Non-redundant roles for LXRα and LXRβ in atherosclerosis susceptibility in low density lipoprotein receptor knockout mice

The liver X receptors LXRα and LXRβ play critical roles in maintaining lipid homeostasis by functioning as transcription factors that regulate genetic networks controlling the transport, catabolism, and excretion of cholesterol. The studies described in this report examine the individual anti-atherogenic activity of LXRα and LXRβ and determine the ability of each subtype to mediate the biological response to LXR agonists. Utilizing individual knockouts of LXRα and LXRβ in the Ldlr−/− background, we demonstrate that LXRα has a dominant role in limiting atherosclerosis in vivo. Functional studies in macrophages indicate that LXRα is required for a robust response to LXR ligands, whereas LXRβ functions more strongly as a repressor. Furthermore, selective knockout of LXRα in hematopoietic cells and rescue experiments indicate that the anti-atherogenic activity of this LXR subtype is not restricted to macrophages. These studies indicate that LXRα plays a selective role in limiting atherosclerosis in response to hyperlipidemia.


Bone marrow transplantation
Bone marrow transplantations were carried as previously described ( 8,20 ). After transplant, animals were allowed to recover on a standard diet (Lab Diet 5001) for 4 weeks before being fed the Western diet for an additional 8 weeks.

Quantitation of atherosclerosis
Atherosclerosis in root sections and en face preparations were quantitated as previously described ( 8,20 ).

Lipid and lipoprotein analyses
Plasma total cholesterol and triglyceride levels were determined by colorimetric enzymatic assays that were adapted to 96well plate as previously described ( 8 ). Terminal plasma samples from mice fed the Western diet for 20 weeks were pooled and fractionated by fast-protein liquid chromatography (FPLC) gel fi ltration on Superose 6 columns. Triglyceride and cholesterol concentrations of each fraction were determined as above.

Measurement of TNF ␣ plasma levels
To measure tumor necrosis factor-␣ (TNF ␣ ) protein levels, mice of the appropriate genotypes were fed the Western diets for 2 weeks and then given a single injection of LPS (25 g). Plasma TNF ␣ levels were determined by ELISA (R and D Systems) 1.5 h after the LPS treatment.

Quantitative real-time PCR
Quantitative analysis of gene expression was carried out as previously described ( 8 ).

Statistical analyses
Results were analyzed by one-way ANOVA (ANOVA) or Student's unpaired t -test using GraphPad Prism (GraphPad Software, Inc.).

