Identification of macrophage liver X receptors as inhibitors of atherosclerosis

Abstract

Recent studies have identified the liver X receptors (LXRα and LXRβ) as important regulators of cholesterol metabolism and transport. LXRs control transcription of genes critical to a range of biological functions including regulation of high density lipoprotein cholesterol metabolism, hepatic cholesterol catabolism, and intestinal sterol absorption. Although LXR activity has been proposed to be critical for physiologic lipid metabolism and transport, direct evidence linking LXR signaling pathways to the pathogenesis of cardiovascular disease has yet to be established. In this study bone marrow transplantations were used to selectively eliminate macrophage LXR expression in the context of murine models of atherosclerosis. Our results demonstrate that LXRs are endogenous inhibitors of atherogenesis. Additionally, elimination of LXR activity in bone marrow-derived cells mimics many aspects of Tangier disease, a human high density lipoprotein deficiency, including aberrant regulation of cholesterol transporter expression, lipid accumulation in macrophages, splenomegaly, and increased atherosclerosis. These results identify LXRs as targets for intervention in cardiovascular disease.

The contribution of elevated cholesterol levels to cardiovascular disease necessitates tight control over cholesterol synthesis and transport. Indeed, classical studies have described the negative feedback loop by which elevations in intracellular cholesterol repress transcription of genes involved in cholesterol synthesis (1). In contrast, recent studies suggest the existence of a positively acting cholesterol-responsive pathway regulated by the liver X receptors (LXRs). LXRα (NR1H3) and LXRβ (NR1H2) are members of the nuclear hormone receptor superfamily of transcription factors and are bound and activated by naturally occurring oxidized forms of cholesterol (2).

Analysis of LXR function using genetic knockouts and synthetic agonists has identified important roles for this family of transcription factors in the control of cholesterol and lipid metabolism including regulating the genes encoding ATP binding cassette (ABC) transporters involved in sterol absorption and cholesterol transport (3–6). In addition, LXRs directly or indirectly regulate a number of genes involved in cholesterol and fatty acid metabolism including the gene encoding the sterol regulatory binding element protein 1c, a master transcriptional regulator of fatty acid synthesis (5, 7).

Although originally described as regulators of entero-hepatic function (8), the identification of LXRs as direct regulators of ABC transporter gene expression in peripheral cells such as macrophages suggests a broad role for these receptors in whole-body cholesterol homeostasis (9). In particular, LXR directly regulates expression of ABCA1 and apolipoprotein E (ApoE) in nonhepatic tissues (4, 5, 10). Both ABCA1 and ApoE have important functions in cellular cholesterol efflux mechanisms that promote transfer of excess intracellular cholesterol to extracellular acceptors such as high density lipoprotein (HDL) particles, a process termed reverse cholesterol transport (9). The importance of reverse cholesterol transport is highlighted by Tangier disease, a rare genetic form of HDL deficiency caused by mutations in the gene encoding ABCA1. Tangier disease patients exhibit reductions in HDL levels, accumulate cholesterol in peripheral tissues, and have an increased risk for atherosclerosis (11–13).

Both LXRα and LXRβ are expressed in macrophages, a cell type that is required for the formation of atherosclerotic lesions and is particularly sensitive to perturbations in cholesterol homeostasis (14). To directly address the role of LXR activity in atherogenesis, we used bone marrow transplantation to create macrophage-selective knockouts in the context of established mouse models of atherosclerosis (15). These studies identify LXRs as antiatherogenic factors in vivo and directly link LXR activity to the pathogenesis of atherosclerosis.

Materials and Methods

Animals.

Homozygous ApoE^−/− mice, low density lipoprotein receptor mice (LDLR^−/−), and C57BL/6 mice were from The Jackson Laboratory. Homozygous LXRαβ^−/− and LXRαβ^+/+ mice in a mixed genetic background (C57BL/6 × 129Sv) were from a breeding colony established and maintained at X-Ceptor Therapeutics. Both the LXRαβ^−/− and LXRαβ^+/+ mice have been backcrossed to each other since their original creation in 1999.

Isolation of Mouse Peritoneal and Bone Marrow-Derived Macrophages.

