Protective Functions of Liver X Receptor α in Established Vulnerable Plaques: Involvement of Regulating Endoplasmic Reticulum–Mediated Macrophage Apoptosis and Efferocytosis

Abstract

Background

Liver X receptor (LXR) belongs to the metabolic nuclear receptor superfamily, which plays a critical regulatory role in vascular physiology/pathology. However, effects of systemic LXR activation on established vulnerable plaques and the potential isotype‐specific role involved remain unclear.

Methods and Results

The 8‐week‐old male apolipoprotein E^−/− mice went through carotid branch ligation and renal artery constriction, combined with a high‐fat diet. Plaques in the left carotid artery acquired vulnerable features 4 weeks later, confirmed by magnetic resonance imaging scans and histological analysis. From that time on, mice were injected intraperitoneally daily with PBS or GW3965 (10 mg/kg per day) for an additional 4 weeks. Treatment with LXR agonists reduced the lesion volume by 52.61%, compared with the vehicle group. More important, a profile of less intraplaque hemorrhage detection and necrotic core formation was found. These actions collectively attenuated the incidence of plaque rupture. Mechanistically, reduced lesional apoptosis, enhanced efferocytosis, and alleviated endoplasmic reticulum stress are involved in the process. Furthermore, genetic ablation of LXRα, but not LXRβ, blunted the protective effects of LXR on the endoplasmic reticulum stress–elicited C/EBP‐homologous protein pathway in peritoneal macrophages. In concert with the LXRα‐predominant role in vitro, activated LXR failed to stabilize vulnerable plaques and correct the acquired cellular anomalies in LXRα^−/− apolipoprotein E^−/− mice.

Conclusions

Our results revealed that LXRα mediates the capacity of LXR activation to stabilize vulnerable plaques and prevent plaque rupture via amelioration of macrophage endoplasmic reticulum stress, lesional apoptosis, and defective efferocytosis. These findings might expand the application scenarios of LXR therapeutics for atherosclerosis.

Nonstandard Abbreviations and Acronyms

7‐KC

7‐ketocholesterol

ApoE

apolipoprotein E

ER

endoplasmic reticulum

LXR

liver X receptor

UPR

unfolded protein response

Clinical Perspective

What Is New?

• Activating liver X receptor (LXR) reduces vulnerable features and prevents plaque rupture in preexisting unstable lesions.

• LXR activation alleviates apoptosis and defective efferocytosis as well as endoplasmic reticulum stress in carotid vulnerable plaques.

• LXRα inhibits the endoplasmic reticulum stress–mediated CHOP pathway in macrophages and vulnerable plaques

What Are the Clinical Implications?

• These results help to advance the understanding of LXR treatment on vulnerable plaques and are promised to facilitate the eventual application of therapies targeting LXRα.

• Taking the dumbness of LXRβ and the adverse effects of LXRα in treating vulnerable plaques into consideration, lesional drug delivery might be preferable to systemic activation of LXRα.

Atherothrombosis and the subsequent occlusion of arteries, which give rise to acute coronary syndrome or stroke, have been the leading causes of death in the world. These adverse clinical events are highly correlated with plaque vulnerability in a thrombosis‐promoting milieu. ¹ Specifically, vulnerable plaques, which have several features, including arterial outward remodeling, intraplaque hemorrhage, necrotic core formation, and inflammatory cell accumulation, ² , ³ are prone to rupture and precipitate thrombotic events.

Macrophages are the key drivers of atherothrombotic processes. Their apoptosis and efferocytosis (the clearance of apoptotic cells) play pivotal roles in determining the rupture or regression fate of plaques. ⁴ Excessive lipids that accumulate in endoplasmic reticulum (ER) can disturb cellular homeostasis and cause ER stress. ⁵ Despite the counteracting effect of cells to mobilize the unfolded protein response (UPR), prolonged ER stress activates the CHOP pathway and leads to cell death. ⁶ These apoptotic macrophages as well as red cells from intraplaque hemorrhage, which fail to be efficiently removed by phagocytes, facilitate the necrotic core formation. Moreover, the nonresolving inflammatory response prevails, contributing to the loss of mechanical stability in lesions. All these events bring about fragile conditions that predispose to plaque rupture. However, the crucial factors in regulating macrophage apoptosis and the efficiency of efferocytosis are not fully known.

Liver X receptor (LXR), a member of the nuclear receptor family, is defined as a regulatory hub of lipid metabolism and inflammation in atherosclerosis. ⁷ Two isotypes of LXRs, LXRα and LXRβ, have similar but not identical properties. Ways of activating LXR vary, compared with its natural sterol ligands; nonsterol agonists like GW3965 have a more potent effect on LXR target genes. ⁸ Our previous studies have revealed that LXR protects cardiomyocytes against myocardial ischemia/reperfusion injury and postmyocardial infarction remodeling, mainly through LXRα. ⁹ , ¹⁰ Then, questions that need to be asked are as follows: (1) What effects can LXR exert on vulnerable plaques before the fatal cardiac ischemia happens? (2) Does LXR have a protective role in ER stress–induced macrophage apoptosis and impaired apoptotic cell clearance? (3) Which isotype of LXR is responsible for the protection?

In the present study, using a mice model of rupture‐prone plaques, we seek to figure out the impacts of LXR activation on vulnerable plaques, and the role of LXRα on macrophage apoptosis and efferocytosis in response to ER stress in vitro/in vivo. These efforts may advance our knowledge of LXR from a therapeutic perspective.

Methods

The authors declare that all supporting data are available within the article and its online supplementary files.

Animal Experiment

The animal protocol was approved by the institutional ethics committee of Renji Hospital (RJ2018‐1030), and the study was performed according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

ApoE^−/− mice (20–25 g) on a C57BL/6 background (GemPharmatech, China) were randomly enrolled into 3 groups: baseline, vehicle, and GW3965. All 3 groups received surgery, as previously described. ¹¹ , ¹² , ¹³ , ¹⁴ In brief, 8‐week‐old male ApoE^−/− mice underwent the branch ligation of the left common carotid artery, with the left superior thyroid artery remaining intact, and the left renal artery was constricted using a pin gauge (0.12 mm in diameter) to preserve renal perfusion. Anesthesia was performed via an intraperitoneal injection of pentobarbital sodium (50 mg/kg), and surface anesthetic agent lidocaine (dilute to 0.5%) was adopted locally before making a surgical incision, or before final skin closure. Levels of anesthesia were assessed by pedal reflex (firm toe pinch), and respiratory rates were monitored at 15‐minute intervals. Then, mice were fed a high‐fat diet (40 kcal% fat, 1.25% cholesterol, and 0.5% cholic acid; Research Diets) for 4 weeks. Magnetic resonance imaging (MRI) scans were applied every 2 weeks. The first assay happened right after the surgery procedure. At 4 weeks after the surgery, the mice in the baseline group were euthanized. The GW3965 and vehicle groups were administered GW3965 (10 mg/kg per day) and 5% dimethyl sulfoxide in PBS, respectively, for another 4 weeks by intraperitoneal injection. LXRα‐deficient mice, generously provided by Dr Ancai Yuan (Shanghai Jiao Tong University), were intercrossed with ApoE^−/− mice to generate LXRα^−/−ApoE^−/− mice. Key resources are listed in Table S1.

