Phosphorylation of C/EBP Homologous Protein (CHOP) by the AMP-Activated Protein Kinase Alpha 1 in Macrophages Promotes CHOP Degradation and Reduces Injury-Induced Neointimal Disruption In Vivo

Abstract

Rationale

Elevated levels of C/EBP homologous protein (CHOP), a member of the C/EBP transcription factor family, in advanced atherosclerotic plaques is reported to be associated with atherosclerotic plaque rupture in humans. However, the molecular mechanism by which CHOP accumulation occurs is poorly defined.

Objective

The aim of this study was to investigate if (1) macrophage AMP-activated kinase (AMPK) regulates cellular CHOP accumulation and (2) whole-body Ampk deletion leads to neointimal disruption.

Methods and Results

In isolated or cultured macrophages, Ampkα1 deletion markedly increased apoptosis and CHOP, whereas pharmacological activation of AMPK dramatically reduced CHOP protein level via promoting CHOP degradation by proteasome. In addition, co-transfection of Chop-specific siRNA, but not control siRNA, markedly reduced apoptosis in macrophages transfected with Ampkα1-specific siRNA. Mechanistically, AMPKα1 was found to co-immunoprecipitate with CHOP and phosphorylate CHOP at serine 30. Furthermore, serine 30 phosphorylation of CHOP triggered its ubiquitination and proteasomal degradation. In a mouse model of plaque stability, deletion of Ampkα1 but not Ampkα2 promoted injury-induced neointimal disruption. This was paralleled by increased CHOP expression and apoptosis in vivo. Finally, transfection of Chop-specific siRNA but not control siRNA reduced both CHOP level and injury-induced neointimal disruption in vivo.

Conclusions

Our results indicate that AMPKα1 mediates CHOP ubiquitination and proteasomal degradation in macrophages by promoting the phosphorylation of CHOP at serine 30. We conclude that AMPKα1 might be a valid therapeutic target in preventing atherosclerotic vulnerable plaque formation.

INTRODUCTION

Rupture of advanced atherosclerotic plaques and subsequent partial or complete thrombotic artery occlusion are the main pathological mechanisms for acute myocardial infarction and stroke.^{1, 2} Increased necrosis, inflammation, matrix degradation, cellular apoptosis, and reduced fibrous caps are the well-established morphological and pathological characteristics of vulnerable atherosclerotic plaques.^3-5 Pathologically, lipid-overloaded macrophages, a major cellular component of advanced atherosclerotic plaques, make plaques vulnerable to rupture because cellular debris from necrotic or apoptotic macrophages triggers inflammation and matrix proteinase activation.^6-9 Thus, higher levels of apoptotic macrophages within advanced plaques may promote plaque vulnerability.

The C/EBP homologous protein (CHOP) is a member of the C/EBP transcription factor family and is ubiquitously expressed. The expression of CHOP is induced under stress conditions such as DNA damage, growth arrest, and nutrient deprivation.¹⁰ CHOP is reported to play a vital role in regulating macrophage apoptosis in advanced atherosclerosis.^{11, 12} Myoishi et al. demonstrated that elevated levels of both ER stress markers and apoptotic cells were observed within ruptured plaques when compared to the stable fibrous plaques.¹³ Mechanistically, CHOP-dependent pathway is activated in unstable plaques, suggesting that CHOP-mediated apoptosis may be involved in triggering coronary artery plaque rupture.¹³ Recently, a noteworthy clinical study reported that apoptosis and CHOP elevate in advanced rupture-prone plaques within human carotid arteries.¹⁴ In mice, genetic deletion of CHOP leads to reduced macrophage apoptosis both in vitro and in vivo, which contributes to improved atherosclerotic plaque stability.^{11, 15} Taken together, it has been suggested that CHOP-mediated macrophage apoptosis is a central contributor of instability of atherosclerotic plaque. However, how CHOP is regulated in atherosclerotic plaques is poorly studied.

AMPK-activated protein kinase (AMPK) is a well characterized regulator of cellular energy status. AMPKα1 is the predominant isoform of AMPKα in vascular cells including endothelial cells, vascular smooth cells, and monocytes/macrophages.¹⁶ However, both AMPKα1 and AMPKα2 have been demonstrated to regulate vascular cell function.^17-21 Genetic deletion of Ampkα2 promotes atherosclerosis in vivo, likely via aberrant endoplasmic reticulum (ER) stress.²² Whether or not dysfunctional AMPK promotes atherosclerotic plaque vulnerability remains unknown. The aim of the present study was to test if AMPK reduces atherosclerotic plaque vulnerability by promoting CHOP ubiquitination and degradation. Here, we report that phosphorylation of CHOP at Ser30 by AMPKα1 triggers CHOP degradation resulting in reduced macrophage apoptosis and subsequent ameliorated plaque vulnerability in vivo.

