Cholesterol Independent Suppression of Lymphocyte Activation, Autoimmunity and Glomerulonephritis by Apolipoprotein A-I in Normocholesterolemic Lupus-Prone Mice

Abstract

Apolipoprotein A-I (ApoA-I), the major lipid-binding protein of high-density lipoprotein (HDL), can prevent autoimmunity and suppress inflammation in hypercholesterolemic mice by attenuating lymphocyte cholesterol accumulation and removing tissue oxidized lipids. However, whether ApoA-I mediates immune suppressive or anti-inflammatory effects in normocholesterolemic conditions and the mechanisms involved remain unresolved. We transferred bone marrow from SLE-prone Sle123 mice into normal, ApoA-I knockout (ApoA-I^−/−) and ApoA-I transgenic (ApoA-I^tg) mice. Increased ApoA-I in ApoA-I^tg mice suppressed CD4⁺ T and B cell activation without changing lymphocyte cholesterol levels or reducing major ApoA-I-binding oxidized fatty acids. Unexpectedly, oxidized fatty acid peroxisome proliferator-activated receptor gamma (PPARγ) ligands 13-hydroxyoctadecadienoic acid (HODE) and 9-HODE were increased in lymphocytes of autoimmune ApoA-I^tg mice. ApoA-I reduced Th1 cells independently of changes in CD4⁺FoxP3⁺ regulatory T cells or CD11c⁺ dendritic cell activation and migration. Follicular helper T cells, germinal center B cells and autoantibodies were also lower in ApoA-I^tg mice. Transgenic ApoA-I also improved SLE-mediated glomerulonephritis. However, ApoA-I deficiency did not have opposite effects on autoimmunity or glomerulonephritis, possibly due to compensatory increases of ApoE on HDL. We conclude that although compensatory mechanisms prevent pro-inflammatory effects of ApoA-I deficiency in normocholesterolemic mice, increasing ApoA-I can attenuate lymphocyte activation and autoimmunity in SLE independently of cholesterol transport, possibly through oxidized fatty acid PPARγ ligands, and can reduce renal inflammation in glomerulonephritis.

INTRODUCTION

Apolipoprotein A-I (ApoA-I) is the major cholesterol and lipid binding protein component of high-density lipoprotein (HDL) and as such confers many of its protective properties in atherosclerosis (1). Although the ability of ApoA-I to counteract excessive cellular cholesterol accumulation and promote reverse cholesterol transport (RCT) have been linked to anti-inflammatory action of HDL in rodent atherosclerosis models (2, 3), other mechanisms are thought to be additionally involved (4). These include the binding and hydrolysis of oxidized lipids by HDL-associated ApoA-I and paraoxonase-1 (PON-1) enzymatic activity respectively, which contribute to anti-inflammatory effects of HDL in hypercholesterolemic mice (5). For example, oxidized metabolites of linoleic and arachidonic acids (hydroxyoctadecadienoic [HODE] and hydroxyeicosatetraenoic [HETE] acids respectively) that have pro-inflammatory effects on vascular cells are abundantly produced in atherosclerosis by the action of lipoxygenase (LO) enzymes and reactive oxygen species (ROS) (6) and are reduced by ApoA-I in concert with its vaso-protective and anti-atherogenic action in hypercholesterolemic atherosclerosis models (5, 7).

There is considerable evidence from studies in hypercholesterolemic animal models to support the notion that modulating ApoA-I could alter the levels of cholesterol in lymphoid tissue and other organs to affect immune activation and inflammation in autoimmune settings. For example, ApoA-I deficiency in hypercholesterolemic low-density lipoprotein receptor knockout mice causes excessive lymphocyte cholesterol accumulation in lymph nodes, resulting in hyperproliferation of T lymphocytes and the development of systemic autoimmunity resembling systemic lupus erythematosus (SLE) (8). Impaired immune cell cholesterol homeostasis caused by deficiency of the liver X receptor (LXR) pathway or scavenger receptor BI (a receptor involved in hepatic cholesterol ester uptake from HDL) similarly caused lymphocyte hyperproliferation and the development of SLE-like disease (9–11). The common mechanism mediating the autoimmune phenotypes in these hypercholesterolemic settings is the abnormally high immune cell cholesterol accumulation which causes immune hyperactivation at least in part through modulation of membrane raft-dependent receptor signaling (2). It has therefore been suggested that ApoA-I is essential to prevent systemic autoimmunity resulting from excessive immune cell cholesterol accumulation under conditions of hypercholesterolemia or interrupted cholesterol transport in mice. Indeed, the notion that ApoA-I suppresses autoimmunity in hypercholesterolemia by counterbalancing excessive cellular cholesterol accumulation to dampen lymphocyte activation and proliferation is also supported by data in mice showing suppressive effects of genetic disruptions in cholesterol transport pathways on cellular activation and proliferation in other systems, including the hematopoietic stem cell compartment (12).

While data in hypercholesterolemic models have provided important insights into the interactions between HDL cholesterol metabolism and autoimmunity (2), their interpretation with respect to immunomodulatory properties of ApoA-I and HDL in SLE is confounded by the extremely high levels of hypercholesterolemia and disruption of homeostatic mechanisms controlling cellular cholesterol levels in the animal models employed. As a result, questions exist over their physiological relevance, particularly considering the disappointing outcome of clinical trials in coronary heart disease patients of an ApoA-I mimetic peptide, 4F, which can provide robust anti-inflammatory and anti-atherogenic effects in hypercholesterolemic rodent models by recapitulating cholesterol and oxidized lipid binding properties of ApoA-I (13, 14). Indeed, the high expectations for 4F as a therapeutic were based largely on its effects in hypercholesterolemic animal models, with comparatively little data from normocholesterolemic animal models that are not compromised by confounding effects of hypercholesterolemia on inflammation and the immune system. Furthermore, the role of oxylipids like HODEs and HETEs that are typically involved in atherosclerosis and bound by ApoA-I in concert with its anti-inflammatory action in hypercholesterolemic mouse models have not been investigated in autoimmune settings like SLE. There is therefore no direct evidence that modulating ApoA-I can provide immune suppression in autoimmune settings without hypercholesterolemia and it is not known whether ApoA-I can modify cellular and molecular pathways to suppress relevant immune processes such as lymphocyte activation unless they are first pushed in the opposite direction (hyperactivated) by hypercholesterolemia (15).

Further underscoring the need for studies in normocholesterolemic models to test immune suppressive and anti-inflammatory properties of ApoA-I, SLE is associated with lower ApoA-I (HDL) levels and the development of dysfunctional pro-inflammatory HDL which may contribute to premature atherosclerosis in SLE patients (16–18). It has also been suggested that these reductions in ApoA-I (HDL) levels and function in SLE might feedback onto autoimmunity itself, amplifying immune activation, autoantibody production and organ inflammation. However, this has not been addressed in the context of normocholesterolemic SLE models with modified ApoA-I (HDL) levels. We therefore employed a normocholesterolemic bone marrow transfer model of SLE to test whether modulating ApoA-I levels can provide immune suppression in SLE despite an underlying predisposition for HDL to become reduced or dysfunctional, to characterize the cellular and molecular mechanisms involved, and identify lipid based immune suppressive pathways mediated by ApoA-I that could be exploited therapeutically in SLE.

MATERIALS AND METHODS

Mice

C57BL/6J (WT), C57BL/6J ApoA-I^−/−, C57BL/6J ApoA-I^tg (containing approximately twice the normal levels of plasma ApoA-I (19)) and Sle123 mice were obtained from Jackson Laboratory (Bar Harbor, ME) and colonies were maintained at the University of Alabama at Birmingham. Female mice were weaned at four weeks of age and maintained on a standard rodent chow diet for the duration of the study. All studies were conducted in conformity with Public Health Service (PHS) Policy on Humane Care and Use of Laboratory Animals and with approval from the Institutional Animal Care and Use Committee of the University of Alabama at Birmingham.

