Jump to
Abstract
Rationale: Chronic inflammation is central in the development of atherosclerosis. Both innate and adaptive immunities are involved. Although several studies have evaluated the functions of natural killer (NK) cells in experimental animal models of atherosclerosis, it is not yet clear whether NK cells behave as protective or proatherogenic effectors. One of the main caveats of previous studies was the lack of specificity in targeting loss or gain of function of NK cells.
Objectives: We used 2 selective genetic approaches to investigate the role of NK cells in atherosclerosis: (1) Ncr1iCre/+R26lsl−DTA/+ mice in which NK cells were depleted and (2) Noé mice in which NK cells are hyperresponsive.
Methods and Results: No difference in atherosclerotic lesion size was found in Ldlr−/− (low-density lipoprotein receptor null) mice transplanted with bone marrow (BM) cells from Ncr1iCreR26Rlsl−DTA, Noé, or wild-type mice. Also, no difference was observed in plaque composition in terms of collagen content, macrophage infiltration, or the immune profile, although Noé chimera had more IFN (interferon)-γ–producing NK cells, compared with wild-type mice. Then, we investigated the NK-cell selectivity of anti–asialoganglioside M1 antiserum, which was previously used to conclude the proatherogenicity of NK cells. Anti–asialoganglioside M1 treatment decreased atherosclerosis in both Ldlr−/− mice transplanted with Ncr1iCreR26Rlsl−DTA or wild-type bone marrow, indicating that its antiatherogenic effects are unrelated to NK-cell depletion, but to CD8+ T and NKT cells. Finally, to determine whether NK cells could contribute to the disease in conditions of pathological NK-cell overactivation, we treated irradiated Ldlr−/− mice reconstituted with either wild-type or Ncr1iCreR26Rlsl−DTA bone marrow with the viral mimic polyinosinic:polycytidylic acid and found a significant reduction of plaque size in NK-cell–deficient chimeric mice.
Conclusions: Our findings, using state-of-the-art mouse models, demonstrate that NK cells have no direct effect on the natural development of hypercholesterolemia-induced atherosclerosis, but may play a role when an additional systemic NK-cell overactivation occurs.
Introduction
Atherosclerosis is a chronic inflammatory disease characterized by the accumulation of cholesterol, immune cells, and fibrous elements, forming atherosclerotic plaques in large- and medium-sized arteries.1 Over the past 20 years, a large body of evidence supported the implication of innate and adaptive immunities in the development of the disease. Macrophages are the first immune cells to infiltrate developing lesions, as well as dendritic cells and T cells. However, less abundant cells such as NKT and natural killer (NK) cells are also present.2 NK cells are bone marrow (BM)–derived innate lymphocytes characterized by their unique ability to kill aberrant cells without prior sensitization.3 They act by means of direct cytotoxicity against their targets and by producing large array of cytokines and chemokines, the latter contributing also to the initiation of antigen-specific immune responses and thus shaping of adaptive immunity.4,5 Regardless of their activation status and their localization, NK cells express the NKp46 marker encoded by the Ncr1 gene. NK cells have been identified within atherosclerotic plaques in humans and in mice. In advanced atherosclerotic plaques, they have been localized to necrotic cores, deep within the plaques and in shoulder regions.6,7 Chemoattractants such as CCL2 and CX3CL1 may be involved in their recruitment to the developing lesions.8,9 Directly addressing the role of NK cells in immunity and atherosclerosis has been challenging because of the lack of a good animal model of NK-cell deficiency. Earlier studies used the beige mutant mice, an accepted model of defective NK-cell functional activity,10 with opposing results. In the first study, beige mutant mice fed a high-fat diet (HFD) containing cholate showed no differences in atherosclerotic lesions compared with controls.11 However, in a more recent study by Schiller et al,12 NK cells were ascribed an atheroprotective role when beige mutant mice were bred to Ldlr−/− (low-density lipoprotein receptor null) mice. Beige mutation, which involves the Lyst (lysosomal trafficking regulator) gene encoding a protein implicated in lysosomal trafficking, results in a complicated phenotype that goes beyond decreased NK-cell activity.13,14 Defects in cell function may include, in addition to NK cells, neutrophils, macrophages, or smooth muscle cells,15 making it unclear whether the increase in atherosclerosis in LystbeigexLdlr−/− mice was a consequence of defective NK-cell function.