METHODS

Murine ligation and cuff-triggered neointimal disruption model

A well-characterized murine model was used to assay neointimal formation and disruption, as described previously.^{15, 23, 24} Briefly, 9-week-old male Apoe^−/−, Apoe^−/−/Ampkα1^−/−, and Apoe^−/−/Ampkα2^−/− mice were anesthetized with an intraperitoneal injection of 100 mg/kg ketamine-HCl and 10 mg/kg xylazine-HCl, and their right common carotid artery was ligated proximal to the bifurcation. Four weeks after ligation, a polyethylene cuff (427410; BD Biosciences, San Jose, CA) was placed around the ligated right common carotid artery. At 4 days after cuff placement, the right common artery was collected and the intracuff lesions were evaluated by hematoxylin-eosin (HE) staining. The incidence of injury-induced neointimal disruption was calculated. For in vivo siRNA transfection,²⁵ 15 μg of scramble Stealth RNAi™ siRNA (12935-112; Invitrogen, Grand Island, NY) or Stealth Chop-specific siRNA (1320001; Invitrogen, Grand Island, NY) dissolved in 30% pluronic gel solution (P2443; Sigma-Aldrich, St. Louis, MO) was perivascularly delivered to the ligated and cuffed right common carotid artery of the Apoe^−/− mice starting 7 days before tissue collection. After 7 days of siRNA delivery, the carotid arteries were collected to analyze the incidence of neointimal disruption and the efficiency of siRNA delivery to the carotid artery was demonstrated by Western blot analysis.

Statistical analysis

Quantitative values are expressed as the mean ± SEM and represent data from at least three independent experiments. The difference between two groups was analyzed by Student's t-test. The difference among more than two groups was analyzed by one-way analysis of variance, followed by Newman-Keus multiple comparison test, and comparisons of different parameters between each group were made by two-way ANOVA analysis followed by Bonferroni posttests. For incidence of injury-induced neointimal disruption, groups were compared with Chi-square test (three groups) or Fisher exact test (two groups). P values of less than 0.05 were considered statistically significant.

An expanded Materials and Methods are available in the Online Data Supplement.

RESULTS

Ampkα1 deletion promotes macrophage apoptosis

Lipid-overloaded macrophages are a major cellular component of advanced atherosclerotic plaque. Overwhelming evidence suggests that atherosclerotic plaques become vulnerable to rupture when apoptotic macrophages trigger a local inflammatory response and matrix proteinase activation.^6-9 To test whether AMPKα modulates macrophage apoptosis, we first detected the effect of genetic deletion of Ampkα on apoptosis in macrophages. As shown in Figure 1A, the number of terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL)-positive cells in Ampkα1^−/− bone marrow-derived macrophages (BMDMs) was significantly higher compared to wild type (WT). In contrast, the number of TUNEL-positive BMDMs from Ampkα2^−/− mice did not vary from WT mice. Consistent with this observation, the levels of both cleaved caspase 3 and poly(ADP-ribose) polymerase (PARP), two well-characterized markers for apoptotic cells, were significantly higher in the BMDMs isolated from Ampkα1^−/− mice (Figure 1B) compared to either WT or Ampkα2^−/− mice. Taken together, our results indicate that the deletion of Ampkα1 but not Ampkα2 promotes macrophage apoptosis.

Figure 1. Ampkα1 deficiency promotes apoptosis in macrophages.

A. Bone marrow-derived macrophages (BMDMs) from WT, Ampkα1^−/−, and Ampkα2^−/− mice were stained using a TUNEL kit. Nuclei were stained with DAPI (blue). Left, representative immunofluorescence photomicrographs of TUNEL assays in BMDMs. Scale bar, 50 μm. Right, the percentage of TUNEL-positive (green) BMDMs. n = 6. B. Immunoblots of BMDMs isolated from WT, Ampkα1^−/−, and Ampkα2^−/− mice. The protein-specific antibodies are indicated. n = 5. Results are mean ± SEM. *P < 0.05 vs WT.

AMPKα1 inversely regulates CHOP protein levels in macrophages

It has been reported that CHOP-induced macrophage apoptosis promotes atherosclerotic plaque rupture.¹⁵ Given that AMPKα1 is the predominant AMPKα isoform in human and mouse macrophages,¹⁸ we set out to determine if AMPKα1 regulates apoptosis by altering the levels of CHOP in macrophages. To test the regulatory role of AMPKα1 in CHOP-induced macrophage apoptosis, mouse macrophage-like cell line RAW264.7 cells were transfected with scramble, Ampkα1-specific, and/or Chop-specific siRNA. As shown in Online Figure I, transfection of Ampkα1 siRNA markedly increased the levels of CHOP and cleaved caspase 3. Scramble siRNA had no effect. Chop siRNA reversed the increased cleaved caspase 3 level by Ampkα1 siRNA (Online Figure I). Therefore, we demonstrated that elevated caspase 3 cleavage in Ampkα1 siRNA-treated cells is CHOP-dependent by co-transfecting Chop siRNA with Ampkα1 siRNA. In contrast, the transfection of Chop siRNA without co-transfection with Ampkα1 siRNA did not alter cleaved caspase 3 levels compared to those transfected with scramble siRNA. Taken together, these results indicate that CHOP is required for Ampkα1 deficiency-induced apoptosis.

Next, we determined if Ampkα deletion alters CHOP in BMDMs isolated from WT, Ampkα1^−/−, and Ampkα2^−/− mice. Figure 2A showed the deletion of Ampkα1 but not Ampkα2 significantly increased CHOP protein levels in macrophages. Consistent with this, BMDMs from Ampkα1^−/− mice but not WT or Ampkα2^−/− mice exhibited higher mRNA levels of well-known CHOP target genes, such as Wars, Sars, Trib3, and Ero1α,²⁶ suggesting that Ampkα1 deletion increases CHOP activity in macrophages (Online Figure II). Furthermore, we demonstrated that activation of AMPK with either 5-aminoimidazole-4-carboxamide ribonucleoside (AICAR) (Figure 2B) or A769662 (Figure 2C) lowered CHOP levels in macrophages in a time-dependent manner. Conversely, inhibiting AMPK activity with compound C led to an upregulation of CHOP levels in a time-dependent manner (Figure 2D). Overall, our results support the hypothesis that AMPKα1 is an inverse regulator of CHOP.