Bone marrow transplantation (BMT)

Bone marrow (BM) transplantations were conducted as previously described (20). To distinguish donor from recipient BM-derived cells colonies of WT, ApoA-I^−/− and ApoA-I^tg mice expressing the differential Ly5.1 (CD45.1) alloantigen were derived by crossing with the B6.SJL-Ptprca Pep3b/BoyJ strain of C57BL/6J mice (Jackson Laboratories). Ly5.1 C57BL/6J, Ly5.1 ApoA-I^−/− and Ly5.1 ApoA-I^tg recipients were lethally irradiated (two doses of 450 RADS ionizing radiation spaced by three hours) and transplanted at six weeks of age with 5×10⁶ BM mononuclear cells from Ly5.2 WT or Ly5.2 Sle123 donor mice. Peripheral blood samples were collected by retro-orbital puncture and processed for flow cytometric analysis of reconstitution of major hematopoietic lineages using the following conjugated antibodies: anti-CD4 [RM4–5] PerCP, anti-CD8 [53–6.7] FITC, anti-CD11b [M1/70] PE, anti-CD19 PerCP [1D3] (BD Pharmingen, San Jose, CA), anti-Ly5.1 [A20] FITC or PE, anti-Ly5.2 [104] APC (Southern Biotech, Birmingham, AL). No aberrant ApoA-I specific antibody responses were generated as a result of BM transfer into WT or ApoA-I gene targeted mice (by anti-human ApoA-I and anti-mouse ApoA-I ELISA; as in (18); data not shown).

Measurement of plasma autoantibodies

Anti-dsDNA autoantibody levels were measured by ELISA as previously described (18). Ninety-six well MaxiSorp plates (Thermoscientific, Waltham, MA) were coated with dsDNA (0.24 µg/well; Sigma, St. Louis, MO) in bicarbonate buffer (150mM Na₂CO₃, 350mM NaHCO₃ [pH 9.7]) overnight at 4°C. Plates were washed with PBS containing 0.5% Tween 20 (PBST) using an automated plate washer (BioTek, Winooski, VT) and blocked with 3% nonfat milk/PBS. Plates were subsequently incubated with diluted plasma (1:100 in PBS) for 2 hours, washed, and incubated with horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG at 1:5,000 (Southern Biotech, Birmingham, AL) in blocking buffer. Plates were washed, developed with tetramethylbenzidine substrate (Pierce, Rockford, IL), and measured on a Bio-Rad Model 680 microplate reader (Bio-Rad, Hercules, CA) at 450 nm.

Flow cytometry

Single cell suspensions were prepared from collagenase D (Roche, Indianapolis, IN) digested lymph nodes (axillary, superficial cervical, brachial, inguinal, and iliac) as described previously (18). Cells were washed with 10mM EDTA/PBS, resuspended in 5% FBS/PBS, counted and stained for flow cytometric analysis in the presence of anti CD16/32 (Fc Block) (UAB Epitope Recognition and Immunoreagent Core Facility). Data were acquired using a BD FACSCalibur flow cytometer and analyzed using FlowJo software (TreeStar, Ashland, OR). The following conjugated antibodies were used: anti-CD4 [RM4–5] PerCP, anti-CD8 [53–6.7] APC, anti-B220 [RA3–6B2] PE, anti-CD11c [HL3] PE, anti-CD69 [H1.2F3] FITC, anti-CD44 [IM7] FITC, anti-CD62L [MEL-14] PE, anti-CD86 [GL1] FITC (BD Pharmingen, San Jose, CA), F4/80 [MF48024] PE (Life technologies, Grand Island, NY), and anti-MHCII [M5/114.15.2] PerCP (BioLegend, San Diego, CA). Intracellular FoxP3 staining was performed according to the manufacturer’s instructions using a Mouse Regulatory T Cell Staining Kit (eBioscience, San Diego, CA). Intracellular BrdU staining was performed according to the manufacturer’s instructions using an APC BrdU Flow Kit (BD Pharmingen, San Jose, CA); mice were given an i.p. injection of 1 mg BrdU five hours prior to analysis. Intracellular staining for interferon-γ (IFNγ) secreting CD4⁺ Th1 and IL-17 secreting CD4⁺ Th17 cells was performed following a four hour stimulation at 37°C with 1× Cell Stimulation Cocktail in the presence of 1× brefeldin A (eBioscience, San Diego, CA) according to the manufacturer’s instructions and subsequent cell surface staining for CD4. The following antibodies were used: anti-CD4 [RM4–5] PE or FITC (BD Pharmingen, San Jose, CA), anti-IFNγ [XMG1.2] PerCP, anti-IL17 [TC11–18H10.1] FITC (BioLegend, San Diego, CA). Prior to intracellular staining, cells were fixed and permeabilized using eBioscience’s FoxP3/Transcription Factor Staining Buffer Set according to the manufacturer’s instructions (eBioscience, San Diego, CA). Cell surface staining for T follicular helper cells was performed using the following conjugated antibodies: anti-CXCR5 [2G-8] biotin followed by anti-CD4 [RM4–5] PerCP (BD Pharmingen, San Jose, CA), anti-PD1 [J43] FITC (eBioscience, San Diego, CA), and Streptavidin APC (BD Pharmingen, San Jose, CA). Intracellular staining for BCL6 was performed using anti-BCL6 [K112–91] PE (BD Pharmingen, San Jose, CA) following cell fixation and permeabilization using eBioscience’s FoxP3/Transcription Factor Staining Buffer Set according to the manufacturer’s instructions (eBioscience, San Diego, CA). Staining for germinal center B cells was performed using the following conjugated antibodies: anti-CD19 [ID3] APC, anti-CD95 [Jo2] PE (BD Pharmingen, San Jose, CA) and PNA FITC (Sigma, St. Louis, MO).

FITC skin painting assay

Dendritic cell migration assays were conducted as described in Robbiani et al (21). WT, ApoA-I^−/− and ApoA-I^tg mice were anesthetized and their sides were shaved. FITC (8 mg/ml) was dissolved in sensitizer consisting of equal volumes of acetone and dibutyl phthalate (Sigma, St. Louis, MO). Twenty-five µl FITC containing sensitizer was applied to the right dorsal side of the mouse, while 25µl sensitizer alone was applied to the left dorsal side. Eighteen hours later, draining lymph nodes (axillary, brachial, inguinal) from each side (“control LNs” from left and “FITC LNs” from right) were removed and single cell suspensions were prepared by collagenase D digestion as described above for counting and flow cytometric analysis with anti-CD11c [HL3] PE antibody (BD Pharmingen, San Jose, CA).

Kidney morphometric analysis and histopathology

Kidneys were fixed in neutral buffered 10% formalin and processed routinely for paraffin sectioning. Five µm thick sections were stained with periodic acid-Schiff stain (PAS) and hematoxylin. Slides were examined with experimental classifications concealed from the observer. For morphometric analysis, nine glomeruli were evaluated per section as previously described (22). To minimize selection bias, glomeruli were selected beginning at the capsule at the 12 o’clock position of the section and proceeding clockwise, selecting only centrally sectioned glomeruli having open capillary lumina and avoiding collapsed or tangentially sectioned glomeruli. Images of selected glomeruli were made with a SPOT Insight 4 megapixel digital camera (Diagnostic Instruments, Sterling Heights, Michigan) at an objective magnification of 40×. After adjustment for sharpness, contrast, and brightness to maximize visibility of PAS staining, images were analyzed by histomorphometry for total glomerular area, PAS stained area, and nuclear area with Image Pro Plus v6.2 image analysis software (Media Cybernetics, Silver Spring, Maryland) by thresholding grayscaled images, color segmentation of color images, or both. Glomeruli were also evaluated by subjective scoring, without the pathologist’s knowledge of the genotype of the mice, as previously described (22, 23). Glomerular cellularity, neutrophil infiltration, necrosis, hyaline deposits, mesangial expansion, cellular crescent formation, and interstitial inflammatory cell accumulation were evaluated. Each was scored 0, 1, 2, or 3 for normal, mild, or severe, respectively. Equal numbers of glomeruli from superficial, middle, and deep cortex were examined. Only glomeruli sectioned through the approximate center of the tuft and including the base of the tuft were included. Overall glomerular activity scores for each mouse were calculated as the mean of summed individual scores for each glomerulus, with scores for necrosis and crescent formation each weighted by a factor of 2 as follows (proliferation [cellularity] + neutrophil infiltration + necrosis² + hyaline deposits + mesangial expansion + cellular crescent formation² + interstitial inflammatory cell infiltration).

