Inflammasome Activation Aggravates Cutaneous Xanthomatosis and Atherosclerosis in ACAT1 (Acyl-CoA Cholesterol Acyltransferase 1) Deficiency in Bone Marrow

Introduction

ACAT (Acyl-CoA cholesterol acyltransferase), also known as SOAT (sterol O-acyltransferase), esterifies cellular free cholesterol (FC) to form cholesteryl esters (CEs) in the endoplasmic reticulum (ER).¹ In mammals, 2 isoforms of ACAT have been identified: ACAT1 and ACAT2. In mice, ACAT1 is expressed in most tissues, including macrophages, as the major isoenzyme, and ACAT2 is expressed exclusively in intestinal epithelial cells and hepatocytes.

In the early stage of atherogenesis, monocytes infiltrate into the vascular wall, take up modified lipoproteins, and transform into CE-laden macrophages, which are called foam cells.² Therefore, targeting ACAT1 to prevent foam cell formation has been considered a promising strategy to reduce atherosclerotic lesions. However, the role of ACAT1 in atherosclerosis remains controversial. Many pharmacological studies in animal models have shown that ACAT inhibitors, both ACAT1-selective inhibitors and nonselective inhibitors, were beneficial in reducing atherosclerosis,^3–8 but opposite results have also been reported.⁹ Similar conflicting results have been reported in studies using genetically altered mice. We have previously reported that elimination of ACAT1 decreased plaque size and cholesterol ester contents in the aorta when introduced into Ldlr (LDL [low-density lipoprotein] receptor)-null or Apoe-null mice fed high-fat diets.¹⁰ Consistently, myeloid cell-specific elimination of ACAT1 decreased atherosclerotic lesion size in Apoe-null mice fed a high-fat diet.¹¹ On the contrary, Ldlr-null mice transplanted with bone marrow cells from Acat1-null mice had somewhat increased atherosclerosis.¹² In humans, attempts to reduce atherosclerotic plaque volume with ACAT inhibitors have not been successful: avasimibe or pactimibe failed to reduce the percent atheroma volume in patients with coronary disease.^13,14 Pactimibe had no effect on maximum carotid intima-media thickness in patients with familial hypercholesterolemia but was associated with an increase in mean carotid intima-media thickness, as well as increased incidence of major cardiovascular events.¹⁵

Despite these conflicting phenotypes in atherosclerosis, ACAT1-deficient mice with hyperlipidemia were consistently reported to exhibit severe skin lesions, such as hair loss and skin hypertrophy, in a systemic model,¹⁰ a bone marrow transplantation model,¹⁶ or a myeloid-specific deletion model.¹¹ These lesions are histologically characterized by massive accumulation of macrophages with necrotic cells and clefts of cholesterol crystals (CCs), resembling cutaneous xanthomas. Unlike conventional cutaneous xanthomas, which primarily contain CE, the skins of ACAT1-null mice with hypercholesterolemia contain substantial amounts of FC instead of CE.¹⁶ CCs have been added to a battery of pathogen- or damage-associated molecular patterns (DAMPs), which trigger the activation of NLRP3 (NACHT, LRR [leucine-rich repeats], and PYD [pyrin domain] domain-containing protein 3) inflammasomes. This activates caspase-1, which in turn promotes the secretion of inflammatory cytokines, such as IL (interleukin)-1β.^17,18 In mice, atherosclerotic lesions often contain CCs,¹⁶ and inhibition of NLRP3 inflammasomes or caspase-1 has been demonstrated to attenuate the progression of atherosclerosis in mice.^17,19,20

Loss of ACAT1 leads to excessive accumulation of nonesterified FC in macrophages when incubated with exogenous sources of cholesterol, such as FC or modified LDL. These FC-enriched macrophages are prone to activate ER stress²¹ and increase the secretion of inflammatory cytokines, such as TNF-α (tumor necrosis factor-α) and IL-6.²² However, it is not fully understood how excess FC in ACAT1-deficient mice leads to the aggravation of atherosclerosis and cutaneous xanthoma.

Based on these considerations, we hypothesized that NLRP3 inflammasome activation links the loss of ACAT1 to the aggravation of atherosclerosis and cutaneous xanthomatosis. To test whether NLRP3 is involved in the aggravation of atherosclerosis/xanthomatosis when ACAT1 is deleted in macrophages, we transplanted bone marrow from WT (wild type) mice and mice that lacked ACAT1, NLRP3, or both to Ldlr-null mice and compared the degree of atherosclerosis and skin phenotypes after feeding them high-cholesterol diets.

Materials and Methods

The authors declare that all supporting data are available within the article and its online-only Data Supplement.