MRI Scans

The 7.0‐T MRI (Bruker, Germany) was applied for in vivo lesion detection. A gas anesthesia machine with an oxygen source and a precision vaporizer was adopted. Mice were anesthetized by isoflurane inhalation (3% for induction, and 1.5% for maintenance) and placed in a prone position. Three MRI sequences (PD, T2, and T2 star) were applied. Detailed parameters are available in Data S1.

Tissue Collection

Mice were anesthetized with sustained isoflurane inhalation (3%), and blood was collected by cardiac puncture after rib removal. Freshly obtained EDTA blood was used to measure complete blood counts using the XT‐2000i hematology analyzer (Sysmex, Germany). To obtain plasma, blood was placed for 15 minutes at room temperature, then centrifuged at 1500g for 15 minutes at 4°C. Plasma levels of total cholesterol and triglycerides were measured with Hitachi 7180 autoanalyzer (Hitachi High‐Technologies Corp, Japan). Exsanguination was used as a means of euthanasia in this condition. Detailed method descriptions are available in Data S1.

Histological Staining of Murine Left Common Carotid Artery

These experiments were performed as described, ¹⁵ and details are shown in Data S1.

Immunofluorescence

These experiments were performed as described, ¹⁵ and details are shown in Data S1.

In Situ Lesional Apoptosis and Efferocytosis Assay

Detailed method descriptions are available in Data S1.

In Vivo Efferocytosis Assay

Detailed method descriptions are available in Data S1.

Cell Culture

Raw 264.7 (mouse macrophage) cells were maintained in DMEM (Gibco), and Jurkat (human T‐lymphocyte) cells were maintained in Roswell Park Memorial Institute 1640 medium (Gibco), supplemented with 10% (vol/vol) fetal bovine serum (Gibco) and 10 U/mL penicillin and 100 mg/mL streptomycin (Gibco). Cells were cultured in a humidified CO₂ incubator at 37°C. Cells were treated with 7‐ketocholesterol (7‐KC; Sigma‐Aldrich) and GW3965 (MedChemExpress, China), as indicated. 7‐KC was dissolved in ethanol, and GW3965 was dissolved in dimethyl sulfoxide.

Primary Macrophage Experiment

Primary peritoneal macrophages from LXRα knockout and LXRβ knockout mice were obtained. In brief, 1 mL of 4% Brewer thioglycollate medium (Sigma‐Aldrich) was injected into the peritoneal cavity of each mouse. The mice were euthanized with anesthetic overdose after 4 days of stimulation. Peritoneal exudate cells were obtained by rinsing the peritoneal cavity with 10 mL of cold PBS 3 times in a sterile condition, centrifuged at 200g for 10 minutes at 4°C, and plated in culture media (DMEM containing 10% fetal bovine serum). After 4 hours, cells were washed with warm PBS, and new culture medium was added.

Real‐Time Quantitative Polymerase Chain Reaction

Detailed method descriptions are available in Data S1.

Western Blotting Assay

Detailed method descriptions are available in Data S1.

Macrophage Apoptosis Experiment

Detailed method descriptions are available in Data S1.

In Vitro Efferocytosis Assay

Detailed method descriptions are available in Data S1.

Statistical Analysis

Data were analyzed with GraphPad Prism (Version 7.0). Normality was tested with the Shapiro‐Wilk normality tests. Normally distributed variables are expressed as mean±SEM. The incidence of plaque rupture was evaluated via χ² test with Yates' correction. We used Student t test to assess the effects of parameters between 2 different groups (unpaired test). 2‐way repeated‐measures ANOVA, followed by Bonferroni multiple‐comparisons test; 1‐way ANOVA, followed by Dunnett multiple‐comparisons test or Tukey multiple‐comparisons test; and 2‐way ANOVA, followed by Tukey multiple‐comparisons test, were made as appropriate. The Pearson correlation coefficient was done as appropriate. P<0.05 was considered as being statistically significant.

Results

The Vulnerable Plaque Is Established at 4 Weeks After Surgery in the Left Common Carotid Artery

To establish vulnerable plaques in the left common carotid artery, we adopted artery ligations and fed the ApoE^−/− mice with high‐fat diets (Figure 1A and 1B). At 4 weeks after surgery, mice in the baseline group were examined using MRI scans and hematoxylin and eosin staining. To make it comparable, the first section in each method was selected from the level of the carotid artery bifurcation. Every 2 serial sections were spaced 500 μm in MRI, and 250 μm in the histological staining. These parameters assured us that 1 MRI picture corresponded to 2 hematoxylin and eosin images. The 2500‐μm range of left common carotid artery was presented in line (Figure 1D). The black box in Figure 1C denoted the corresponding gross picture. As we could see, features like intraplaque hemorrhage and necrotic core formation, which were highlighted, occurred at a high frequency. We also found that macrophages prevailed in the subendothelial area, and smooth muscle cells migrated to the intima region (Figure 1G and 1H). Interestingly, LXRα protein was highly expressed in the lesion area, whereas the expression of LXRβ scattered in the subintimal area close to media (Figure 1I and 1J). More important, the Pearson correlation (Figure 1E and 1F) demonstrated that MRI is a valuable tool to trace individual plaque growth; the correlation between in vivo MRI lumen area and histology lumen area is high (r=0.81; P=0.0082); and either is lesion area (r=0.76; P=0.0186). The preexisting unstable plaque was established to explore drug effects in vulnerable plaque experiments.

Figure 1. Vulnerable features of established atherosclerotic lesions in the left common carotid artery.

The vulnerable plaque has been established after 4 weeks (w) of ligation in the setting of hypercholesterolemia, which is confirmed by magnetic resonance imaging (MRI) scans and histological staining. A, The schematic of the animal experimental design. B, The schematic of the surgical intervention. C, The gross picture of the carotid artery in D. Rectangle denotes the 2500‐μm region descending from the carotid bifurcation. Bar=2000 μm. D, Consecutive carotid artery sections were recorded by MRI T2 star (T2*)‐weighted sequence and stained with hematoxylin and eosin (H&E). The intervals of MRI and H&E images are 500 and 250 μm, respectively. Dotted circles in red denote the artery wall. Red and yellow bars are signs of lesions dominant in intraplaque hemorrhage and necrotic core formation, respectively, and dotted lines in black nearby denote the corresponding features. Bar=1000, 1000, and 100 μm (from left to right in D). E and F, Graphs show correlation analysis (n=9) between in vivo MRI lumen area and histology lumen area (×10⁴ μm²) (E), and lesion area (×10⁴ μm²) (F). G and H, Representative photomicrographs of carotid sections stained with an antibody against cluster of differentiation (CD) 68 (red) or α‐smooth muscle actin (α‐SMA) (green), and counterstained with 4′,6‐diamidino‐2‐phenylindole (DAPI) (blue). Bar=100 μm. I and J, Representative photomicrographs of carotid sections stained with an antibody against liver X receptor (LXR) α (red) or LXRβ (green), and counterstained with DAPI (blue). Dotted lines denote the lumen area. Bar=100 μm. The “n” refers to the number of mice, and P values are shown as indicated (Pearson correlation coefficient for E and F). GW indicates GW3965; HFD, high‐fat diet; ip., intraperitoneal injection; IPH, intraplaque hemorrhage; LECA, left external carotid artery; LICA, left internal carotid artery; LOA, left occipital artery; LSTA, left superior thyroid artery; and NC, necrotic core.