Kidney immunofluorescence analysis

Eight µm cryosections prepared from kidneys in optimum cutting temperature compound (Tissue-Tek) were fixed in acetone and stained with goat anti-mouse IgG-Alexa594 (Life technologies, Grand Island, NY), rat anti-mouse CD4 [GK1.5], rat anti-mouse CD11b [M1/70] (BD Pharmingen, San Jose, CA), or rat anti-murine neutrophil [NIMP-R14] (Thermo Fisher) (24) antibodies. Goat anti-rat IgG-Alexa555 conjugated secondary antibody (Life technologies, Grand Island, NY) was used for sections stained with anti-CD4, anti-CD11b and anti-murine neutrophil primary antibodies. Color images were captured on an Olympus BX60 microscope using a 40× objective. For quantification of IgG and CD11b, average fluorescence intensity was quantified from six representative 400× microscopic fields per kidney cryosection using the histogram function of Adobe Photoshop CS5.1 (Adobe Systems, San Jose, CA) as previously described (25). For CD4 and neutrophil quantification, the numbers of CD4⁺ T cells and neutrophils were enumerated from six representative 400× microscopic fields per kidney cryosection by counting Hoechst stained nuclei that were positive for cytoplasmic CD4 or NIMPR14 staining.

Urine albumin and creatinine measurement

Urine albumin was quantified using a Mouse Albumin ELISA Quantitation Set (Bethyl Laboratories, Inc.) according to the manufacturer’s instructions at a sample dilution of 1:1000. Urine albumin and creatinine was measured in spot urine samples taken at euthanization for the control and Sle123 BMT animal groups. Urine creatinine was determined by underivatized, stable isotope dilution LC-MS/MS as described in Young et al (26). Briefly, samples were separated using a Waters 2795 separations module (LC) (Waters, Milford, MA) with a TSK-Gel Amide 80 column (Tosoh Bioscience, Tokyo, Japan) with an isocratic flow of 10mM ammonium acetate in 65% acetonitrile, and creatinine was subsequently detected by Waters Quattro micro API (MS/MS) in the positive ion mode.

Lipoprotein and lipid analyses

Plasma lipids were processed for measurement of total cholesterol, esterified cholesterol, HDL cholesterol, LDL-cholesterol, triglycerides and free fatty acids by enzymatic procedures as previously described (27). To obtain FPLC plasma cholesterol profiles, plasma (200 µl pooled from 2 mice) was fractionated using a Superose 6 column (Amersham Biosciences, GE Life Sciences, Pittsburgh, PA) on a Biologic DuoFlow FPLC system (Bio-Rad, Hercules, CA) as previously described (20). One-half ml fractions were collected and analyzed for total cholesterol content using enzymatic cholesterol reagents (Infinity Cholesterol Reagent TR13421, Thermo Scientific). Total plasma and fractionated plasma PON1 paraoxonase activity were measured using 1 mM Paraoxon (Supelco Analytical, Sigma Aldrich, St. Louis, MO) in 50 mM Tris/HCl pH 8.0, 1mM CaCl₂ as previously described (18).

Cholesterol Analysis by Gas Chromatography Mass Spectrometry (GC/MS)

GC-MS cholesterol analyses were performed in the Mass Spectrometer Facility (MJT) of the Comprehensive Cancer Center of Wake Forest School of Medicine. Lipids were extracted from ten million lymph node lymphocyte cells prepared by forcing lymph nodes through 100µm strainers as described by Wilhelm et al (8). Lipids were resuspended in 200µl hexane and 100µl analyzed on a Finnigan TSG Quantum XLS mass spectrometer interfaced to a trace gas chromatograph as described by Wilhelm et al (8). To quantify esterified cholesterol levels, a 50µl hexane lipid aliquot was also saponified at room temperature for 2 hours with methanolic KOH as described by Nordskog et al (28) and then extracted with hexane as described by Wilhelm et al (8) prior to GC-MS.

SDS-PAGE and immunoblotting

Plasma FPLC fractions (5µl) were separated on 4–15% mini-Protean TGX gels (Bio-Rad, Hercules, CA) and transferred to Immobilon P membranes (Millipore, Billerica, MA) as previously described (20). Membranes were blocked with PBST containing 5% nonfat milk by weight, incubated with primary antibodies diluted 1:2,000 in blocking buffer [Rabbit anti-mouse ApoA-I, rabbit anti-mouse ApoE, and/or goat anti-human ApoA-I (Meridian Life Sciences, Inc., Memphis, TN)] followed by HRP-conjugated goat anti-rabbit IgG (Bio-Rad, Hercules, CA) or donkey anti-goat IgG (Southern Biotech, Birmingham, AL) secondary antibodies diluted 1:4,000 in blocking buffer. Immunoreactive proteins were visualized using enhanced chemiluminescence (ImmunStar, Bio-Rad, Hercules, CA). Neat plasma (1µl) samples were separated on 4–15% mini-Protean TGX gels (Bio-Rad, Hercules, CA) and transferred to Immobilon FL membranes (Millipore, Billerica, MA). Membranes were blocked with PBST containing 5% nonfat milk by weight, incubated with primary antibodies diluted 1:2,000 in blocking buffer [Rabbit anti-mouse ApoA-I, rabbit anti-mouse ApoE, goat anti-human ApoA-I (Meridian Life Sciences, Inc., Memphis, TN) or rabbit anti-mouse ApoA-II (29)] followed by IRDye 800CW-conjugated donkey anti-rabbit IgG or IRDye 680RD-conjugated donkey anti-goat IgG secondary antibodies (Li-Cor, Lincoln, NE) diluted 1:40,000 in blocking buffer. Immunoreactive proteins were visualized using Li-Cor’s Odyssey CLx Infrared Imaging System (Li-Cor, Lincoln, NE). Acquired images were adjusted for optimal brightness and contrast, and band intensities were quantified using Image Studio Analysis Software v3.1 (Li-Cor, Lincoln, NE).

