Nrf2 is essential for cholesterol crystal-induced inflammasome activation and exacerbation of atherosclerosis



Oxidative stress and inflammation — two components of the natural host response to injury — constitute important etiologic factors in atherogenesis. The pro-inflammatory cytokine interleukin (IL)-1 significantly enhances atherosclerosis, however, the molecular mechanisms of IL-1 induction within the artery wall remain poorly understood. Here we have identified the oxidative stress-responsive transcription factor NF-E2-related 2 (Nrf2) as an essential positive regulator of inflammasome activation and IL-1-mediated vascular inflammation. We show that cholesterol crystals, which accumulate in atherosclerotic plaques, represent an endogenous danger signal that activates Nrf2 and the NLRP3 inflammasome. The resulting vigorous IL-1 response critically depended on expression of Nrf2, and Nrf2-deficient apolipoprotein E (Apoe)−/− mice were highly protected against diet-induced atherogenesis. Importantly, therapeutic neutralization of IL-1α and IL-1β reduced atherosclerosis in Nrf2+/−Apoe−/− but not in Nrf2−/−Apoe−/− mice, suggesting that the pro-atherogenic effect of Nrf2-signaling was primarily mediated by its permissive role in IL-1 production. Our studies demonstrate a role for Nrf2 in inflammasome activation, and identify cholesterol crystals as disease-relevant triggers of the NLRP3 inflammasome and potent pro-atherogenic cytokine responses. These findings suggest a common pathway through which oxidative stress and metabolic danger signals converge and mutually perpetuate the chronic vascular inflammation that drives atherosclerosis.


Atherosclerosis is a complex, multi-factorial disease characterized by the progressive accumulation of lipids and infiltrating leukocytes in the arterial wall 1, 2. Although major risk factors have been identified and provide targets for therapeutic intervention, much less is known about the molecular mechanisms that initiate and sustain the underlying inflammatory process 1, 3, 4. The pro-inflammatory cytokine IL-1 conveys many atherogenic effects by inducing the expression of adhesion molecules on endothelial cells, the production of chemokines and cytokines, as well as vascular smooth muscle cell proliferation 5. As a consequence, IL-1-signaling severely enhances atherosclerosis and vascular inflammation 6–10. Monocytes and vascular cells can produce IL-1, but the mechanism of IL-1 induction in the arterial wall remains unclear.

Oxidative stress is a central aspect of atherogenesis and oxidized lipids contribute to disease progression via an array of pro-atherogenic and pro-inflammatory properties 11, 12. The transcription factor NF-E2-related 2 (Nrf2) limits the detrimental effects of oxidative stress by inducing key antioxidant enzymes. Nrf2-signaling triggered by oxidized lipids 13–15, shear stress 16, or targeted gene transfer 17–19 protects vascular cells from oxidative stress in vitro and in vivo. Moreover, some Nrf2-induced genes have been associated with cardiovascular disease by genetic studies 20, 21, and ameliorate atherosclerosis in animal models 18, 22–24. Accordingly, Nrf2-signaling is considered to protect from atherosclerosis by inducing the expression of antioxidant enzymes. Paradoxically, the overall effect of Nrf2-signaling has been reported to enhance atherogenesis in mice 25, which remains unexplained. Here we have investigated the role of Nrf2-signaling in atherogenesis and found that despite being the transcriptional master regulator of the antioxidant response, Nrf2 drastically aggravated atherosclerosis. We show that this effect was not related to an influence of Nrf2 on serum lipids, lipid peroxidation or lipoprotein profiles, but was attributed to a previously unknown function of Nrf2 in inflammasome activation and IL-1 production in response to cholesterol crystals. Our results suggest that while Nrf2 is induced as a defense mechanism against oxidative stress it at the same time further enhances atherogenesis by supporting IL-1-mediated vascular inflammation, and may therefore represent a suitable strategic target for therapy.