Editorial, see p 6
Meet the First Author, see p 3
Subsequently, different results were observed using an Ly49A transgenic mouse model of NK-cell deficiency.7 In these transgenic mice, the Ly49A cDNA was expressed under control of the mouse granzyme A genomic sequence, inducing the expression of the receptor on all NK cells and one half of T cells.16 Ly49A is an inhibitory receptor, which recognizes MHC (major histocompatibility complex) class I molecules.17 In this transgenic model, the number of NK (NK1.1+CD3−) cells was markedly reduced in the spleen and other peripheral tissues, and the number of NK1.1+ T cells was slightly reduced in the spleen.16 However, this phenotype, which is not totally understood, is linked to the expression of the ubiquitous transcription factor ATF2 (activating transcription factor 2), raising the possibility of additional defects.18 Smaller size lesions were observed in Ldlr−/− recipients reconstituted with the BM of ly49A transgenic mice.7 Although the authors concluded that NK-cell deficiency was responsible for the observed effect, they did not exclude the possibility that T cells, specifically NKT and CD8 subsets, which are known to express granzyme A and whose functions are influenced by Ly49A,19,20 could contribute to this proatherogenicity. It is worth mentioning that NKT and CD8 T cells have been classified as proatherogenic cells.21,22
Lately, Selathurai et al23 performed loss- and gain-of-function studies to assess NK-cell function in atherosclerosis. In the loss-of-function experiments, treatment of Apoe−/− mice with anti–asialo-GM1 antibody significantly reduced atherosclerotic lesion development. This study was supported by gain-of-function experiments where adoptive transfer of NK cells (wild type [WT], IFN (interferon)-γ−/−, or perforin−/−/granzyme−/−) into lymphocyte-deficient Apoe−/−×Rag2−/−×Il2rg−/− mice confirmed that its cytotoxic effects are proatherogenic and promote necrotic core development.23 However, the glycolipid asialoganglioside M1 (GM1) is also expressed by several cell types, including myeloid cells, epithelial cells, and T-cell subsets. Hence, the selectivity of NK-cell depletion with anti–asialo-GM1 antiserum has been questioned.24
Here, we used 2 state-of-the-art genetic approaches: (1) the Ncr1iCreR26Rlsl−DTA mice in which NK cells are selectively depleted25 and (2) the Noé mice in which NK cells are hyperresponsive.26 BM from these mice was used to repopulate the hematopoietic system of lethally irradiated Ldlr−/− mice fed an HFD. Using these 2 genetic models, we provide strong evidence that NK cells do not significantly affect atherosclerosis development unless an additional inflammatory environment is induced.
Methods
Animals
Experiments were conducted according to the guidelines formulated by the European Community for experimental animal use (L358-86/609EEC) and were approved by the Ethical Committee of INSERM and the French Ministry of Agriculture (agreement A75-15–32). Ncr1iCreR26Rlsl−DTA mice (deficient in NK cells)25 and Noé mice (high IFN-γ–producing and CD107a-expressing NK cells)26 have been generated and characterized by Eric Vivier laboratory (CIML, Marseille, France).
The Ncr1iCre knock-in mice27 were bred with R26Rlsl−DTA mice to generate NK46iCre/+R26lsl−DTA/+ mice referred to as Ncr1iCreR26Rlsl−DTA mice thereafter. Ncr1iCre knock-in mice were generated from C57BL/6 ES cells, and R26Rlsl−DTA mice were obtained from Jackson. This line was backcrossed to C57BL/6 for 10 generations by the donating laboratory.27 In this model, the Cre-mediated removal of the floxed STOP codon leads to the expression of the diphtheria toxin fragment A and consequent cell death. A near to complete absence of NK cells in the BM and every tested organ (including the blood, thymus, lymph nodes, spleen, liver) was observed in these mice.25 Of note, a deficiency in ILC1 (immune lymphoid cell 1; which also express NKp46) in the liver and in NKp46+ ILC3 in the small intestine was also observed in these mice. Aside from NKp46+ cell deficiency, the immune status of the mice is strictly normal at steady state, in terms of cell populations and activation status.25
The Noé mice were generated by random mutagenesis, in which a point mutation in the Ncr1 gene impairs the cell surface expression of NKp46 receptor inducing higher capacity to produce IFN-γ and to degranulate. Once generated, Noé mice were backcrossed to C57BL/6J.26
Ten-week-old male C57BL/6J Ldlr−/− mice were subjected to medullar aplasia by lethal total body irradiation (9.5 Gy). The mice were repopulated with an intravenous injection of BM cells isolated from femurs and tibias of sex-matched Ncr1iCreR26lsl−DTA, Noé, or control C57BL/6J mice. After 4 weeks of recovery, mice were fed a proatherogenic HFD containing 15% fat, 1.25% cholesterol, and 0% cholate for 8, 12, or 15 weeks.
In some experiments, 14-week-old chimeric male Ldlr−/− mice, which received BM from Ncr1iCreR26lsl−DTA or control C57BL/6J mice, were treated every 5 days by intravenous injections of anti–asialo-GM1 antibody (54 µg/injection), Wako chemicals, Richmond, VA) or control serum (normal rabbit serum) for 8-week HFD. We used the same experimental protocol as previously described by Selathurai et al.23
For NK-cell activation experiments, 14-week-old chimeric male Ldlr−/− mice, which received Ncr1iCre R26lsl−DTA or control C57BL/6J BM, were treated twice a week by intraperitoneal injection of 100 µg polyinosinic:polycytidylic acid [(poly(I:C)], a synthetic analogue of dsRNA (double-strand RNA), TLR3 (Toll-like receptor 3) ligand (Invivogen, SD), or saline (as a control) during 8-week HFD.
For tracking NK cells in mouse lesions, lethally irradiated Ldlr−/− mice were reconstituted with BM of Ncr-1gfp/gfp mice and put on HFD for 8 weeks. The generation of Ncr-1gfp/gfp mice was previously described by Gazit et al.28
Quantification of Atherosclerotic Lesions
Mice were anesthetized with isoflurane before euthanize. Plasma cholesterol was measured using a commercial kit (DiaSys Cholesterol FS*, Germany). Quantification of lesion size and composition was performed as previously described.29 In brief, the heart and ascending aorta were removed, perfusion-fixed in situ with 4% paraformaldehyde, then placed in PBS 30% sucrose solution overnight, before being embedded in frozen optimal cutting temperature compound, and frozen at −70°C. Afterward, 10-μm serial sections of aortic sinus were obtained (cut on cryostat). Lipids were detected using oil red O (Sigma-Aldrich, St. Louis, MO) coloration and quantified by a blinded operator using HistoLab software (Microvisions Instruments, Paris, France),30 which was also used for morphometric studies. En face quantification was used for atherosclerotic plaques along thoracoabdominal aorta, as previously described.29
Collagen was detected using Sirius red staining. The presence of macrophages was determined using monoclonal rat anti-mouse macrophage/monocyte antibody (specifically MAB1852).