HODE/HETE measurement

For plasma HODE/HETE measurement, butylated hydroxytolune (BHT) was added to freshly isolated 100µl plasma aliquots to a final BHT concentration of 100µM and overlayed with argon gas prior to storage at −80°C. Lipids were extracted from 100µl plasma aliquots and analyzed by LC-MS/MS as described in Imaizumi et al (7) using a Shimadzu prominence HPLC system and an AB/Sciex API-4000 Q TRAP mass spectrometer. Aliquots (100µl) of plasma were acidified by addition of 1.7ml of pH2.3 H₂O, vortexed briefly and incubated on ice for 15 minutes. Deuterated standards for each measured HODE and HETE (20ng of each) (Cayman Chemical, Ann Arbor, MI) were spiked into samples prior to extraction. Samples were subsequently added to solid phase extraction (SPE) columns (Oasis® HLB 30mg 1cc; Waters, Milford, MA #WAT094225) while pulling a slow vacuum so that each sample took approximately 2 minutes to pass through. After washing SPE columns with 1ml 5% methanol, lipids were eluted with 1ml 100% methanol, dried under argon and resuspended in 100µl 80% methanol. For HODEs and HETEs, lipids (50µl) were injected onto a Luna C18(2)-HST reverse-phase LC column (2×100mm, 2.5µ; Phenomenex, Torrance, CA) using a Shimadzu auto-sampler with gradient elution at 0.3 ml/minute (mobile phase A: 0.1% formic acid in H₂O, mobile phase B: 0.1% formic acid in acetonitrile). Gradient elution was as follows: from 0 min to 2 min hold at 50% B, from 2–3 min increase to 60% B, from 3–15 min an increase to 65% B, from 15–17 min hold at 65% B, from 17–19 min an increase to 100% B, from 19–21 min hold at 100% B, from 21–23 min decrease back to 50% B and a hold for the system to equilibrate to initial conditions until 27 min. For Linoleic acid and arachidonic acid, lipids (50µl) were injected onto a Luna C12 Proteo reverse-phase LC column (2×50mm, 4µ; Phenomenex, Torrance, CA) using a Shimadzu auto-sampler with gradient elution at 0.25 ml/minute (mobile phase A: 10mM ammonium acetate in H₂O, mobile phase B: 10mM ammonium acetate in isopropyl alcohol). Gradient elution was as follows: from 0 min to 10 min increase from 35% B to 100% B, from 10–11 min decrease from 100% B to 35% B, from 11–17 min hold at 35% B for the system to equilibrate to initial conditions. A standard curve (0.1, 1, 2.5, 5, 7.5, 10, 25, 50, 75, 100, 250, 500, 750, 1000ng/ml) of a mixture of unlabeled linoleic acid, arachidonic acid, HODEs and HETEs (13-HODE, 9-HODE, 15-HETE, 5-HETE, 12-HETE) were also run for quantification of individual fatty acid and HODE/HETE species. Each point of this standard curved was spiked with deuterated linoleic acid, arachidonic acid, HODEs/HETEs standards exactly as described above for samples (20ng of each deuterated fatty acid and HODE/HETE per standard curve point). The transitions monitored were the following mass-to-charge ratio (m/z): m/z 295.1→194.8 for 13-HODE; 295.0→171.0 for 9-HODE; 319.1→219.0 for 15-HETE; 319.1→115.0 for 5-HETE; 319.1→179.0 for 12-HETE; 299.0→197.9 for 13(S)-HODE-d₄; 299.1→172.0 for 9(S)-HODE-d₄; 327.1→226.1 for 15(S)-HETE-d₈; 327.1→116.0 for 5(S)-HETE-d₈; 327.1→184.0 for 12(S)-HETE-d₈, 303.0→59.0 for arachidonic acid, 311.0→267.0 for arachidonic acid-d₈, 279.0→261.0 for linoleic acid, 283.0→265.0 for linoleic acid-d₄.

Statistics

Statistical analysis was performed using SigmaPlot 11 (Systat Software, Inc., San Jose, CA). One-way ANOVA was used for multiple comparisons and Student t-test was used for single comparisons. For all statistical analyses, p<0.05 was considered significant.

RESULTS

Transgenic expression of ApoA-I suppresses lymphocyte activation and autoimmunity

We transplanted bone marrow (BM) from Ly5.2 SLE-prone Sle123 mice into 6 week old Ly5.1 wild-type (WT), Ly5.1 ApoA-I knockout (ApoA-I^−/−) or Ly5.1 human ApoA-I transgenic (ApoA-I^tg) mice (Sle123→WT, Sle123→ApoA-I^−/− and Sle123→ApoA-I^tg). BM transplanted (BMT) animals were subsequently bled at intervals and analyzed at 38 weeks of age (32 weeks post-BMT) to measure autoimmune phenotypes and glomerulonephritis (GN). Groups of Ly5.1 WT, Ly5.1 ApoA-I^−/− and Ly5.1 ApoA-I^tg mice were also transplanted with Ly5.2 WT BM (WT→WT, WT→ApoA-I^−/− and WT→ApoA-I^tg control BMT groups) in order to control for effects of lethal irradiation and hematopoietic reconstitution. A schematic representation of this BM transplantation (BMT) scheme is shown in Figure 1A. Myeloid and lymphoid repopulation was equivalent between WT, ApoA-I^−/− and ApoA-I^tg mice transplanted with Sle123 BM (Table I) such that any differences in autoimmune phenotypes could not be ascribed to differences in hematolymphoid reconstitution with Sle123 cells.

Figure 1.

(A) Schematic of BMT scheme used to generate chimeric autoimmune mice (animals were transplanted with WT or Sle123 BM at 6 weeks of age). (B) Plasma HDL-C levels in the indicated control, control WT BMT and Sle123 BMT mice (38 weeks of age, 32 weeks post-BMT, n=8). (C) Plasma PON1 activity against paraoxon substrate in the indicated control, control WT BMT and Sle123 BMT mice (38 weeks of age, 32 weeks post-BMT, n=8). Note autoimmune-mediated reductions in HDL-cholesterol (HDL-C) and HDL-associated PON1 activity in Sle123→WT mice (in dashed box). (D) Plasma levels of mouse ApoA-I (mApoA-I) and human ApoA-I (hApoA-I) in 4 control and 4 Sle123 BMT mice (38 weeks of age, 32 weeks post-BMT). Graph showing band intensities for plasma mApoA-I is shown in (E). Note reduction in plasma mApoA-I in Sle123→WT mice compared to their control WT counterparts. ApoA-I^tg animals have markedly lower endogenous mApoA-I levels as previously reported (19). (All data mean ± SD, *p<0.05 by one way ANOVA within WT, ApoA-I^−/− or ApoA-I^tg group, ‡p<0.05 by T test compared to control group).

Table I.

Hematopoietic reconstitution (%Ly5.2 in peripheral blood cells of the indicated lineages) in WT, ApoA-I^−/− and ApoA-I^tg BM transplanted mice (38 weeks of age).

			Recipient (Ly5.1)
			WT	ApoA-I−/−	ApoA-Itg
Donor (Ly5.2)	WT	4 weeks post-BMT	23.0 ±10.7 (n=6)	31.9 ±15.6 (n=6)	28.0 ±15.2 (n=6)	CD4+
			29.2 ± 8.9 (n=6)	30.8 ±13.5 (n=6)	32.5 ±14.1 (n=6)	CD8+
			93.7 ±3.1 (n=6)	99.0 ±0.4 (n=6)	96.8 ±0.4 (n=6)	CD19+
			92.5 ±5.2 (n=6)	97.3 ±1.7 (n=6)	93.4 ±1.6 (n=6)	CD11b+
		32 weeks post-BMT	88.8 ±4.8 (n=10)	92.0 ±1.9 (n=10)	92.3 ±1.8 (n=10)	CD4+
			83.8 ±6.5 (n=10)	89.9 ±3.4 (n=10)§	82.6 ±5.5 (n=10)	CD8+
			97.6 ±1.9 (n=10)	99.0 ±0.4 (n=10)	95.0 ±1.3 (n=10)	CD19+
			93.1 ±5.1 (n=10)	97.3 ±3.1 (n=10)	95.0 ±1.3 (n=10)	CD11b+
	Sle123	4 weeks post-BMT	46.6 ±7.3 (n=12)‡	51.1 ±7.1 (n=12) ‡	40.4 ±6.3 (n=12)*	CD4+
			45.7 ±9.6 (n=12)‡	52.3 ±8.7 (n=12)‡	39.5 ±8.5 (n=12)*	CD8+
			96.7 ±1.3 (n=12)	97.1 ±1.0 (n=12)	97.0 ±0.6 (n=12)	CD19+
			95.8 ±3.1 (n=12)	96.3 ±1.5 (n=12)	97.1 ±0.9 (n=12)	CD11b+
		32 weeks post-BMT	88.1 ±5.9 (n=12)	88.3 ±4.1 (n=12)‡	89.5 ±4.6 (n=12)	CD4+
			89.0 ±4.8 (n=12)	89.3 ±3.3 (n=12)	87.2 ±6.4 (n=12)	CD8+
			94.5 ±2.8 (n=12)‡	97.8 ±0.9 (n=12)	97.4 ±0.7 (n=12)	CD19+
			97.5 ±1.2 (n=12)‡	96.0 ±2.6 (n=12)	94.1 ±3.5 (n=12)	CD11b+
			%Ly5.2

^*p<0.05 compared to ApoA-I^−/− mice (but not WT mice) of the same BMT donor group.