Figure 2. AMPKα1 downregulates CHOP protein levels in macrophages.

A. Immunoblots of CHOP in WT, Ampkα1^−/−, and Ampkα2^−/− BMDMs. n = 5. B, C, and D. Immunoblots of CHOP in RAW264.7 cells treated with AICAR (1 mmol/L), A769662 (100 μmol/L) or Compound C (10 μmol/L) for indicated time. n = 5. Results are mean ± SEM. *P < 0.05 vs WT or control.

To assess whether increased CHOP protein levels resulting from AMPK inhibition are due to an increase in Chop mRNA levels, we conducted quantitative real-time RT-PCR to determine Chop mRNA in BMDMs isolated from WT, Ampkα1^−/− and Ampkα2^−/− mice. As shown in Online Figure III, similar levels of Chop mRNA existed in all three genotypes. Next, we examined if AMPK activation with AICAR altered the half-life (t_1/2) of Chop mRNA. To test this, RAW264.7 cells were incubated with actinomycin D and treated with or without AICAR for the indicated time. As depicted in Online Figure IV, AICAR did not accelerate Chop mRNA degradation in RAW264.7 cells. These data indicate that AMPK does not directly affect Chop mRNA at both the transcriptional and post-transcriptional levels.

AMPK decreases CHOP protein stability

To explore how AMPK activation decreases CHOP protein levels, cycloheximide (CHX)-pretreated macrophages were exposed to AICAR and the steady-state levels of CHOP were measured by Western blot analysis. As shown in Figure 3A, AMPK activation by AICAR significantly increased the degradation of CHOP. Transfection of Ampkα1-specific siRNA but not scramble siRNA prevented CHOP degradation (Figure 3B). Consistent with these results, genetic deletion of Ampkα1 in BMDMs inhibited CHOP degradation when compared to WT BMDMs (Figure 3C), indicating that AMPKα1 is an important regulator of CHOP protein stability in macrophages.

Figure 3. AMPKα1 decreases CHOP protein stability.

A. RAW264.7 cells were pretreated with CHX (5 μg/mL) for 30 min before being treated with or without AICAR (1 mmol/L) for the indicated time, followed by Western blot analysis to detect CHOP. n = 5. B. RAW264.7 cells were transfected with scramble or Ampkα1 siRNA for 48 h and then treated with CHX (5 μg/mL) for the indicated time, followed by Western blot analysis to detect CHOP. n = 5. C. BMDMs isolated from WT and Ampkα1^−/− mice were treated with CHX (5 μg/mL) for the indicated time, followed by Western blot analysis to detect CHOP. n = 5. Results are mean ± SEM. *P < 0.05 vs control, scramble siRNA or WT.

AMPKα1 phosphorylates CHOP at Ser30

CHOP has been reported to be phosphorylated by p38 MAP Kinase (p38 MAPK) and casein kinase 2 (CK2).^{27, 28} AMPK is a well-known threonine/serine protein kinase. Therefore, we hypothesized that AMPKα1 phosphorylates CHOP to control its stability. Computer alignment studies indicated that Ser30 within CHOP to be a potential optimal AMPKα1 substrate motif (Figure 4A)²⁹. As shown in Figure 4B, AMPKα1 co-immunoprecipitates with CHOP in macrophages. Furthermore, we observed costaining of AMPKα1 and CHOP in BMDMs (Figure 4C). Finally, an in vitro kinase assay showed that both endogenous immunoprecipitated AMPKα1 and exogenous recombinant protein AMPKα1 phosphorylates CHOP (p-CHOP) at Ser30 (Figure 4D and E).

Figure 4. Phosphorylation of CHOP Ser30 by AMPKα1 is required for AICAR-induced CHOP downregulation.

A. Alignment of peptide sequences flanking human and mouse CHOP Ser30. B. Interaction of AMPKα1 and CHOP. C. Co-staining of AMPKα1 (red) and CHOP (green) in BMDMs. Nuclei were stained with DAPI (blue). Scale bar, 20 μm. D. AMPKα1 was immunoprecipitated (IP) from lysed RAW264.7 cells with a specific antibody. Human recombinant GST-CHOP (100 ng) and immunoprecipitated AMPKα1 were incubated in a kinase reaction buffer in the presence of AMP at 37 °C for 30 min. Immunoblotting for phosphorylation of CHOP at Ser30 was shown. E. Human recombinant GST-CHOP (500 ng) and human recombinant AMPKα1 (500 ng) were incubated in the kinase reaction buffer in the presence of AMP at 37 °C for 30 min. Immunoblotting for phosphorylation of CHOP at Ser30 was shown. F, G, and H. Immunoblots of p-CHOP (Ser30) in RAW264.7 cells treated with AICAR (1 mmol/L), A769662 (100 μmol/L) or Compound C (10 μmol/L) for indicated time. n = 5. I. Immunoblots of p-CHOP (Ser30) in WT, Ampkα1^−/−, and Ampkα2^−/− BMDMs. n = 5. J. WT and Ampkα1^−/− BMDMs were treated with 100 μmol/L A769662 for 12 h. p-CHOP (Ser30) was determined by Western blot. K and L. RAW264.7 cells were transfected with full-length WT DDK-Chop (WT) or the indicated site-directed mutants, DDK-Chop S30A or DDK-Chop S30E, for 48 h and then treated with or without AICAR for 12 h, followed by Western blot analysis to detect DDK. n = 5. Results are mean ± SEM. *P < 0.05 vs control or WT. ^#P < 0.05 vs as indicated. NS, not significant.