Thioglycolate-elicited peritoneal macrophages were isolated from mice 4 days after peritoneal injection of thioglycolate broth media. Macrophages were stained with oil red O by rinsing adherent cells with 50% isopropanol for 1 min and then with 0.5% oil red O for 5 min. To isolate bone marrow-derived macrophages, femurs and tibias from LXRαβ^+/+ and LXRαβ^−/− mice were flushed with DMEM containing 10% FBS. After lysis of red blood cells, bone marrow cells were cultured in DMEM containing 30% L929 cell conditioned media and 10% lipid-depleted serum. RNA was isolated after 24 h of ligand treatment.

RNA Isolation and Analysis of Gene Expression by Quantitative Reverse Transcription–PCR.

Real-time PCR was performed by using a Perkin–Elmer/ABI 7700 Prism. Probes and primers were designed by using Primer Express (Applied Biosystems). Levels of cyclophilin were measured in all samples, and the results are presented as number of target transcripts per cyclophilin transcript.

Bone Marrow Transplantation.

Recipient ApoE^−/− and LDLR^−/− mice (10 weeks of age) were lethally irradiated with 900 rads (9 Gy) and transplanted with bone marrow cells (3 × 10⁶) from 6- to 8-week-old donor mice via tail vein injection. For transplantations into ApoE^−/− mice two independent bone marrow transplantations were carried out. Male donors with female recipients were used for the 8-week experiment (n = 7 for LXRαβ^+/+ → ApoE^−/− and LXRαβ^−/− → ApoE^−/− groups, n = 6 for ApoE^−/− → ApoE^−/−). In contrast, female donors with male recipients were used for the 16-week experiment (n = 6 for LXRαβ^+/+ → ApoE^−/− group, n = 10 for the LXRαβ^−/− → ApoE^−/− group, and n = 7 for the ApoE^−/− → ApoE^−/− group. Similarly, two independent transplantations into LDLR^−/− mice were evaluated. Male donors with female recipients were used for the 6-week experiment (n = 11 for LXRαβ^+/+ → LDLR^−/−, n = 12 for LXRαβ^−/− → LDLR^−/−, and n = 6 for LDLR^−/− → LDLR^−/−). In the 20-week experiment female donors with male recipients were used (n = 5 for LXRαβ^+/+ → LDLR^−/− and n = 8 for the LXRαβ^−/− → LDLR^−/−).

Lipid and Lipoprotein Analyses.

Free and esterified cholesterol, plasma cholesterol, and triglyceride levels were determined by enzymatic assays by using the supplier's protocols (Sigma).

Analysis of Atherosclerosis.

The extent of atherosclerosis in en face mouse aortas was quantitated by computer-assisted image analysis (16, 17). Immunohistochemical analysis of aortic root sections was performed by using procedures as described (17).

Statistical Analyses.

Results were analyzed by one-way ANOVA and/or Student's unpaired t test by using GraphPad (San Diego) prism.

Results

LXRs Are Antiatherogenic Factors in Vivo.

To investigate the role of LXR activity in macrophages, peritoneal macrophages isolated from LXRαβ^+/+ and LXRαβ^−/− mice were cultured in vitro. Microscopic examination revealed a striking accumulation of oil red O-positive droplets in LXRαβ^−/− macrophages, indicative of lipid accumulation (Fig. 1a). Little or no oil red O staining is observed in LXRαβ^+/+ macrophages (Fig. 1b). Quantitative analysis of cholesterol levels indicates that LXRαβ^−/− macrophages contained a 2.7-fold increase in free cholesterol and a 2.4-fold increase in total cholesterol compared with LXRαβ^+/+ (Fig. 1c). No difference in triglyceride content is observed between LXRαβ^+/+ and LXRαβ^−/− macrophages. The elevated cholesterol levels observed in LXRαβ^−/− macrophages are consistent with reports demonstrating that LXR regulates genes involved in cholesterol transport including ABCA1 and ApoE (5–7, 10).

Figure 1.

Cholesterol accumulation in LXRαβ^−/− macrophages. Oil red O-stained peritoneal macrophages from LXRαβ^−/− (a) and LXRαβ^+/+ (b) mice. (c) Total and free cellular cholesterol content of LXRαβ^+/+ (empty bars) and LXRαβ^−/− (filled bars) macrophages. * indicates significantly different from LXRαβ^+/+ controls (total cholesterol P < 0.002, free cholesterol P = 0.05).

