Liver X Receptors (LXR) as Therapeutic Targets in Dyslipidemia

Transcription

1 REVIEW Liver X Receptors (LXR) as Therapeutic Targets in Dyslipidemia Jerzy Bełtowski Department of Pathophysiology, Medical University, Lublin, Poland Keywords Atherosclerosis; Lipogenesis; Liver X receptor; Oxysterols; Reverse cholesterol transport. Correspondence Jerzy Bełtowski, Department of Pathophysiology, Medical University, ul. Jaczewskiego 8, Lublin, Poland. Tel: ; Fax: ; jerzy.beltowski@am.lublin.pl doi: /j x Liver X receptors (LXR) α and β belong to a family of nuclear receptors which form heterodimers with the retinoid X receptor (RXR) and, upon ligand binding, stimulate the expression of target genes. LXR were initially described as orphan receptors and later oxidized cholesterol derivatives (oxysterols) were identified as their natural ligands. In addition, several synthetic LXR agonists such as T and GW3965 were synthesized. Oxysterols are formed in amounts proportional to cholesterol content in the cell and therefore LXR operate as cholesterol sensors which protect from cholesterol overload by inhibiting intestinal cholesterol absorption, stimulating cholesterol efflux from cells to high-density lipoproteins (HDL), its transport to the liver, conversion to bile acids, and biliary excretion. In addition, LXR agonists activate fatty acid synthesis by stimulating the expression of a lipogenic transcription factor, sterol regulatory element-binding protein-1c (SREBP-1c), leading to the elevation of plasma triglycerides and liver steatosis. Lipogenic effect seems is the most important negative feature of LXR agonists considered as potential hypolidemic drugs. Some of currently used drugs also affect LXR signaling. For example, statins may impair LXR signaling by inhibiting oxysterol synthesis, whereas fibrates and thiazolidinediones increase LXR expression and activity. Introduction Liver X receptors (LXR) belong to a family of nuclear receptors which bind to the regulatory region of target genes and upon ligand binding regulate their transcription. LXR was initially isolated from a human liver cdna library as an orphan receptor since its natural ligand was unknown [1]. Later, oxidized cholesterol derivatives (oxysterols) were identified as specific LXR agonists. Studies performed during the last decade suggest that LXR are cholesterol sensors which, in response to excess cholesterol, stimulate its transport to the liver and biliary excretion. Therefore, stimulation of LXR is a potentially useful therapeutic option for patients with dyslipidemia. Herein, I briefly summarize the current knowledge about structure and function of LXR, their ligands, and effect of LXR agonists on lipid metabolism at molecular and whole-organism levels. Apart from discussing perspectives of application of LXR agonists as hypolipidemic drugs, I also characterize the effect of currently used therapies on LXR signaling system. Liver X Receptors An Overview There are two LXR isoforms in mammals: LXRα (NR1H3) and LXRβ (NR1H2). LXRα is abundantly expressed in the liver, intestine, kidney, spleen, and adipose tissue, whereas LXRβ is ubiquitously expressed at a lower level [2]. Both isoforms share almost 80% identity of their aminoacid sequences. LXR molecule consists of four domains: (1) N-terminal ligand-independent activation function domain (AF-1) which may stimulate transcription in the absence of ligand, (2) DNA-binding domain (DBD) containing two zinc fingers, (3) hydrophobic ligand-binding domain (LBD) required for ligand binding and receptor dimerization, and (4) C-terminal liganddependent transactivation sequence, also referred to as Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd 297

2 Liver X Receptors Bełtowski activation function-2 (AF-2), which stimulates transcription after ligand binding. LXRα exists in three variants originating from alternative promoter usage and mrna splicing: LXRα1, LXRα2, and LXRα3 [3].LXRα1 is a major variant in most tissues except the testis where LXRα2 predominates. LXRα3 is expressed at low level in the lung, thyroid gland, and spleen. All three variants bind DNA, but LXRα2 isless active in stimulating transcription than LXRα1. Both LXRα and LXRβ form heterodimers with the retinoid X receptor (RXR), a common partner for several nuclear receptors including peroxisome proliferatoractivated receptor (PPAR), vitamin D receptor (VDR), thyroid hormone receptor (TR), and farnesoid X receptor (FXR). LXR/RXR is a permissive