We next established if AMPK-dependent Ser30 phosphorylation of CHOP occurred in intact cells. Macrophages were exposed to either AICAR or A769662 at the times indicated. As depicted in Figure 4F and 4G, AMPK activator AICAR or A769662 markedly increased p-CHOP at Ser30 at 8 hours. Further, prolonged incubation (up to 24 hours) increased levels of p-CHOP at Ser30. Conversely, inhibition of AMPK with compound C, a potent AMPK inhibitor, time-dependently lowered the levels of p-CHOP at Ser30 (Figure 4H). In line with these results, BMDMs isolated from Ampkα1^−/− mice exhibited significantly lower levels of p-CHOP at Ser30 compared with those isolated from WT or Ampkα2^−/− mice (Figure 4I). Furthermore, A769662 increased p-CHOP (Ser30) in WT BMDMs but not Ampkα1−/− BMDMs, suggesting AMPKα1 is required for A769662-mediated CHOP phosphorylation at Ser30 (Figure 4J).

CHOP phosphorylation at Ser30 is required for AICAR-induced downregulation of CHOP

To determine if phosphorylation of CHOP at Ser30 is involved in the AICAR-induced downregulation of CHOP, we generated a phosphorylation-deficient Chop mutant (Ser30 is replaced by alanine, S30A) and a phosphomimetic Chop mutant (Ser30 is replaced by glutamic acid, S30E). RAW264.7 cells were transfected with full-length WT Chop (WT), phosphorylation-deficient Chop, S30A, or phosphomimetic Chop, S30E. 48 hours after the transfection, cells were treated with or without AICAR for 12 h.

The Chop-S30A mutant alone did not alter CHOP protein levels (Figure 4K). As expected, AICAR reduced CHOP levels in WT transfected macrophages, whereas AICAR did not alter CHOP protein levels in macrophages transfected with Chop-S30A (Figure 4K). Conversely, mutation of the phosphomimetic Chop-S30E variant markedly lowered the levels of CHOP compared with CHOP in WT transfected macrophages (Figure 4L). AICAR treatment did not cause further reduction of CHOP in cells transfected with the Chop-S30E mutant (Figure 4L). Taken together, these data indicate that phosphorylation of CHOP at Ser30 is required for AICAR-induced degradation of CHOP.

AMPK increases CHOP ubiquitination

Three major protein degradation pathways exist in mammalian cells: the proteasome, the lysosome, and the autophagosome. Cellular inhibitor of apoptosis protein-1 (cIAP1) is reported to promote CHOP degradation through the proteasome pathway in β-cells.³⁰ We next determined which pathway is involved in AMPK-mediated CHOP degradation in macrophages. To assess which pathway mediates CHOP degradation in macrophages, cells were treated with either the proteasome inhibitor MG-132, the lysosome inhibitor chloroquine (CQ), or the autophagy inhibitor 3-MA. As shown in Figure 5A, Neither CQ nor 3-MA altered CHOP levels. However, MG-132 led to significantly increased CHOP expression in macrophages, indicating a role for proteasome degradation. Consistent with this, MG-132 abated the AICAR-induced reduction of CHOP (Figure 5B). Taken together, these results support the hypothesis that CHOP degradation is mediated by the proteasome in macrophages.

Figure 5. AMPKα1 increases CHOP ubiquitination, and phosphorylation of CHOP Ser30 positively regulates CHOP ubiquitination.

A. RAW264.7 cells were treated with the proteasome inhibitor (MG-132, 1 μmol/L), lysosome inhibitor (chloroquine, CQ, 10 μmol/L), or autophagy inhibitor (3-Methyladenine, 3-MA, 10 mmol/L) for 12 h and CHOP levels were detected by Western blot analysis. n = 5. B. RAW264.7 cells were incubated with or without AICAR (1 mmol/L) and MG-132 (1 μmol/L) for 12 h, and then CHOP levels were detected by Western blot analysis. n = 5. C. RAW264.7 cells were pretreated with MG-132 (1 μmol/L) for 30 min and then treated with or without AICAR (1 mmol/L) for 6 h. CHOP was immunoprecipitated from lysed RAW264.7 cells with a specific antibody and analyzed by Western blot analysis with an anti-ubiquitin antibody. n = 4. D. CHOP was immunoprecipitated from WT or Ampkα1^−/− BMDMs with a specific antibody and analyzed by Western blot analysis with an anti-ubiquitin antibody. n = 4. E. RAW264.7 cells were transfected with full-length WT Chop (WT) or the indicated site-directed mutants, Chop S30A or Chop S30E. DDK was immunoprecipitated (IP) from lysed RAW264.7 cells with a specific antibody and analyzed by immunoblotting with an anti-ubiquitin antibody. n = 4. F. RAW264.7 cells were transfected with indicated site-directed mutant of Chop (S30E) or (S30A) for 48 h and then treated with CHX (5 μg/mL) for the indicated times, followed by Western blot analysis to detect DDK. n = 5. Results are mean ± SEM. *P < 0.05 vs control or Chop S30A. ^#P < 0.05 vs control plus AICAR.