Nrf2-signaling aggravates atherosclerosis

To elucidate the role and mechanism of Nrf2-signaling in atherogenesis, we examined the effect of Nrf2-deficiency on the atherosclerosis-susceptible apolipoprotein E (Apoe)−/− background. The absence of Nrf2 protected from atherosclerosis and markedly reduced the lesion size in Nrf2−/−Apoe−/− mice by 50% as compared with their heterozygous Nrf2+/−Apoe−/− littermates (Fig. 1A and B and Supporting Information Fig. 1), which was in stark contrast to the well-established function of Nrf2 in the antioxidant response 15, 18, 26. Nrf2-deficiency did not affect lipid metabolism, as Nrf2+/−Apoe−/− and Nrf2−/−Apoe−/− mice exhibited comparable serum lipoprotein profiles (Fig. 1C) and plasma cholesterol levels (Fig. 1D). Furthermore, both strains had similar levels of circulating lipoperoxides in serum (Fig. 1E). Nrf2-signaling induces scavenger receptor expression 14, and could thus promote cellular lipid uptake and foam cell formation. To investigate this possibility, we exposed Nrf2-expressing and Nrf2-deficient macrophages to atherogenic lipids in vitro (Fig. 1F). Both Nrf2−/−Apoe−/− and Nrf2+/−Apoe−/− macrophages were efficiently transformed into foam cells by copper-oxidized low density lipoprotein (OxLDL) as well as by dyslipidemic murine serum containing oxidized lipid moieties generated on minimally modified LDL in vivo (Fig. 1F). Nrf2-deficient macrophages were also equally susceptible to cholesterol loading with β-methyl-cyclodextrin (βMCD), which excluded increased cholesterol efflux as a mechanism. Thus, Nrf2 affected neither lipid metabolism nor susceptibility to foam cell transformation, and therefore enhanced atherogenesis through a yet-unidentified pathway.

Figure 1.

Nrf2 exacerbates atherosclerosis without altering lipid metabolism or susceptibility to foam cell transformation. (A and B) Extent of atherosclerosis in Nrf2+/−Apoe−/− (n=11) and Nrf2−/−Apoe−/− mice (n=21) after being fed on a high fat diet containing 0.5% cholesterol (HFCD) for 20 wk. (A) Selected examples of en face preparations of the aorta and (B) data for all mice. Circles represent individual mice and gray lines indicate mean±SEM. ***p<0.001 as determined by the two-tailed t-test. (C) Serum lipoprotein profiles of diet-fed Nrf2−/−Apoe−/− (closed circles) and Nrf2+/−Apoe−/− (open triangles) mice. Pooled serum of five mice from each group was separated by gel filtration chromatography before protein and cholesterol concentrations were determined in individual fractions. The VLDL/LDL (very low density lipoprotein/low density lipoprotein) and HDL (high density lipoprotein) peaks are indicated. (D) Cholesterol levels in plasma of Nrf2+/−Apoe−/− (n=11) and Nrf2−/−Apoe−/− mice (n=21) at the end of the study. Data are shown as mean+SEM. (E) Levels of circulating lipoperoxides were determined as TBARS (thiobarbituric acid reactive substances) in serum. Data are shown as mean+SEM. (F) Foam cell transformation of thioglycollate-elicited Nrf2+/−Apoe−/− and Nrf2−/−Apoe−/− macrophages upon challenge with the indicated lipids for 48 h as revealed by Oil Red O staining. OxLDL, copper-oxidized LDL; HFCD serum, dyslipidemic serum isolated from cholesterol diet-fed Ldlr−/− mice; Chol./βMCD, cholesterol/β-methyl-cyclodextrin complexes. Micrographs were taken at 40× magnification. One of the three independent experiments is shown.

Cholesterol crystals induce inflammatory responses in vitro and in vivo

cholesterol crystals are consistently found in atherosclerotic plaques, and anecdotal evidence links their presence to inflammatory responses 27–30. We hypothesized that cholesterol crystals represent an atherosclerosis-relevant “injuring agent” that triggers and perpetuates vascular inflammation, and tested the response of BM-derived macrophages to cholesterol crystals in vitro. Compared with the monosodium urate (MSU) crystals implicated in gouty inflammation 31, 32, cholesterol crystals induced higher levels of IL-1α, less IL-1β and IL-6, and similar amounts of IL-12 (Fig. 2A). Moreover, exposure to cholesterol crystals transformed macrophages into foam cells, which resembled those induced by OxLDL (Fig. 2B). Similar results were obtained using RAW264.7 macrophages, thereby confirming the ability of cholesterol crystals to trigger these pro-atherogenic responses (Fig. 2C and D). Consistent with the potent induction of IL-1 responses in vitro, intraperitoneal injection of cholesterol crystals elicited serious peritonitis in vivo, as indicated by massive infiltration of not only neutrophils, monocytes but also eosinophils (Fig. 2E–G). Altogether, cholesterol crystals induced a quantitatively and qualitatively distinct inflammatory response in vitro and in vivo.