Increased atherosclerosis in Lxr
To defi ne the anti-atherogenic activity of the individual LXR subtypes, single LXR ␣ and LXR ␤ knockouts were crossed into the Ldlr Ϫ / Ϫ background to create Ldlr and Ldlr Ϫ / Ϫ / Lxr ␤ Ϫ / Ϫ double knockout strains. Animals from the three strains were placed on a Western diet (21% fat, 0.15% cholesterol), and plasma lipid levels were measured biweekly for 20 weeks ( Fig. 1 ). Genetic deletion of LXR ␣ in the Ldlr Ϫ / Ϫ background resulted in a dramatic decrease in topoietic cells is necessary for the anti-atherogenic activity of LXR agonists ( 8 ), implicating macrophages as a critical site of action for the LXRs. Each LXR subtype demonstrates a unique but overlapping expression pattern. LXR ␣ is highly expressed in cell types that modulate cholesterol and fatty acid synthesis, including the liver and intestine, whereas LXR ␤ is ubiquitously expressed (10)(11)(12). In the liver, LXR ␣ is the predominant subtype expressed, and studies indicate that LXR ␣ controls fatty acid synthesis via regulation of the gene encoding the sterol-regulatory element binding protein 1c (SREBP1c) ( 13,14 ). Additionally LXR ␣ modulates cholesterol homeostasis by controlling expression of cholesterol 7 ␣ -hydroxylase ( 11 ), the rate-limiting enzyme in the conversion of cholesterol to bile acids, and the cholesterol transporters ABCG5 and ABCG8 ( 15 ). In other tissues, including the intestine and macrophages, both subtypes are coexpressed and appear competent to regulate LXR target genes ( 4,16,17 ).
The ability to induce reverse cholesterol transport spurred great interest in the LXRs as drug targets for cardiovascular disease ( 13,14,18 ). Increases in plasma triglycerides and cholesterol observed in animals treated with synthetic LXR agonists, however, has raised concerns regarding the therapeutic potential of LXR-based drugs ( 18 ). As described above, the effects on lipid levels in animals treated with LXR agonists appear to be selectively mediated by LXR ␣ in the liver ( 10,15 ). In contrast, treatment with an LXR agonist reduced atherosclerosis in apoE Ϫ / Ϫ / Lxr ␣ Ϫ / Ϫ mice ( 19 ), suggesting that LXR ␤ alone is suffi cient to mediate the anti-atherogenic functions of LXR activation. Not surprisingly, several investigators have suggested that LXR ␤ -selective activators may have an optimized therapeutic potential ( 10,12 ). To our knowledge, the anti-atherogenic activity of such selective modulators has not yet been described.
Although the case for LXR ␤ -selective agonists has been put forward, the individual anti-atherogenic potential of LXR ␣ and LXR ␤ has not yet been addressed. To examine the anti-atherogenic activity of each LXR subtype in vivo, genetic knockouts of LXR ␣ or LXR ␤ were crossed into the LDL receptor knockout mouse ( Ldlr Ϫ / Ϫ ) background. Analysis of atherosclerosis in these strains indicates that LXR ␣ plays a dominant role in limiting disease in response to elevated lipid levels and that this subtype is required for the full anti-atherogenic activity of LXR agonists. Experiments that selectively rescue LXR ␣ function in bone marrow-derived or nonbone marrow-derived cells suggest that LXR activity may be required in cell types other than macrophages to limit diet-induced cardiovascular disease.

Animals and diets
All animal experiments were approved by the Exelixis Institutional Animal Care and Research Advisory Committee. Ldlr when normalized to the lesion area (data not shown). To extend the analysis to additional areas beyond the aortic root, en face quantitation was carried out ( Fig. 2B ). Analysis of the arch and thoracic regions indicates there is an 80% increase in Ldlr with the most dramatic differences distal to the aortic arch ( Fig. 2C ). Deleting LXR ␤ , however, had no effect. A signifi cant increase in atherosclerosis is also selectively detected in Ldlr

LXR ␣ is required for the anti-atherogenic activity of LXR agonists Synthetic LXR agonists have been shown to decrease atherosclerosis and promote the regression of lesions in Ldlr
Ϫ / Ϫ mice ( 7,8 ). To explore the contribution of each LXR subtype to the anti-atherogenic activity of LXR agonists, mice of the appropriate genotypes were exposed to the Western diet for 12 weeks and concurrently treated plasma triglyceride levels at all time points, confi rming previous data identifying this receptor subtype as the principal regulator of fatty acid synthesis ( 10, 12 ) ( Fig. 1A ). Quantitation of atherosclerosis by serial section analysis of the aortic root ( Fig. 2A ) demonstrated that deletion of LXR ␣ signifi cantly increases atherosclerosis by 25%. There was also a trend toward an increase in Ldlr Ϫ / Ϫ / Lxr ␤ Ϫ / Ϫ animals that did not reach statistical signifi cance ( P = 0.09). Immunohistochemical staining did not detect signifi cant differences in macrophage or collagen content  alone can mediate an anti-atherogenic response to an LXR agonist. Once again, however, en face analysis uncovered a large increase in atherosclerosis in Ldlr Ϫ / Ϫ / Lxr ␣ Ϫ / Ϫ animals that was unaffected by treatment with T0901317 ( Fig.  3D ). Thus when atherosclerosis is measured throughout the aorta, LXR ␣ is necessary for the anti-atherogenic activity of LXR agonists. In contrast, even at a lower dose, the LXR agonist remained effective in animals lacking LXR ␤ ( Fig. 3D ).