Mice

Animal care and experimental procedures were performed according to the regulations of the Animal Care Committees of Jichi Medical University. Ldlr^−/−23 and Acat1^−/− mice¹⁰ were generated as described previously. Nlrp3^−/− mice were kindly provided by Dr Vishva M. Dixit (Genentech, South San Francisco, CA).²⁴ Mice lacking both ACAT1 and NLRP3 (Acat1^−/−;Nlrp3^−/−) were generated by mating these mice. All mice used in this study had a C57BL/6J genetic background.

All mice (C57BL/6J [WT], Ldlr^−/−, Acat1^−/−, Nlrp3^−/−, and Acat1^−/−;Nlrp3^−/− mice) were maintained in a temperature-controlled (25°C) facility with a 12-hour light/dark cycle and given free access to food and water. Unless otherwise stated, mice were fed a normal chow diet (CE-2), containing 4.4% (w/w) fat and 25.3% (w/w) protein (CLEA Japan).

Bone Marrow Transplantation

Female Ldlr^−/− mice (8 weeks of age) were lethally radiated (9 Gy) by a cesium gamma source and transplanted with 5×10⁶ bone marrow cells, isolated from male WT, Nlrp3^−/−, Acat1^−/−, or Acat1^−/−;Nlrp3^−/− mice (8 weeks of age). The transplanted mice were fed a chow diet for the first 4 weeks and then fed Western-type diet containing 0.21% (w/w) cholesterol and 20% (w/w) milk fat (Research Diets) or atherogenic diet containing 1.25% (w/w) cholesterol, 7.5% (w/w) cocoa butter, and 0.5% (w/w) cholic acid (Oriental Yeast Company). The recipient mice had free access to water that was acidified to pH 2.6 for infection prevention. In addition, 1 week before and 2 weeks after radiation, 100 mg/L neomycin (Cat No. N-6386; Sigma) and 10 mg/L polymyxin B sulfate (Cat No. P-4932; Sigma) were added to the water. After the indicated period on the high-cholesterol diets (8 weeks for Western-type diet and 12 weeks for the atherogenic diet), the transplanted mice were euthanized for the assessment of the atherosclerotic lesions.

Plasma Lipid Analyses

Plasma lipids were evaluated before and after the high-cholesterol diet feeding. After a 16-hour fast, blood was collected into tubes containing EDTA for separation of plasma. Plasma levels of total cholesterol and triglycerides were determined enzymatically using kits (Determiner TC II, Kyowa Medex and L-Type TG M; Wako). High-performance liquid chromatography analyses of plasma were performed as described previously.²⁵

Computed Tomographic Scans

Computed tomographic scans were performed using LaTheta LCT-200 (Hitachi Aloka Medical, Tokyo, Japan) and LaTheta software (Hitachi Aloka Medical). Mice were anesthetized with isoflurane (Cat No. 099-06571; Wako) and fixed to base before imaging to avoid motion artifacts. Axial images of both feet were obtained at a pitch of 504 µm each (96-μm slice thickness), and cross-sectional areas excluding bones were measured, and the degree of hypertrophy was expressed as average area sizes of 10 slices around ankle joints.

Histology

Whole foot and back skins of the mice were trimmed after euthanasia and fixed with 10% neutral buffered formalin, decalcified with formic acid/citric acid solution, embedded in paraffin. Sections were stained with hematoxylin and eosin.

For immunostaining, the frozen sections of the aortic roots were blocked with 1.5% goat serum (Cat No. S-1000; Vector Labs), and endogenous avidin and biotin activity was blocked using Avidin/Biotin Blocking Kit (Cat No. SP-2001; Vector Labs), and endogenous peroxidase activity was quenched by treating with 0.3% hydrogen peroxide. The sections were then incubated with the primary antibody of Moma-2 (0.8 µg/mL; Cat No. MCA519G) or αSMA (α-smooth muscle actin; 1 µg/mL; Cat No. ab5694; Abcam) overnight at 4°C in buffer. This was followed by incubation with biotinylated secondary anti-rat antibody (7.5 µg/mL; Cat No. BA-9400; Vector Labs) for Moma-2 staining or biotinylated secondary anti-rabbit antibody (7.5 µg/mL; Cat No. BA-1000; Vector Labs) for αSMA staining for 2 hours at 37°C and then with avidin-biotin peroxidase complex (Cat No. PK-6100; Vector Labs) for 30 minutes. Lastly, the sections were developed with 3,3′-diaminobenzidine tetrahydrochloride (Cat No. D-4293; Sigma) and counterstained with hematoxylin.

Frozen sections of the aortic roots were also stained with Movat pentachrome as described previously.^26,27

Apoptotic cells were detected by the terminal transferase-mediated dUTP-biotin nick end labeling (TUNEL) method using in Situ Apoptosis Detection Kit (Cat No. MK-500; Takara), and the nuclei were stained with methyl green according to the protocol provided by the manufacturer.