We next determined if AMPK activation promotes CHOP ubiquitination. Macrophages were pretreated with MG-132 for 30 min prior to AICAR treatment. AICAR promoted CHOP ubiquitination with or without MG-132 (Figure 5C). In contrast, the levels of ubiquitinated CHOP in BMDMs of Ampkα1^−/− mice were markedly lower than their counterparts from WT mice (Figure 5D). Collectively, these data indicate that AMPK activation enhances CHOP ubiquitination.

Phosphorylation of CHOP at Ser30 positively regulates CHOP ubiquitination

We assayed if AMPK-mediated CHOP phosphorylation at Ser30 is required for CHOP ubiquitination. RAW264.7 cells were transfected with either full-length WT Chop (WT) or Chop mutants (Chop-S30A, Chop-S30E). Compared to WT Chop, phosphomimetic Chop-S30E showed elevated ubiquitination, whereas the phosphorylation-deficient Chop-S30A mutant exhibited reduced ubiquitination (Figure 5E). These data suggest that CHOP phosphorylation at Ser30 is important for CHOP ubiquitination and subsequent degradation.

To further examine the key role of Ser30 phosphorylation on CHOP stability, CHX pulse-chase experiments were done with both the S30E and S30A mutants of Chop in RAW264.7 cells. As shown in Figure 5F, the half-life of Chop-S30A (>8 hours) was twice as long as that of the Chop-S30E. These data further confirm that CHOP Ser30 phosphorylation leads to CHOP ubiquitination and degradation.

Ampkα1 deletion aggravates injury-induced neointimal disruption in Apoe^−/− mice

Given that CHOP plays an important role in promoting atherosclerotic instability¹⁵ and our above-mentioned results showed AMPKα1 down-regulates CHOP level in vitro, we reasoned that AMPKα1 activation could reduce injury-induced neointimal disruption by lowering CHOP in mice in vivo. To test the effects of AMPK-CHOP pathway in plaque stability, we generated the right carotid artery ligation plus cuff placement model, a well-characterized for studying injury-induced neointima stability.^{15, 23, 24} Four weeks after the ligation of the right carotid arteries of 9-week-old male Apoe^−/−, Apoe^−/−/Ampkα1^−/−, and Apoe^−/−/Ampkα2^−/− mice, cuffs were placed in the ligated right carotid arteries for 4 days before mice were sacrificed for further analysis. As shown in Online Figure V, serum triglycerides and total cholesterol were comparable among the three groups. After ligation and cuff placement, the right carotid arteries of most of the mice in all groups developed neointimal plaque and mean lesion areas were similar between the groups (Table 1), suggesting that macrophage AMPKα1 deletion did not impact the sizes of injury-induced neointimal lesions.

Table 1.

Characterization of injury-induced neointimal plaques in Apoe^−/−, Apoe^−/−Ampkα1^−/−, and Apoe^−/−/Ampkα2^−/− mice.

Mouse genotype (total number)	Neointimal plaque	Mean plaque area (×103 μm2)	Neointimal disruption with thrombus
			mural	occlusive	total
Apoe−/− (n=24)	21	15.74 ± 1.66	6 (25.0%)	7 (29.2%)	13 (54.2%)
Apoe−/−/AmpkαT−/− (n=21)	19	17.18 ± 2.43	4 (19.0%)	14 (66.7%)*	18 (85.7%)*
Apoe−/−/Ampkα2−/−(n=19)	15	16.34 ± 1.69	4 (21.1%)	6 (31.5%)	10 (52.6%)

Neointimal disruption are defined as plaques that show disruptions in the neointima with mural or occlusive thrombus. Data represented as the mean ± SEM.

^*P < 0.05 vs Apoe^−/− or Ampkα1^−/− mice.

Next, we assayed the roles of macrophage AMPKα1 in maintaining neointimal plaque stability. It is commonly believed that in neointimal lesions, both collagen and vascular smooth muscle cells (VSMC) promotes stabilization of the neointima whereas macrophage destabilize the plaques.^{31, 32} To this end, we measured the contents of collagen, VSMC, and macrophages in neointimal plaques. As depicted in Figure 6A, collagen contents, measured by picrosirius red staining, were significantly lower in the neointima of Apoe^−/−/Ampkα1^−/− mice than those in either the Apoe^−/− or Apoe^−/−/Ampkα2^−/− mice. In addition, vascular smooth muscle cell (VSMC) contents showed no difference among the three groups (Online Figure VI). On the country, the macrophage infiltration in neointimal lesion, detected by anti-CD68 immunofluorescence staining, was higher in Apoe^−/−/Ampkα1^−/− mice than those detected in either the Apoe^−/− or Apoe^−/−/Ampkα2^−/− mice (Figure 6B). Taken together, these results indicate that AMPKα1 deletion promotes instability of the neointimal lesions.

Figure 6. Ampkα1 deletion aggravates injury-induced neointimal disruption and promotes apoptosis in injury-induced neointimal lesion.