LXR ␣ selective regulation of gene expression in macrophages
We previously demonstrated that macrophages are an important site of action for the anti-atherogenic activity of LXR agonists ( 8,20 ). Therefore to explore a mechanistic basis for the selective effect of LXR ␣ deletion on atherosclerosis, the expression of LXR target genes was examined in bone marrow-derived macrophages isolated from As shown in Fig. 4A and 4B , the ability of T0901317 to induce expression of ABCA1 and ABCG1 was selectively impaired in Lxr ␣ Ϫ / Ϫ cells (compare solid black lines to dotted lines). These results indicate that LXR ␣ is required to maximally induce genes involved in reverse cholesterol transport. The subtype selective regulation of SREBP1c in response to agonist treatment is somewhat different with the LXR agonist T0901317. As expected from previous studies, plasma triglycerides were signifi cantly lower in Ldlr Ϫ / Ϫ / Lxr ␣ Ϫ / Ϫ mice, regardless of agonist treatment, while T0901317 elevated triglyceride levels in strains expressing LXR ␣ ( Fig. 3A ). Importantly, Ldlr Ϫ / Ϫ / Lxr ␤ Ϫ / Ϫ mice treated with 10 mg/kg T0901317 rapidly developed massive hypertriglyceridemia (data not shown) that could not be accurately measured, and 3 out of 10 mice died within 4 weeks of initiating treatment. At week 4, the dose for this group was dropped to 3 mg/kg, and the rest of the animals survived until study completion. Nevertheless, given the small number of animals that remained in this group, the conclusions regarding the effects of T0901317 on Ldlr Ϫ / Ϫ / Lxr ␤ Ϫ / Ϫ mice should be interpreted with caution. The LXR agonist-dependent hypertriglyceridemia observed in these animals suggests, at least in the Ldlr Ϫ / Ϫ background, that LXR ␤ partially represses the ability of LXR ␣ to regulate triglyceride levels. Little or no effect of T0901317 on total plasma cholesterol levels was detected ( Fig. 3B ). Consistent with previous studies ( 7,8,22 ), treatment of Ldlr Ϫ / Ϫ mice with T0901317 signifi cantly reduced atherosclerosis when quantitated by root section ( Fig. 3C ) or en face analysis ( Fig. 3D ). T0901317 also signifi cantly reduced atherosclerosis at the aortic root in the absence of either of LXR ␣ or LXR ␤ ( Fig. 3C ), indicating that each LXR subtype ability of LXRs to regulate reverse cholesterol transport and infl ammation in these cells is thought to underlie their anti-atherogenic activities ( 9 ). To determine if it is simply the absence of LXR ␣ in the hematopoietic system that lead to the increased atherosclerosis observed in the Ldlr Ϫ / Ϫ background, bone marrow transplantation experiments were used to either selectively eliminate or selectively express LXR ␣ in hematopoietic cells ( Fig. 5 ). As expected, transplantation of Ldlr Ϫ / Ϫ mice with bone marrow from Ldlr Ϫ / Ϫ / Lxr ␣ Ϫ / Ϫ animals (hematopoietic system knockout) increased atherosclerosis relative to controls (in Fig. 5 Fig. 5 , compare groups 4 and 5; P = 0.0001) but not to the levels measured in Ldlr Ϫ / Ϫ controls (in Fig. 5 , compare groups 1 and 5; P = 0.01). Taken together, the results of the bone marrow transplantation experiments indicate that LXR ␣ function in cell types other than macrophages may contribute to the antiatherogenic activity of this receptor.
( Fig. 4C ). Deletion of LXR ␣ has only a small effect on the agonist-dependent induction of this target gene, whereas deletion of LXR ␤ actually allows a super-induction, suggesting that LXR ␤ functions to repress the ability of LXR ␣ to induce SREBP1c. Nuclear receptors are also known to inhibit gene expression. In the absence of activating ligands LXR can repress transcription by recruiting corepressors to promoters that contain LXR binding sites ( 21 ). Interestingly, when the mRNA levels of ABCA1, ABCG1, and SREBP1c are examined in the absence of T0901317 ( Fig. 4D-F ), deletion of LXR ␣ has no effect on the basal/ligand-independent levels of the three target genes. On the other hand, all three mRNAs are increased in Lxr ␤ Ϫ / Ϫ and Lxr ␣ Ϫ / Ϫ / Lxr ␤ Ϫ / Ϫ cells, indicating that LXR ␤ plays an important role in repressing target gene expression in the absence of activating signals. LXRs can also inhibit pro-infl ammatory gene expression in the presence of agonists by inhibiting the activity of nuclear factor kappa ␤ (NF ␤ ) ( 23 ). We did not, however, observe LXR subtype-specifi c differences in the ability of T0901317 to inhibit expression of monocycte chemotactic protein 1 (MCP-1), TNF ␣ , or inducible nitric oxide synthase (iNOS) in lipopolysaccharide stimulated bone marrow derived macrophages (data not shown).