Activating LXR Reduces Vulnerable Features and Prevents Plaque Rupture in Preexisting Unstable Lesions

Confirming the existence of unstable plaques, mice were administered GW3965 (an agonist of LXR) or vehicle solution daily by intraperitoneal injection for another 4 weeks. Longitudinal MRI scans were applied to each mouse. As we collected images from 3 MRI sequences (PD, T2, and T2 star), the plaque could be traced and outlined in an integrated and dynamic way (Figure S1A). On 4 weeks of GW3965 administration, the arterial outward remodeling was ameliorated (vehicle 4 weeks versus GW3965 4 weeks: 26.17±0.98×10⁴ versus 13.83±0.40×10⁴ μm²; P<0.0001), with the wall area decreased (27±1.03×10⁴ versus 19±0.37×10⁴ μm²; P<0.0001), whereas the lumen area was increased (0.83±0.31×10⁴ versus 5.17±0.31×10⁴ μm²; P<0.0001; Figure 2A through 2D). As the plaque grew, the vehicle group observed 9 of 10 greenish black left common carotid arteries, whereas the GW3965 group had only 2 cases of plaque rupture (a pooled result from 2 batches of mice). Subsequent histological analysis revealed that the abnormal gross appearance was an indication of mural or occlusive thrombi with fibrous cap ruptures (Figure 2E and 2G). To get a grasp of the overall plaque burden and detailed plaque composition, consecutive carotid frozen sections were stained with hematoxylin and eosin, oil red O, and Masson trichrome staining (Figure 2F). Serial lesion area analysis indicated LXR activation had broad protection on sections calculated (Figure 2H). As a result, lesion volume and mean lesion area reduced by 52.61% and 52.01%, respectively, compared with the vehicle group (vehicle versus GW3965: 65.62±2.34×10⁷ versus 31.10±2.56×10⁷ μm³ [P<0.0001]; 26.32±0.96×10⁴ versus 12.63±1.00×10⁴ μm² [P=0.014]; Figure 2I and 2J). Accordingly, lesional lipid deposition was reduced by 53.67% (20.88±1.31% versus 9.67±1.28% of lesion area; P=0.004; Figure 2K), and collagen content increased by 186.42% (10.6±0.89% versus 30.36±1.92% of lesion area; P<0.0001; Figure 2L). Blood tests were applied at the end of the indicated period as well, and the total cholesterol decreased (767.4±56.6 versus 614.5±38.6 mg/dL; P=0.0496; Figure S2B), whereas triglycerides increased (109.5±6.2 versus 148.9±8.6 mg/dL; P=0.004; Figure S2C). Intriguingly, a profile of less intraplaque hemorrhage detection and necrotic core formation was found (19.76±2.31% versus 6.58±0.79% of lesion area [P<0.0001]; 28.64±2.41% versus 12.37±1.71% of lesion area [P<0.0001]; Figure 2M and 2N). These results suggested that activation of LXR improved the stability of vulnerable plaques.

Figure 2. Activating liver X receptor (LXR) reduces vulnerable features and prevents plaque rupture in preexisting unstable lesions.

Ligand‐activated LXR reduces arterial outward remodeling, intraplaque hemorrhage, and necrotic core formation in the left common carotid artery. As vulnerable plaques were established, mice were administered GW3965 (GW) (10 mg/kg per day) or 5% dimethyl sulfoxide in PBS for another 4 weeks (w) by intraperitoneal injection. A, Magnetic resonance imaging (MRI) T2 star (T2*)‐weighted images of the carotid artery (the same level from one individual) were taken at 0, 2, and 4 weeks in the vehicle (Veh) group and the GW3965 group. Red dotted circles show the artery wall. Bar=1000 μm. B through D, Graphs show the in vivo MRI lumen area, wall area, and lesion area (×10⁴ μm²; n=6) measurement at the time points mentioned in A, when treated with GW3965 or not. E, Representative photomicrographs of carotid frozen sections stained with hematoxylin and eosin and Masson trichrome. Arrowheads denote fibrous cap rupture with a mural thrombus or an occlusive thrombus. Bar=100 μm. F, Representative gross pictures and photomicrographs of carotid frozen sections stained with hematoxylin and eosin (H&E), oil red O (ORO), and Masson trichrome. Asterisks represent the necrotic core areas, and dotted lines denote the areas of intraplaque hemorrhage. Bar=1000 (first column in F) and 100 μm. G, Graph shows the measurements of the incidence of plaque rupture (percentage; n=10). H through J, Graphs show the measurements of (1) serial lesion area (×10⁴ μm²; n=6); (2) lesion volume (×10⁷ μm³; n=6); and (3) mean lesion area (×10⁴ μm²; n=6). K and L, Graphs show the measurements of (1) lipid deposition (percentage of lesion area; n=6); and (2) collagen content (percentage of lesion area; n=6). M and N, Graphs show the measurements of (1) intraplaque hemorrhage (percentage of lesion area; n=6); and (2) necrotic core (percentage of lesion area; n=6). Each circle represents the average specific parameter of 8 to 11 sections in each mouse or the average lesion area of 6 mice at the indicated level in H. Data are presented as mean±SEM, and “n” refers to the number of mice. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (2‐way repeated‐measures ANOVA, followed by Bonferroni multiple‐comparisons test [B–D]; 2‐sided χ² test with Yates' correction [G]; 2‐tailed unpaired Student t test [I–N]). n.s. indicates not significant.

LXR Activation Alleviates Apoptosis and Defective Efferocytosis as Well as ER Stress in Carotid Vulnerable Plaques