^‡p<0.05 compared to the same host genotype transplanted with WT BM at the same time-point.

^§p<0.05 (One Way ANOVA) compared to WT and ApoA-I^−/− mice of the same BMT donor group.

As expected, because BM-derived cells do not express ApoA-I (30) and ApoA-I is important for stabilizing PON-1 activity on HDL particles (31), plasma levels of HDL-cholesterol (HDL-C) and HDL-associated PON-1 activity were decreased in transplanted as well as control non-BMT ApoA-I^−/− mice and increased in transplanted as well as control non-BMT ApoA-I^tg animal groups (Figure 1B and C and Table II). As previously reported (32), plasma LDL-C and triglyceride levels were increased (Table II) and endogenous mouse ApoA-I (mApoA-I) levels were significantly reduced by post-transcriptional mechanisms in the ApoA-I^tg animals (19) (Figure 1D–E). Despite these reductions in endogenous mApoA-I, total ApoA-I levels (mApoA-I plus human transgenic ApoA-I) are increased approximately 2-fold in ApoA-I^tg mice (19).

Table II.

Lipid profiles in control, control WT BMT and Sle123 BMT WT, ApoA-I^−/− and ApoA-I^tg mice (38 weeks of age) (n=8).

	WT	ApoA-I−/−	ApoA-Itg	WT → WT	WT → ApoA-I−/−	WT → ApoA-Itg	Sle123 → WT	Sle123 → ApoA-I−/−	Sle123 → ApoA-Itg
HDL-C (mg/dl)	78.6 ±8.9	26.5 ±7.1‡	159.5 ±22.6‡	69.4 ±8.6	22.8 ±4.4‡	181.4 ±35.0‡	62.6 ±7.7*	28.7 ±6.1‡	167.2 ±22.7‡
LDL-C (mg/dl)	30.5 ±7.2	19.6 ±5.1	143.1 ±51.5‡*	29.2 ±6.8	12.5 ±2.1	62.1 ±27.9‡	23.6 ±4.7	26.5 ±19.4	50.8 ±16.8‡
TG (mg/dl)	24.7 ±11.1	27.9 ±15.3	72.5 ±24.0‡	36.4 ±10.6	26.8 ±5.9	46.7 ±30.0	38.8 ±50.4	25.0 ±13.3	62.5 ±20.8‡
FFA (mg/dl)	24.2 ±6.2	19.6 ±2.6	33.0 ±8.6	28.0 ±6.2	15.5 ±6.4	20.3 ±6.0	27.5 ±10.2	20.8 ±4.7	22.2 ±6.3
PON-1 (nmol/min/ml)	153.3 ±34.3	39.6 ±9.0‡	222.6 ±22.8‡	164.3 ±10.6	62.2 ±17.6‡	338.7 ±82.7‡*	110.9 ±28.3*	47.6 ±15.2‡	209.2 ±53.1‡

HDL-C = high-density lipoprotein cholesterol; LDL-C = low-density lipoprotein cholesterol; TG = triglycerides; FFA = free fatty acids; PON-1=paraoxonase-1. Values are the mean ± SD.

^*p<0.05 (One Way ANOVA) within the same host genotype group (WT, ApoA-I^−/− or ApoA-I^tg).

^‡p<0.05 (One Way ANOVA) within the same control non-BMT, control WT BMT or Sle123 BMT group.

Plasma levels of anti-dsDNA IgG autoantibodies were significantly lower in Sle123→ApoA-I^tg mice compared to their Sle123→WT and Sle123→ApoA-I^−/− counterparts at 32 and 38 weeks of age (26 and 32 weeks post-BMT respectively) (Figure 2A). However, Sle123→ApoA-I^−/− animals had comparable levels of autoantibodies to those in Sle123→WT animals at all time-points measured (Figure 2A). Although control WT BMT animals had reduced numbers of lymph node B cells and CD4⁺ T cells compared to their control non-BMT counterparts, presumably due to lethal irradiation and transplantation, autoimmune associated increases in lymphocytes were evident in the WT mice transplanted with Sle123 BM (Figures 2B and 3B). B cell numbers and activation were reduced in lymph nodes of Sle123→ApoA-I^tg mice by 41% and 57% respectively (Figure 2B). Lymph node CD4⁺ (but not CD8⁺) T cell numbers and activation were reduced in Sle123→ApoA-I^tg mice by 38% and 43% respectively (Figure 3). Similar suppressive effects of transgenic ApoA-I expression were observed in splenic B and CD4⁺ T cells from Sle123→ApoA-I^tg mice (B cells reduced by 47%, CD4⁺ T cells reduced by 57%, CD8⁺ T cells not reduced). ApoA-I deficiency did not result in opposite effects, demonstrating that, unlike in hypercholesterolemic SLE models (8, 10, 11, 15), ApoA-I deficiency does not exacerbate autoimmunity under normal (normocholesterolemic) conditions in mice.

Figure 2.

Transgenic ApoA-I expression reduces B cell activation and anti-ds DNA IgG autoantibodies in Sle123→ApoA-I^tg mice. (A) Anti-dsDNA autoantibody levels in control (n=7), WT BMT control (n=7) and Sle123 BMT (n=10) animal groups. Ages of animals in weeks on y-axis. (B) Flow cytometric analysis of lymph node B220⁺ B lymphocyte activation (measured by CD69 expression) in 38 week old control (n=12–15), control WT BMT (n=10–12) and Sle123 BMT (n=12–15) mice. Reduced numbers of B cells and lower B cell activation in Sle123→ApoA-I^tg mice. (All data mean ± SD, *p<0.05 by one way ANOVA within control, control WT BMT or Sle123 BMT groups, ‡p<0.05 by T test compared to control).

Figure 3.

Reduced numbers and activation of CD4⁺ T lymphocytes in Sle123→ApoA-I^tg mice. (A) Flow cytometric analysis (showing gates) of lymph node CD4⁺ T lymphocyte activation (measured by CD69 and CD44^highCD62L^low expression) in 38 week old control, control WT BMT and Sle123 BMT mice. (B) Numbers of CD4⁺ T lymphocytes in lymph nodes of the indicated mice (n=12–15 control, n=10–12 control WT BMT, n=10–12 Sle123 BMT mice). (C–D) Percent and absolute numbers of CD69 expressing CD4⁺ T lymphocytes in lymph nodes of indicated mice. (E) CD44^highCD62L^low CD4⁺ T lymphocytes in lymph nodes of indicated mice. Note that ApoA-I deficiency in this normocholesterolemic setting does not augment T cell activation. (All data mean ± SD, *p<0.05 by one way ANOVA within control, control WT BMT or Sle123 BMT groups, ‡p<0.05 by T test compared to control).

Consistent with fewer activated CD4⁺ T cells in Sle123→ApoA-I^tg mice, numbers of proliferating CD4⁺ (but not CD8⁺) T cells were also reduced (by 58%) (Figure 4A). Importantly, this was not associated with a decreased rate of proliferation, as the level of BrdU incorporation into the CD4⁺ T cells as measured by BrdU mean fluorescence intensity (MFI) was equivalent between Sle123→WT, Sle123→ApoA-I^−/− and Sle123→ApoA-I^tg mice (59±9, 56±14 and 51±11 respectively). Thus, unlike in hypercholesterolemic mice in which the proliferative rate of T lymphocytes was increased by ApoA-I deficiency (8), loss or transgenic expression of ApoA-I in our normocholesterolemic Sle123 BMT model had no significant effect on the rate of T lymphocyte proliferation.