Digital images were captured using a light microscope AX-80 (Olympus).

Quantitation of Atherosclerotic Lesions

After feeding with a high-cholesterol diet, the mice were euthanized and their hearts were isolated, and the atherosclerotic lesions were quantified by the analysis of serial cross sections through the aortic root. The cross-sectional lesion area was evaluated according to the method described previously^28,29 with slight modifications. In brief, the heart was perfused with saline containing 4% (w/v) formalin and fixed for >48 hours in the same solution. The basal half of the heart was embedded in cryo embedding media OCT (Cat No. 4583; Sakura Finetek), and serial sections were embedded in Cryostat (6-µm thick). Four sections, each separated by 60 µm, were used to evaluate the lesions; 2 at the end of the aortic sinus and 2 at the junctional site of sinus and ascending. Sections were stained with Oil Red O and counterstained with hematoxylin, and the stained area was quantified as the atherosclerotic lesion by using Adobe Photoshop 6 software (Adobe Systems).

Preparation of Lipoproteins and CCs

For preparation of lipoproteins, blood was collected from normolipidemic volunteers after an overnight fast. LDLs (d=1.019–1.063 g/mL) were isolated from the plasma by sequential density ultracentrifugation.³⁰ LDL was acetylated by repetitive additions of acetic anhydride,³¹ and the protein concentrations of acetylated LDL (acLDL) were measured by using Pierce BCA protein assay kit (Cat No. 23225; Thermo Scientific).

For preparation of CC, cholesterol (Cat No. C-8667; Sigma) was solubilized in ethanol (10 mg/mL) with presence of 20% endotoxin-free water to obtain hydrated CC. After evaporation of the organic solvent by using centrifugal evaporator at room temperature, resulting CCs were collected by addition of medium followed by suspension with the use of a sonicator.

Cell Culture Experiments

Mouse peritoneal macrophages (MPMs), derived from WT, Acat1^−/−, Nlrp3^−/−, or Acat1^−/−;Nlrp3^−/− mice (female; 8–12 weeks of age), were obtained 3 days after an intraperitoneal injection of 2 mL of 5% (w/v) thioglycollate broth (Cat No. 399-00143; Wako). Macrophages were plated on 12- or 24-well plates and cultured in DMEM supplemented with 10% (v/v) fetal bovine serum and antibiotics. After being cultured for 3 hours, nonadherent cells were removed, and the adherent macrophages were maintained in the medium. For the measurement of IL-1β, MPMs were primed with 100 ng/mL lipopolysaccharide (Cat No. L-2630; Sigma) for 10 hours before the addition of stimulus. CC or 5 mmol/L ATP (Cat No. A-2383; Sigma) were applied 6 or 1 hour, respectively, before the supernatant was collected. MPMs were treated/untreated with acLDL for the indicated periods following by the medium was switched to serum-free RPMI1640 medium for 24 hours (for the purpose of starvation).

Determination of Cholesterol Content

For extracting lipids, skin tissues were snap-frozen in liquid nitrogen, pulverized, and homogenized with a mixture of water/chloroform/methanol according to the method of Bligh and Dyer, with minor modifications.³² Cellular lipids of MPMs treated/untreated with acLDL were extracted by hexane/isopropyl alcohol as described previously.³³ Cholesterol content of lipids extracted from skin tissues and cells was determined enzymatically using kits (Determiner TC II and Determiner L FC; Kyowa Medex).

Quantitative Real-Time Polymerase Chain Reaction

Total RNA was prepared using a guanidium thiocyanate-phenol-chloroform extraction procedure.³⁴ cDNA was synthesized using the high-capacity cDNA reverse transcriptase kit (Cat No. 4368813; Applied Biosystems). All reactions were done in triplicates, and relative amounts of mRNA were calculated using a standard curve or comparative computed tomographic method on the StepOnePlus RealTime PCR instrument (Applied Biosystems), according to the manufacturer’s protocol. Mouse β-actin mRNA was used as the invariant control. The primer sets for real-time polymerase chain reaction are listed in Table I in the online-only Data Supplement.

ELISA

Levels of IL-1β were assessed by using a mouse ELISA kit (Cat No. 559603; Becton Dickinson) according to the manufacturer’s instructions.

Analysis for cell culture experiments was performed in serum-free supernatants of culture medium, and analysis for skin tissues was performed in lysis buffer used for protein extraction. For extracting proteins, skin tissues were snap-frozen in liquid nitrogen, pulverized, and resuspended in a protein lysis buffer (25 mM Tris [pH 2.8], 2 mM EDTA, 1 mM dithiothreitol, 10% glycerol, 1% Triton X-100, 1 mM PMSF [phenylmethylsulfonyl fluoride], and 20 kIU/mL aprotinin). Protein concentrations of skin extract were measured by using Pierce BCA protein assay kit.