The amounts of apoptotic cells and their disposal are important factors in necrotic core formation. To detect apoptosis and efferocytosis in plaque areas and investigate whether LXR activation was involved in these processes, sections of carotid plaques were costained for DNA fragmentation with terminal deoxynucleotidyl transferase–mediated biotin–deoxyuridine triphosphate nick‐end labeling (TUNEL) staining and macrophage surface marker with cluster of differentiation (CD) 68 antibody (Figure 3A). We found that the relative number of lesional apoptotic cells decreased on GW3965 intervention, compared with the vehicle group (vehicle versus GW3965: 35.35±4.02% versus 19.62±3.34% of total cells; P=0.0132; Figure 3B). Meanwhile, the percentage of apoptotic cells undergoing efferocytosis increased in plaque areas (1.05±0.13 versus 2.40±0.22; P=0.0003; Figure 3C). We then conducted flow cytometry assays evaluating peritoneal uptake of labeled apoptotic cells. The GW3965 group showed superior phagocytic capacity to the vehicle group (42.57±9.44% versus 81.90±2.54% of F4/80⁺ cells; P=0.0158; Figure S3A and S3B). Restored cell viability may enhance cell interaction and provide chances for efferocytosis. These macrophages may acquire additional antiapoptosis properties when processing cell debris. Further immunofluorescence assay on cleaved caspase‐3, a canonical apoptotic effector caspase, recapitulates the above findings, referring apoptosis as a key target in the LXR treatment (3.37±0.77% versus 1.29±0.39% of lesion area; P=0.0372; Figure 3D and 3E). Then, we examined 2 apoptotic initiator caspases, mitochondrial‐related caspase‐9 and ER‐related caspase‐12, for the underlying mechanism. LXR activation profoundly reduced caspase‐12 expression (7.30±0.65% versus 2.44±0.53% of lesion area; P=0.0002; Figure 3D and 3E), whereas the inhibition of caspase‐9 expression was not significant (7.25±1.39% versus 4.32±1.03% of lesion area; P=0.1195; Figure 3D and 3E). As we know, ER stress could be a determinant of cell and plaque fate; we next analyzed BiP (binding‐immunoglobulin protein), ATF4 (activating transcription factor 4), and CHOP (C/EBP‐homologous protein) expressions within the plaque, which are recognized as ER stress markers. The vehicle group was stained positive for these 3 molecules, whereas they were barely found in the GW3965 group (7.73±1.16% versus 1.74±0.66% of lesion area [P=0.0012]; 6.32±0.95% versus 1.97±0.60% of lesion area [P=0.0032]; 5.37±0.72% versus 0.97±0.34% of lesion area [P=0.0002]; Figure 3F and 3G). Taken together, these findings showed LXR reduced apoptosis, defective efferocytosis, and ER stress in the vessel wall.

Figure 3. Liver X receptor (LXR) activation alleviates apoptosis and defective efferocytosis as well as endoplasmic reticulum (ER) stress in carotid vulnerable plaques.

Activating LXR through GW3965 (GW) reduces apoptosis and enhances efferocytosis in carotid arteries, and the benefits rely on modulating ER stress. A, Representative immunofluorescence staining of apoptosis and efferocytosis in carotid sections is shown. Macrophages were stained with an anti–cluster of differentiation (CD) 68 antibody (red), and apoptotic cells were stained with terminal deoxynucleotidyl transferase–mediated biotin–deoxyuridine triphosphate nick‐end labeling (TUNEL) (green). Dotted squares (second column in A) indicate the corresponding part of the enlarged pictures (third column in A). Yellow asterisks show examples of macrophage‐associated apoptotic cells, whereas green asterisks indicate free apoptotic cells. Bar=100 μm. B and C, Graphs show the quantification of lesional apoptotic cells (percentage of total cells; n=6) (B) and the measurement of efferocytosis index (n=6) (C). The efferocytosis index is the macrophage‐associated apoptotic cells/free apoptotic cells. D, Representative photomicrographs of carotid sections stained with an antibody against cleaved caspase‐3 (Cl‐Caspase‐3) (red; left), caspase‐9 (red; middle), or caspase‐12 (red; right), and counterstained with 4′,6‐diamidino‐2‐phenylindole (DAPI) (blue). E, Graph shows the quantification of caspase‐positive staining area (percentage of lesion area; n=6) in D. F, Representative photomicrographs of carotid sections stained with an antibody against BiP (binding‐immunoglobulin protein) (red; left), ATF4 (activating transcription factor 4) (red; middle), or CHOP (C/EBP‐homologous protein) (red; right), and counterstained with DAPI (blue). G, Graph shows the quantification of positive staining area of ER stress markers (percentage of lesion area; n=6) in F. Dotted lines denote the lumen areas. Data are presented as mean±SEM, and “n” refers to the number of mice. *P<0.05, **P<0.01, ***P<0.001 (2‐tailed unpaired Student t test [B, C, E, and G]). n.s. indicates not significant; and Veh, vehicle.

LXR Activation Inhibits 7‐KC–Induced ER Stress in Macrophages

To further explore the role and potential mechanism of LXR in regulating ER stress, we did subsequent experiments focusing on macrophage apoptosis and efferocytosis in vitro. 7‐KC, which belongs to the oxysterol family, is a bioactive cholesterol derivative. A recent study has shown that high plasma 7‐KC levels are associated with adverse clinical outcomes in patients with stable coronary artery disease. ¹⁶ In human late‐stage lesions, the 7‐KC abundance is only second to 27‐hydroxycholesterol ¹⁷ ; both of them are regulators of apoptosis. First, we examined UPR‐related gene expressions in Raw 264.7 macrophages. Wherever in concentration‐gradient or time‐point experiment, CHOP expression began to increase significantly during 12 hours of 7‐KC incubation at the dose of 40 μmol/L and reached its peak at 24 hours of dose 80 μmol/L (Figure S4A through S4D). Then, we adopted the combination of 40 μmol/L and 24 hours for further experiments. To investigate whether LXR plays a role in regulating ER stress in macrophages, we activated LXR with GW3965, as in the animal experiment. Abca1 and Abcg1 are LXR‐responsive genes involved in cholesterol efflux. With the concentration of GW3965 added up to 1 μmol/L, the gene expressions surged (Figure S4E). UPR pathways were again examined, and the preincubation with GW3965 for 24 hours almost offset the increased expressions of Chop, Atf4, and spliced X‐box‐binding protein 1 (sXBP), which were induced by 7‐KC (Figure S4F). Next, an array of molecules was tested by Western blotting, from the upstream ER membrane protein PRKR‐like endoplasmic reticulum kinase (PERK) and inositol‐requiring protein‐1α (IRE1α) to the apoptosis effector CHOP (Figure S4G; quantification data, see Figure S5A through S5E). As we have demonstrated LXR protection on established plaques, similarly, we primed macrophages with 7‐KC for 12 hours to see whether GW3965 treatment could protect irritated macrophages. The introduction of GW3965 exhibited effective protection on stressed cells at an incubating duration of 24 hours (Figure S6A and S6B). The results suggested that LXR activation silenced the decompensated UPR pathway.

The Favorable Effects of LXR Agonists on ER Stress–Related CHOP Pathway Are Lost in the Setting of Genetic Ablation of LXRα, but Not LXRβ, Isotype

LXRs have 2 isotypes, LXRα and LXRβ, which possess different expression patterns. We then evaluated the ability of activated LXRα or LXRβ to counteract the CHOP pathway. Primary macrophages were obtained from the peritoneal cavity of isotype‐specific knockout mice (Figure 4A). The expression of LXRα was detectable in LXRβ‐deficient macrophages, and the subsequent GW3965 intervention was sufficient to suppress 7‐KC–induced CHOP pathway (BiP, P=0.0330; ATF4, P=0.0034; and CHOP, P=0.0178; Figure 4C), whereas depletion of LXRα made LXR agonists futile (Figure 4B). We conclude that GW3965 exerted ER protection mainly through LXRα.

Figure 4. The favorable effects of liver X receptor (LXR) agonists on endoplasmic reticulum (ER) stress–related CHOP pathway are lost in the setting of genetic ablation of LXRα, but not LXRβ.