To specifically examine the role of LXR in macrophages, we first used bone marrow-derived macrophages to further define the role of LXRs in the regulation of macrophage gene expression. As shown in Fig. 2, treatment of LXRαβ^+/+ macrophages with the LXR pan-agonist T0901317 (7) produces a 3- to 5-fold induction of the mRNAs encoding ABCA1, ABCG1 (a second ABC transporter implicated in cholesterol transport) (18), and ApoE. This ligand-dependent induction is enhanced when an agonist for LXR's heterodimeric partner the retinoid X receptor (RXR) (LG268) (19) is also included. Importantly, the response of all three genes to LXR and RXR agonists is completely eliminated in LXRαβ^−/− macrophages (Fig. 2).

Figure 2.

Regulation of macrophage gene expression by LXR. Bone marrow-derived macrophages isolated from LXRαβ^+/+ and LXRαβ^−/− mice were cultured for 24 h in the absence (empty bars) or presence of 1.0 μM T0901317 (LXR pan-agonist, gray bars), 1.0 μM LG268 (RXR agonist, striped bars), or 1.0 μM T0901317 + 1.0 μM LG268 (filled bars). Expression of ABCA1 (a), ABCG1 (b), and ApoE (c) were quantitated by real-time PCR and normalized to cyclophilin levels.

Because LXRs are active in bone marrow-derived macrophages, we generated a selective knockout of LXR activity in an atherogenic genetic background by transplanting bone marrow cells from LXRαβ^−/− mice into lethally irradiated ApoE knockout mice (ApoE^−/−). ApoE^−/− mice develop spontaneous hyperlipidemia and extensive atherosclerosis, and these mice are widely used to evaluate the impact of therapeutic agents on atherogenesis (15). Importantly, studies have shown that reconstitution of recipient ApoE^−/− mice with ApoE^+/+ bone marrow results in normalization of plasma cholesterol levels and protection from atherosclerosis caused by expression of ApoE in donor macrophages (15). Thus in this experimental paradigm only cells derived from bone marrow precursors, including macrophages, are LXR^−/−. In contrast, LXRs are present in the liver and intestine where they are known to be crucial for proper cholesterol and fatty acid metabolism (9).

Evaluation of plasma lipid levels in recipient mice biweekly after transplantation indicated that, as expected because of macrophage ApoE expression, plasma cholesterol levels in LXRαβ^+/+ → ApoE^−/− mice decline by week two and are reduced by 40% 16 weeks after transplantation (Fig. 3a and Table 1, which is published as supporting information on the PNAS web site, www.pnas.org). This result is consistent with published reports demonstrating that reconstitution with ApoE^+/+ macrophages provides sufficient macrophage-derived ApoE to promote clearance of plasma apolipoprotein B-containing lipoproteins (15). Interestingly, plasma cholesterol levels in LXRαβ^−/− → ApoE^−/− mice are significantly higher than the levels observed in LXRαβ^+/+ → ApoE^−/− mice (Fig. 3a and Table 1). Previous work has demonstrated that the ApoE gene is itself a target for LXR regulation in macrophages (10) and immunohistochemical analysis (Fig. 3 b–d) indicates that ApoE is expressed at lower levels in aortic root lesions from LXRαβ^−/− → ApoE^−/− mice when compared with LXRαβ^+/+ → ApoE^−/− lesions 16 weeks after transplantation. The decrease in ApoE staining suggests that the inability of LXRαβ^−/− bone marrow to normalize cholesterol levels in the ApoE^−/− mice may be caused, at least in part, by aberrant regulation of macrophage ApoE expression.

Figure 3.

Plasma cholesterol levels in ApoE^−/− mice after bone marrow transplantation. (a) Plasma cholesterol levels measured at 2-week intervals. (b–d) ApoE staining of aortic root sections from LXRαβ^+/+ → ApoE^−/− (b), LXRαβ^−/− → ApoE^−/− (c), and ApoE^−/− → ApoE^−/− (d) mice.