Statistical Analyses

Results are presented as the mean±SD. Mann-Whitney U test was used to compare the mean values between 2 groups, and Kruskal-Wallis test and the post Steel-Dwass test were used for multiple comparisons. All analyses were performed using GraphPad Prism, version 6.0 (GraphPad Prism Software). A P value of <0.05 was considered statistically significant.

Results

Effects of Bone Marrow Transplantation on Body Weight and Plasma Lipids

Bone marrow isolated from male WT, Nlrp3^−/−, Acat1^−/−, or Acat1^−/−Nlrp3^−/− mice was transplanted into female Ldlr^−/− mice. Male to female donor-recipient transplant pairs are commonly used in bone marrow transplantation studies because the successful donor engraftment and repopulation can be easily demonstrated by detecting the presence of genes on the Y chromosome in recipients’ bone marrow. Although it is known that female Ldlr^−/− mice have larger aortic root lesion areas than male Ldlr^−/− mice even in C57BL/6J background especially fed diets with low cholesterol contents,^35,36 we did not use male Ldlr^−/− recipients in this study.

Four weeks after transplantation, the recipient mice were fed a Western-type diet for 8 additional weeks or an atherogenic diet for 12 additional weeks (Figure IA in the online-only Data Supplement).

Successful reconstitution of recipients with cells of donor origin after bone marrow transplantation was verified by polymerase chain reaction–assisted amplification of the Sry gene (a male marker) and either the Acat1 or Nlrp3 mutant gene (Figure IB in the online-only Data Supplement).

The bone marrow-specific inactivation of the Acat1 or Nlrp3 gene did not significantly affect the body weight, plasma lipids (Tables 1 and 2), or the lipoprotein fractions (Figure II in the online-only Data Supplement) in Ldlr^−/− mice after feeding with a high-cholesterol diet.

Loss of NLRP3 Partially Reverses the Worsening of Atherosclerosis in Mice Lacking ACAT1 in Bone Marrow Cells

After feeding the mice with a Western-type diet for 8 weeks, atherosclerotic lesions were evaluated at the aortic roots. Oil Red O staining of the lesions of the recipients of Acat1^−/− and Acat1^−/−;Nlrp3^−/− bone marrow were less intense than the staining in the WT and Nlrp3^−/− bone marrow, indicating that their lesions contained smaller amounts of CEs (Figure 1A). Figure 1B compares the lesion sizes estimated by Oil Red O staining between the 4 types of mice. As expected, the recipients of Nlrp3^−/− bone marrow had 38% smaller (P<0.05) lesion areas than the recipients of WT bone marrow. Conversely, lesion areas were 47% larger (P<0.05) in the recipients of Acat1^−/− bone marrow than in the recipients of WT bone marrow. Lesion areas in the recipients of Acat1^−/−;Nlrp3^−/− bone marrow were 25% smaller (P<0.05) than those in the recipients of Acat1^−/− bone marrow, indicating that NLRP3 inflammasomes are responsible at least in part for the increase of lesion sizes in the recipients of Acat1^−/− bone marrow. However, lesion areas in the recipients of Acat1^−/−;Nlrp3^−/− bone marrow were not as small as those of Nlrp3^−/− bone marrow; therefore, the worsening of atherosclerosis in the recipients of Acat1^−/− bone marrow cannot be attributable only to NLRP3 inflammasomes.

• Download figure
• Download PowerPoint

The degree of infiltrated macrophages was assessed by Moma-2 staining (Figure 2A and 2B). Unexpectedly, the lesion areas positively stained for Moma-2 in the mice transplanted with Acat1^−/− bone marrow were 44% smaller (P<0.05) than those of the mice transplanted with WT bone marrow. Similarly the Moma-2–positive lesion areas of the mice transplanted with Acat1^−/−;Nlrp3^−/− bone marrow were 40% smaller (not significant) than those of the mice transplanted with Nlrp3^−/− bone marrow. Loss of NLRP3 did not significantly affect the Moma-2–positive areas.