LXR ␣ functions in bone marrow-and nonbone marrow-derived cells
Macrophages as well as other immune cells play a critical role in the development of atherosclerosis ( 2 ), and the

Ldlr
Ϫ / Ϫ / Lxr ␣ Ϫ / Ϫ mice (i.e., LXR ␤ alone can mediate an anti-atherogenic response). On the other hand, en face analysis of the aortic arch and thoracic region indicated that the agonist had no effect in the absence of LXR ␣ . Additionally, the en face analysis uncovered a large increase in atherosclerosis distal to the aortic arch region in these animals. It is not unusual to see differences in atherosclerosis measured by these two methods. Serial sections provide cross-sectional information from a single area whereas en face analysis measures the surface area of the entire segment covered by lesions. In mouse models of atherosclerosis, lesions fi rst develop in the aortic root and arch areas and then progress distally down the segment as animals age ( 25 ). The large increase in atherosclerosis observed in Lxr ␣ Ϫ / Ϫ animals suggests that LXR ␣ may selectively infl uence the kinetics of early lesion initiation and/or disease progression. After exposure to the Western diet for 20 weeks, such an acceleration of lesion progression may be easily visible at areas distal to the arch while less apparent at the older, more advanced areas measured at the aortic root. Zhu et al. ( 26 ) have also recently observed that LXR ␣ is expressed at higher levels in the thoracic aorta compared with the arch region, raising the possibility that this subtype plays a unique or quantitatively dominant role in the thoracic region. LXR ␤ expression in these regions was not examined.
The function of LXR ␤ is more diffi cult to discern from these studies. After 12 weeks on the Western diet, there was a general failure of the Ldlr Ϫ / Ϫ / Lxr ␤ Ϫ / Ϫ mice to thrive.
Although there was no obvious pathology at necropsy 8 weeks later, lymphoid hyperplasia has been observed in older Lxr ␤ Ϫ / Ϫ mice ( 27 ), which, when coupled with the Western diet, may have contributed to the poor health of these animals. Interestingly, in Lxr ␤ Ϫ / Ϫ macrophages, the basal levels of ABCA1 and ABCG1 were elevated, suggesting a role for this receptor in limiting reverse cholesterol transport in the absence of activating signals. Additionally, treatment of Ldlr Ϫ / Ϫ / Lxr ␤ Ϫ / Ϫ mice with T0901317 produced a massive increase in plasma triglycerides, indicating that LXR ␤ functions to limit the ability of LXR ␣ to induce fatty acid synthesis (at least under hyperlipidemic conditions). A similar inhibitory effect of LXR ␤ on the agonist-dependent induction of SREBP1c was detected in macrophages. These results suggest opposing functions for the two subtypes, with LXR ␣ functioning to drive transcription in the presence of agonists and LXR ␤ functioning to repress transcription when ligands are active.