To evaluate the role of LXR in atherosclerotic lesion development, atherosclerosis was quantified by en face analysis of aortas (16) from recipient mice at 8 weeks and at 14 weeks after bone marrow transplantation (Fig. 4). Interestingly, aortas from LXRαβ^−/− → ApoE^−/− mice exhibit 3- to 8-fold more atherosclerotic lesions throughout the entire aorta compared with LXRαβ^+/+ → ApoE^−/− mice. As expected from previous studies (15), a substantial reduction of atherosclerosis is observed when LXRαβ^+/+ → ApoE^−/− mice are compared with ApoE^−/− → ApoE^−/− controls (2- to 4-fold). The results of the en face analysis were confirmed by quantitation of oil red O-stained aortic root sections 8 weeks after transplantation (see Fig. 7, which is published as supporting information on the PNAS web site). Immunohistochemical staining of aortic root sections with antibodies to the macrophage-specific marker Moma-2 demonstrated that the cellular composition of the lesions is not significantly different in the three transplant genotypes (see Fig. 8, which is published as supporting information on the PNAS web site).

Figure 4.

Increased atherosclerosis in LXRαβ^−/− → ApoE^−/− mice. (a–g) Sudan IV-stained aortic preparations showing atherosclerotic lesions (red) from ApoE^−/− recipient mice 8 weeks (a–c) and 16 weeks (e–g) after transplantation. (d and h) Quantitation of atherosclerosis in en face aortic preparations from LXRαβ^+/+ → ApoE^−/− (empty bar), LXRαβ^−/− → ApoE^−/− (filled bar), and ApoE^−/− → ApoE^−/− (striped bar) mice 8 weeks (d) and 16 weeks (h) after transplantation. Lesion areas are expressed as percent of total aortic surface area. * indicates significantly different from LXRαβ^+/+ → ApoE^−/− controls (P < 0.02). ** indicates significantly different from ApoE^−/− → ApoE^−/− control (P < 0.01).

Antiatherogenic Activity in LDLR^−/− Mice.

To further explore the antiatherogenic activity of LXRs we turned to a second animal model of atherosclerosis, the LDLR^−/− mouse. The LDLR^−/− model differs from the ApoE^−/− model in several respects that are particularly relevant to the analysis of macrophage LXR activity. First, in contrast to ApoE, there is no evidence that the LDLR gene is directly regulated by LXR. Second, transplantation of LDLR^+/+ bone marrow into LDLR^−/− recipients does not lower cholesterol levels or reduce the severity of atherosclerosis (15), allowing LXRαβ^+/+ and LXRαβ^−/− macrophages to be compared under conditions of equivalent cholesterol loads. Third, significant atherosclerosis is observed only in LDLR^−/− mice fed a high-fat diet (15). This requirement for both genetic and environmental risk factors to induce atherosclerosis in LDLR^−/− mice is thought to better reflect the interplay between genetic and lifestyle contributions in human cardiovascular disease.

To investigate the contribution of macrophage LXRs to atherosclerosis in LDLR^−/− mice, LDLR^−/− recipients were irradiated and reconstituted with LXRαβ^+/+-, LXRαβ^−/−-, and LDLR^−/−-derived bone marrow. One week after transplantation mice were placed on a Western diet (21% fat, 0.15% cholesterol) and maintained for an additional 5 weeks. As expected, the Western diet induced a significant time-dependent increase in serum cholesterol levels in all three groups of mice (Fig. 5a). In contrast to the ApoE^−/− study, however, there were no differences in cholesterol levels among the three experimental groups (Fig. 5a), nor were there differences in lipoprotein profiles analyzed by FPLC (see Table 2, which is published as supporting information on the PNAS web site). Importantly, in this study all of the donors and recipients have normal hepatic expression of ApoE, which accounts for more than 90% of circulating ApoE levels (15). Thus differences in macrophage ApoE expression among the LXRαβ^+/+ → LDLR^−/−, LXRαβ^−/− → LDLR^−/−, and LDLR^−/− → LDLR^−/− groups should not influence serum cholesterol levels as they would in the ApoE^−/− system. Although all of the animals in the study experienced a similar cholesterol burden, reconstitution with LXRαβ^−/− bone marrow (LXRαβ^−/− → LDLR^−/−) resulted in a significant increase in atherosclerosis compared with either LXRαβ^+/+ → LDLR^−/− (3.2-fold) or LDLR^−/− → LDLR^−/− mice (2.3-fold) (Fig. 5b and Fig. 9, which is published as supporting information on the PNAS web site).