Figure 4.

Reduced frequency of proliferating CD4⁺ T cells, lower interferon-γ (IFNγ) secreting TH1 (but not IL-17 secreting TH17) and follicular T helper (Tfh) cells in Sle123→ApoA-I^tg mice. (A) Flow cytometric analysis of in vivo BrdU incorporation into lymph node CD4⁺ T cells 6 hours following IP BrdU injection (n=8) (‡p<0.05 by T test compared to control non-BMT, *p<0.05 by one way ANOVA within control, control WT BMT or Sle123 BMT groups). (B) Numbers of IFNγ secreting and IL-17 secreting CD4⁺ T cells in lymph nodes from the indicated 38 week old mice (representative flow cytometric plots shown alongside). (C) CXCR5⁺PD-1⁺BCL-6⁺ follicular T helper (Tfh) cells and (D) CD19⁺CD95⁺PNA⁺ germinal center B (GC B) cells in 38 week old WT, WT→WT, Sle123→WT, Sle123→ApoA-I^−/− and Sle123→ApoA-I^tg mice. (n=6) (All data mean ± SD, ‡p<0.05 by one way ANOVA compared to WT and WT→WT control groups, *p<0.05 by one way ANOVA within Sle123 BMT group).

Similarly to a previous study in SLE-prone gld mice (18), plasma HDL-C levels, mApoA-I and PON-1 activity were reduced as a result of autoimmunity in Sle123→WT mice (Figure 1B–D, Supplemental Figure IA–B, and Table II). Indeed, reductions in plasma HDL-C in autoimmune Sle123→WT mice were inversely related to certain parameters of autoimmunity, including anti-dsDNA autoantibodies and CD4⁺ T cell activation (Supplemental Figure ID). Autoimmune mediated reduction of endogenous mApoA-I was also seen in Sle123→ApoA-I^tg animals (Figure 1D–E and Supplemental Figure IA–B).

Interestingly, we observed effects of ApoA-I deficiency in control non-autoimmune WT BMT mice, including elevations in plasma PON-1 activity (↑57%) (Figure 1C and Table II) and an approximately 5-fold increase in numbers of lymph node B cells (Figure 2B). CD4⁺ T cells were increased in WT→ApoA-I^−/− mice (3-fold) and non-BMT ApoA-I^−/− mice (2-fold) (Figure 3B) compared to their WT counterparts, suggesting this was an intrinsic property of ApoA-I^−/− mice. However, CD4⁺ lymphocyte expansion associated with the development of autoimmunity obscured this phenotype in Sle123 BMT mice (Figure 3B). In WT→ApoA-I^tg animals, we observed slightly increased plasma HDL-C (↑14%) and higher PON-1 activity (↑52%) compared to their control non-BMT ApoA-I^tg counterparts (Figure 1B–C and Table II).

Reduced Th1 and Tfh cells in Sle123→ApoA-I^tg mice

Because the reduction in CD4⁺ T cells in Sle123→ApoA-I^tg mice was relatively mild (38%↓), we reasoned that it might reflect a suppressive effect on the activation and/or differentiation of select CD4⁺ T cell populations, as opposed to a generalized reduction in all CD4⁺ cells. Interferon-γ secreting Th1 cells, but not IL-17 secreting Th17 cells, were reduced by half in Sle123→ApoA-I^tg mice (Figure 4B). CD4⁺CXCR5⁺PD1⁺BCL6⁺ Tfh cells provide help to B cells for germinal center (GC) B cell formation and antibody production in response to foreign antigens as well as for pathogenic autoantibody production in SLE (33). Tfh cells were also significantly reduced in Sle123→ApoA-I^tg mice (↓65%) (Figure 4C). Consistent with their role in GC formation (33), reduced Tfh cells resulted in lower numbers of GC B cells (↓47%) (Figure 4D). Again, ApoA-I deficiency did not result in opposite effects to those in ApoA-I^tg mice. In addition, immune suppressive effects of ApoA-I were mediated independently of changes in the numbers of CD4⁺FoxP3⁺ regulatory T cells (data not shown). Finally, although ApoA-I was shown to modulate CD11c⁺ DC lymph node migration in hypercholesterolemic settings as measured by a FITC skin painting assay (34), we did not observe any difference in CD11c⁺ DC migration between WT, ApoA-I^−/− and ApoA-I^tg mice using the same assay (Supplemental Figure IIA).

Immune suppressive action of ApoA-I is associated with modulation of cellular oxidized lipids but not cholesterol

ApoA-I deficiency in hypercholesterolemic mouse models and causes systemic autoimmunity by promoting lymphocyte cholesterol accumulation, resulting in modulation of lipid raft dependent antigen receptor-mediated activation (2, 8, 10, 11, 35). Conversely, increasing ApoA-I in hypercholesterolemic mice may regulate antigen receptor-mediated lymphocyte activation by reducing cellular cholesterol to modulate lipid rafts and antigen receptor activation in the opposite direction (2, 8, 35). However, cholesterol levels in lymphocytes from autoimmune and control WT, ApoA-I^−/− and ApoA-I^tg mice were not significantly different as measured by GC-MS (Figure 5A). Thus, immune suppression by ApoA-I in normocholesterolemic SLE mice was independent of changes in lymphocyte cholesterol homeostasis.

Figure 5.

(A) Immune suppression in Sle123→ApoA-I^tg mice is not associated with reduced lymphocyte cholesterol content. Total (free plus esterified) and free cholesterol content of lymphocytes from the indicated control, WT BMT and Sle123 BMT mice (mean ±SD, n=5). (B) Levels of the indicated HODEs and linoleic acid in lymphocytes from the indicated WT BMT and Sle123 BMT mice (All data mean ± SD, n=5). Note that increases in lymphocyte 13-HODE and 9-HODE levels in Sle123→ApoA-I^tg mice are not simply due to elevated levels of its precursor, linoleic acid. (‡p<0.05 by T test compared to control non-BMT, *p<0.05 by one way ANOVA within Sle123 BMT group).

HODEs and HETEs derived from 12/15-LO enzymatic and ROS-mediated oxidation of linoleic and arachidonic acids respectively have been widely used as measures of oxidative stress (36, 37) and the reduction of their levels in plasma and tissues has been used as a surrogate measure of the anti-inflammatory action of ApoA-I in hypercholesterolemic rodent models (5, 7). Surprisingly, ApoA-I mediated immune suppression in Sle123→ApoA-I^tg mice was associated with increased lymphocyte levels of 13-HODE and 9-HODE (Figure 5B). Linoleic acid was not significantly different between WT, ApoA-I^−/− and ApoA-I^tg animals (Figure 5B), suggesting that the increases in 13-HODE and 9-HODE were not simply due to higher precursor fatty acid levels. Lymphocyte levels of 5-HETE, 12-HETE, 15-HETE and their precursor, arachidonic acid, were not significantly different between WT, ApoA-I^−/− and ApoA-I^tg mice (data not shown).