At least 4 sections per mouse were examined for each immunostaining, and appropriate negative controls were used.
For NK-cell tracking, frozen sections of spleen and aortic sinus of Ncr-1gfp/gfp→ Ldlr−/− mice were stained with rabbit anti-GFP (green fluorescent protein) antibody (NK cells) and counterstained with DAPI (nuclear dye) and monocytes/macrophages antibody.
To detect NK cells in human plaques, immunostaining studies were performed in atherosclerotic plaques from carotid arteries obtained after surgical thromboendarterectomy (Pathology Department, HEGP Hospital, Paris). We used a specific antibody against human NKp46 (generous gift of Innate Pharma, Marseille, France) in formalin-fixed paraffin-embedded tissues after antigen retrieval by heating in citrate buffer and ABC (avidin–biotin–peroxidase complex) peroxidase technique as previously described.31 Spleen was used as an internal control in the 2 separate experiments.
Spleen Cell Culture and Cytokine Assays
Cells were cultured in RPMI 1640 medium supplemented with Glutamax (Thermo Fischer Scientific), 10% fetal calf serum, 0.02 mmol/L β-mercaptoethanol, and antibiotics. For cytokine measurements, splenocytes were cultured at 5×105 cells per well for 24 hours and stimulated with lipopolysaccharide (1 μg/mL) and IFN-γ (100 UI/mL; Sigma). IL-6 (interleukin 6), IL-12, TNF-α (tumor necrosis factor α), IL-1β, IL-10, and IFN-γ productions in the supernatants were measured using specific ELISA (R&D Systems).
Flow Cytometry
NK cells were identified as CD3−NK1.1+ or CD3−NK1.1+NKp46+. Monocytes were identified as CD11b+Ly6G− or CD11b+CD115high. Among them, classical monocytes were Gr1high (or Ly6Chigh) and nonclassical monocytes were Gr1low (or Ly6Clow). Neutrophils were identified as CD11b+CD115−Ly6G+. Regulatory T cells were considered as CD3+CD4+CD25highFoxp3+ (forkhead box P3). CD4+, CD8+, and NKT (CD3+NK1.1+) lymphocyte subsets were also analyzed.
Cells were labeled with fluorescein isothiocyanate–conjugated or PE-Cy7 (phycoérythrine-Cy7)–conjugated anti-NKp46 (29A1.4), APC-conjugated anti-NK1.1 (PK136), PerCp-Cy5.5 (peridinin chlorophyll protein-Cy5.5)–conjugated anti-CD3e (145-2C11), fluorescein isothiocyanate–conjugated anti-CD4 (RM4-5), eFluor-450–conjugated anti-CD11b (M1/70), PE-conjugated anti-CD69 (H1.2F3), PE-Cy7–conjugated anti-CD115 (AFS98), PE-Cy7–conjugated anti-Foxp3 (FJK-16s), APC-conjugated anti-CD25 (PC61.5), V500-conjugated anti-B220 (RA3-6B2), PE-conjugated anti–IL-10 (JES5-16E3), PerCp-Cy5.5–conjugated anti–MHC II (M5/114.15.2) from eBiosciences, Alexa fluor 700–conjugated anti-CD8 (53–6.7), Brilliant Violet 421–conjugated anti–IFN-γ (XMG1.2) and fluorescein isothiocyanate–conjugated anti-CD107a (1D4B) from Biolegend, fluorescein isothiocyanate–conjugated anti-Ly6C (AL-21), PE-conjugated anti-Ly6G (1A8), PerCp-Cy5.5–conjugated Gr-1 (RB6-8C5), PE-Cy7–conjugated anti-CD11c (HL3), and APC-conjugated anti-IgM (11/41) from BD Biosciences. For blood staining, erythrocytes were lysed using BD fluorescence-activated cell sorter lysing solution (BD Biosciences). For intracellular cytokine staining, lymphocytes were stimulated in vitro with leukocyte activation cocktail (BD Biosciences) according to the manufacturer’s instructions for 4 hours. Surface staining was performed before permeabilization using Foxp3 staining buffer kit (eBiosciences) and intracellular staining.
Forward scatter and side scatter were used to gate live cells excluding red blood cells, debris, and cell aggregates in total blood cells, splenocytes, and BM. Single-cell suspensions stained with fluorophore-conjugated antibodies were acquired using an LSRII Fortessa (BD) flow cytometer and analyzed with FlowJo software (Miltenyi).
NK-Cell Stimulation
For specific NK-cell stimulation, spleen cell suspensions were dispensed into 96-well 2HB Immulon plate previously coated with 25 μg/mL of purified anti-NK1.1 antibody or 10 μg/mL of anti-NKp46 antibody. Anti-CD107a antibody was added in the presence of monensin for 4 hours.
For IL-12/18 stimulation, spleen cells were incubated with 25 ng/mL IL-12 and 20 ng/mL IL-18 (R&D systems) for 4 hours in the presence of monensin. Cells were surface stained and intracellular IFN-γ was revealed.
LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Thermo Fischer Scientific, L34957) was used as a viability dye, and the percentage of IFN-γ–producing and CD107a-expressing NK cells was detected by Flow cytometry as mentioned above.