Figure 5.

Cholesterol levels and atherosclerosis in LDLR^−/− mice 6 weeks after bone marrow transplantation. (a) Total plasma cholesterol levels measured at 2-week intervals. (b) Quantitation of atherosclerosis in en face aortic preparations from LXRαβ^+/+ → LDLR^−/− (empty bar), LXRαβ^−/− → LDLR^−/− (filled bar), and LDLR^−/− → LDLR^−/− (striped bar) mice. Lesion areas are expressed as percent of total aortic surface area. * indicates significantly different from LXRαβ^+/+ → LDLR^−/− controls (P < 0.002). ** indicates significantly different from LDLR^−/− → LDLR^−/− control (P < 0.001).

To further evaluate the long-term effects of macrophage LXR deficiency in vivo, the LDLR^−/− transplant described above was repeated with recipient mice maintained on a Western diet for 19 weeks. Quantitation of atherosclerosis once again demonstrated a 3-fold increase in lesion area when LXRαβ^−/− → LDLR^−/− mice are compared with LXRαβ^+/+ → LDLR^−/− animals (see Fig. 10, which is published as supporting information on the PNAS web site). Furthermore, morphological examination of mice at necropsy revealed that the spleens of LXRαβ^−/− → LDLR^−/− mice are dramatically enlarged (Fig. 6a; average spleen weight for LXRαβ^−/− → LDLR^−/− mice = 154 mg, average spleen weight for LXRαβ^+/+ → LDLR^−/− = 80 mg). Histologic analysis of frozen sections shows increased oil red O staining in spleens from LXRαβ^−/− → LDLR^−/− indicating excess lipid accumulation (Fig. 6 b and c). Splenomegaly was not observed in the 8-week experiment, suggesting that a prolonged exposure to high serum cholesterol levels is required to manifest this phenotype. The observation of splenomegaly resulting, at least in part, from lipid accumulation is reminiscent of Tangier disease patients who present with splenomegaly caused by the accumulation of lipid-rich macrophages (20). In conclusion, the results presented here identify LXRs as endogenous inhibitors of atherosclerosis and key determinants of macrophage lipid accumulation and foam cell formation.

Figure 6.

Atherosclerosis and splenomegaly in LXRαβ^−/− → LDLR^−/− mice 20 weeks after bone marrow transplantation. (a) Spleens from LXRαβ^+/+ → LDLR^−/− and LXRαβ^−/− → LDLR^−/− mice. (b and c) Oil red O staining of spleen sections from LXRαβ^+/+ → LDLR^−/− (b) and LXRαβ^−/− → LDLR^−/− mice (c) to visualize lipids. (Magnification: ×4.)

Discussion

The observation that LXRs mediate the sterol-dependent induction of reverse cholesterol transport in macrophages via direct regulation of target genes such as ABCA1 and ApoE (9) prompted an examination of the role of these receptors in atherosclerotic lesion development. To examine the relationship between LXR and atherosclerosis, bone marrow transplantations were used to selectively eliminate LXR activity in bone marrow-derived cells. This selective knockout allowed us to investigate the contribution of macrophage LXR activity to atherogenesis without complications arising from the loss of LXR functions in other tissues. Selective loss of bone marrow LXR activity increases atherosclerotic lesion development in two different mouse models, suggesting that these receptors function as endogenous inhibitors of atherogenesis. Thus, these observations provide a direct link between LXR activity and cardiovascular disease. Furthermore, the ability to regulate LXR activity by the direct binding of synthetic small molecules suggests that LXR ligands may prove useful in the treatment of atherosclerosis.