• Download figure
• Download PowerPoint

To evaluate the proportion of vascular smooth muscle cells (VSMCs) in the atherosclerotic lesions, immunostaining for αSMA was performed (Figure 2A and 2B). In the recipients of WT or Nlrp3^−/− bone marrow, αSMA-positive areas were restricted to the superficial subintimal regions. In the recipients of Acat1^−/− or Acat1^−/−;Nlrp3^−/− bone marrow, however, αSMA-positive areas were larger distributing from superficial to deep subintimal regions. αSMA-positive areas in the mice transplanted with Acat1^−/− bone marrow were 1.8-fold larger (P<0.05) than those in the mice transplanted with WT bone marrow. The recipients of Acat1^−/−;Nlrp3^−/− bone marrow also had 2.6-fold larger (P<0.05) αSMA-positive areas than the recipients of Nlrp3^−/− bone marrow. Loss of NLRP3 did not affect the αSMA-positive areas.

Other features, such as the amounts of ECM (extracellular matrix), were analyzed with Movat pentachrome staining (Figure 2A and 2B). The lesions of the recipients of Acat1^−/− marrow contained 1.9-fold larger acellular areas than the recipients of WT bone marrow. The acellular areas were largely filled with mucins without evidence of nuclei. Similarly, the lesions of the recipients of Acat1^−/−;Nlrp3^−/− bone marrow contained 2.1-fold larger acellular areas than the recipients of Nlrp3^−/− bone marrow. Furthermore, the recipients of WT or Nlrp3^−/− bone marrow had less TUNEL-positive cells in the atherosclerotic lesions, whereas the recipients of Acat1^−/− or Acat1^−/−;Nlrp3^−/− bone marrow contained significantly larger numbers of TUNEL-positive cells in the lesions (Figure 2A and 2B). Loss of NLRP3 did not significantly affect either the acellular areas or the number of TUNEL-positive cells. Taken together, these results suggest that loss of ACAT1 accelerates cell death mostly via apoptosis of bone marrow-derived cells regardless of the presence or absence of NLRP3 inflammasome, thereby promoting the accumulation of extracellular mucins.

Loss of NLRP3 Almost Completely Reverses the Worsening of Cutaneous Xanthomatosis in Mice Lacking ACAT1 in Bone Marrow Cells

The mice transplanted with Acat1^−/− bone marrow fed Western-type diet for 8 weeks did not develop the macroscopic skin lesions that were observed in the previous studies.^10,11,16 We speculated that the amount and the feeding period of Western-type diet were not sufficient to induce the skin phenotype. To develop skin lesions, we fed the different sets of mice with an atherogenic diet containing more cholesterol for 12 weeks. Increasing the exposure to cholesterol by feeding an atherogenic diet for 12 weeks did not significantly affect body weight or plasma lipids in any of the mice (Table 2).

The recipients of WT and Nlrp3^−/− bone marrow did not develop obvious skin lesions (except for hair loss possibly caused by the radiation damage). In contrast, the recipients of Acat1^−/− bone marrow showed remarkably thick skin, especially around ankle joints (Figure 3A). Notably, the recipients of Acat1^−/−;Nlrp3^−/− bone marrow did not show similar skin lesions, at least during the observation period. Computed tomographic scans showed that the thicknesses of the skins around the ankle joints of the recipients of WT, Nlrp3^−/−, and Acat1^−/−;Nlrp3^−/− bone marrow was almost the same, whereas the thickness in the recipients of Acat1^−/− bone marrow was 2.5-fold thicker (P<0.05) than those in the other types of mice (Figure 3B and 3C).

• Download figure
• Download PowerPoint

The skins of the recipients of Acat1^−/− bone marrow had extensive infiltration of inflammatory cells throughout the entire dermal layer and consequently were thickened (Figure 4A). The thickened dermis of the recipients of Acat1^−/− bone marrow was filled with a number of inflammatory cells containing some cells with foamy appearance or partially multinucleated giant cells and many cholesterol clefts (Figure 4B). Very few subcutaneous adipose tissues were observed. These features are characteristic of cutaneous xanthoma. In contrast, the recipients of WT, Nlrp3^−/−, or Acat1^−/−;Nlrp3^−/− bone marrow exhibited almost normal skin structures. Oil Red O stained the hair follicles and residual adipose tissues but only a few of the inflammatory cells (Figure III in the online-only Data Supplement).

• Download figure
• Download PowerPoint

In lipids extracted from the skins, FC content was 4-fold higher (P<0.05) in the skin from the recipients of Acat1^−/− bone marrow, whereas CE content was not significantly different in the 4 types of mice (Figure 5A). The expressions of a macrophage marker (CD [cluster of differentiation] 68), inflammatory genes (MCP-1 [monocyte chemoattractant protein-1], TNFα, IL-6, and IL-1β), and an ER stress marker (CHOP [C/EBP (CCAAT/enhancer binding protein) homologous protein]) were remarkably increased (3–73-fold increases) in the skin of the recipients of Acat1^−/− bone marrow compared with their expressions in the other types of mice (Figure 5B). These findings suggest that elimination of ACAT1 in bone marrow cells induced skin lesions mainly via inflammatory changes of the infiltrated macrophages. Furthermore, the expressions of neutrophil marker (Ly6G [lymphocyte antigen 6 complex locus G6D]) and CD8 were observed only in the skin of the recipients of Acat1^−/− bone marrow (data not shown), and there were also higher expression levels of T-cell marker (CD3e) and CD4 in the skin of the recipients of Acat1^−/− bone marrow (Figure 5B). This suggests that inflammatory macrophages induce the infiltration of other inflammatory cells, such as neutrophils and T lymphocytes in the skin of those mice. IL-1β protein levels in skin tissues of the recipients of Acat1^−/− bone marrow were significantly higher than those of the other types of mice (Figure 5C).