Male 9-week-old Apoe^−/−, Apoe^−/−/Ampkα1^−/−, and Apoe^−/−/Ampkα2^−/− mice were treated as described in the Methods. A. Left, picrosirius red staining viewed by polarized light. Scale bar, 100 μm. Right, quantitative analysis of collagen content. n = 6. B. Representative immunofluorescence staining of frozen cross sections of the intracuff carotid artery with a CD68 (red) antibody in Apoe^−/−, Apoe^−/−/Ampkα1^−/−, and Apoe^−/−/Ampkα2^−/− mice. Sections were counterstained with DAPI (blue). Scale bar, 100 μm. Quantification of the CD68-positive staining is shown on the right. n = 5. C. Representative immunofluorescence photomicrograph of the TUNEL assay. Nuclei were stained with DAPI (blue). Scale bar, 100 μm. The percentage of TUNEL-positive (green) cells is shown on the right. n = 5. D. Representative immunohistochemical staining of frozen cross sections of the intracuff carotid artery with a cleaved caspase 3 antibody in Apoe^−/−, Apoe^−/−/Ampkα1^−/−, and Apoe^−/−/Ampkα2^−/− mice. Sections were counterstained with hematoxylin. Scale bar, 50 μm. Quantification of the cleaved caspase 3-positive staining is shown on the right. n = 5. Lumen (L), neointima (NI), and media (M) in injured vessels were labeled. Results are mean ± SEM. *P < 0.05 vs Apoe^−/− mice.

It was important to assess if macrophage AMPKα1 deletion promoted neointimal disruptions in vivo. Disrupted neointima is defined as neointima with mural or occlusive thrombus.²⁴ Although incidence of neointimal disruption was similar between Apoe^−/− and Apoe^−/−/Ampkα2^−/− mice (54.2% vs 52.6%, respectively, Table 1), the incidence of disrupted neointima was significantly increased in the Apoe^−/−/Ampkα1^−/− mice compared with Apoe^−/− mice (54.2% vs 85.7%, respectively, Table 1), suggesting that AMPKα1 plays an essential and specific role in neointimal stability but not neointimal lesion size. Taken together, deletion of Ampkα1 but not Ampkα2 promotes the injury-induced neointimal disruption in Apoe^−/− mice.

To strengthen our conclusion, we developed another murine atherosclerotic model. Apoe^−/− and Apoe^−/−/Ampkα1^−/− mice were fed Western diet for 10 weeks and the stability of atherosclerotic plaques in mouse brachiocephalic arteries was determined. As shown in Online Figure VIIA, the collagen content in the brachiocephalic arteries of Apoe^−/−/Ampkα1^−/− mice was significantly reduced when compared to those in Apoe^−/− mice. In contrast, the necrotic core area of the brachiocephalic arteries of Apoe^−/−/Ampkα1^−/− mice was significantly greater than those in Apoe^−/− mice (Online Figure VIIB). However, mean lesion areas showed no significant difference between groups (Online Figure VIIB). Taken together, these results suggest that Ampkα1 deletion reduces the stability of brachiocephalic artery atherosclerotic plaques in Apoe^−/− mice in vivo.

Ampkα1 deletion promotes apoptosis in neointimal plaque

To detect if apoptosis was involved in the AMPKα1-dependent neointimal disruption, we next assessed apoptosis in the injury-induced neointimas of Apoe^−/−, Apoe^−/−/Ampkα1^−/−, and Apoe^−/−/Ampkα2^−/− mice. TUNEL staining showed that the number of TUNEL-positive cells was significantly higher in the atherosclerotic plaques of Apoe^−/−/Ampkα1^−/− mice compared with either Apoe^−/− or Apoe^−/−/Ampkα2^−/−

Macrophages are reported to be the primary apoptotic cell type involved in generating rupture-prone plaques of the human carotid artery.¹⁴ Since macrophages are ones of the main cellular components in injury-induced neointima, we next stained the ruptured plaque sections with both TUNEL and macrophage marker CD68. As shown in Online Figure VIII, the TUNEL-positive staining co-localized with the CD68 staining and overlap coefficient for TUNEL with CD68 is >90%, indicating that apoptosis mainly occurred in macrophages within the ruptured plaque. Furthermore, the levels of cleaved caspase 3 were significantly higher in the atherosclerotic plaques of Apoe^−/−/Ampkα1^−/− mice compared with those in the Apoe^−/− or Apoe^−/−/Ampkα2^−/− mice (Figure 6D). Consistently, cleaved caspase 3 in the atherosclerotic plaque of the brachiocephalic arteries of Apoe^−/−/Ampkα1^−/− mice was significantly higher than those in Apoe^−/− mice, suggesting increased apoptosis in atherosclerotic plaque of the brachiocephalic arteries of Apoe^−/−/Ampkα1^−/− mice (Online Figure IX). Overall, these results indicate that Ampkα1 deficiency triggers atherosclerotic plaque rupture via increased apoptosis.

Chop knockdown by siRNA ameliorates carotid artery neointimal disruption in Apoe^−/− mice

