less fluidity [47,48]. The instability and reduced fluidity of 22A-POPC would have caused reduced LPS neutralization capability. 22A-DMPC, in contrast, exhibits a fluid liquid crystalline phase yet is relatively stable at physiological temperature since its T_m was close to physiological temperature. Biophysical characterizations should be further investigated to validate the fluidity of each sHDL through changes in fluorescence polarization anisotropy and how the fluidity influences the interaction with LPS.

Next, we sought to explore the disruption of lipid raft integrity, thereby decreasing TLR4 on the surface of the cell as TLR4 presentation is localized within lipid rafts. Numerous studies have reported differences in cholesterol efflux capability for PCs of different saturation and fatty acid chain length. For example, saturated long-chain phospholipids such as DPPC and DSPC have higher cholesterol efflux capabilities and higher physical binding affinity to cholesterols than POPC [49–52]. Our result was marginally in discordance with previous reports, as we focused on the fluidity of sHDL rather than physical cholesterol binding affinity. Analogous to our LPS binding results, we observed the greatest cholesterol efflux capacity from 22A-DMPC followed by 22A-POPC, and the least capacity from 22A-DPPC and 22A-DSPC based on its fluidity to efflux cholesterol.

Despite the significant cholesterol depletion observed with sHDLs, none of them notably reduced the lipid raft content. Nevertheless, the modest lipid raft reduction was sufficient to result in a significant decrease of TLR4 recruitment on the cell surface (Figure 4). Several past studies reported that reduction of lipid raft cholesterol occurs through ABCA1, a key transporter that primarily interacts with ApoA-I to efflux cholesterol from macrophages [8,53,54]. Therefore, one explanation for why the high cholesterol efflux from sHDLs did not translate to a dramatic decrease in lipid rafts is that sHDLs initiate cholesterol efflux through another transporter, such as ABCG1, SR-BI or by passive diffusion. ABCG1 transporters interact with mature HDL to efflux cholesterol while the SR-BI transporter is located in the caveolar region of macrophage and promotes bidirectional cholesterol efflux. Validation of 22A-DMPC or other sHDL transporter-specific efflux in macrophages would explain the relative minimal lipid raft disruption compared with the dramatically observed cholesterol efflux capacity.

We also showed stimulation of ATF3 expression by sHDL, consistent with initial reports by De Nardo et al. [9], and further demonstrated that the expression of ATF3 is critically dependent on sHDL phospholipid composition. Activation of ATF3 leads to the recruitment of histone deacetylase 1 to the promoter region of the proinflammatory cytokine gene and assists in deacetylating to limit transcriptional binding [55,56]. In addition, a recent study demonstrated that ATF3 can directly interact with the p65 subunit of NF-κB to attenuate the NF-κB activity, thus modulating the inflammatory response, rather than via indirect histone deacetylase 1 pathway [57]. We demonstrated that 22A-DMPC induced prominent expression of ATF3 mRNA and protein while other sHDLs had no effect (Figure 5). We propose that the sHDL phospholipid composition and fluidity of sHDL can also influence the ATF3 pathway in macrophages causing altered expression of ATF3.

Interestingly, we were surprised by the results from our in vivo endotoxemia model with sHDL administration. While we expected to see enhanced suppression of inflammatory response from the 22A-DMPC group due to its promising results in vitro, we did not expect that the 22A-DSPC group would also show a prominent suppression of the inflammatory response (Figure 7). 22A-DSPC in vitro was the least effective among sHDL treatments, and we believe that 22A-DSPC efficacy in the endotoxemia model owes to its exceptional half-life. The half-life of 22A-DSPC was nearly twofold longer than 22A-DMPC, observed from a pharmacokinetic study [19], which may have allowed for greater exposure and neutralization of LPS. A credible explanation is that sHDL is known to be dissociated and remodeled upon administration in vivo [20,27], thus, 22A-DSPC is likely remodeled at a slower rate due to its rigid gel phase phospholipid compared with the fluid liquid crystalline phase phospholipid of 22A-DMPC [19]. Nevertheless, the prolonged circulation of 22A-DSPC did not translate to an increase in survival following lethal doses of LPS (Figure 8) and failed to prevent organs from inflammatory cell infiltrations (Figure 8). The efficacy of 22A-DMPC was further demonstrated in a lethal endotoxemia model which increased the survival rate and time (Supplementary Table 2) along with the prevention of inflammation-induced organ injury (Supplementary Table 3). The Endotoxemia model we used in this study via direct injection of LPS is a simple and instantaneous protocol to mimic the physiology of severe sepsis and has been widely utilized to represent human sepsis for decades, nevertheless, it is also accompanied by several limitations. All sHDL formulations were well-tolerated in animal studies, which were consistent with previous preclinical and clinical studies. The endotoxemia model is consist of only a single component of LPS among the complex components from gram-negative bacteria release [58] and no source for continuous exposure of endotoxin release [59]. Therefore, it would be worthwhile to explore more physiologically relevant models such as the cecal ligation and puncture (CLP) or colon ascendants stent peritonitis (CASP).

Overall, our studies demonstrated that the sHDL formulated with DMPC, whose T_m is closer to physiological temperature, produced the most enhanced anti-inflammatory activity through LPS neutralization, reduced TLR4 recruitment and induced ATF3.

Conclusion

In conclusion, for the first time, we demonstrate that phospholipid composition drastically alters the anti-inflammatory effect of sHDL on LPS-induced inflammation both in vitro and in vivo. Our data suggested that the fluidity of sHDL due to structural variances of phospholipids critically determines the anti-inflammatory effect by promoting different anti-inflammatory mechanisms. In this study, 22A-DMPC exhibited the most fluid sHDL at physiological temperature, displaying the greatest anti-inflammatory activities through multiple mechanisms including LPS neutralization, disruption of lipid raft integrity and activation of ATF3 in vitro but also protected mice against mortality and organ injury from lethal endotoxemia. Therefore, we suggest that 22A-DMPC may be a potential therapeutic effect against LPS-induced sepsis.

Summary points.

• Synthetic high-density lipoprotein (sHDL) nanodiscs exert anti-inflammatory effects against lipopolysaccharide (LPS)-induced inflammation through various mechanisms including LPS neutralization, lipid raft disruption, NF-κB pathway inhibition and ATF3 activation.

• The phospholipids components of sHDLs affect the physiochemical properties of sHDL particles.

• All sHDLs, including 22A-POPC, 22A-DMPC, 22A-DPPC and 22A-DSPC, presented LPS binding capacity and inhibited the production of proinflammatory cytokines on LPS-treated macrophages.

• 22A-DMPC reduced TLR-4 expression on cell membranes, which may be attributed to its strong cholesterol efflux and lipid raft depletion capacity.

• 22A-DMPC greatly activated ATF3 expression, suggesting an anti-inflammatory mechanism independent of LPS binding.

• 22A-DMPC protected mice against mortality and organ injury in LPS-induced sepsis mice model.

• Overall, sHDL nanodiscs composed with DMPC, whose T_m is closest to physiological temperature, exhibited the most potent anti-inflammatory effects, which may be a potential therapeutic effect against LPS-induced sepsis.