• Download figure
• Download PowerPoint

Treatment of ACAT1-Deficient MPMs With CCs Did Not Potentiate Inflammasome Activation

CCs can function as a danger signal to activate NLRP3 inflammasomes in macrophages.^17,18 We suspected that Acat1^−/− macrophages are more sensitive to CCs in terms of activation of NLRP3 inflammasomes than WT macrophages because of the presence of numerous cholesterol clefts and the selective increase in FC content in the skin of the Acat1^−/− bone marrow recipients. We prepared MPMs from WT, Nlrp3^−/−, Acat1^−/−, or Acat1^−/−;Nlrp3^−/− mice and treated them with CCs after priming with lipopolysaccharide and measured IL-1β levels in the supernatants of culture media. Addition of CCs robustly increased IL-1β in media of WT and Acat1^−/− MPMs in a dose-dependent manner (Figure 6A). The degree of IL-1β elevation was not significantly different in the 2 types of MPMs. On the contrary, secretion of IL-1β was totally abolished in Nlrp3^−/− and Acat1^−/−;Nlrp3^−/− MPMs, indicating that loss of ACAT1 does not affect the sensitivity of macrophages to CCs in terms of NLRP3 inflammasome activation.

• Download figure
• Download PowerPoint

Cholesterol Loading of ACAT1-Deficient MPMs by Treatment With acLDL Activated ER Stress, but It Did Not Activate Inflammasomes

Next, to investigate whether FC accumulation induced by the loss of ACAT1 activates NLRP3 inflammasomes, we treated MPMs from the 4 types of mice with acLDL after priming with lipopolysaccharide. Oil Red O staining confirmed that CE deposition, which was observed after treatment of MPMs from WT or Nlrp3^−/− mice with acLDL, was hardly detectable in MPMs from Acat1^−/− or Acat1^−/−;Nlrp3^−/− mice (Figure 6B). Consistently, loss of ACAT1 almost completely reversed the accumulation of CE induced by treatment with acLDL in WT MPMs (Figure 6C). More importantly, loss of ACAT1 increased the content of FC in MPMs by 2-fold (P<0.05) regardless of the presence of NLRP3. Ten hours after priming with lipopolysaccharide, IL-1β was detectable in the media (Figure 6D). Treatment with acLDL did not affect the secretion of IL-1β in either WT or Acat1^−/− MPMs. The secretion of IL-1β was almost totally abrogated in both Nlrp3^−/− and Acat1^−/−;Nlrp3^−/− MPMs, indicating that NLRP3 is required for the secretion of IL-1β. Together, these results show that FC accumulation per se was not sufficient to activate NLRP3 inflammasomes.

Finally, we measured the expression of Chop and Tnfa in MPMs after incubation with or without 100 µg/mL of acLDL (Figure 6E). Incubation with acLDL increased the expression of Chop by 2-fold (P<0.05) in Acat1^−/− and Acat1^−/−;Nlrp3^−/− MPMs but not in WT or Nlrp3^−/− MPMs. The increased expression of Chop was sustained for 24 hours. Similarly, incubation with acLDL transiently increased the expression of Tnfa by 2-fold (P<0.05) in Acat1^−/− and Acat1^−/−;Nlrp3^−/− MPMs but not in WT or Nlrp3^−/− MPMs. These results indicate that excessive FC accumulation in Acat1^−/− MPMs induced ER stress and the expression of inflammatory genes.

Discussion

The present results show that NLRP3 is involved in the aggravation of cutaneous xanthomatosis and atherosclerosis to a lesser degree in hyperlipidemic mice whose ACAT1 was selectively deleted in bone marrow. ACAT1-deficient MPMs did not activate NLRP3 inflammasomes but induced the expression of Chop when loaded with cholesterol by incubation with acLDL in vitro. These results suggest that ACAT1 deficiency promotes cell death via sustained activation of ER stress under excessive influx of cholesterol. Death of the cells enriched with FC may promote the formation of DAMPs, including CC, in the extracellular milieu, which in turn activates NLRP3 inflammasomes of the surrounding myeloid cells. Inflammation recruits more inflammatory cells into the lesions and promotes formation of DAMPs, thereby constituting a vicious cycle.