Quantitative Real-Time Polymerase Chain Reaction
Total RNA from spleen or abdominal aorta was extracted using Trizol reagent (Invitrogen). Quantitative real-time polymerase chain reaction was performed on an ABI Prism 7700 Sequence Detection System (Thermo Fisher Scientific, Inc) in duplicates. Cycle threshold for GAPDH was used to normalize gene expression of samples. Relative expression was calculated using the 2-δ computed tomography method followed by geometric average, as recommended.32,33 The following primer sequences were used: IL-10 (F: 5′ AAG TGA TGC CCC AGG CA 3′; R: 5′ TCT CAC CCA GGG AAT TCA AA 3′), perforin (F: 5′ ACA GTA GAG TGT CGC ATG TAC AGT TT 3′; R: 5′ GAG GGC TCT GAG CGC CTT TTT GAA 3′), granzyme B (F: 5′ ACT CTT GAC GCT GGG ACC TA; R: 5′ AGT GGG GCT TGA CTT CAT GT 3′), IFN-γ (F: 5′ TCA AGT GGC ATA GAT GTG GAA AGA A 3′; R: 5′ TGG CTC TGC AGG ATT TTC ATG 3′), TGF-β (F: 5′ CGG CCG GAA ATT CCC AGC TTC T 3′; R: 5′ GTG AGA CGG GCT TCG GGG TCA T 3′), TNF-α (F: 5′ GAT GGG GGG CTT CCA GAA CT 3′; R: 5′ CGT GGG CTA CAG GCT TGT CAC 3′), IL-1β (F: 5′ GAA GAG CCC ATC CTC TGT GA 3′; R: 5′ GGG TGT GCC GTC TTT CAT TA 3′), and IL-6 (F: 5′ AAA GAC AAA GCC AGA GTC CTT CAG AGA GAT 3′; R: 5′ GGT CTT GGT CCT TAG CCA CTC CTT CTG T 3′). Polymerase chain reaction conditions were 10 minutes at 95°C; 42 cycles of 95°C for 15 s, 60°C for 30 s, and a final extension of 72°C for 30 s.
Statistical Analysis
Values are expressed as mean±SEM. Differences between values were evaluated using nonparametric Mann–Whitney test or 1-way ANOVA with Bonferroni correction to compare >2 groups. All these analyses were performed using GraphPad Prism version 5.0b for Mac (GraphPad Software), and values were considered significant at P<0.05.
Results
Genetic NK-Cell Depletion Does Not Alter the Development of Atherosclerosis
To assess the direct role of NK cells in the development of atherosclerosis, we performed BM transplantation experiments using either control (WT) or Ncr1iCreR26lsl−DTA BM cells to repopulate lethally irradiated male atherosclerosis-prone Ldlr−/− mice.
After 4 weeks of recovery, mice were put on HFD for 8, 12, or 15 weeks. First, we confirmed NK-cell depletion in spleen (Figure 1A and 1B), blood (Online Figure I), and BM (data not shown) of chimeric Ncr1iCreR26lsl−DTA→ Ldlr−/− mice. As expected, the frequencies of other leukocyte populations in the spleen (Figure 1C), blood (Online Figure I), and BM (data not shown) were not different between groups of chimeric mice. Data shown in Figure 1 correspond to 8-week HFD-fed mice. Data were similar in 12- and 15-week HFD-fed mice (data not shown). Circulating cholesterol levels and animal weights were similar between the 2 groups of mice (Online Table). Interestingly, NK-cell deficiency did not alter lesion size in aortic root and thoracic aorta after 8-week (Figure 2A and 2B), 12-week (Figure 2C and 2D), or 15-week (Figure 2E and 2F) HFD. Similarly, no effect was observed on atherosclerosis development in female Ldlr−/− mice reconstituted with BM of control or Ncr1iCreR26lsl−DTA mice and put on HFD for 8 weeks (Online Figure IIA through IIC). To determine whether plaque composition was affected by NK-cell depletion, we analyzed macrophage content (Online Figure IID and IIE), collagen content (Online Figure IIF and IIG), and necrotic core (data not shown), but found no significant differences between groups.
Natural killer (NK) cells are selectively depleted in irradiated Ldlr–/– mice transplanted with bone marrow (BM) from Ncr1iCreR26lsl−DTA mice. A and B, Efficiency of NK cell (CD3−NK1.1+NKp46+) depletion in spleen (gated on total viable splenocytes) of male Ldlr−/− mice reconstituted with the BM of NK-cell–deficient (Ncr1iCreR26lsl−DTA → Ldlr−/−; n=10) compared with control mice (wild type [WT] → Ldlr−/−; n=8) after 8 wk of high-fat diet (HFD). C, Representative quantitative analysis of the percentage of myeloid (neutrophils, monocytes [classical and nonclassical], and dendritic cells) and lymphoid (CD4+, CD8+, and NKT cells) populations in the spleen of control and NK-cell–deficient chimeric mice assessed by flow cytometry, after 8-wk HFD. ***P<0.001. Ldlr−/− indicates low-density lipoprotein receptor null.
Natural killer (NK) cell depletion does not affect atherosclerosis. Representative photomicrographs of oil red O staining and quantification of lesion size in aortic sinus (A) or thoracic aorta (B) of male controls (wild type [WT] → Ldlr−/−; n=8) and NK-cell–deficient (Ncr1iCreR26lsl−DTA → Ldlr−/−; n=10) mice after 8 wk (A, B), 12 wk (n=7–6, respectively; C, D), or 15 wk (n=5–3, respectively) HFD (E, F). Ldlr−/− indicates low-density lipoprotein receptor null.