The increased atherosclerosis generated by transfer of LXR-deficient bone marrow into ApoE^−/− and LDLR^−/− mice highlights macrophages and fatty atherosclerotic lesions themselves as direct sites of action for potential LXR-based therapeutic agents. In support of this conclusion we and others have shown that LXRs regulate gene expression and cholesterol transport in macrophages cultured in vitro (6, 21). In contrast to the antiatherogenic activities of LXRs in macrophages, treatment of experimental animals with LXR agonists elevates serum triglycerides, most likely resulting from induction of sterol regulatory binding element protein 1c and other genes involved in fatty acid synthesis in the liver (7). Because hypertriglyceridemia can be a contributing factor for cardiovascular disease, tissue or cell type-specific LXR ligands will most likely be needed to maximize the therapeutic potential of LXR-based drugs. Thus, the identification of macrophages as critical targets for LXR action should facilitate the development of effective LXR-based therapies. Although macrophage function is absolutely required for atherosclerotic lesion development (14), the possibility that the absence of LXR activity in other bone marrow-derived cell types contributes to the observed phenotypes cannot yet be excluded.

The ability of LXRs to regulate expression of genes encoding proteins that participate in reverse cholesterol transport including ABCA1 and ApoE provides a straightforward explanation for the increase in atherosclerosis observed with LXRαβ^−/− macrophages. One would predict that elimination of LXR activity would at least partially mimic an ABCA1 deficiency. Consistent with this hypothesis, we and others have shown that LXRαβ^−/− macrophages accumulate cholesterol and exhibit defects in ABCA1 gene expression and reverse cholesterol transport (6, 21). Similarly Aiello et al. (22) have shown that ABCA1-deficient macrophages increase atherosclerosis when transplanted into ApoE^−/− mice. The dramatic splenomegaly observed when LXRαβ^−/− bone marrow is introduced in LDLR^−/− mice also supports a role for LXRs in cholesterol transport. Accumulation of lipid-enriched cells in the spleen, as we observed in LXRαβ^−/− → LDLR^−/− mice, is associated with Tangier disease and other forms of lipid storage diseases (20). Nevertheless, we cannot rule out the possibility that LXR-dependent pathways distinct from reverse cholesterol transport also impact atherosclerosis. For instance a recent report has implicated LXR in the regulation tumor necrosis factor α, an inflammatory cytokine (23). We have not, however, observed consistent differences in cytokine levels when macrophages from LXR^+/+ and LXR^−/− are compared. Analysis of quadruple LXRαβ^−/−/ABCA1^−/−/ApoE^−/− mice will be needed to determine the contribution of ABCA1 regulation to the antiatherogenic activity of LXR.

Recent studies have associated two other nuclear receptors, RXR and the peroxisome proliferator-activated receptor γ (PPARγ), with the pathogenesis of atherosclerosis. Treatment of atherogenic mouse models with RXR or PPARγ agonists decreases atherosclerotic lesions (21, 24), and both RXR and PPARγ can impinge on LXR-regulated pathways. Like many other nuclear receptors, LXRs bind to DNA and activate transcription as heterodimers with RXR. RXR-LXR heterodimers are known to respond to agonists for both receptors, and not surprisingly RXR agonists mimic many of the effects of LXR activators including induction of ABCA1 and reverse cholesterol transport in macrophages (5, 21). Similarly, activation of PPARγ leads to a direct increase in the expression of LXRα via a PPARγ binding site in the LXRα promoter (25, 26). Consistent with our results demonstrating the antiatherogenic effects of LXR, transplantation of PPARγ^−/− bone marrow into LDLR^−/− mice also increases atherosclerosis (25). These observations suggest that LXR is downstream of PPARγ with regard to the antiatherogenic effects of PPARγ ligands.

The identification of LXRs as antiatherogenic factors suggests that synthetic LXR ligands may further increase the receptors' antiatherogenic potential by stimulating reverse cholesterol transport, particularly in macrophages. Importantly, current drug treatment for cardiovascular disease and hypercholesterolemia generally use statins to decrease LDL cholesterol by enzymatic inhibition of 3-hydroxy-3-methylglutaryl CoA reductase in the liver. Thus the combination of statins with mechanistically distinct LXR-based drugs provides an exciting opportunity for synergy, particularly in patients who do not fully respond to mono-therapy with statins alone.

Abbreviations

LXR

liver X receptor

ABC

ATP binding cassette

ApoE

apolipoprotein E

LDLR

low density lipoprotein receptor

RXR

retinoid X receptor

PPARγ

peroxisome proliferator-activated receptor γ