To determine whether AMPK inhibition leads to CHOP accumulation in vivo, we examined the CHOP levels in plaques of Apoe^−/−, Apoe^−/−/Ampkα1^−/− and Apoe^−/−/Ampkα2^−/− mice. CHOP expression was significantly increased in the injury-induced neointimal lesion of Apoe^−/−/Ampkα1^−/− mice compared to that in Apoe^−/− or Apoe^−/−/Ampkα2^−/− mice (Figure 7A). In line with this, the presence of p-CHOP (Ser30) in the atherosclerotic plaques of Apoe^−/−/Ampkα1^−/− mice was less than that observed in plaques in either the Apoe^−/− or Apoe^−/−/Ampkα2^−/− mice (Figure 7B), suggesting a negative association between CHOP and p-CHOP (Ser30) in vivo. Consistently, a negative association was also found between CHOP and p-CHOP (Ser30) in atherosclerotic plaque of the brachiocephalic arteries in Apoe^−/− and Apoe^−/−/Ampkα1^−/− mice fed Western diet for 10 weeks (Online Figure XA and B). Finally, we tested if genetic inhibition of Chop is able to prevent carotid artery atherosclerotic plaque rupture in Apoe^−/− mice. Ligation and cuff placement was described in Methods. Pluronic gel containing scramble or Chop siRNA was perivascularly applied to the ligated and cuffed carotid arteries for 7 days. As expected, Chop siRNA but not scramble siRNA abated CHOP protein expression in the ligated and cuffed carotid arteries (Figure 7C). However, compared with scramble siRNA-treated carotid arteries, the Chop siRNA markedly decreased the incidence of neointimal disruption in Apoe^−/− mice (Figure 7D and Online Table III). Compared with scramble siRNA-treated carotid arteries, the transfection of Chop-specific siRNA in treated-Apoe^−/− mice markedly increased collagen contents and suppressed macrophage accumulation and apoptosis (Online Figure XIA, B and Online Figure XII). As expected, in vivo knockdown of Chop by siRNA had no effect on VSMC contents (Online Figure XIC). Taken together, these results indicate that CHOP induces carotid artery neointimal disruption of Apoe^−/− mice.

Figure 7. In vivo knockdown of Chop by siRNA reduces carotid artery injury-induced neointimal disruption in Apoe^−/− mice.

A. Representative immunohistochemical staining of frozen cross sections of the intracuff carotid artery with a CHOP-specific antibody in Apoe^−/−, Apoe^−/−/Ampkα1^−/−, and Apoe^−/−/Ampkα2^−/− mice. Sections were counterstained with hematoxylin. Scale bar, 50 μm. Quantification of the CHOP-positive staining is shown on the right. n = 5. B. Representative immunohistochemical staining of frozen cross sections of the intracuff carotid artery with p-CHOP (Ser30) antibody in Apoe^−/−, Apoe^−/−/Ampkα1^−/−, and Apoe^−/−/Ampkα2^−/− mice. Sections were counterstained with hematoxylin. Scale bar, 50 μm. Quantification of the p-CHOP (Ser30)-positive staining is shown on the right. n = 5. Lumen (L), neointima (NI), and media (M) in injured vessels were labeled. C. Representative western blots of scramble siRNA and Chop siRNA knockdown of CHOP protein in carotid arteries 7 days after delivery. n = 6. Both scramble siRNA and Chop siRNA were perivascularly applied to ligated arteries for 7 days. D. A bar graph shows the incidence of neointimal disruption (n = 18 in the group of Apoe^−/− mice treated with scramble siRNA, n = 18 in the group of Apoe^−/− mice treated with Chop siRNA). E. Proposed scheme for accelerated injury-induced neointimal disruption in Apoe^−/−/Ampkα1^−/− mice. Ampkα1 deletion decreases CHOP phosphorylation, which leads to reduced CHOP ubiquitination/CHOP degradation. As a result, CHOP levels increase, which subsequently triggers increased macrophage apoptosis and accelerates injury-induced neointimal disruption.

DISCUSSION

In this study, we have demonstrated for the first time that AMPK activation promotes CHOP degradation via ubiquitination and that genetic deletion of Ampkα1 causes CHOP accumulation. This results in aberrant macrophage apoptosis and consequent injury-induced neointimal disruption in vivo. Mechanistically, AMPKα1 binds with CHOP and phosphorylates CHOP at Ser30. This phosphorylation triggers CHOP ubiquitination and consequent degradation by the proteasome. The principal findings of this study are: 1) AMPKα1 inhibition increases macrophage apoptosis; 2) AMPKα1 inhibition-induced macrophage apoptosis is CHOP-mediated; 3) AMPK activation reduces, and AMPK inhibition increases CHOP protein levels in macrophages; 4) AMPKα1 phosphorylates CHOP at Ser30; 5) CHOP phosphorylation at Ser30 initiates CHOP ubiquitination and subsequent CHOP degradation in the proteasome; 6) genetic inhibition of Ampkα1 promotes in vivo injury-induced neointimal disruption in mice via increase in CHOP accumulation, and 7) local inhibition of Chop by siRNA decreases the incidence of neointimal disruption in Apoe^−/− mice. Thus, we conclude that AMPKα1 activation might prevent injury-induced neointimal disruption by promoting proteasomal degradation of CHOP in macrophages (Figure 7E).

The most important finding of this study is that we have demonstrated for the first time that AMPKα1 activation suppresses injury-induced nerointima disruption by suppressing CHOP-dependent macrophage apoptosis. Compared with either Apoe^−/− or Apoe^−/−/Ampkα2^−/− mice, the incidence of neointimal disruption with thrombus was significantly increased in Apoe^−/−/Ampkα1^−/−. Macrophage apoptosis in the advanced lesion is one of the most important inducers of atherosclerotic plaque vulnerability.^{33, 34} Kolodgie et al. reported an increased number of apoptotic macrophages at sites of plaque disruption compared with those in intact fibrous caps.⁸ We are able to show that macrophage infiltration and macrophage apoptosis in injured carotid arteries was significantly higher in Apoe^−/−/Ampkα1^−/− mice compared to those in Apoe^−/− or Apoe^−/−/Ampkα2^−/− mice, that apoptosis-suppressing effects of AMPK in macrophages is important in maintaining the stability of injury-induced neointimas. These results are consistent with a published clinical study in which lesion macrophages are the primary apoptotic cell type involved in generating rupture-prone plaques of the human carotid artery.¹⁴