Peritoneal macrophages from LXRα knockout or LXRβ knockout mice were isolated and treated as indicated. Then, proteins were collected for the Western blotting assay (n=3). A, Isotype‐specific gene knockout in peritoneal exudate cells was confirmed by Western blotting (n=3). The graph shows the relative protein levels to β‐actin. B and C, GW3965 (GW)–activated LXR in LXRα‐deficient macrophage had minor effects on 7‐ketocholesterol (7‐KC)–induced C/EBP‐homologous protein (CHOP) pathway, including binding‐immunoglobulin protein (BiP), activating transcription factor 4 (ATF4), and CHOP protein expressions, whereas ligand activation of LXR in the absence of LXRβ is sufficient to suppress CHOP‐mediated ER stress. Graphs show the relative protein levels to β‐actin. Data are presented as mean±SEM, and “n” refers to the number of independent experiments taken. *P<0.05, **P<0.01 (2‐tailed unpaired Student t test [A], 1‐way ANOVA, followed by the Tukey multiple‐comparisons test [B and C]). n.s. indicates not significant.

LXRα Mediates the Effects of LXR Activation on Apoptosis Alleviation and Efferocytosis Enhancement in Macrophages

We next explore which isotype of LXRs accounts for regulating apoptosis and efferocytosis of macrophages in vitro. For TUNEL staining, activating LXR in LXRα‐deficient macrophages had no apparent impact on DNA fragmentation induced by 7‐KC (7‐KC versus GW3965+7‐KC, 42.70±4.01% versus 36.32±3.58% of total cells; P=0.3998; Figure 5A and 5B). As mentioned above, caspase‐3 activation is a classic biological marker for apoptosis, and full‐length caspase‐3 is cleaved into a 17‐kDa fragment during apoptosis. Therefore, we examined the protein levels of cleaved caspase‐3. GW3965 intervention did not reduce the protein expression of cleaved caspase‐3 in the setting of genetic ablation of LXRα (P=0.9998; Figure 5C; quantification data, see Figure S5G). However, LXR activation in wild‐type controls or with LXRβ deficiency was able to alleviate macrophage apoptosis in TUNEL staining and in Western blotting assay (Figures S7A through S7C and S8A through S8C). The lipid overloading stress is decisive not only in the cell fate but also in the ability of macrophages to remove apoptotic cells. Apoptotic cells were generated by a UV irradiating on Jurkat cells. Fluorescein‐labeled apoptotic cells were added to cell culture media of macrophages. Using a combination of bright‐field and fluorescent‐field or flow cytometry analysis, efferocytosis was evaluated and depicted as the association between macrophages and apoptotic cells. The results revealed that phagocytosis decreased in 7‐KC stimulation (Figure S7D through S7G). LXRα knockout, but not LXRβ knockout, damaged the capability of LXR activation to reverse the efferocytosis impairment (10.01±0.70% versus 11.86±1.27% [P=0.8686]; 8.51±0.55% versus 9.04±0.68% [P=0.9797]; Figure 5D through 5G; Figure S8D through S8G). In sum, LXRα mediated the protective effects of LXR activation on fixing macrophage malfunction, whereas LXRβ was dispensable.

Figure 5. Genetic ablation of liver X receptor (LXR) α blunts the beneficial effects of LXR activation on macrophage apoptosis and defective efferocytosis.

GW3965 (GW) was added to 7‐ketocholesterol (7‐KC)–flamed peritoneal macrophages from LXRα^−/− mice. A and B, Terminal deoxynucleotidyl transferase–mediated biotin–deoxyuridine triphosphate nick‐end labeling (TUNEL) staining (n=4) of macrophage apoptosis was performed in different treatment conditions described previously. Representative photomicrographs show that LXR agonists in LXRα‐deficient macrophages scarcely reduced TUNEL‐positive cells (green) induced by 7‐KC. Dotted squares (first row in A) indicate the corresponding part of enlarged pictures (second row in A). Bar=100 μm. The graph shows apoptotic cell measurements. C, Western blotting of cleaved caspase‐3 protein expression (n=3) was also adopted to access the protective effects of LXR activation on apoptosis. D and E, In vitro efferocytosis assay (n=4) was conducted using peritoneal macrophages from LXRα knockout mice. Apoptotic cells (green) that colocated with macrophages in the bright field were reckoned as effective efferocytosis. Dotted squares (first row in D) indicate the corresponding part of the enlarged pictures (second row in D). Bar=100 μm. Representative photomicrographs show that more free macrophages (apoptotic cells without macrophage association) were observed on 7‐KC stimulation, and benefits of LXR treatment were rarely found. Flow cytometry analysis (n=4) was adopted as well. F and G, Graphs show the phagocytic index measurements. The phagocytic index is defined as the ratio of apoptotic cell–associated macrophages/total cells. Data are presented as mean±SEM, and “n” refers to the number of independent experiments taken. *P<0.05, **P<0.01, ****P<0.0001 (1‐way ANOVA, followed by the Tukey multiple‐comparisons test [B, F, and G]). Acs indicates apoptotic cells; DAPI, 4′,6‐diamidino‐2‐phenylindole; Mφ, macrophages; and n.s., not significant.

LXRα Knockout Renders GW3965 Invalid in Reducing Plaque Vulnerability

Prompted by our in vitro data, we generated LXRα^−/−ApoE^−/− mice to further evaluate if LXRα plays a crucial role in stabilizing plaques. First of all, LXRα protein expression was rarely found in the vessel wall of LXRα^−/−ApoE^−/− mice (Figure S9A). After 4 weeks of plaque progression, only 1 of 9 individuals in the LXRα^−/−ApoE^−/− group and none of the ApoE^−/− mice had a trace of plaque rupture (Figure S9B), and no mouse died in both groups. Compared with ApoE^−/− mice using histological methods (Figure S9C), mice of additional general knockout of LXRα might acquire more diffused and bigger plaque alongside the vessel bed (Figure S9D through S9F), with lipid deposition increased, and collagen content was not evidently deteriorated (Figure S9G and S9H). At the time point of 4 weeks, intraplaque hemorrhage and necrotic core areas were not significantly changed (Figure S9I and S9J). Leaving the plaque in LXRα^−/−ApoE^−/− mice to grow, each group had 2 mice that died (death cases were excluded from the further statistical evaluation; Figure 6A). Evidence of occlusive thrombosis was found in the autopsy. The plaques progressed to a more advanced stage, and each individual in the vehicle group exhibited signs of plaque rupture. When it comes to GW3965 intervention, there were still 8 of 9 mice that experienced plaque rupture (LXRα^−/−ApoE^−/− vehicle versus LXRα^−/−ApoE^−/− GW3965: 9/9 versus 8/9; P>0.9999; Figure 6B). Specifically, plaque burden was similar in these 2 groups (73.98±2.90×10⁷ versus 73.27±4.52×10⁷ μm³ [P=0.9989]; 29.72±1.19×10⁴ versus 29.57±1.79×10⁴ μm² [P=0.9998]; Figure 6E and 6F); neither lipid/collagen profile (23.16±1.77% versus 21.59±1.73% of lesion area [P=0.8823]; 13.84±1.51% versus 14.81±1.42% of lesion area [P=0.9644]; Figure 6G and 6H) nor vulnerable features of intraplaque hemorrhage or necrotic core showed noticeable differences (19.35±1.37% versus 19.54±1.13% of lesion area [P=0.9997]; 30.93±2.19% versus 29.77±2.14% of lesion area [P=0.9713]; Figure 6I and 6J). However, when compared with GW3965 group in ApoE^−/− mice, the protection of LXR activation in LXRα^−/−ApoE^−/− mice was largely damaged (ApoE^−/− GW3965 versus LXRα^−/−ApoE^−/− GW3965: 40.43±1.89×10⁷ versus 73.27±4.52×10⁷ μm³ [P=0.0001]; 16.28±0.78×10⁴ versus 29.57±1.79×10⁴ μm² [P<0.0001] [Figure 6E and 6F]; 11.06±1.35% versus 21.59±1.73% of lesion area [P=0.0029]; 25.73±2.15% versus 14.81±1.42% of lesion area [P=0.0013] [Figure 6G and 6H]; 7.27±0.76% versus 19.54±1.13% of lesion area [P=0.0004]; 14.78±1.36% versus 29.77±2.14% of lesion area [P=0.0007] [Figure 6I and 6J]). On the whole, silencing LXRα impairs plaque protection originated from LXR activation.

