Impaired Development of Atherosclerosis in Abcg1^−/− Apoe^−/− Mice

Characterization of Abcg1^−/−Apoe^−/− Mice

Endothelial cells of Abcg1^−/− mice fed a Western diet accumulate 7-ketocholesterol, a nonenzymatic product of cholesterol autoxidation.¹³ Abcg1^−/− mice also exhibit decreased endothelium-dependent vasodilatation and decreased endothelial nitric oxide synthase activity.¹³ Macrophages from Abcg1^−/− mice also exhibit increased expression of inflammatory genes and accumulate intracellular 7-ketocholesterol.^9,11 In addition, loss of ABCG1 from macrophages has been reported to result in increased secretion of ApoE protein,¹⁹ an antiatherosclerotic protein.²⁸

Consequently, to better understand the effect of loss of function of ABCG1 from all cell types, including macrophages and endothelial cells, and to remove any confounding effects that could arise from altered secretion of ApoE from macrophages, we generated Abcg1^−/− Apoe^−/− double-knockout (DKO) mice. Analysis of the plasma showed that, compared with Apoe^−/− mice, DKO mice had increased hemoglobin (14.00±0.67 versus 12.68±0.96; P<0.04; mean±SEM n=10) and hematocrit (40.42±1.94 versus 36.80±2.14; mean±SEM, n=10; P<0.03) values; however, lipid and lipoprotein levels and a broad array of hematology values did not significantly differ between the 2 genotypes (data not shown).

To accelerate the development of atherosclerosis, DKO and Apoe^−/− mice were challenged with a Western diet (21% fat and 0.2% cholesterol). After 16 weeks, analysis of plasma from DKO and Apoe^−/− mice indicated that there was no significant difference in lipid levels (Supplemental Table I) or the lipoprotein profile (Supplemental Figure IA). Lungs of the DKO, but not the Apoe^−/− mice, appeared white and stained positive with oil red O, especially in areas enriched in LacZ-expressing macrophages (Supplemental Figure II; data not shown). In addition, the red and white pulp of spleens of the DKO, but not Apoe^−/− mice, contained cells that were positive for both LacZ and oil red O (Supplemental Figure II). Thus, multiple tissues of the DKO mice exhibited evidence of neutral lipid accumulation.

Abcg1^−/−Apoe^−/− Mice Have Decreased Atherosclerotic Lesions Containing Increased Numbers of Apoptotic Macrophages

After 16 weeks on the Western diet, atherosclerotic lesions were determined both by en face analysis of the Sudan IV–stained descending aorta or after analysis of stained frozen sections (25 to 30 sections per mouse) of the aortic root.

The data show that Abcg1^−/− Apoe^−/− mice had significantly smaller lesions than Apoe^−/− mice in both the aortic root (Figure 1A and B, panel b versus a) and in the descending aorta (Figure 1D and E). The lesions of the DKO mice contained numerous LacZ-positive macrophages (Supplemental Figure III), thus excluding the possibility that DKO macrophages do not enter the subendothelial space. Calcified deposits in the lesions of the aortic root, identified after staining of sections with either von Kossa or oil red O, hematoxylin, and fast green, were significantly reduced in the Abcg1^−/− Apoe^−/− mice (Figure 1C; and Figure 1B, panel d versus c and b versus a), consistent with the smaller lesions in the DKO mice. Photomicrographs taken at a higher magnification illustrate that the calcium deposition occurs within the aortic lesions, adjacent to the medial layer (Supplemental Figure IV).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Atherosclerosis and lesion calcification are reduced in Abcg1^−/− Apoe^−/− compared with Apoe^−/− mice. DKO and Apoe^−/− mice (6 to 8 mice per group) were fed a Western diet for 16 weeks. A, Frozen sections (25 to 30 sections per mouse; n=6 mice per group) from the aortic root were stained with oil red O and counterstained with hematoxylin and fast green before lesion areas were determined, as described in the “Methods” section. Each point represents an individual mouse. **P<0.01. B, Representative oil red O–stained sections counterstained with hematoxylin and fast green (a and b) and adjacent sections stained with von Kossa (c and d). Calcification/calcium phosphate deposits are indicated by arrows in panels a and c. C, Quantification of lesion calcification. *P<0.05. D, Lesions in the descending aorta were identified by en face analysis and quantified as described in the “Methods” section (n=8 mice per group). Data represent mean±SEM. **P<0.01. E, Representative Sudan IV–stained aortas are shown from the 2 genotypes.

Loss of ABCG1 From Hematopoietic Cells Delays the Development of Atherosclerosis and Increases Apoptotic Macrophages in the Lesions, Independent of ApoE

To determine the relative importance of hematopoietic and nonhematopoietic Abcg1^−/− cells on the observed changes in atherosclerosis noted in the Abcg1^−/− Apoe^−/− mice (Figure 1), we performed bone marrow transplantation studies in which bone marrow from either Apoe^−/− or DKO mice was transplanted into recipient Apoe^−/− animals. After a 4-week recovery period, the mice were fed a Western diet for 12 weeks. An analysis of the lungs of the Apoe^−/− recipients indicated that mice that had been transplanted with DKO donor cells, but not those receiving Apoe^−/− cells, contained LacZ-positive cells and white patches consistent with lipid deposition in macrophages (data not shown).