A key finding of this study is that AMPK promotes CHOP protein degradation by enhancing CHOP ubiquitination. First, we found that AMPK activation accelerated CHOP protein decay. Furthermore, this AMPK-accelerated CHOP degradation was abolished with Ampkα1 knockdown or genetic deletion. In addition, we found AMPKα1 phosphorylated CHOP Ser30. CHOP degradation is mediated by the proteasome pathway,³⁰ and we confirmed this in macrophages. Next, we addressed the question whether CHOP phosphorylation at Ser30 promoted its ubiquitination and then degradation. The phosphomimetic CHOP-S30E showed elevated levels of ubiquitination compared to WT CHOP, while phosphorylation-deficient mutant CHOP-S30A showed reduced ubiquitination compared to WT CHOP. This suggests that phosphorylation of CHOP at Ser30 positively regulates CHOP ubiquitination. Furthermore, CHOP expression was significantly higher in atherosclerotic plaques isolated from Apoe^−/−/Ampkα1^−/− mice compared to those isolated from Apoe^−/− or Apoe^−/−/Ampkα2^−/− mice. Transfection of Chop-specific siRNA markedly decreased the incidence of neointimal disruption in Apoe^−/− mice. This result strongly suggests that Ser30 phosphorylation of CHOP by AMPKα1 promotes CHOP ubiquitination and degradation. Importantly, CHOP accumulation promotes the disruption of the injury-induced carotid neointima in Apoe^−/− mice.

AMPK might also be important in maintaining atherosclerotic plaque stability. Although mouse atherosclerotic plaques are in general stable and there is no commonly accepted mouse models to study atherosclerotic plaque stability, the atherosclerotic plaques in brachiocephalic arteries is considered as a well characterized site for studying plaque stability.^{35, 36} Indeed, we found that the aortic lesions of brachiocephalic arteries of Apoe^−/−/Ampkα1^−/− mice exhibited more characteristic features including decreased collagen levels, increased necrotic cores, and cleaved caspase 3 when compared to those from Apoe^−/− mice. In the same model, CHOP expression was significantly increased in the atherosclerotic plaques of brachiocephalic arteries of Apoe^−/−/Ampkα1^−/− mice compared to those in Apoe^−/− mice. Conversely, the levels of p-CHOP (Ser30) in the atherosclerotic plaques of Apoe^−/−/Ampkα1^−/− mice were less than that in plaques of the Apoe^−/− mice, suggesting a negative association between CHOP and p-CHOP (Ser30) in vivo in Western diet-fed mouse atherosclerotic mice. Overall, these results strongly support Ampkα1 deficiency exacerbates atherosclerotic plaque vulnerability.

Atherosclerotic plaque rupture is a common pathogenic mechanism for coronary heart diseases and stroke. Our results strongly support the hypothesis that AMPK activation is a novel target in preventing atherosclerotic plaque rapture. Beside its beneficial effects on atherosclerotic plaque stability, AMPK activation inhibits the initiation and progression of atherosclerosis in vivo^{22, 37, 38} via its protective effects in endothelial cells³⁹ and vascular smooth muscle cells.⁴⁰ Consistently, metformin, a well characterized AMPK activator, has been demonstrated to effectively reduce adverse cardiovascular events and mortality in patients with type II diabetes.⁴¹

In summary, our study has demonstrated that Ampkα1 deficiency increases injury-induced neointimal disruption in Apoe^−/− mice, and AMPKα1 phosphorylates CHOP at Ser30 to promote CHOP ubiquitination and degradation. Our results indicate that AMPK may be a novel therapeutic target for promoting atherosclerotic plaque stability and the prevention of acute coronary heart disease and stroke.

Novelty and Significance.

What Is Known?

• Chop deletion leads to reduced macrophage apoptosis and inhibits plaque rupture in mice.

• AICAR (5-aminoimidazole-4-carboxamide ribonucleoside) reduces CHOP protein levels in several cell types.

What New Information Does This Article Contribute?

• Ampkα1 deficiency promotes macrophage apoptosis, CHOP accumulation, and injury-induced neointimal disruption.

• Inhibition of CHOP reduces Ampkα1 knockdown-elevated macrophage apoptosis.

• Phosphorylation of CHOP at Ser30 by AMPKα1 promotes CHOP ubiquitination and degradation in macrophages.

We show that Ampkα1 deficiency aggravates injury-induced neointimal disruption in Apoe^−/− mice. In macrophages AMPKα1 phosphorylates CHOP at Ser30 to promote its ubiquitination and degradation. Our results imply that AMPK may be a novel therapeutic target for promoting atherosclerotic plaque stability and the prevention of acute coronary heart disease and stroke.

Nonstandard Abbreviations and Acronyms

CHOP

C/EBP homologous protein

AMPK

AMP-activated protein kinase

ER

endoplasmic reticulum

TUNEL

terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling

BMDMs

bone marrow-derived macrophages

WT

wild type

PARP

poly(ADP-ribose) polymerase

AICAR

5-aminoimidazole-4-carboxamide ribonucleoside

CHX

cycloheximide

p38 MAPK

p38 MAP kinase

CK2

casein kinase 2

cIAP1

cellular inhibitor of apoptosis protein-1

CQ

chloroquine

DDK

DYKDDDDK tag

LDLr

low density lipoprotein receptor