Figure 6. Liver X receptor (LXR) α knockout renders GW3965 (GW) invalid in reducing plaque vulnerability.

LXRα^−/− apolipoprotein E (ApoE)^−/− mice were generated to further detect vulnerable features in our rupture‐prone model when losing LXRα functions. Deletion of the isotype LXRα exacerbates plaque progression and blunts plaque stabilization benefits exerted by LXR agonist administration. A, Graph shows the survival curves of 4 groups. B, Graph shows the measurements of the incidence of plaque rupture (percentage). C, Representative gross pictures and photomicrographs of carotid frozen sections stained with hematoxylin and eosin (H&E), oil red O (ORO), and Masson trichrome. Asterisks represent the necrotic core areas, and dotted lines denote the areas of intraplaque hemorrhage. Bar=1000 (first column in C) and 100 μm. D through F, Graphs show the measurements of (1) serial lesion area (×10⁴ μm²); (2) lesion volume (×10⁷ μm³); and (3) mean lesion area (×10⁴ μm²). G and H, Graphs show the measurements of (1) lipid deposition (percentage of lesion area); and (2) collagen content (percentage of lesion area). I and J, Graphs show the measurements of (1) intraplaque hemorrhage (percentage of lesion area); and (2) necrotic core (percentage of lesion area). For A through J, n=4 in both ApoE^−/− vehicle (Veh) and ApoE^−/− GW3965 groups; n=7 in both LXRα^−/−ApoE^−/− vehicle and LXRα^−/−ApoE^−/− GW3965 groups. Each circle/triangle represents the average specific parameter of 8 to 11 sections in each mouse or the average lesion area of 4 or 7 mice at the indicated level in D. Data are presented as mean±SEM, and “n” refers to the number of mice. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (2‐sided χ² test with Yates' correction [B]; 2‐way ANOVA, followed by the Tukey multiple‐comparisons test [E through J]). ApoE KO indicates ApoE^−/−; DKO, LXRα^−/−ApoE^−/−; and n.s., not significant.

Protective Effects of LXR Activation Against Lesional Macrophage Apoptosis, Impaired Efferocytosis, and Excessive ER Stress Are Abrogated in LXRα; ApoE^−/− Mice

We next scrutinize biological processes in the plaque area (Figure 7A through 7C). Although no distinguishing differences in vulnerable features were found, macrophage infiltration increased in LXRα^−/−ApoE^−/− mice at 4 weeks compared with ApoE^−/− mice (Figure S10D). Intriguingly, lesional macrophage apoptosis and efferocytosis varied little (Figure S10E and S10F), whereas cleaved caspase‐3–positive macrophages (Figure S10G) and CHOP‐positive macrophages increased (Figure S10H). At 8 weeks, ER stress prevailed in plaque areas in LXRα^−/−ApoE^−/− mice, and administration of GW3965 shows low potency to alleviate macrophage abundance (LXRα^−/−ApoE^−/− vehicle versus LXRα^−/−ApoE^−/− GW3965: 82.62±2.67% versus 82.4±2.12% of total cells; P=0.9999; Figure 7D), apoptosis (46.78±2.83% versus 45.65±3.81% of CD68⁺ cells; P=0.9939; Figure 7E), defective efferocytosis (0.95±0.11 versus 1.02±0.15; P=0.988; Figure 7F), cleaved caspase‐3 expression (56.16±3.82% versus 53.01±2.34% of CD68⁺ cells; P=0.8664; Figure 7G), and CHOP expression (16.52±1.34% versus 16.11±1.27% of CD68⁺ cells; P=0.9944; Figure 7H). When compared with GW3965 group in ApoE^−/− mice, the above‐mentioned benefits of LXR activation were impaired in LXRα^−/−ApoE^−/− mice (ApoE^−/− GW3965 versus LXRα^−/−ApoE^−/− GW3965: 67.70±2.33% versus 82.4±2.12% of total cells [P=0.0044] [Figure 7D]; 22.48±2.97% versus 45.65±3.81% of CD68⁺ cells [P=0.0015] [Figure 7E]; 2.66±0.30 versus 1.02±0.15 [P<0.0001] [Figure 7F]; 26.18±3.03% versus 53.01±2.34% of CD68⁺ cells [P=0.0001] [Figure 7G]; 7.90±1.03% versus 16.11±1.27% of CD68⁺ cells [P=0.0026] [Figure 7H]). These data collectively demonstrate the importance of LXRα throughout the formation and rupture of vulnerable plaques.

Figure 7. Protective effects of liver X receptor (LXR) activation against lesional macrophage apoptosis, impaired efferocytosis, and excessive endoplasmic reticulum stress were abrogated in LXRα^−/− apolipoprotein E (ApoE)^−/− mice.