NLRP3 deficiency in the recipients of Acat1^−/− bone marrow had a much greater effect on cutaneous xanthoma than on atherosclerosis. The histology of cutaneous xanthoma had features similar to those in human xanthoma.³⁷ It was also largely similar to that observed in older Ldlr^−/− mice (13 months of age) fed the same diet for a longer time (8 months)³⁸ except for the near absence of Oil Red O staining (Figure III in the online-only Data Supplement). Similar premature severe xanthoma has been reported in Ldlr^−/− mice lacking ACAT1 generally, as well as in Ldlr^−/− mice whose Acat1 was deleted in a myeloid cell-specific manner.¹¹ Together, these results indicate that the skin phenotype was largely generated by ACAT1 deficiency in bone marrow cells, more specifically in myeloid cells. The involvement of cutaneous cells, such as acinar cells of sebaceous glands, should be negligible because they express high level of ACAT1.¹⁰ Preferential retention of apoB (apolipoprotein B)-containing lipoproteins in the ECM of the dermis may be the initial event, as has been suggested for the development of atherosclerosis.³⁹ Although the recipients of Acat1^−/− bone marrow did not develop obvious skin lesions after feeding with a Western-type diet for 8 weeks, they developed more advanced atherosclerotic lesions than the recipients of WT bone marrow after feeding with the same diet for the same period. We would ascribe less efficient leakage of lipoproteins into the ECM of the dermis to the milder skin phenotypes in ACAT1 KO (knockout) mice.

The gene expression results indicate that the lesions were filled with inflammatory cells comprised mostly of macrophages (CD68) and to a lesser degree neutrophils and T lymphocytes (CD3e, CD4, and CD8; Figure 5B). In agreement with the extensive infiltration of inflammatory cells, the expression of inflammatory cytokines, such as IL-1β, MCP-1, IL-6, and TNF-α, was highly elevated. In parallel, the tissue content of IL-1β in the skin from the recipients of Acat1^−/− bone marrow was increased, although the increase was much less significant than the increase of mRNA (Figure 5C), probably owing to incomplete recovery of IL-1β from the tissues, as well as homeostatic cleavage of pro–IL-1β by inflammasome-independent pathways.⁴⁰ These findings suggest that Acat1^−/− bone marrow cells that infiltrated into the skin elicit inflammation under hypercholesterolemia, thereby forming a vicious cycle: inflammatory cells promote vascular permeability and lipoprotein retention in subcutaneous tissues, which facilitates the recruitment of more inflammatory cells.

Since the recipients of Acat1^−/−;Nlrp3^−/− bone marrow had almost normal skin (Figures 3 and 4), NLRP3 is required for the development of the severe cutaneous xanthoma in the recipients of Acat1^−/− bone marrow. Consistently, loss of NLRP3 completely normalized the highly induced expression of inflammatory genes (Figure 5B and 5C). These results indicate that NLRP3 inflammasome activation in bone marrow-derived cells is critical for the development of severe cutaneous xanthoma in ACAT1-deficient hyperlipidemic mice.

How does ACAT1 deficiency cause inflammasome activation in the skin of hypercholesterolemic mice? First, we tested the hypothesis that ACAT1-deficient macrophages are more sensitive to the signals that trigger NLRP3 inflammasome activation (CCs and ATP¹⁷), which can be generated by necrotic cell death. The increases of IL-1β secretion in response to these DAMPs were not different between WT and Acat1^−/− MPMs. Next, we examined whether FC accumulation activates NLRP3 inflammasomes. Although Acat1^−/− MPMs treated with acLDL accumulated intracellular FC, however, no difference was found in the IL-1β secretion between WT and Acat1^−/− MPMs. Instead, Acat1^−/− MPMs expressed higher mRNA levels of Chop and Tnfa than WT MPMs, corroborating the results of MPMs treated with ACAT inhibitors.^21,22 Because both CHOP activation⁴¹ and TNF-α are cytotoxic,⁴² the elevation of these expressions is thought to induce cell death in Acat1^−/− MPMs with FC accumulation. Exposure to excess cholesterol could occur via both an SRA (scavenger receptor A)-dependent pathway, which selectively recognizes acLDL, and an SRA-independent pathway, including CD36 and LDLR pathways. Protracted exposure by these pathways would likely lead to more FC accumulation, ER stress, TNF-α hypersecretion, and eventual cell death. Extracellular buildup of necrotic debris containing DAMPs, such as CCs, and ATP might induce the activation of NLRP3 inflammasomes in surrounding macrophages. CD36 may be engaged in the latter process as well.⁴³