Furthermore, we investigated the immune-inflammatory response in chimeric Ldlr−/− mice because NK cells are known to produce cytokines in response to infection or inflammation.34 The expression levels of IL-6, IL-10, IL-1β, IL-12p70, TNF-α, and TGF-β mRNA measured in the spleen were not different between control and NK-cell–deficient chimeric mice (Online Figure IIIA). Lipopolysaccharide and IFN-γ–stimulated splenocytes from chimeric control or Ncr1iCreR26lsl−DTA mice also showed no difference in inflammatory phenotype with same levels of IL-6, IL-10, and TNF-α (Online Figure IIIB).
Hyperresponsive NK Cells Do Not Promote Atherosclerosis Development
We next addressed the role of NK-cell hyperresponsiveness in atherosclerosis by using Noé mice, in which a point mutation in the Ncr1 gene impairs the cell surface expression of the NKp46 receptor inducing a higher capacity to produce IFN-γ and to degranulate.26 We performed BM transplantation experiments using Noé or control (C57/B6J) mice to reconstitute the BM of male Ldlr−/− mice. After 8-week HFD, an intact NK-cell population (CD3−NK1.1+) was detected in Noé chimeric mice compared with controls (Figure 3A through 3C). The frequencies of other leukocyte populations were also the same in spleen and blood of mice reconstituted with the WT or Noé BM (Online Figure IVE and IVF). Interestingly, a higher frequency of IFN-γ–producing NK cells was observed in chimera mice reconstituted with the Noé BM. However, no difference in plaque size, aortic root, and thoracic aorta (Figure 3D through 3G) or plaque composition in terms of macrophage accumulation or collagen content (Online Figure IVA through IVD) was observed between groups. No difference in immune-inflammatory response was detected either (data not shown). This series of experiments was repeated twice, and similar results were found. There were no differences in plasma cholesterol levels or mice body weight between Noé and control chimeric mice (Online Table).
Hyperresponsive natural killer (NK) cells do not promote atherosclerosis. Representative flow cytometry plots (A) and quantitative (B) analysis of the percentage of (B) NK cells in the splenocytes of controls (wild type [WT] → Ldlr−/−) and (Noé → Ldlr−/−) chimeric mice gated on CD3−NK1.1+ cells and (C) IFN (interferon)-γ–producing NK cells gated on CD3−NK1.1+ cells. Representative photomicrographs of oil red O–stained atherosclerotic lesions in the aortic sinus (D) and thoracic aorta (F) of male controls (n=8) and Noé chimeric mice (n=7), along with their corresponding quantifications (E and G, respectively) after 8-wk high-fat diet (HFD). **P<0.01. FSC indicates forward scatter. Ldlr−/− indicates low-density lipoprotein receptor null.
These data show that NK-cell hyperresponsiveness observed in Noé mice is not sufficient to affect the development of atherosclerosis.
Anti–Asialo-GM1–Mediated Decrease in Atherosclerosis Is NK Cell Independent
The findings with the above 2 experiments were unexpected, given that the latest study on the role of NK cell in atherosclerosis came up with the conclusion that NK cells are proatherogenic through their production of perforin and granzymes.23 Hence, we revisited the experimental model of Selathurai et al,23 in which the authors used anti–asialo-GM1 antibody to deplete NK cells in Apoe−/− mice. We treated control (WT→ Ldlr−/−) or NK-cell–deficient (Ncr1iCreR26lsl−DTA→ Ldlr−/−) chimeric mice either with anti–asialo-GM1 or control serum every other 5 days during the 8-week HFD.
Initially, we examined the effectiveness of anti–asialo-GM1 treatment for depleting NK cells in peripheral blood. A single dose of anti–asialo-GM1 antibody was sufficient to deplete NK cells within 5 days by >80% in WT chimeric mice (Table). NK cells are absent in Ncr1iCreR26lsl−DTA chimeric mice. However, anti–asialo-GM1 treatment also depleted CD8+ T and NKT populations in both control and Ncr1iCreR26lsl−DTA chimeric mice compared with control groups treated with control serum (Table).
Quantitative Analysis of the Percentage of Blood Cells 5 Days After Starting Treatment With Anti–Asialo-GM1 and After 8 Weeks HFD in Controls (WT → Ldlr−/−) and NK-Cell–Deficient (Ncr1iCreR26lsl−DTA → Ldlr−/−) Mice Treated With Either Anti–Asialo-GM1 or Control Serum
Similar effects in blood and spleen cells were observed in anti–asialo-GM1–treated chimeric mice after 8-week HFD. Anti–asialo-GM1 treatment led to almost 80% reduction of NK cells in control mice, between 60% and 80% reduction in NKT cells in control and NK-cell–deficient chimeric mice, respectively, and 30% reduction in CD8 T cells in both groups (Table).
Although anti–asialo-GM1 treatment seemed to be less effective in depleting NKT and CD8+ T-cell populations after 8-week HFD in the spleen, the trend toward a smaller size population persisted (Online Figure VA). There were no differences in the percentage of monocytes, but we observed a trend of decrease in blood (Table) and spleen (data not shown) neutrophils.
Importantly, IFN-γ–producing CD8+ T-cell population was significantly decreased by almost 50% in anti–asialo-GM1–treated Ncr1iCreR26lsl−DTA chimeric mice with a trend also in control anti–asialo-GM1–treated chimeric mice (P=0.077), whereas no significant differences were observed in IFN-γ–producing CD4+ T cells (Online Figure VB).
Lesion size decreased significantly by 42% in anti–asialo-GM1–treated control chimeric mice compared with mice treated with control serum (Figure 4). However, more interestingly, similar effects were observed in NK-cell–deficient chimeric mice with 68% reduction in aortic root lesion size in anti–asialo-GM1–treated Ncr1iCreR26lsl−DTA chimeric mice compared with mice treated with control serum (Figure 4), despite similar plasma cholesterol levels (Online Table). Therefore, we concluded that reduction of atherosclerotic lesions by anti–asialo-GM1 treatment was NK cell independent.