Macrophage viability and function in vulnerable plaques remain defective in the case of LXRα gene ablation, even when supplemented with LXR pan‐agonist GW3965 (GW). A, Representative immunofluorescence staining of macrophage apoptosis and efferocytosis in carotid sections is shown. Macrophages were stained with an anti–cluster of differentiation (CD) 68 antibody (red), and apoptotic cells were stained with terminal deoxynucleotidyl transferase–mediated biotin–deoxyuridine triphosphate nick‐end labeling (TUNEL) (green). Yellow asterisks show examples of macrophage‐associated apoptotic cells, whereas green asterisks indicate free apoptotic cells. Bar=100 μm. B, Representative photomicrographs of carotid sections stained with antibodies against cleaved caspase‐3 (Cl‐Caspase‐3) (red) and CD68 (green), and counterstained with 4′,6‐diamidino‐2‐phenylindole (DAPI) (blue). Bar=100 μm. C, Representative photomicrographs of carotid sections stained with antibodies against C/EBP‐homologous protein (CHOP) (red) and CD68 (green), and counterstained with DAPI (blue). Dotted squares (second column in A through C) indicate the corresponding part of the enlarged pictures (third column in A through C). Bar=100 μm. D, Graph shows the quantification of CD68⁺ cells (percentage of total cells). E and F, Graphs show the quantification of lesional apoptotic CD68⁺ cells (percentage of CD68⁺ cells) (E) and the measurement of efferocytosis index (F). The efferocytosis index is the macrophage‐associated apoptotic cells/free apoptotic cells. G, Graph shows the quantification of cleaved caspase3 (Cl‐Cas3)⁺CD68⁺ cells (percentage of CD68⁺ cells). H, Graph shows the quantification of CHOP⁺CD68⁺ cells (percentage of CD68⁺ cells). Dotted lines denote the lumen areas. For D through H, n=4 in both ApoE^−/− vehicle (Veh) and ApoE^−/− GW3965 groups; n=7 in both LXRα^−/−ApoE^−/− vehicle and LXRα^−/−ApoE^−/− GW3965 groups. Data are presented as mean±SEM, and “n” refers to the number of mice. *P<0.05, **P<0.01, ***P<0.001 (2‐way ANOVA, followed by the Tukey multiple‐comparisons test [D–H]). ApoE KO indicates ApoE^−/−; DKO, LXRα^−/−ApoE^−/−; and n.s., not significant.

Discussion

In the current study, we assessed LXR protection on mice vulnerable plaques and lipid‐stressed macrophages. We found that LXR activation alleviated vulnerable characteristics, such as arterial outward remodeling, intraplaque hemorrhage, and necrotic core formation, mainly through LXRα. Consequently, plaques acquire resistance to rupture.

On the basis of our previous work on this nuclear receptor as well as the mouse model of plaque rupture, we supposed LXR might exert focal protection apart from its lipid‐lowing effect, which is important to stabilize rupture‐prone plaques, as the late‐stage lesions not only experienced a poor systemic lipid profile but also the local inflammation and stress. Herein, we modified a well‐described mice model of plaque rupture to get vulnerable plaques. ¹¹ , ¹² , ¹⁸ , ¹⁹ Endogenous renovascular hypertension combined with carotid low shear stress was achieved by artery ligations. Then, a high‐fat diet was added (to accelerate the progression of early‐phase plaques to a vulnerable stage), so that vulnerable plaques that originally required 8 weeks to form basically have the relevant characteristics at 4 weeks (except for the low incidence of rupture), as shown in Figure 1. In the real‐world practice, patients stratified as high‐risk individuals with atherosclerotic cardiovascular disease were told to stop excessive fat intake and start lipid‐lowering therapy. Our pilot study revealed that plaques with sustained lesional lipid burden (continued high‐fat diet supply) commonly coexist with large necrotic core areas. They together result in reduced cell number. The low cellularity may cause difficulties in detecting other parameters of interest (ie, ER stress, apoptosis, and efferocytosis) in the mechanistical exploration. Thus, we adopted a high‐fat diet retraction strategy as the drug intervention period began. More important, after 8 weeks of plaque progression (high‐fat diets are retracted during the last 4 weeks), short‐term events happen in the vehicle group and are concomitant with evident phenotype anomalies. We thus obtained established plaques at 4 weeks to facilitate our drug intervention for vulnerable plaque experiments.

LXRs have 2 isotypes, LXRα and LXRβ, that are encoded by Nr1h3 and Nr1h2, respectively. LXRα is expressed in metabolically active tissues or cells, like the liver, intestines, and macrophages, whereas LXRβ is ubiquitously expressed. ²⁰ Previous work has shown LXRα protein is highly expressed in human lesional foam cells positive for CD68. ²¹ In our study, evaluation of LXRα and LXRβ proteins in plaque areas provided some interesting information; LXRα expression diffused in the lesion area, whereas LXRβ was mainly located in the peri‐intimal region (Figure 1). Lesional expression of LXR proteins was a prerequisite of LXR activation in vivo, whereas distinct expression pattern might indicate their differential roles in plaque progression. On 4 weeks of treatment on unstable lesions by LXR agonist GW3965, the progression of plaque burden and lipid deposition were retarded. In contrast, the collagen content was elevated. More important, activating LXR alleviated intraplaque hemorrhage and necrotic core formation. At last, the incidence of plaque rupture in the GW3965 group decreased to a low level (Figure 2).

As a result of lacking plaque rupture mice model, scientific evidence of LXR treatment on clinically relevant plaques is insufficient. Our current study may shed some light on this field. Previous studies have shown that LXR agonists retard lesion growth in ApoE knockout, low‐density lipoprotein receptor knockout, or ApoE*3‐Leiden mice mainly through lipid modulation. ⁸ , ²² , ²³ In these experiments, efficient reverse cholesterol transport and elevated high‐density lipoprotein were detected. ²⁴ Despite the atheroprotective effects, severe hypertriglyceridemia and hepatic steatosis are barriers to their clinical use. ²⁰ , ²⁵ Treatment on macrophage LXR of target organs may be a decent solution, as macrophage overexpression of LXRα ameliorates atherosclerosis as well as hypertriglyceridemia, ²⁶ , ²⁷ whereas deletion of macrophage LXR brings harm. ⁷ , ²⁸ In addition, activated LXRβ is enough to provide resistance to early lesion progression in the setting of hypercholesterolemia, ²⁹ although LXRα is required for a robust response to LXR ligands. ³⁰ More nonlipid effects of LXR have been revealed. LXR agonists can still be atheroprotective when lacking ABCA1/G1 in macrophages, mainly through the transrepression of inflammatory genes. ³¹ In a regression model, aortic arches from ApoE^−/− mice with bone marrow deficiency of LXRα or LXRβ were transplanted into wild‐type recipients. Plaques from both LXRα‐deficient mice and LXRβ‐deficient ApoE^−/− mice exhibited impaired regression. ³² Intriguingly, a study performed in New Zealand white rabbits has shown that LXR agonist LXR‐623 exerts significant effects on plaque regression in combination with simvastatin. ³³ These pieces of evidence show that LXR may exert the atheroprotection apart from lipid‐lowering effects. However, when it comes to vulnerable plaques, the underlying mechanisms are still little known.