Figure 2.

Cholesterol crystals (CC) induce inflammatory responses in vitro and in vivo. (A) To evaluate the crystal-induced inflammatory response, WT BM-derived macrophages were stimulated with MSU and CC for 18 h before the concentration of indicated cytokines in culture supernatants was determined by ELISA. Dashed lines indicate the cytokine levels measured in non-stimulated cultures. (B) Foam cell formation in macrophages pulsed with CC or OxLDL for 48 h; the intracellular lipid accumulation was revealed by Oil Red O staining. Micrographs were taken at 40× magnification. (C) Inflammatory cytokine responses and (D) foam cell transformation of RAW264.7 macrophage cells exposed to CC in vitro. n.d., none detected. Micrographs were taken at 40× magnification. (A and C) Data shown represent mean and SEM of triplicate cultures from one of two similar experiments. (E–G) Characterization of crystal-elicited peritonitis. (E) Cellularity, (F) differential counts, and (G) flow cytometric analysis of the inflammatory infiltrate in peritoneal exudates of mice injected with CC or MSU crystals 18 h earlier. The experiment was performed twice with four mice per group. Data are shown as mean+SEM.

Cholesterol crystals accumulate in atherosclerotic lesions in vivo

Large, extracellular cholesterol crystals are frequently observed as “cholesterol clefts” in advanced plaques, but removal of the crystals by lipid solvents prevents their direct visualization with prevalent methods. Conversely, in vitro studies indicate that cholesterol crystallizes in macrophage-foam cells during the very initial stages of atherosclerosis 33, 34. Using in situ fixation followed by cryo-sectioning of atherosclerotic lesions, we detected a uniform layer of needle-shaped cholesterol crystals at the bottom of the necrotic core (Fig. 3A and B). Moreover, we frequently observed isolated crystals embedded within single Oil Red O-positive lipid droplets (Fig. 3C and D) that likely originated within the lipid droplet of a cell, which disappeared after rupture of its membranes by the growing crystal 34, 35. Due to their anisotropic nature, cholesterol crystals exhibit typical birefringence under cross-polarized light. Analysis of atherosclerotic lesions for birefringent crystals revealed the high density of large, extracellular cholesterol crystals within the advanced plaque (Fig. 3G and H), which was not appreciable by bright field microscopy (Fig. 3F). In addition, it also indicated the presence of smaller, amorphic intracellular cholesterol crystals in less advanced lesions (Fig. 3I and J). Thus, cholesterol crystals were much more abundant in lesions than assumed by conventional staining techniques, and accumulated at high levels within atherosclerotic plaques.

Figure 3.

CC accumulate in atherosclerotic plaques in vivo. CC were observed in Oil Red O- and HE-stained atherosclerotic lesions of HFCD-fed Apoe−/− mice as (A and B) a layer of needle-shaped crystals at the bottom of the necrotic core, (C and D) frequently in the form of single needles embedded within Oil Red O-positive droplets of neutral fat, and (E and F) in part forming stacks of crystals. (G–J) Cross-polarized light microscopy reveals CC by their birefringence as (G) white crystals or (H–J) depending on their orientation within the sections as blue and yellow crystals. Compared with analysis of the same section in (F) bright field illumination only, analysis under (G) cross-polarized light reveals the high density of CC present in advanced plaques. CC are present at high density in the form of (G and H) large extracellular crystals in advanced plaques as well as (I and J) small intracellular crystals within less advanced lesions.