Anti-asialoganglioside M1 treatment decreases atherosclerosis in both control and natural killer (NK)-cell–deficient chimeric mice. Representative photomicrographs of oil red O staining (A and B) and quantification of lesion size in left aortic sinus (C) after 8 wk of high-fat diet (HFD) of male controls (A) and natural killer (NK) cell–deficient mice (B) treated with either α-asialoganglioside M1 (GM1) or control serum (n=10 per group). *P<0.05, ***P<0.001. WT indicates wild type.
NK Cells Exacerbate Atherosclerosis Development in Inflammatory Conditions
Finally, we addressed the role of NK cells in the development of atherosclerosis in inflammatory conditions in which these cells are known to be activated. To do so, we used poly(I:C), which mimics a viral infection. Control (WT→ Ldlr−/−) or NK-cell–deficient (Ncr1iCreR26lsl−DTA → Ldlr−/−) chimeric mice were treated repeatedly with 100 μg poly(I:C) twice a week, whereas another control group (WT→ Ldlr−/− mice) was treated in parallel with saline over 8-week HFD.
Interestingly, after 8-week HFD, poly(I:C) treatment of control mice significantly increased CD69 and CD107a expression on NK cells compared with control mice that received saline, although the increase in frequency of IFN-γ–producing NK cells was not significant. As expected, the percentage of NK cells significantly decreased by almost 50% in the spleen of poly(I:C)-treated compared with saline-treated chimeric mice reconstituted with WT BM (Figure 5A). Moreover, the expression of perforin, granzyme B, and IFN-γ mRNA was significantly enhanced in the spleen after poly(I:C) treatment, which also holds true for granzyme B in the aorta (Figure 5B). Interestingly, the expression of granzyme B mRNA, but not IFN-γ, in the spleen and aorta was reduced in poly(I:C)-treated NK-deficient mice. This further confirms NK-cell activation in response to poly(I:C) treatment.
Natural killer (NK) cells are activated by poly(I:C) treatment. A, Representative quantitative analysis of flow cytometry–based staining of NK cells (gated on CD3−NK1.1+), CD69+ NK cells (gated on CD3−NK1.1+), IFN (interferon)-γ+ NK cells (gated on CD3−NK1.1+), and CD107a+ NK cells (gated on CD3−NK1.1+) in the 3 groups of mice: controls (wild type [WT] → Ldlr−/−) treated with saline (n=6) or polyinosinic:polycytidylic acid (poly(I:C); n=11) and NK-cell–deficient (Ncr1iCreR26lsl−DTA → Ldlr−/−) mice treated with poly(I:C) (n=13). B, Quantification of perforin, granzyme B, and IFN-γ mRNA expression in the spleens and abdominal aorta of the 3 groups of chimeric mice. *P<0.05, **P<0.01, ***P<0.001. N.D indicates not determined. Ldlr−/− indicates low-density lipoprotein receptor null.
Although no change in aortic root plaque size was observed in poly(I:C)-treated compared with saline-treated (WT→ Ldlr−/−) chimeric mice, NK-cell deficiency in poly(I:C)-treated chimeric mice markedly decreased lesion development compared with poly(I:C)-treated WT controls (Figure 6). Reduction in atherosclerosis was associated with reduction in macrophage infiltration in (Ncr1iCreR26lsl−DTA → Ldlr−/−) poly(I:C)-treated mice compared with poly(I:C)-treated control mice (Online Figure VIA). No change in collagen content was detected (Online Figure VIB). No differences in body weight or serum cholesterol levels were observed between the 3 groups of mice (Online Table). Hence, NK cells might be proatherogenic when activated in a systemic inflammatory context.
Natural killer cells exacerbate atherosclerosis in poly(I:C)-induced inflammatory conditions. Representative photomicrographs of oil red O staining (A) and quantification of lesion size in left aortic sinus (B) after 8-wk HFD in the 3 groups. **P<0.01. poly(I:C) indicates polyinosinic:polycytidylic acid; and WT, wild type.
NK Cells Are Scarcely Detected in Human and Mouse Atherosclerotic Lesions
To specifically track NK-cell homing to atherosclerotic lesions in hypercholesterolemic mice, we reconstituted lethally irradiated Ldlr−/− mice with BM from Ncr-1gfp/gfp mice. After 8-week HFD, staining of atherosclerotic lesions with anti-GFP antibody showed very low number of NK cells, 1 to 2 cells per lesion examined (Figure 7A). In contrast, NK cells were abundant in the spleen.
Natural killer (NK) cells are barely detected in human and mouse atherosclerotic plaques. A, Photomicrographs of NK cells (anti-GFP+) detected in spleen and aortic sinus of Ncr-1gfp/gfp → Ldlr−/− mice (n=3) at 8-wk high-fat diet (HFD); the dashed line marks the intima, lumen (L), and adventitia (A). B, Immunohistochemistry in human atherosclerotic plaques (n=17). Few NKp46+ NK cells (brown) are detected in early fatty streak lesions (n=5), and none in advanced carotid artery plaques (n=12). GFP indicates green fluorescent protein; Ldlr−/−, low-density lipoprotein receptor null; and MOMA-2, macrophage/monocyte antibody.
In human, we also found very few NK cells in atherosclerotic plaques. Only 1 or 2 cells were detected in the plaques of uncomplicated atherosclerosis, whereas none were observed in advanced plaque lesions (Figure 7B). Specificity of the anti-human NKp46 antibody was tested in human spleen samples.