Zooming into the biological processes that happened in the plaque, LXR agonists might exert focal protection in that lesional cell death was reduced while efferocytosis was enhanced. As we found LXR activation led to reduced plaque area, there might be other contributors involved. Apart from apoptosis and efferocytosis, monocyte recruitment and macrophage egress can also change lesion size. We then conducted complete blood counts and observed a decrease in the blood monocyte number/ratio after GW3965 intervention (Figure S11). Although the results according to published data indicate that LXR might normalize hypercholesterolemia‐induced monocytosis in ApoE knockout mice, ³⁴ , ³⁵ recent studies suggest that although monocyte recruitment to atherosclerotic lesions primarily contributes to macrophage accumulation in early lesion development, the local proliferation of lesional macrophages predominates in established atherosclerotic lesions. ³⁶ Hence, whether decreases in blood monocytes alter macrophage accumulation in plaques and to what extent in our model merits further investigation. For macrophage egress, monocyte‐derived cells expressing C‐C chemokine receptor type 7 (CCR7) (mainly macrophages and dendritic cells) can migrate out of the plaque in an LXR‐dependent manner, reducing the cellularity in the plaque and thus the plaque size. ³² Meanwhile, CCR7 is found to be upregulated during the phenotype shift to M1 macrophages. ³⁷ We found cumulative CCR7⁺CD68⁺ cells in the subintimal and vessel wall area of vulnerable plaques, whereas scarce CCR7⁺ cells were presented in the GW3965 group (Figure S12). The CCR7⁺CD68⁺ cell population may be composed of detained monocyte‐derived cells and proinflammatory M1 macrophages. Loss of CCR7⁺ cell detection is likely a consequence of cell egress or the resolution of inflammation after GW3965 treatment. As we did not see a notable difference in the ratio of lesional CD68⁺ cells between the 2 groups, we infer that CCR7‐mediated cell egress may play an active role in protective effects of LXR in vulnerable plaques, but not a dominant one. We thus focused on lesional apoptosis/efferocytosis of macrophages and the participation of subcellular organelles in this circumstance. Reduced expression of basal caspase‐12, found in the GW3965 group, was intriguing enough for us to explore the ER stress pathway. Mechanistically, LXR activation alleviated the expression of BiP, ATF4, and CHOP, 3 ER stress markers, and thus the lesional stress (Figure 3). This finding is consistent with our previous work on myocardial ischemia/reperfusion injury, ⁹ highlighting ER stress as a pivotal participant in cardiovascular diseases. In addition, it supports previous research into the vessel area, which links LXR and ER stress inhibition. In the hyperplastic neointima, LXR agonist T0901317 significantly decreases CD68⁺TUNEL⁺ double‐positive macrophages and attenuates neointimal hyperplasia. ³⁸ In another hypercholesteremia mouse model, LXRα is able to increase the production of bioactive lipids and provide resistance to macrophage ER stress in atherosclerosis. ³⁹ There are also studies exploring the protective role of CHOP gene knockout in atherosclerotic mice models. ⁴⁰ , ⁴¹ , ⁴² Referring to efferocytosis, previous reports have demonstrated that apoptotic cells activate LXR in phagocytes for their own clearance and lead to immune tolerance, and GW3965 treatment ameliorates lupus‐like autoimmunity in mice. ⁴³ In atherosclerosis, a combination of LXR/retinoic acid receptor α (RARα) agonists may cooperate in the control of efficient phagocytosis. ⁴⁴

LXRα and LXRβ synergize to regulate genes that encode proteins involved in cholesterol uptake, efflux, transport, and excretion, named reverse cholesterol transport. Are both of these 2 isotypes equally essential to inhibit ER stress? Therefore, we conducted in vitro experiments to figure it out. We used 7‐KC stimulation in cell experiments as a reflection of what happens in vulnerable plaques. ⁴⁵ LXRs are activated by certain oxysterol ligands but not 7‐KC. Meanwhile, nonsterol synthetic agonists (ie, GW3965 used in our study) have potent transcriptional activities; thus, they are useful tools for LXR research. ²⁰ Unlike T0901317, GW3965 has no undesirable effects on farnesoid X receptors and androstane receptors. As 7‐KC induced ER stress, LXR activation through GW3965 inhibited the UPR cascade (Figure S4). The dephosphorylation of 2 ER sensors, PERK and IRE1α, as well as the downregulation of BiP, which binds to unfolded or misfolded proteins, might indicate the clearance of ER stress triggers, such as aberrant protein synthesis, lipid overloading, and oxidative stress. ⁶ The peritoneal macrophage is a valuable tool for accessing specific protein functions in vitro. We preincubated LXRα knockout or LXRβ knockout macrophages with LXR agonists before 7‐KC stimulation in experiments. Activating LXR in the setting of LXRα genetic ablation blunted effects of GW3965 on ER stress inhibition, apoptosis alleviation, and efferocytosis enhancement, whereas the loss of LXRβ had no apparent detrimental impact on LXR activation (Figures 4 and 5, Figures S7 and S8). A battery of genes is LXR‐responsive, biological processes involved in cholesterol regulation, membrane phospholipid remodeling, innate immune response, phagocytosis, and so on. Considering the powerful restoration of cell viability, we suppose that LXR activation motivates several antiapoptotic ways. The interaction network is worth detailed study in our future work.

To further explore if the plaque‐stabilizing effect of LXR activation relies on intact LXRα function, general LXRα deletion was introduced in ApoE^−/− mice, and 2 comparisons were made: (1) After 4 weeks of plaque growth, LXRα^−/−ApoE^−/− mice are more susceptible to lipid retention, with a concurrent trend of higher plasma cholesterol level and heavier plaque burden, compared with ApoE^−/− mice. This stage of vulnerable plaque seems relatively quiet, but with ongoing recruitment/proliferation of macrophages and the failed dissipation of lipid pool, molecular changes occur before cellular defects. (2) After 8 weeks of plaque progression, short‐term events happen and are concomitant with evident phenotype anomalies, even though high‐fat diets are retracted during the last 4 weeks. We infer that lipid toxicity may not be a determinant in the formation of advanced plaques. Effects of improved lipid profile attributable to GW3965 administration are partly preserved with solely activated LXRβ, but plaques are still prone to rupture irrespective of LXR activation in LXRα^−/−ApoE^−/− mice, and the macrophage dysfunction remained the same (apoptosis, impaired efferocytosis, and decompensated ER stress–related CHOP pathway). These findings highlight the importance of LXRα in treating vulnerable plaques (Figures 6 and 7). Taking the dumbness of LXRβ and the adverse effects of LXRα into consideration, lesional drug delivery might be preferable to systemic activation of LXRα.

In summary, LXRα has a protective role in vulnerable plaques, and the details are described in Figure 8. Ligand activation of LXR reduces arterial outward remodeling, intraplaque hemorrhage, and necrotic core formation. These effects synergize to prevent plaque rupture. Mechanistically, activating LXRα alleviates apoptosis and enhances efferocytosis. And LXRα is able to correct the aberrant ER stress–mediated CHOP pathway in macrophages and vulnerable plaques. We suppose all these efforts will help to elucidate LXR biology and facilitate the eventual application of LXR therapies in cardiovascular diseases.

Figure 8. Protective functions of liver X receptor (LXR) α in established vulnerable plaques.

Activating LXRα alleviates apoptosis and enhances efferocytosis in macrophages. In mice, LXRα reduces arterial outward remodeling, intraplaque hemorrhage, and necrotic core formation, and eventually stabilizes vulnerable plaques. Endoplasmic reticulum (ER) stress–mediated C/EBP‐homologous protein (CHOP) pathway is involved in these processes. ATF4 indicates activating transcription factor 4.

Sources of Funding

This work was supported by the National Natural Science Foundation of China (Nos. 81870338, 81570390, 81270374, and 81801650).

Disclosures

None.

Supporting information

Data S1

Table S1

Figures S1–S12

References 43, 46, 47

(J Am Heart Assoc. 2021;10:e018455. DOI: 10.1161/JAHA.120.018455.)