ApoA-I improves glomerulonephritis in Sle123 BMT mice

Glomerulonephritis (GN) was significantly improved in autoimmune Sle123→ApoA-I^tg mice but not exacerbated in their Sle123→ApoA-I^−/− counterparts (Figure 6). Histological analysis of glomeruli from control and BMT groups showed significant increases in glomerular area and overall GN activity scores in Sle123 BMT mice compared to their control non-BMT and control WT BMT counterparts (Figure 6A–B). These measures of GN were significantly improved in Sle123→ApoA-I^tg mice, while ApoA-I deficiency had no significant effect on GN (Figure 6A–B). GN scores were also moderately increased in control WT BMT groups compared to their control non-BMT counterparts (Figure 6B), reflecting deleterious effects of lethal irradiation and/or BM reconstitution on renal pathology. Albuminuria in Sle123→ApoA-I^tg mice was slightly reduced, but this was only statistically significant in a pair-wise comparison with Sle123→WT animals (p=0.052 by one-way ANOVA between the three Sle123 BMT groups) (Figure 6C). Improvement of GN in Sle123→ApoA-I^tg mice was associated with marked reductions in glomerular IgG deposition (Figure 7A), consistent with the lower plasma autoantibodies in these animals (Figure 2A). Numbers of kidney infiltrating CD4⁺ T cells (Figure 7B), CD11b⁺ cells and NIMPR4⁺ neutrophils (Figure 8A–B) were also reduced in Sle123→ApoA-I^tg mice, consistent with reduced renal inflammation.

Figure 6.

Protection against glomerulonephritis (GN) in Sle123→ApoA-I^tg mice. (A) Periodic Acid Schiff (PAS) stained kidney sections showing representative glomeruli from the indicated 38 week old non-BMT control, control WT BMT and Sle123 BMT mice. (B) Average glomerular area (upper panel) and overall GN histological scores lower panel) in the indicated control (n=5), control BMT (n=5) and Sle123 BMT (n=6) groups (9 glomeruli scored per mouse) (‡p<0.05 by T test compared to indicated control, *p<0.05 by one way ANOVA within Sle123 BMT group). (C) Increases in urine albumin/creatinine ratio (ACR) in the indicated 38 week-old control (n=4), control BMT (n=4) and Sle123 BMT (n=6–8) animals (all data mean ± SD) (*p=0.052 by one way ANOVA, ‡p=0.049 compared to Sle123→WT by T test).

Figure 7.

Reduced glomerular IgG deposition and renal CD4⁺ T lymphocyte infiltration in Sle123→ApoA-I^tg mice. Representative anti-IgG (A) and CD4 (B) stained kidney sections from the indicated 38 week old control (n=5), control WT BMT (n=5) and Sle123 BMT (n=7–8) mice. Graphs below show glomerular IgG immunofluorescence staining intensity and CD4⁺ T cell numbers in the indicated mice (*p<0.05 by one way ANOVA within its control or BMT group, ‡p<0.05 by T test compared to both non-BMT and BMT control groups of the same host genotype, or non-BMT control group as indicated). Each data point represents the average of 9 glomeruli per mouse.

Figure 8.

Reduced renal CD11b⁺(A) and NIMPR14⁺ neutrophil (B) inflammatory cell infiltration in Sle123→ApoA-I^tg mice. Representative CD11b and NIMPR14 stained kidney cryosections from the indicated 38 week old control, control WT BMT and Sle123 BMT mice. Graphs below shows CD11b staining intensity and NIMPR14⁺ neutrophil numbers in the indicated mice (*p<0.05 by one way ANOVA within its control or BMT group, ‡p<0.05 by T test compared to non-BMT control mice of the same host genotype). Each data point represents the average of 9 glomeruli per mouse.

The lack of a potentiating effect of ApoA-I deficiency on autoimmunity may be due to compensatory increases in HDL ApoE

We reasoned that the absence of an exacerbatory effect of ApoA-I deficiency on autoimmunity in Sle123→ApoA-I^−/− mice might be due to compensatory modifications of other major apolipoproteins on HDL in the absence of ApoA-I (29, 38). Like ApoA-I, ApoE can bind cholesterol and oxidized lipids and promotes cholesterol transport (39). Plasma levels of ApoE (Supplemental Figure IA and C) and ApoE levels on fast performance liquid chromatography (FPLC) separated HDL fractions (Figure 9B) were significantly increased in ApoA-I^−/− mice. ApoA-II, on the other hand, was present at low levels in all mice (Supplemental Figure IA and C), consistent with the very low levels of this apolipoprotein in the C57Bl/6J strain (40). Thus, compositional changes on HDL in ApoA-I^−/− mice that compensate for the loss of those functions of ApoA-I that mediate immune suppression include increased ApoE.

Figure 9.

(A) Cholesterol content of fast performance liquid chromatographically (FPLC) separated plasma lipoproteins from the indicated 38 week old mice (FPLC fraction number on x-axis). (B) Representative anti-mouse ApoA-I (mApoA-I), anti-mouse ApoE (mApoE) and anti-human ApoA-I (hApoA-I) western blots of FPLC HDL fractions shown below. Note reduction in HDL mApoA-I in Sle123→WT mice compared to their control WT counterparts (left panels in B). Note marked increases in HDL mApoE content in ApoA-I^−/− mice and reductions in HDL mApoA-I content in ApoA-I^tg mice. Data are representative of 3 independent experiments.

DISCUSSION

Using a normocholesterolemic murine SLE bone marrow transfer model, we found that increasing ApoA-I levels through transgenic human ApoA-I expression suppressed lymphocyte activation, reduced pathogenic interferon-γ secreting CD4⁺ Th1 cells and Tfh cells, lowered GC B cell numbers and autoantibody production, and improved GN. However, it did not reduce the effects of Sle123 induced autoimmunity to zero. Protection against SLE-mediated GN in Sle123→ApoA-I^tg mice was associated with reductions in kidney infiltrating CD4⁺ T cells (Figure 7B) as well as lower CD11b⁺ monocytic and NIMPR14⁺ neutrophil infiltrates (Figure 8). Although reduced levels of autoantibodies leading to less glomerular IgG deposition could be responsible for this reduced renal inflammation (Figures 2A and 7A), direct anti-inflammatory action of ApoA-I may also be a major contributing mechanism, a possibility that could be addressed using a non-autoimmune anti-glomerular basement membrane (GBM) antibody-mediated nephritis model (41).

Other immune processes previously shown to be modified by ApoA-I in hypercholesterolemic mice, such as CD4⁺FoxP3⁺ Treg generation (42) (data not shown) or CD11c⁺ DC migration (Supplemental Figure IIA) (34), were unaffected in this normocholesterolemic model. Unlike in hypercholesterolemic mice (8), ApoA-I deficiency did not result in opposite effects on autoimmunity and modulation of ApoA-I did not affect immune cell cholesterol levels (Figure 5A). Taken together, these results show that potentiating effects of hypercholesterolemia on lymphocyte activation (and possibly other immune mechanisms) are required in order for ApoA-I to become important for preventing systemic autoimmunity (8) and that a function of ApoA-I other than cholesterol transport is the major mechanism accounting for its immune suppressive action in normal mice. However, it is possible that spatial rather than quantitative changes of cholesterol in lymphocyte membrane rafts might play a role.