Neither plasma lipid levels (Supplemental Table II) nor plasma lipoprotein profiles (Supplemental Figure IB) were significantly different between the 2 groups of recipient Apoe^−/− mice. Compared with wild-type mice, all transplanted mice contained elevated levels of very low-density lipoprotein and LDL and decreased levels of high-density lipoprotein independent of the genotype of the donor cells (Supplemental Figure IB; data not shown).

Quantification of the atherosclerotic lesions showed that they were significantly smaller in mice transplanted with DKO compared with Apoe^−/− donor bone marrow (Figure 2A; and Figure 2B, panel b versus a). Interestingly, and in agreement with the studies using whole-body DKO mice, calcification in the lesions of the aortic root was also significantly decreased in mice transplanted with Abcg1^−/− Apoe^−/− donor bone marrow (Figure 2C; and Figure 2B, panel d versus c and b versus a), consistent with smaller lesions in the latter mice.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Apoe^−/− mice lacking ABCG1 in hematopoietic cells have reduced atherosclerotic lesions. Apoe^−/− mice were transplanted with bone marrow from Apoe^−/− or DKO mice before being fed a Western diet for 12 weeks. All analyses were performed as described in Figure 1. A, C, and D, Lesion size (A) and calcification (C) in the aortic root sections (10 to 13 mice per group) and lesion size in the descending aorta (D) (13 to 16 mice per group) are shown, with each point representing 1 mouse. B, Representative sections from the aortic root after staining with oil red O and counterstaining with hematoxylin and fast green (a and b) or adjacent sections stained with von Kossa (c and d). Arrows identify calcium deposits. E, Representative Sudan IV–stained sections of the descending aorta. Data represent mean±SEM. **P<0.01 and ***P<0.001.

En face analysis of the descending aorta indicated a trend toward lower lesions in those mice receiving bone marrow from Abcg1^−/− Apoe^−/− mice, although the difference failed to reach statistical significance (Figure 2D and E). However, lesion coverage in the thoracic and abdominal sections, but not the proximal sections, was significantly smaller in mice transplanted with DKO cells (Supplemental Figure V), consistent with slower lesion progression in mice receiving DKO donor cells.

Increased Macrophage Apoptosis in Lesions of DKO Mice

After 16 weeks on the Western diet, the aortic root lesions of DKO mice contained significantly more TUNEL-positive cells, often present as multicell aggregates (Figure 3A). A similar difference was seen in the bone marrow transplantation studies in which we observed a 22-fold increase in TUNEL-positive cells in lesions of Apoe^−/− mice transplanted with DKO compared with Apoe^−/− donor cells (Figure 3C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. ABCG1 deficiency results in increased numbers of apoptotic macrophages in atherosclerotic lesions. A through D, The indicated whole-body knockout mice (A and B) or bone marrow–transplanted mice (C and D) were fed a Western diet, as described in the legends to Figure 1 and Figure 2. TUNEL- and 4′,6-diamidino-2-phenylindole–positive cells (green and blue, respectively) were determined in adjacent sections of the aortic roots (A and C) of the indicated mice. Aggregated TUNEL-positive cells are indicated by arrows. Graphs show the percentage of TUNEL-positive cells in the lesions. B and D, Adjacent frozen sections were also stained with either antibody to cleaved caspase 3 (green foci marked by arrows) or macrophages (red). The merged images are also shown (B and D). Data are representative of multiple stained sections (n=15 per section per mouse; 3 mice per genotype). Data are expressed as mean±SEM. ***P<0.001.

One of the late events in apoptosis involves cleavage of the precursor form of caspase 3 to form an active protease.²⁹ To identify cells undergoing apoptosis within the lesions of the aortic root, frozen sections from the aortic roots of mice were immunostained with antibodies to macrophages and to the cleaved form of caspase 3. An analysis of multiple stained sections indicated that lesions of Apoe^−/− mice had few active caspase 3–positive cells, whereas numerous active caspase 3–positive cells, often present as aggregates, were present in the lesions of DKO mice (Figure 3B) and in Apoe^−/− mice that were the recipients of the DKO bone marrow (Figure 3D). These tissue sections also stained positive for macrophages when costained with anti-Mac3 (Figure 3B, D). An analysis of the merged Figure showed that cleaved caspase 3–positive and anti-Mac3–positive cells colocalized in the lesions of mice lacking ABCG1 (Figure 3B and D), thus identifying the apoptotic cells as macrophages. Interestingly, cleaved caspase 3– or TUNEL-positive endothelial cells were never observed in any section, suggesting that loss of ABCG1 from endothelial cells did not result in accelerated apoptosis in vivo (data not shown).