CONCLUSION
The ability to regulate reverse cholesterol transport and infl ammation at the site of atherosclerotic lesions has made macrophages the focus of LXR studies ( 8,20 ). To address the sites of LXR ␣ -dependent anti-atherogenic activity, we used bone marrow transplantations to selectively eliminate or selectively express LXR ␣ in the hematopoietic system. The introduction of LXR ␣ positive bone marrow into Ldlr Ϫ / Ϫ / Lxr ␣ Ϫ / Ϫ mice did partially rescue the athero-

DISCUSSION
The ability of LXRs to regulate reverse cholesterol transport in macrophages has stimulated great interest in understanding the anti-atherogenic activity of the two LXR subtypes. Analysis of single genetic deletions of each receptor in the Ldlr Ϫ / Ϫ background reveals a dominant effect for LXR ␣ in the control of cardiovascular disease. Quantitation of atherosclerosis in animals exposed to a Western diet uncovered an 80% increase in Lxr ␣ Ϫ / Ϫ animals, with dramatic increases in the lesion area in regions distal to the aortic arch. Little or no effect of deleting LXR ␤ was observed. Our results are consistent with reports by Bradley et al. ( 19 ) and Teupser et al. ( 24 ) that detected increased atherosclerosis in apoE Ϫ / Ϫ / Lxr ␣ Ϫ / Ϫ mice and decreased atherosclerosis in Ldlr Ϫ / Ϫ mice over expressing LXR ␣ , respectively. Neither of these studies, however, directly compared the two LXR subtypes.
Functional experiments using bone marrow-derived macrophages demonstrate that LXR ␣ is quantitatively a stronger transcriptional activator of the ABCA1 and ABCG1 promoters in response to agonists than is LXR ␤ . The subtype selective response to agonist treatment measured in macrophages suggests an impaired ability of Lxr ␣ Ϫ / Ϫ macrophages to induce gene expression and reverse cholesterol transport in response to elevated cholesterol levels. In support of the gene expression data, the anti-atherogenic activity of the agonist T0901317 is muted in the absence of LXR ␣ . LXR ␤ , on the other hand, appears to be more responsible for setting the basal level of macrophage ABCA1 and ABCG1 expression by limiting transcription in the absence of activating signals. The effect of the LXR agonist T0901317 on atherosclerosis in Ldlr Ϫ / Ϫ / Lxr ␣ Ϫ / Ϫ mice differed depending on the method of analysis used. Serial section of the aortic root indicated that T0901317 reduced atherosclerosis in Fig. 5. LXR ␣ activity is required in bone marrow derived and nonbone marrow cells. Recipient mice (as defi ned in the fi gure) were irradiated and reconstituted with bone marrow from mice of the appropriate genotype. Four weeks after recovery from the transplant, mice were exposed to the Western diet, and atherosclerosis was quantitated by en face analysis after an additional 8 weeks. genic phenotype. The level of atherosclerosis in these "rescued" mice, however, was still signifi cantly greater than the level of observed in mice expressing LXR ␣ throughout the body. The mice used in the transplantation studies were irradiated and reconstituted with bone marrow at 8 weeks of age. Thus we cannot rule out the possibility that the absence of LXR ␣ in the hematopoietic system prior to irradiation contributed to increased atherosclerosis observed in the rescued mice relative to the controls expressing LXR ␣ everywhere. Nevertheless, the results of this study also suggest the possibility of an important site of action for LXR ␣ that may be distinct from macrophages. Obvious candidates for such nonbone marrow-derived cells are vascular endothelial cells and smooth muscle cells, which are also components of atherosclerotic lesions. Additionally, the liver may be a critical site for the anti-atherogenic activity of LXR ␣ . Analysis of plasma lipids in Ldlr Ϫ / Ϫ / Lxr ␣ Ϫ / Ϫ mice revealed relatively high levels of cholesterol and relatively low levels of tri glycerides, raising the possibility that an aberrant pro-atherogenic lipoprotein particle may circulate in these animals. LXR ␣ is also a critical regulator of apoE expression ( 5 ), and studies have suggested that apoE-defi cient lipoproteins enhance macrophage foam cell formation ( 28 ). Finally, the elimination of hepatic cholesterol via excretion into the bile and catabolism to bile acids is controlled selectively by LXR ␣ ( 11,15 ). Ultimately additional cell-specifi c knockouts will be required to further defi ne the critical sites of LXR ␣ activity.