The inflammatory cytokine response to cholesterol crystals is regulated by Nrf2, NLRP3, and caspase-1 (Casp1)

Having identified cholesterol crystals as potent pro-inflammatory stimuli with direct pathophysiological significance for atherosclerosis, we next asked whether Nrf2 impacted the inflammatory response to cholesterol crystals. While exposure to cholesterol crystals triggered potent IL-1α and IL-1β production in Nrf2+/+ DCs, this response was almost completely abrogated in the absence of Nrf2. In contrast, Nrf2-deficient DCs produced higher amounts of IL-12 (Fig. 4A). Similar results were obtained upon crystal stimulation of Nrf2-expressing and Nrf2-deficient macrophages (Fig. 4B and C). Thus, Nrf2 heavily modulated the physiological cytokine response to cholesterol crystals. Given that IL-1-signaling substantially potentiates atherosclerosis and vascular inflammation 6–10, 36, and provided that we observed vigorous Nrf2-dependent IL-1α and IL-1β responses upon cholesterol crystal-stimulation, we further investigated the cholesterol crystal-induced IL-1 release as a proatherogenic, Nrf2-dependent pathway. The impaired IL-1 responses of Nrf2-deficient cells did not represent a defective generation of the protein precursors as both Nrf2−/− and Nrf2+/+ macrophages produced comparable amounts of intracellular pro-IL-1α and pro-IL-1β (Supporting Information Fig. 2). Release of mature IL-1β requires proteolytic activation of Casp1 by the inflammasome 37, 38. Indeed, IL-1β release was abrogated in NLRP3-deficient macrophages, suggesting that cholesterol crystals triggered Casp1 activation via the NLRP3 inflammasome (Fig. 4D). Moreover, cholesterol crystal-induced IL-1β but not IL-1α responses were drastically reduced in Casp1−/− macrophages, thus demonstrating a direct requirement for Casp1 (Fig. 4E). Complementing these results, crystal-stimulated Nrf2-deficient macrophages also exhibited impaired Casp1 activity as revealed by the reduced secretion of processed, mature IL-1β (p17) (Fig. 4F). Nonetheless, expression of Casp1, NLRP3, or Nrf2 was not required for inducing foam cell formation in macrophages (Supporting Information Fig. 3). While these data established that Nrf2 regulates cholesterol crystal-elicited cytokine responses in vitro, we next tested the importance of Nrf2 for the inflammatory response to cholesterol crystals in vivo. Nrf2-deficiency did not influence the numbers of eosinophils, lymphocytes, or monocyte/macrophages present in the peritoneal exudates of crystal-challenged mice. In contrast, the absence of Nrf2 significantly reduced the numbers of infiltrating neutrophils, as would be expected from the impaired IL-1 production of Nrf2-deficient cells in vitro (Fig. 4G).

Figure 4.

The inflammatory cytokine response to CC is regulated by Nrf2, NLRP3, and Casp1. (A) Production of the indicated pro-inflammatory cytokines by Nrf2-expressing (open circles) and Nrf2-deficient (closed circles) BM DCs after exposure to CC for 18 h. (B–E) IL-1α and IL-β release from LPS-primed thioglycollate-elicited macrophages in response to indicated stimuli. Analysis of (B) Nrf2−/− and Nrf2+/− macrophages, (C) Nrf2−/−Apoe−/− and Nrf2+/−Apoe−/− macrophages, (D) NLRP3−/− and NLRP3+/+ macrophages, and (E) Casp1−/− and Casp1+/+ macrophages. (F) Western blot analysis of LPS-primed Nrf2-deficient and WT macrophages for secretion of processed, mature IL-1β in response to cholesterol and MSU crystal stimulation. Actin was used as loading control. (G) Characterization of crystal-elicited peritonitis in Nrf2−/− and Nrf2+/− mice 18 h post cholesterol crystal injection. Absolute cell counts are shown. Data are shown as mean+SEM and are representative of (A, C, and G) two independent experiments and (B, D, and E) at least three independent experiments. *p<0.05 as determined by two-tailed t-test.

Cholesterol crystals induce Nrf2-signaling

It appeared from these results that Nrf2 expression was essential for the inflammatory response to cholesterol crystals. We therefore next tested whether cholesterol crystals directly activated Nrf2 and monitored the expression of Nrf2-dependent genes in macrophages upon exposure to cholesterol crystals in vitro (Fig. 5). Indeed, stimulation with cholesterol crystals significantly increased the expression of Nqo1, Hmox1, as well as Prdx1, albeit with different efficiency. Compared with the peak expression levels observed in response to tert-butylhydrochinone (tBHQ), a potent activator of Nrf2-signaling, the cholesterol crystal-induced expression of Nqo1, Hmox1, and Prdx1 reached 50, 33, and 100% of the maximal level for the respective genes (Fig. 5A–C). An equivalent induction of Nrf2-dependent gene expression was, however, not observed in response to MSU stimulation. These in vitro data suggested that cholesterol crystals not only triggered Nrf2-dependent IL-1α and IL-1β responses, but also could potentially signal through Nrf2.