Taken together, the data from studies with whole-body DKO mice and after bone marrow transplantation demonstrate that loss of ABCG1 from hematopoietic cells alone is sufficient to slow the progression of atherosclerotic lesions. This is associated with an increase in the number of apoptotic cells in the lesions and decreased calcification within the lesion. All these changes occur by mechanisms that are independent of ApoE.

Identification of Specific Oxysterols Accumulating in Abcg1^−/−Apoe^−/− Macrophages

The identification of specific sterols that accumulate within macrophages in atherosclerotic lesions is complicated by the inability to obtain sufficient numbers of cells. Consequently, we performed bronchoalveolar lavage on Apoe^−/− and Abcg1^−/− Apoe^−/− mice and recovered alveolar macrophages. Analyses of these samples using isotope dilution mass spectrometry identified a number of oxysterols, including 24-, 25-, and 27-hydroxycholesterols that accumulate in the Abcg1^−/− Apoe^−/− and Abcg1^−/− cells compared with wild-type or Apoe^−/− cells (Table). Also, compared with Apoe^−/− mice, 25- and 27-hydroxycholesterol levels are significantly increased in the brain specimens of the DKO mice (Table). Therefore, the increase in the levels of these enzymatically synthesized oxysterols is not limited to macrophages.

Abcg1^−/− Bone Marrow–Derived Macrophages Display a Proapoptotic Phenotype and an Altered Sensitivity to Oxysterols

The data of Figure 4A show that after 7 days in culture, Abcg1^−/− and Abcg1^−/− Apoe^−/− bone marrow–derived macrophages (BMDMs) exhibited a 3- to 6-fold increase in TUNEL-positive cells, compared with wild-type or Apoe^−/− cells. Although exposure of all these cells to oxLDL for 8 hours increased TUNEL staining, the highest levels of apoptosis/TUNEL staining were seen when cells lacked ABCG1 (Figure 4A), consistent with the proposal that ABCG1 is critical for limiting apoptosis in response to lipid loading. As expected, oxLDL treatment of wild-type, Apoe^−/−, Abcg1^−/−, or Abcg1^−/− Apoe^−/− BMDMs increased the expression of the Liver X Receptor (LXR) target genes Abca1 and Srebp1c and the antiapoptotic gene Aim (Supplemental Figure VIA-C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. Macrophages lacking ABCG1 exhibit increased apoptosis in response to oxLDL or oxysterols and enhanced induction of proapoptotic genes. BMDMs in quadruplicate were differentiated in L929-conditioned media containing 10% fetal bovine serum. After 7 days, the media were replaced with media containing 0.2% BSA with or without oxLDL, 50 μg/mL (A), or the indicated oxysterol, 10 μmol/L (B–D). After 8 hours, the number of TUNEL-positive apoptotic cells (total, >1000 cells per field; 6 fields per genotype) (A and B) or the mRNA levels of Bid and Bok (C and D) were determined. Data are expressed as mean±SEM and are representative of 2 experiments. In A, * indicates significantly different from controls that were incubated with buffer (P<0.001); # indicates significantly different from buffer-treated WT and Apoe^−/− cells (P<0.001). In B, # indicates significantly different from dimethylsulfoxide-(DMSO) treated Apoe^−/− cells (P<0.001) and $ indicates significantly different from DMSO-treated DKO cells (P<0.001). In C and D, bars with different letters (a, b, c, d) are significantly different from one another at P<0.001.

Based on the finding that specific oxysterols accumulate in alveolar macrophages and the brain specimens of Abcg1^−/− Apoe^−/− mice (Table), we next investigated whether cells lacking ABCG1 and/or ApoE are particularly sensitive to these same oxysterols. In the absence of added oxysterols, the number of BMDMs undergoing apoptosis was 4-fold greater in Abcg1^−/− Apoe^−/− compared with Apoe^−/− cells (Figure 4B). The addition of 10 μmol/L 7-ketocholesterol, 25-hydroxycholesterol, or 27-hydroxycholesterol increased the number of TUNEL-positive cells (Figure 4B). More important, the percentage of apoptotic cells was greatest after the addition of oxysterols to the DKO BMDMs (Figure 4B).

The increased sensitivity of cells lacking ABCG1 to oxysterol-induced apoptosis suggested that these cells might also exhibit altered expression of genes involved in apoptosis. Consequently, we performed a PCR-based screen to identify apoptotic genes that were altered after exposure of cells to 50 μg/mL oxLDL (data not shown). Confirmation of altered gene expression came from subsequent quantitative RT-PCR analysis that showed that incubation of cells with specific oxysterols increased the expression of Bid and Bok, 2 members of the Bcl-2 proapoptotic gene family (Figure 4C and D). More important, the expression of Bid and Bok was higher in DKO or Abcg1^−/− BMDMs compared with wild-type or Apoe^−/− cells (Figure 4C and D).