Discussion
Using 2 state-of-the-art genetic approaches, our study provides strong evidence that, in contrast to most previous studies, NK cells have no role in atherosclerosis development, unless a systemic inflammation is induced. Transfer of the BM of selectively depleted (Ncr1iCreR26lsl−DTA) or hyperresponsive (Noé) mice into lethally irradiated Ldlr−/− mice had no effect on plaque size (aortic root and thoracic aorta) and composition, or immune-inflammatory response, when compared with control mice. However, treating NK-cell–deficient chimeric mice with poly(I:C), in an attempt to mimic viral infection, significantly reduced lesion size, which suggests that NK cells might be proatherogenic only when activated by virus infection or inflammatory cytokines or in the context of tumor growth.
Earlier studies on the role of this innate immune cell population in atherosclerosis led to conflicting results. The very first studies to examine the role of NK cells in atherosclerosis used the beige mutant mice. Accelerated atherosclerosis was observed in beige Ldlr−/− mice, and NK cells were accredited to be antiatherogenic.12 However, an alternative NK-cell–independent hypothesis is most likely because the beige mutation which leads to functional changes in lysosome trafficking may involve other cells such as lesion macrophages or smooth muscle cells leading to increased foam cell formation or apoptosis.10,35 In contrast to these studies, Ldlr−/− mice repopulated with Ly49A transgenic mice BM and which present an NK-cell deficiency exhibited reduced atherosclerosis.7 However, the expression of the inhibitory receptor Ly49A, which is regulated by granzyme A promoter in this model, is induced not only in NK cells but also in subsets of CD4+ and CD8+ T cells.16 Hence, this expression on T cells could be responsible for the decrease in lesion size in Ly49A transgenic BM recipients. Lately, experiments in Apoe−/− mice fed an HFD treated with anti–asialo-GM1 antibody to deplete NK cells attenuated atherosclerosis. In addition, adoptive transfer of ex vivo activated NK cells increased lesion size in lymphocyte-deficient Apoe−/− mice, an effect that was prevented by transferring perforin- or granzyme-deficient NK cells, which suggests that NK cells augment atherosclerosis via cytotoxic dependent mechanisms.23 However, the expression of asialo-GM1 is not strictly confined to NK cells among hematopoietic cells and is detected in subpopulations of NKT, CD8 T, and γδ T cells36,37 and in some activated forms of CD4+ T cells, macrophages, eosinophils, and basophils under certain experimental conditions.38–40 Our data suggest that the discrepancy observed in former studies was most likely because of effects on other cell populations targeted in addition to NK cells. Indeed, our study clearly demonstrates that NK cells have no role in atherosclerosis development.
NKp46 is also expressed in ILC1 and a subset of ILC3 (NKp46+ ILC3s). These cells are also deleted in the Ncr1iCreR26RDTA mice.25 In our experimental protocol, recipient mice were irradiated before BM transplantation. Both NK cells and ILC1s are sensitive to irradiation and should thus originate from donor BM. About the ILC3 population, NKp46− ILC3 have been shown to be resistant to irradiation,41,42 but we have no information about the radioresistance of NKp46+ ILC3. Therefore, in our model of BM transplantation of Ncr1iCreR26RDTA BM into irradiated Ldlr−/− mice, not only NK cells but also ILC1s and potentially NKp46+ ILC3s should have been depleted, whereas other lymphocytes and myeloid cells were unaffected, suggesting that these other ILCs subset do not play a major role in atherosclerosis either.
Moreover, we questioned the specificity of using anti–asialo-GM1 treatment to target NK cells in the context of atherosclerosis. We clearly demonstrated that its effect was NK cell independent as atherosclerosis was reduced by anti–asialo-GM1 treatment to the same extent in control and NK-deficient mice. We observed that the anti–asialo-GM1 treatment not only depleted NK cells but also strongly reduced NKT and CD8+ T-cell population after 8-week HFD. Among CD8+ T cells, the IFN-γ–producing CD8 T cells were also reduced. This is consistent with a previous study showing that the asialo-GM1+ CD8 population is high in IFN-γ–producing cells in response to stimulus.36 These activated CD8+ T cells which are targeted by anti–asialo-GM1 antibodies might thus be involved in the phenotype observed in atherosclerosis. This is consistent with the fact that CD8+ T-cell activation has been shown to exacerbate atherosclerosis.43 About the NKT cell population, although previous studies indicate that NK-cell depletion with antibody against asialo-GM1 does not always drastically affect the size or function of NKT cell population,44 NKT cell population was significantly decreased in hypercholesterolemic mice treated with anti–asialo-GM1 antibody. NKT cells are considered proatherogenic.45 Their depletion could thus also contribute to the protective phenotype in anti–asialo-GM1–treated mice.
We used 2 models in which NK cells functions are increased: the Noé model in which NK cells are known to be intrinsically hyperresponsive and the poly(I:C) model in which NK cells are activated because of an inflammatory environment.