Figure 5.

CC induce transcription of Nrf2-dependent genes. The induction of expression of the Nrf2-dependent genes (A) Nqo1, (B) Hmox1, and (C) Prdx1 in BM-derived macrophages as assessed by quantitative PCR after treatment with indicated stimuli for 4 h. Data represent the mean+SEM of triplicate cultures from one of two similar experiments. *p<0.05; **p<0.01 as determined by two-tailed t-test.

Nrf2 aggravates atherosclerosis by promoting cholesterol crystal-induced IL-1 production

Collectively, our experiments had thus far established that Nrf2 promoted diet-induced atherosclerosis, and at the same time was required for the cholesterol crystal-induced inflammasome activation and IL-1 production during foam cell formation. We therefore sought to determine whether the pro-atherogenic effect of Nrf2 was directly related to its permissive role in the IL-1 response to cholesterol crystals and assessed the effect of IL-1 neutralization on atherogenesis in both mouse strains. To efficiently neutralize endogenous IL-1α and IL-1β in diet-fed Nrf2−/−Apoe−/− and Nrf2+/−Apoe−/− mice, we induced high titers of neutralizing antibodies against both cytokines using a virus-like particle (VLP)-based vaccine 39. Nrf2−/−Apoe−/− and Nrf2+/−Apoe−/− mice received three biweekly immunizations with VLP displaying murine IL-1α and IL-1β (IL-1.VLP) or with empty VLP (Qβ.VLP) before diet feeding (Fig. 6A). All immunized mice developed high titers of Qβ-specific antibodies, indicating that both vaccines provided comparable immune stimulation (Fig. 6B). As expected, only IL-1.VLP induced antibodies against IL-1α and IL-1β, and mice of both genotypes responded equally well to this vaccination (Fig. 6B). To assure efficient IL-1 neutralization, we administered two booster immunizations throughout the study. Vaccination with control Qβ.VLP neither influenced atherogenesis in Nrf2+/−Apoe−/− or Nrf2−/−Apoe−/− mice (Fig. 6C and D) and the reduction of atherosclerosis in Qβ.VLP-immunized Nrf2−/−Apoe−/− versus Nrf2+/−Apoe−/− mice was similar to that observed for the respective non-immunized mice (Fig. 1). Importantly, while IL-1 neutralization conferred significant protection against atherosclerosis to Nrf2+/−Apoe−/− mice and reduced atherosclerosis in these mice to the level of control-immunized Nrf2−/−Apoe−/− mice, it had no such effect in Nrf2−/−Apoe−/− mice (Fig. 6C and D). Thus, Nrf2-expression was required for protection by IL-1 neutralization, indicating that the pro-atherogenic effect of Nrf2 was imparted by its permissive role in IL-1 production.

Figure 6.

Nrf2 aggravates atherosclerosis by promoting the cholesterol crystal-induced IL-1 production. (A) Nrf2+/−Apoe−/− and Nrf2−/−Apoe−/− mice were immunized with either VLPs displaying murine IL-1α and IL-1β (IL-1.VLP) or with empty control VLP (Qβ.VLP) before atherogenic diet was started. (B) IgG antibody titers in serum of VLP-immunized Nrf2+/−Apoe−/− and Nrf2−/−Apoe−/− mice. (C and D) Extent of atherosclerosis in IL-1.VLP-immunized and control VLP-immunized Nrf2+/−Apoe−/− and Nrf2−/−Apoe−/− mice determined as surface atherosclerosis. (C) En face preparations of aortae of representative mice and (D) data for all mice. (B and D) Data represent the mean+SEM of groups of 6–8 mice. *p<0.05; **p<0.01 as determined by two-tailed t-test.