Other than cholesterol transport and maintenance of cellular cholesterol homeostasis, “anti-oxidant” properties of HDL involving ApoA-I-mediated removal and PON1-mediated hydrolysis of atherogenic oxidized lipids contribute to its anti-inflammatory action in hypercholesterolemic atherosclerosis models (1, 43). The major classes of oxidized lipids bound by ApoA-I include the oxidized metabolites of linoleic acid and arachidonic acid (HODEs and HETEs respectively) (7, 44). Although several HODE and HETE species can recapitulate pro-inflammatory effects of oxidized LDL on vascular cells (45–47), they may have different properties depending on the cell/tissue context and disease setting (48–50). Indeed, we did not observe increases in lymphocyte HODE and HETE levels with the development of autoimmunity in Sle123→WT mice, or their reduction with immune suppression by ApoA-I in Sle123→ApoA-I^tg mice (Figure 5B) (data not shown for arachidonic acid and HETEs). Surprisingly, however, levels of the major oxidative metabolites of linoleic acid, 13-HODE and 9-HODE, were significantly higher in lymphocytes from autoimmune Sle123→ApoA-I^tg animals and this was not due to higher precursor linoleic acid levels (Figure 5B). Concentrations of 13-HODE and 9-HODE in the plasma were not significantly different between Sle123→WT and Sle123→ApoA-I^tg mice (Supplemental Figure IIC), suggesting that the higher levels of these oxidized fatty acids in immune cells from autoimmune Sle123→ApoA-I^tg animals was probably not due to an increased circulating plasma pool in ApoA-I^tg animals. Further work using flow sorted immune cell preparations in conjunction with chiral phase LC-MS approaches to distinguish (R)-HODEs versus (S)-HODEs will be useful to determine how augmenting ApoA-I levels increases lymphocyte 13-HODE and 9-HODE levels in SLE and why it does so only in the context of autoimmunity (control ApoA-I^tg animals did not have the same increases; Figure 5B). Nevertheless, it is noteworthy that 13-HODE and 9-HODE are agonists for peroxisome proliferator-activated receptor gamma (PPARγ), a ligand-dependent nuclear receptor expressed in multiple cell-types including T cells, B cells, macrophages and DCs (51, 52). Anti-inflammatory activities have been attributed to ligand-activated PPARγ, including suppression of T cell activation, proliferation, Th1 polarization and autoimmunity (51, 53–56). For example, 13-HODE can regulate T lymphocyte function by blocking T cell IL-2 production through PPARγ, reducing T cell differentiation and proliferation (53). Higher levels of PPARγ have been reported in SLE lymphocytes (57) and PPARγ agonist treatment of T cells from SLE patients induces transcriptional repression of genes involved in T cell activation, with those related to Th1 differentiation most affected (58). Mice lacking PPARγ only in CD4⁺ T cells develop an autoimmune phenotype characterized by expansion of Tfh cells, resulting in increased GC B cells, production of anti-DNA autoantibodies, and glomerular inflammation (59). Based on our finding that Th1 and Tfh cells are reduced in Sle123→ApoA-I^tg mice (Figure 4B–C), it is possible that PPARγ activation by 13-HODE and/or 9-HODE in lymphocytes may be a mechanism for ApoA-I mediated immune suppression, reducing T cell activation and subsequent expansion of Th1 and Tfh CD4⁺ T cells. Indeed, effects of ApoA-I on the lipid environment of CD4⁺ T lymphocytes and other immune cell-types in SLE may have a significant influence over the repertoire of transcription factors and cytokine signals that control CD4⁺ T cell differentiation into Th and effector subsets (33, 60). It is therefore important to determine whether 13-HODE and 9-HODE are generated intrinsically by the lymphocytes themselves or if they are derived from nearby innate immune cells such as macrophages which have been reported to produce these oxylipids in response to inflammatory stimuli through 12/15-LO enzyme activation to inhibit T cell activation and differentiation through PPARγ (53).

Immune suppression by ApoA-I occurred despite the predisposition of animals to developing reduced HDL-C and ApoA-I levels (see Sle123→WT in Figure 1B and D–E, Supplemental Figure IA–B and Figure 9B), and lower HDL-associated PON-1 activity (see Sle123→WT in Figure 1C). Indeed, there was a significant inverse correlation between HDL-C levels in autoimmune Sle123→WT mice and certain parameters of autoimmunity, including anti-dsDNA autoantibodies and CD4⁺ T lymphocyte activation (Supplemental Figure ID), reflecting autoimmune-mediated HDL reduction as previously described in normocholesterolemic gld mice (18). This is an important finding because it suggests that the predisposition of SLE to lower HDL/ApoA-I levels and PON-1 activity (16) may not necessarily make it refractory to immune suppressive or anti-inflammatory therapeutic strategies targeting ApoA-I, including ApoA-I mimetic peptides (14). Furthermore, autoimmune-mediated reductions in ApoA-I such as those observed in Sle123→WT animals are considered not only to contribute to premature atherosclerosis in lupus patients (61, 62), but also to amplify lymphocyte activation and autoimmunity itself (8, 10). However, unlike in the setting of hypercholesterolemia (8), we found no evidence for an exacerbatory effect of ApoA-I deficiency on any parameter of autoimmunity (something that would be expected if autoimmune mediated HDL reduction and dysfunction amplified autoimmunity itself). Similarly, because ApoA-I is important for stabilizing PON-1 activity on HDL particles (31), PON-1 activity was about 3-fold lower in ApoA-I^−/− mice compared to their WT counterparts (Figure 1C), yet this did not affect autoimmunity. Therefore, our cumulative data suggest that autoimmune-mediated reductions in ApoA-I (HDL) levels and PON-1 activity in SLE patients (61, 63) may not necessarily amplify autoimmunity and inflammation. However, it remains to be determined to what extent the higher PON-1 activity in ApoA-I^tg mice contributed to the observed immune suppressive and anti-inflammatory effects of increased ApoA-I in our study.

ApoA-I deficiency is pro-inflammatory and can promote systemic autoimmunity in hypercholesterolemic mouse models by deregulating lymphocyte cholesterol homeostasis to promote lymphocyte activation (8, 10, 11, 15). The absence of a significant effect of ApoA-I deficiency on any parameter of autoimmunity measured (Figures 2–4) may therefore seem surprising based on the immune suppressive effects of transgenic human ApoA-I expression that we observed in the same mouse SLE model system. While it will be important to rule out possible species differences in mouse versus human ApoA-I in this and other autoimmune models, it is noteworthy that compositional alterations compensating for ApoA-I deficiency may have contributed to a lack of any potentiating effect of ApoA-I deficiency on autoimmunity. For example, we found marked increases in the ApoE content of HDL in ApoA-I^−/− mice (Supplemental Figure IA and C and Figure 9B) which may have compensated for the loss of important ApoA-I functions involved in immune suppression such as lipid binding, but not maintenance of PON-1 activity on HDL (31) considering the markedly lower HDL-associated PON-1 activity in ApoA-I^−/− animals (Figure 1C).

Bioactive lipid mediators with established immune regulatory roles that are bound and carried by HDL, such as sphingosine-1-phosphate (S-1-P), also warrant investigation with respect to mechanisms of ApoA-I mediated immune suppression and anti-inflammatory action in SLE (2, 64, 65). Furthermore, the ability of HDL to modulate lipid homeostasis can affect immunity and renal inflammation through Toll-like receptors (TLRs) (66), perhaps even those mediating aberrant recognition of nucleic acids leading to pathogenic autoantibody production in SLE like TLR7 and TLR9 (67). It will therefore be important to perform broader lipidomic profiling in order to identify lipid metabolites that are associated with immune suppressive and anti-inflammatory actions of ApoA-I, as some of these are likely to be viable targets for therapeutic development. In this respect, it will be important to determine whether oxidative changes to which ApoA-I is inherently susceptible in chronic inflammatory settings such as SLE may be limiting its immunosuppressive potential such that using mutant forms of ApoA-I resistant to oxidative loss of function (68) may provide the most robust platform for these lipidomic approaches. Future studies characterizing the lipidome in tissues from autoimmune animal models with experimentally manipulated ApoA-I levels, HDL composition, or treated with apolipoprotein mimetic peptides, could therefore be used to identify novel lipid-based mechanisms of immune suppression and anti-inflammatory action in SLE.

Abbreviations used in this article

ApoA-I

apolipoprotein A-I

ApoE

apolipoprotein E

ApoA-II

apolipoprotein A-II

GC B cell

germinal center B cell

GC-MS

gas chromatography mass spectrometry

GN

glomerulonephritis

HDL

high-density lipoprotein

HETE

hydroxyeicosatetraenoic acid

HODE

hydroxyoctadecadienoic acid

LC-MS

liquid chromatography mass spectrometry

LO

lipoxygenase

PON-1

paraoxonase-1

ROS

reactive oxygen species

Tfh cell

follicular helper T cell

WT

wild-type