Using the Noé mice, we showed that NK-cell hyperresponsiveness in sterile conditions does not affect the atherosclerotic process, although hypercholesterolemia modestly increased IFN-γ–producing NK-cell population. In contrast, a nonredundant role of NK cells was revealed in an inflammatory context induced by chronic poly(I:C) injection. Along this line, previous studies reported an aggravation of atherosclerosis in hypercholesterolemic mice infected with mouse cytomegalovirus.46 NK cells are one of the first innate immune cells to respond to and contain the mouse cytomegalovirus infection through their direct perforin-mediated cytotoxicity function and IFN-γ release.47 In our study, the viral mimic poly(I:C) significantly augmented NK-cell activation marker CD69 and CD107a degranulation marker. CD107a is a lysosomal-associated membrane protein lining the membrane of cytotoxic granules.48 Its expression is correlated with degranulation and target cell lysis.49 In addition, in our experiments, we noticed an upregulation of the mRNA of perforin, granzyme B, and IFN-γ in the spleens of poly(I:C)-treated control mice, which highlights the marked activation of NK cells in response to poly(I:C). Interestingly, only the expression of granzyme B was significantly decreased in both spleen and aorta of poly(I:C)-treated NK-deficient mice, which points out to a role of NK cytolysis in the proatherogenic effects of NK cells, in agreement with previous studies by Selathurai et al.23 In control mice, we observed no effect of poly(I:C) on atherosclerosis in the aortic root. The discrepancy observed could be attributed to the multiple effects of poly(I:C) in atherosclerosis context. Zimmer et al50 demonstrated that intravenous administration of poly(I:C) induced endothelial dysfunction and increased atherosclerotic lesion development. In contrast, Cole et al51 described poly(I:C)-mediated atheroprotection. Intraperitoneal administration of poly(I:C) attenuated neointima formation and reduced injury-induced media damage.51 Further studies are required to fully elucidate how poly(I:C) influence atherosclerosis. Nonetheless, this series of experiments clearly establish that even though resting NK cells have no effects on atherosclerosis, NK cells might exacerbate atherosclerosis in case of activation by an inflammatory context like viral infections. The ability of activated NK cells to influence other immune cells in the atherosclerotic process should be considered. In immune-competent mice, NK-cell–derived IFN-γ promotes CD4+ TH1 (type 1 T helper) priming.52 It is well established that IFN-γ–producing TH1 subset is a major proatherogenic subset of the adaptive immune system and may activate proatherogenic properties of both vascular and immune cells.53 Moreover, perforin and granzymes, which are highly augmented after poly(I:C) treatment, lead to target cell apoptosis and can accelerate atherosclerosis by increasing necrosis.23
In conclusion, our findings indicate that NK cells have no role in atherosclerosis development in hypercholesterolemic mice, unless a superimposed inflammatory status mimicking viral infection is induced. Further studies are needed for detailed characterization of the mechanisms, whereby NK cells’ response to viral infection would influence the atherosclerotic process.
Sources of Funding
This work was supported by Inserm, the Fondation pour la Recherche Medicale, CNRS (E. Vivier, S. Ugolini), and Aix-Marseille University (E. Vivier, S. Ugolini).
Acknowledgments
W. Nour-Eldine has received a scholarship from the Association of Specialization and Scientific Guidance (ASSG).
Disclosures
None.
Footnotes
In September 2017, the average time from submission to first decision for all original research papers submitted to Circulation Research was 13 days.
The online-only Data Supplement is available with this article at http://circres.ahajournals.org/lookup/suppl/doi:10.1161/CIRCRESAHA.117.311743/-/DC1.
- Nonstandard Abbreviations and Acronyms
- asialo-GM1
- asialoganglioside M1
- ATF2
- activating transcription factor 2
- BM
- bone marrow
- FITC
- fluorescein isothiocyanate
- GFP
- green fluorescent protein
- HFD
- high-fat diet
- IL
- interleukin
- Ldlr−/−
- low-density lipoprotein receptor null
- NK
- natural killer
- poly(I:C)
- polyinosinic:polycytidylic acid
- TLR
- Toll-like receptor
- TNF
- tumor necrosis factor
- WT
- wild type
- Received July 21, 2017.
- Revision received October 12, 2017.
- Accepted October 17, 2016.
- © 2017 American Heart Association, Inc.
References
Novelty and Significance
What Is Known?
Adaptive immunity contributes to atherosclerosis.
The role of natural killer (NK) cells in atherosclerosis has been investigated in murine models in the past, but the results have been mixed.
The previous mouse models are limited by a lack of selectivity with regard to deletion of NK cells.
What New Information Does This Article Contribute?
Specific NK deletion or hyperresponsiveness does not affect atherosclerosis development in hypercholesterolemic mice.
The absence of NK cells under simulated viral infection reduces atherosclerosis.
In general, previous studies suggested that NK cells play a proatherogenic role; however, 1 recent study reported that NK cells are atheroprotective. These studies used loss of NK-cell function in several mouse models, including beige mutant mice, Ly49A transgenic mice, and antibody-mediated depletion (anti–asialoganglioside M1 antibody) in vivo approach. These models are limited by a lack of selectivity with regard to deletion of NK cells; they may involve loss of function for other immune cells, including neutrophils, macrophages, or smooth muscles cells. In this study, we used 2 novel mouse models to investigate the role of NK cells in atherosclerosis: (1) Ncr1iCreR26IsI−DTA model, where NK cells are specifically deleted and (2) the Noé mice in which NK cells are hyperresponsive. Our study provides novel evidence for the lack of a role for NK cells in atherosclerotic development in bone marrow transplantation studies in Ldlr−/− (low-density lipoprotein receptor null) mice. However, data using the viral mimic polyinosinic:polycytidylic acid demonstrate a significant reduction of plaque size in NK-cell–deficient chimeric mice, supporting a proatherogenic role for NK cells when activated by inflammatory conditions.
This Issue
Jump to
Article Tools
Share this Article
- Genetic Depletion or Hyperresponsiveness of Natural Killer Cells Do Not Affect Atherosclerosis DevelopmentNovelty and Significance