The importance of IL-1 receptor signaling 6, 8–10 and the potential involvement of inflammasomes 40 in atherogenesis have been recognized. However, we now identify cholesterol crystals as the relevant metabolic danger signal that triggers and sustains atherogenic inflammation through the NLRP3 inflammasome and induction of IL-1. During the preparation of this manuscript, the cholesterol crystal-triggered inflammasome activation and IL-1β induction was reported 41, 42, whereas we here show cholesterol crystal-induced Casp1-independent IL-1α production as well as NLRP3/Casp1-dependent IL-1β responses, and demonstrate that both critically depend on Nrf2 to promote atherosclerosis. Thus, despite a partial overlap with the study of Duewell et al. 41, our results significantly extend these shared observations and put them into a completely novel perspective. In particular, our findings establish Nrf2 as a novel essential regulator of the cholesterol crystal-elicited inflammatory response that directly relays oxidative stress to vascular inflammation.

Oxidative stress and inflammation form a common signature of chronic inflammatory and metabolic diseases, several of which represent risk factors for atherosclerosis and are linked to inflammasome activation 43–45. Our data suggest that these conditions may signal via the Nrf2-linked pathway that is triggered by cholesterol crystals, and could contribute to atherogenesis through increasing oxidative stress-induced Nrf2-signaling or by potentiating the inflammasome-IL-1 axis. Similarly, oxidative stress, perturbations of cellular lipid homeostasis 46, 47, cholesterol crystal formation 48, and inflammasome activation 49 occur in microbial infection, which may therefore amplify cholesterol crystal-driven inflammation via these shared pathogenic mechanisms. It will thus be important to determine how Nrf2 regulates such potential crosstalk between chronic inflammatory diseases, infections, and atherogenesis.

At this time we can only speculate on how Nrf2 may simultaneously regulate both IL-1α and IL-1β responses. Besides the possibility that completely unrelated functions of Nrf2 independently regulate IL-1α and IL-1β we perceive several potential explanations. First, both cytokines depend on the unconventional protein secretion pathway, which might require functional Nrf2-signaling. However, the release of mature IL-1α via unconventional protein secretion has been associated to Casp1 activation 50, whereas we found the cholesterol crystal-elicited IL-1α response to be Casp1-independent (Fig. 4E). Thus, the release of other factors that rely on unconventional protein secretion, such as FGF-2, galectin-3, or HMGB1, will have to be examined in Nrf2-deficient cells in order to further investigate this possibility. Second, Nrf2 regulates a large battery of genes that predominantly mediate cyto-protective responses against oxidant or electrophilic stress, but also conserve cellular organelle integrity. Both cholesterol and MSU crystals likely trigger inflammasome activation by causing phagosomal destabilization 51. It is thus conceivable that Nrf2-signaling could facilitate physiological IL-1α and IL-1β responses by sensing loss of organelle integrity similar to that shown for the NLRP3 inflammasome 51–53, and that this function of Nrf2 is required to initiate the signaling cascade that leads to the processing and secretion of bioactive IL-1α and IL-1β. Notably, Nrf2-signaling is linked to autophagy 54, a cellular process that recently has not only been demonstrated to regulate the inflammasome activity for functional IL-1β responses 52 but also appears to contribute to the unconventional protein secretion pathway 55 that is important for cellular release of both, IL-1α and IL-1β.

In conclusion, the results presented in this study identify the oxidative stress-inducible transcription factor Nrf2 as a novel positive regulator of the inflammasome and demonstrate that Nrf2-signaling severely aggravates atherosclerosis by enhancing the IL-1-mediated vascular inflammation. Our findings suggest that combined IL-1α/IL-1β neutralization may provide a valid therapeutic approach for cardiovascular disease that selectively limits the Nrf2-regulated inflammatory consequences of oxidative stress without interfering with the beneficial antioxidant defenses induced by Nrf2-signaling in vascular cells.

Materials and methods


Apoe-deficient mice (N12 on the C57BL/6 background) were obtained from Jackson Laboratories (Bar Harbor, ME, USA). Nrf2−/− mice 26, crossed to C57BL/6 for more than eight generations, were provided by the RIKEN BioResource Center, Japan. Nrf2+/−Apoe−/− and Nrf2−/−Apoe−/− mice were generated by intercrossing Nrf2−/− and Apoe−/− strains. For induction of atherosclerosis, age- and sex-matched littermates were fed a high-fat, high-cholesterol diet (D12107C containing 0.5% cholesterol; Research Diets, New Brunswick, NJ, USA) for 20 wk. NLRP3−/− mice 32 on the C57BL/6 background were provided by Dr. Jürg Tschopp, University of Lausanne, Epalinges, Switzerland. Casp1−/− mice 56 were provided by Dr. Wolf-Dietrich Hardt, Federal Institute of Technology, Zurich, Switzerland. All animal experiments were performed according to institutional guidelines and Swiss federal regulations, and were approved by the veterinary office of the Kanton of Zurich (permission no. 148/2008).