Compared with the results of cutaneous xanthoma, the effects of abrogation of NLRP3 on atherosclerotic lesions were more modest. The present study showed a worsening of atheroma in the recipients of Acat1^−/− bone marrow, which is consistent with the results of other investigations using the same method.¹² The results show that Moma-2–positive macrophages comprised a large portion of the lesions of the mice transplanted with WT bone marrow, whereas a major part of the lesions in the recipients of Acat1^−/− bone marrow was filled with αSMA-positive areas and acellular areas containing mucins. Mucins, which are complex mixtures of glycosaminoglycans and proteoglycans in ECM, are the secretory products of VSMCs.⁴⁴ These results suggest that VSMCs have a significant role in the atherogenesis that occurs in ACAT1 deficiency. VSMCs are thought to predominate over macrophages, because ACAT1-deficient macrophages were prone to die because of excessive FC accumulation. Indeed, there were significantly larger numbers of TUNEL-positive cells in the lesions of recipients transplanted with Acat1^−/− or Acat1^−/−;Nlrp3^−/− bone marrow (Figure 3). Macrophage-like cells, which can be transformed from VSMCs in atherosclerotic lesions,⁴⁵ are not as efficient as activated macrophages at phagocytosis of dying macrophages.⁴⁶ Therefore, it is conceivable that the atherosclerotic lesions of the recipients of Acat1^−/− bone marrow grew more easily than the recipients of WT bone marrow by accumulating necrotic cells.

However, as mentioned above, atherosclerosis was attenuated in both Acat1^−/−;Ldlr^−/− mice¹⁰ and myeloid cell-specific Acat1-null mice.¹¹ The VSMCs in Acat1^−/−;Ldlr^−/− mice also lack ACAT1. Inhibition of ACAT in VSMCs is nontoxic and potentially helps to protect against atheroma expansion.⁴⁷ We confirmed the atheroprotective role of ACAT1 deficiency in vascular cells not derived from bone marrow in an experiment using Acat1^−/−;Ldlr^−/− mice transplanted with Ldlr^−/− bone marrow (Figure IV in the online-only Data Supplement). It is unclear why myeloid-specific Acat1-null mice were also resistant to atheroma expansion. Bone marrow transplantation transfers not only myeloid cells but also lymphoid cells. Therefore, it is possible that ACAT1-deficient lymphocytes are also critically involved in the aggravation of atherosclerosis. Indeed, it has been recently reported that ACAT1 deficiency leads to potentiation of effector function and proliferation of proatherogenic CD8⁺ T cells.^48,49 Further studies are needed to define the role of ACAT1 in the development of atherosclerosis in each cell lineage.

Consistent with our results, inflammasomes have been reported to have an atherogenic role in other models of inflammasome inactivation.²⁰ CCs have been regarded as the major culprit, but it remains possible that ATP derived from dead cells also activates inflammasomes in atherosclerotic lesions. Reduction of atheroma was also observed in the recipients of Acat1^−/−;Nlrp3^−/− bone marrow compared with the recipients of Acat1^−/− bone marrow. However, it is noteworthy that the attenuation of the size of the lesions was not as complete as was observed in the cutaneous xanthoma. Moreover, the proportions of macrophages and acellular areas were remarkably similar between Acat1^−/−;Nlrp3^−/− and Acat1^−/− bone marrow recipients. These results underscore the different pathogeneses underling the development of xanthoma and atherosclerosis. For example, the different transmigratory behaviors of inflammatory cells may explain the different responses between skin and aorta.⁵⁰ In the presence of ACAT1, transmigration (emigration) of cholesterol-loaded macrophages into the circulation may be less efficient in the aorta than the skin, thereby enhancing accumulation of foam cells. On the contrary, In the absence of ACAT1, transmigration (immigration) of circulating monocytes into the skin in response to inflammation may be more efficient in the skin than in the aorta, thereby enhancing the vicious cycle of xanthoma formation. Further studies are needed to identify the tissue determinants of transmigratory behavior of inflammatory cells. Hydrostatic pressure, density, and permeability of capillaries or lymphatics or their responsiveness to inflammatory cytokines may be potential determinants.

In conclusion, elimination of ACAT1 in bone marrow-derived cells aggravates cutaneous xanthoma and atherosclerosis. The development of cutaneous xanthoma is induced mainly via the activation of NLRP3 inflammasome pathway.

Acknowledgments

We thank Mika Hayashi, Nozomi Takatsuto, and Mihoko Sejimo for excellent technical assistance and Dr Vishva M. Dixit for providing Nlrp3^−/− mice. We also thank Biopathology Institute, Co, Ltd (Oita, Japan), for assistance with the histology techniques.

Footnotes