Evaluation of atherosclerosis

Analysis of atherosclerosis was performed as published previously 57. Mice were sacrificed by CO2 overdose. After collection of a blood sample from the vena cava inferior, mice were perfused with PBS for 5 min via a cannula inserted into the right ventricle, followed by an initial in situ fixation with PBS containing 4% formalin, 5% sucrose, and 2 mM EDTA for additional 5 min. After perfusion and fixation, organ samples were harvested, frozen in Tissue-Tek O.C.T. compound (Sysmex-Digitana AG, Horgen, Switzerland) and stored at −80°C until analysis by immunohistochemistry. The heart and aorta were dissected under a Leica Wild M3Z stereo dissecting microscope (Leica Microsystems, Heerbrugg, Switzerland). In brief, surrounding tissue was carefully removed to expose the heart and ascending aorta. The ascending aorta was then cut just above the heart and the upper half of the heart was removed, rinsed with PBS and frozen in O.C.T. compound for analysis of atherosclerosis at the aortic origin. Next, the aorta was completely exposed from the arch to the iliac bifurcation; adhering fat tissue and adventitia as well as the branching small arteries were removed. While still attached to the large branching cervical, renal and iliac arteries, the aorta was cut open longitudinally starting from a small incision in one of the iliac arteries. The aorta was then harvested by cutting of the remaining branching arteries and pinned against a black silicone surface (Dow Corning, Wiesbaden, Germany) with Minutien Pins (Fine Science Tools GmbH, Heidelberg, Germany) in a 15 cm petri dish. Aortae were then fixed with 4% paraformaldehyde in PBS for additional 20 min, washed two times with PBS, and stained with Oil Red O to reveal atherosclerotic lesions. For this purpose, the aortic en face preparations were rinsed with 60% isopropanol and stained with a 0.3% Oil Red O (Sudan IV; Sigma O0625) solution in 60% isopropanol for 30 min, followed by a 2 min destaining in 60% isopropanol and two further washes with PBS. The en face preparations were photographed using a Nikon D50 digital camera, and the extent of atherosclerosis was evaluated by morphometric image analysis using the Motic Images Plus 2.0 software (Motic, Wetzlar, Germany).

Cholesterol crystals

cholesterol crystals were produced in vitro from a 10 mg/mL cholesterol solution in ethanol. Precipitated crystals were washed twice with PBS, resuspended in complete RPMI 1640 at a concentration of 5 mg/mL, and shortly sonicated in a water bath to obtain a homogeneous crystal solution. From this stock solution, cholesterol crystals were further diluted for in vitro stimulations as indicated. To assess responses to cholesterol crystals in vivo, washed crystals were resuspended in PBS, and 600 μL of a 5 mg/mL solution were i.p. injected.

Analysis of peritonitis

The in vivo response to cholesterol crystals and control MSU crystals (prepared as in 31) was assessed at 18 h post i.p. injection. Peritoneal exudate cells were harvested by lavage with 10 mL ice-cold PBS and characterized by Diff-Quik staining (Medion Diagnostics AG, Düdingen, Switzerland) of cytospins as well as by flow cytometry analysis after staining with 7/4-FITC, CCR3-PE, CD11b-PerCP, and Gr-1-APC (all from eBioscience, San Diego, CA, USA) using a BD FACSCalibur flow cytometer (BD Biosciences, San Jose, CA, USA) and the FlowJo software (Tree Star, Ashland, OR, USA).

Serum lipids and lipoprotein profiles

Plasma total cholesterol and triglycerides were measured on a Roche-Hitachi Modular Clinical Chemistry analyzer using commercial reagents (Roche Diagnostics, Rotkreuz, Switzerland). Lipoprotein profiles of Nrf2+/−Apoe−/− and Nrf−/−Apoe−/− mice were assessed by FPLC. Serum of five Nrf2+/−Apoe−/− and of five Nrf−/−Apoe−/− mice was pooled, and fractionated over a HiPrep 16/60 Sephacryl S-200 HR gel filtration column on a ÄKTA FPLC (both GE Healthcare, Glattbrugg, Switzerland). The concentrations of cholesterol and protein in individual serum fractions were then measured with the BCA Protein Assay and the Infinity Cholesterol Reagent (both from Perbio Science Switzerland, Lausanne, Switzerland) according to the manufacturer's instructions. The level of lipoperoxides was determined as thiobarbituric acid reactive substances (TBARS) in serum of cholesterol diet-fed mice using the OXItek TBARS kit (Zeptometrix, Buffalo, NY, USA).


Organs were removed from in situ paraformaldehyde-fixed animals and frozen in O.C.T. medium. Cryosections were cut at 7 μm thickness from frozen tissue using a Microm HM520 cryostat (Histocom AG, Zug, Switzerland). Sections were air dried, rehydrated in PBS, and fixed for 5 min in 4% paraformaldehyde in PBS. After a quick rinse in 60% isopropanol, sections were stained with Oil Redo for 30 min at RT, followed by two short rinses with 60% isopropanol to remove unbound ORO. Sections were then washed with ddH2O and counterstained with Mayer's hematoxylin according to the manufacturer's instructions. Sections were then embedded with fluorescence mounting medium (Dako Schweiz AG, Baar, Switzerland) and analyzed under a Zeiss Axioplan 2 imaging microscope (Carl Zeiss AG, Feldbach, Switzerland) using the Improvision OpenLab 4.0.1 software (PerkinElmer, Schwerzenbach, Switzerland).

In vivo neutralization of IL-1α and IL-1β

High-titered neutralizing antibodies against IL-1α and IL-1β were induced by vaccination with VLPs displaying the murine cytokines on their surface 39.

Cells and reagents

All reagents were from Sigma (Sigma-Aldrich Chemie GmbH, Buchs, Switzerland) except otherwise stated. Thioglycollate-elicited macrophages were harvested by peritoneal lavage four days after i.p. injection of 3.8% thioglycollate (Becton Dickinson AG, Basel, Switzerland). BM of various mouse strains was cultured in vitro in presence of L-cell supernatant for 8 days to generate BM-derived macrophages. Cells were washed with PBS, counted and plated in flat bottom tissue culture plates overnight. On the following day, macrophages were primed with 30 ng/mL LPS for 5 h, washed twice with medium, before the indicated stimuli were added in complete medium to induce cytokine production. BM-derived DCs were differentiated in the presence of recombinant 2 ng/mL granulocyte-macrophage colony stimulating factor (R&D Systems Europe, Abingdon, UK) in vitro for 10 days.

To assess foam cell formation, macrophages or DCs were cultured on acid-treated and ethanol-sterilized coverslips (Huber AG, Reinach, Switzerland) overnight. Adherent cells were washed with medium and then cultured in presence of indicated stimuli at 37°C for 24, 48, and 72 h. After fixation in 4% paraformaldehyde/PBS, the coverslips were washed twice with PBS, rinsed in 60% isopropanol, and stained with 0.3% Oil Red O in 60% isopropanol for 30 min to visualize lipid droplets. After two washes in PBS, coverslips were counterstained with Mayer's hematoxylin according to the manufacturer's protocol and embedded in mounting medium (Dako Schweiz AG).


Cytokine concentrations in cell culture supernatants were determined by sandwich ELISA using the following antibody pairs: IL-1α (14-7011-85 and 13-7111-85), IL-1β (14-7012-85 and 13-7112-85), IL-6 (14-8061-62 and 13-7062-85), and IL-12 (14-7125-85 and 13-7123-85) (all from eBioscience).

Statistical analysis

Statistical analysis was performed with the two-tailed t-test using the Prism software (GraphPad Software, La Jolla, CA, USA). Differences were considered significant for p<0.05.


We thank J. Tschopp for NLRP3−/− mice, W. Hardt for Casp1−/− mice, and members of the animal facility for technical support. This research was supported by grants from the Swiss National Science Foundation (310030-124922/1, to M. K.) and the Swiss Federal Institute of Technology Zurich (ETH-18 09-1, to M. K. and S. F.).

Conflict of interest: The authors declare no financial or commercial conflict of interest.