heterodimer which may be activated by either LXR agonist or 9-cis retinoic acid (9cRA), a specific RXR ligand. Simultaneous application of the agonists of both partners usually elicits a stronger response than either agonist alone. LXR/RXR complex binds to a liver X receptor response element (LXRE) in the regulatory region of target genes. The ideal LXRE sequence is a direct repeat-4 (DR-4) DNA fragment consisting of two AGGTCA hexameric motifs separated by 4-nucleotide spacer. Analysis of functional LXREs in the known target genes revealed that the consensus sequences contain a number of invariant nucleotides in each half-site but the nucleotides at other positions may vary. Although LXRα and LXRβ have similar affinities for the ideal LXRE, they exhibit different affinities for specific nonideal LXRE sequences in certain genes [4]. In the absence of ligand, LXR may stimulate transcription to some extent, however, may also actively repress expression of some genes. Therefore, the expression of some target genes is higher in LXR-deficient mice than in wild-type animals, although it is unresponsive to LXR agonists. LXR ligands may also inhibit transcription of certain genes through several mechanisms including: (1) stimulation of an intermediate regulatory protein which inhibits transcription or destabilizes mrna, (2) binding of LXR to other transcription factors and inhibiting their activity, (3) competition between LXR and other transcription factors for a common pool of RXR, (4) competition between LXR/RXR heterodimers and other transcription factors for the same coactivators, (5) binding of LXR/RXR heterodimer to LXRE sequence overlapping with the response element for other nuclear receptors. In renal juxtaglomerular apparatus, cyclic AMP increases renin gene expression through the LXRαdependent mechanism by stimulating the binding of LXRα to a specific DNA sequence referred to as cyclic AMP response element and a negative response element (CNRE) within a promoter region of the renin gene. This effect is not mimicked by classic LXR or RXR agonists and, vice versa, camp does not induce LXR binding to LXRE [5]. This is an additional mechanism through which LXR may regulate gene expression. Agonists and Antagonists of LXR Endogenous LXR Ligands Monooxygenated cholesterol derivatives (oxysterols) seem to be the major natural LXR ligands [6]. The EC 50 value of LXR for most oxysterols is within low micromolar range, which is close to normal level of these compounds in plasma and tissues. The metabolic fate of oxysterols is faster than that of cholesterol and, therefore, their concentration better reflects short-term changes in cholesterol balance, which makes oxysterols the good candidates for regulatory molecules. In contrast to oxysterols, cholesterol has no affinity for LXR. There are three sources of oxysterols in the animal body: (1) endogenous production in enzymatic reactions, (2) endogenous nonenzymatic production through reactive oxygen species (ROS)-dependent oxidation of cholesterol, and (3) delivery from alimentary sources. In general, oxysterols produced in enzymatic reactions (Fig. 1) are potent LXR agonists, whereas nonenzymatically-generated ones have weak or no agonistic activity. Interestingly, natural oxysterol enantiomers have much higher affinity for LXR than stereoisomers which are not endogenously generated. Enzymatically generated oxysterols recognized as LXR ligands may be divided into three groups: (1) intermediates of the cholesterol biosynthetic pathway; until now 24(S),25-epoxycholesterol (24(S),25-EC) is the main representant of this group, (2) intermediary compounds in the synthesis of steroid hormones from cholesterol, that is, 22(R)-hydroxycholesterol (22(R)-HC) and, to a lesser extent, 20(S)-hydroxycholesterol, and (3) several hydroxysterol compounds formed from cholesterol by sterol hydroxylases ; many of them being various isoforms of cytochrome P450 (CYP). 24(S),25-EC is a potent LXR agonist with an EC 50 of about 300 nm [6]. It is synthesized from dioxidosqualene in the shunt pathway of the mevalonate cascade (Fig. 2), and is detected in high amounts in the liver and, to a lesser extent, in many other tissues [7]. In contrast to oxysterols formed from cholesterol, the level of 24(S),25-EC is reduced by cholesterol overload and stimulated by cholesterol depletion due to negative feedback regulation by cholesterol of the expression and activity of several enzymes localized proximally to 24(S),25- EC in the mevalonate cascade such as 3-hydroxy-3- methylglutarylcoenzyme A (HMG-CoA) reductase. 298 Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd

3 Bełtowski Liver X Receptors HO HO OH cholesterol 22(R)-hydroxycholesterol OH OH HO HO 24(S)-hydroxycholesterol O 27-hydroxycholesterol HO 24(S),25-epoxycholesterol HO OH 4 -hydroxycholesterol OH HO desmosterol HO 25-hydroxycholesterol OH O F3C F3C N O S O F3C Cl N O HO CF3 T GW3965 Figure 1 Structure of cholesterol and selected natural and synthetic LXR agonists. Conversion of cholesterol to pregnenolone is a first and rate-limiting step in steroidogenesis, that is, formation of steroid hormones from cholesterol. This reaction is catalyzed by CYP11A1 also referred to as cytochrome P450 side chain cleavage enzyme (P450scc) or cholesterol desmolase ; the enzyme contained in inner mitochondrial membrane of steroidogenic tissues including adrenal cortex, ovaries, testes, and placenta. CYP11A1 is also detected at low level in the nervous system, heart, pancreas, and skin [8]. This enzyme catalyzes three consecutive reactions leading to the conversion of cholesterol to pregnenolone with 22(R)-HC and 20,22- dihydroxycholesterol as intermediates; both of them being potent LXR activators. Pregnenolone has no affinity to LXR, whereas nonnatural enantiomer 22(S)-HC is a potent LXR antagonist [6]. Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd 299

4 Liver X Receptors Bełtowski acetyl-coa + acetoacetyl-coa HMG-CoA HMGCR mevalonate statins mevalonate~p mevalonate~pp isopentenyl~pp dimethylallyl~pp isopentenyl~pp geranyl~pp geranylgeranyl~pp farnesyl~pp farnesylated proteins geranylgeranylated proteins squalene SQS OSC DHCR24 24,25-dihydrolanosterol lanosterol 24(S),25-epoxylanosterol CYP51 24,25-dihydro-FF-MAS FF-MAS 24(S),25-epoxy FF-MAS (MAS412) DHCR14 24,25-dihydro-T-MAS T-MAS 24(S),25-epoxy T-MAS (MAS414) SQE 2,3-oxidosqalene (2,3-monoepoxysqualene, MOS) 2,3;22,23-dioxidosqalene (2,3;22,23-diepoxysqalene, DOS) 24,25-dihydroxzymosterol zymosterol 24(S),25-epoxyzymosterol lathosterol cholesta-7,24-dien-3 -ol 24(S),25-epoxylathosterol 7-dehydrocholesterol 7-dehydrodesmosterol 24(S),25-epoxy-7-dehydrocholesterol DHCR7 cholesterol desmosterol 24(S),25-epoxycholesterol DCR24 Figure 2 Mevalonate cascade. Recognized LXR agonists are presented on the gray background. Geranylgeranyl-pyrophosphate, an LXR antagonist, is presented in the white frame. Enzymes relevant for this review are also presented. HMGCR, 3-hydroxy-3-methylglutaryl-coenzyme A reductase (EC ); SQS, squalene synthase (EC ); SQE, squalene epoxidase (EC ); OSC, oxidosqualene:lanosterol cyclase (EC ); DHCR24, 3β-hydroxysterol 24 -reductase (desmosterol reductase, EC ); CYP51, lanosterol 14α-demethylase (sterol 14- demethylase, EC ); DHCR14, sterol 14α-reductase (EC ); DHCR7, 7-dehydrocholesterol reductase (EC ); FF-MAS, follicular fluid meiosis-activating sterol (LXR agonist isolated from the ovary); T-MAS, testis meiosis-activating sterol (LXR agonist isolated from testis). The family of hydroxylated cholesterol derivatives known to activate LXR includes 24(S)- hydroxycholesterol (24(S)-HC), 25-hydroxycholesterol (25-HC), and 27-hydroxycholesterol (27-HC). 24(S)-HC, also referred to as cerebrosterol, is the most abundant hydroxysterol in the brain [9]. It is synthesized in neurons but not in glial cells by a cytochrome P450 CYP46 isoform (cholesterol 24-hydroxylase) [10]. 24(S)-HC plays an important role in brain cholesterol homeostasis. Because cholesterol does not cross the blood brain barrier, it cannot be removed from this organ through the classical reverse transport mechanism involving high-density lipoproteins (HDL). In contrast, 24(S)-HC crosses the blood brain barrier and thus conversion of cholesterol to 24(S)-HC is a main route of cholesterol efflux from the brain [11]. 27-HC is generated by the mitochondrial cholesterol 27-hydroxylase (CYP27) expressed in the liver and other 300 Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd

5 Bełtowski Liver X Receptors cells including macrophages, fibroblasts, and endothelial cells. CYP27 is unique among other cytochrome P450 isoforms in that it catalyzes three consecutive oxidation reactions of a C-27 methyl group to alcohol (27-HC), aldehyde, and carboxylic acid (3β-hydroxy-5-cholestenoic acid) [12]. In the liver, CYP27 is a first enzyme of a so called alternative or acidic pathway of bile acid synthesis, which accounts for the formation of about 4% of bile acid pool in humans [13]. In other cells, 27-HC is not only the LXR ligand but also may be involved in cholesterol efflux, beause both 27-HC and cholestenoic acid are removed from the cells in a HDL-independent manner [12]. This mechanism may be especially important for cells which have no direct contact with plasma and thus are isolated from HDL. Deficiency of CYP27 causes cerebrotendinous xanthomatosis (CTX) an autosomal recessive disorder associated with premature atherosclerosis and increased cholesterol content in the vascular wall despite apparently normal plasma lipid profile [14]. Cholestenoic acid is even a more potent LXR activator than 27-HC [15]. 25-HC is a minor oxysterol formed by a specific microsomal enzyme not belonging to the CYP family, cholesterol 25-hydroxylase, which contains two nonheme iron atoms [16]. 4β-hydroxycholesterol (4β-HC) is formed in the liver and possibly in the small intestine by CYP3A4 and is one of the major circulating oxysterols [17]. Plasma concentration of 4β-HC is about 5-fold higher than the concentration of 4α-HC which is a product of nonenzymatic cholesterol autooxidation. Triple knockout mice lacking CYP46, CYP27, and cholesterol 25-hydroxylase exhibit an impaired response of hepatic LXR target genes to dietary cholesterol load, which confirms an essential role of endogenous oxysterols in LXR signaling [18]. Yang et al. [19] have demonstrated that desmosterol (24-dehydrocholesterol), one of the intermediates in cholesterol biosynthesis, is a potent LXR agonist. In contrast, several other intermediates such as lathosterol, lanosterol, and 7-dehydrocholesterol are inactive. Inhibition of desmosterol reductase by triparamol leads to the upregulation of LXR target genes due to the accumulation of endogenous desmosterol. It should be mentioned that not all endogenous oxysterols are LXR ligands. For example, 7αhydroxycholesterol produced by hepatic 7α-hydroxylase (CYP7A1), a rate-limiting enzyme of bile acid synthesis, is a weak LXR agonist, whereas its oxidation products, cholic and chemodeoxycholic acids, are completely inactive [20]. In addition, most nonenzymatically formed oxysterols such as 7-keto, 7α-hydroxy-, 7βhydroxycholesterol, or 5,6-epoxycholesterol are not LXR ligands. Mitro et al. [21] have recently demonstrated that glucose and glucose-6-phosphate bind to and activate LXRα and LXRβ. In contrast, other simple carbohydrates have no effect. In addition, feeding mice with glucose results in the upregulation of LXR target genes in the liver and intestine. Although this study suggested the role of LXR as a glucose sensor, such interpretation was questioned by more recent data for several reasons [22]: (1) L-glucose, the enantiomer not occurring in mammalian cells, was as efficient as D-glucose, (2) glucose stimulated LXR within the milimolar range; although this is physiological level in plasma, intracellular glucose concentration is much lower, (3) stimulatory effect of glucose on LXR was not confirmed in other experimental systems [23], and (4) LXR is not required to cells responses to changes of glucose concentration [23]. Exogenous LXR Ligands Two nonsteroid synthetic LXR agonists, T and GW3965, are commonly used in experimental studies. T activates both LXRα and LXRβ withanec 50 of about 20 nm [24]. GW3965 has a greater affinity for LXRβ than for LXRα; however, the difference is insufficient to differentiate between both isoforms in experimental studies. T is not a specific LXR ligand since it activates also FXR [25], and pregnane X receptor [26]. Thus, not all effects of T observed in experimental studies can be attributed to LXR stimulation. Plant cells do not synthesize cholesterol but several other sterols with similar chemical structure such as sitosterol, stigmasterol, campesterol, brasicasterol, and ergosterol. Plant sterols are poorly absorbed from the intestine and inhibit cholesterol absorption by displacing it from bile micelles. Supplementation of plant sterols may reduce plasma cholesterol level and is considered as a potential antiatherosclerotic therapy [27]. It has been demonstrated that stigmasterol is a potent LXR ligand [28]. Plat et al. [29] observed that both sitosterol and campesterol effectively stimulated LXR in cultured intestinal cells. In addition, certain oxidized derivatives of phytosterols (oxyphytosterols) are more effective LXR activators than 24(S),25-EC [30]. LXR Antagonists Geranylgeranylpyrophosphate (GGPP), one of the products of the mevalonate cascade, inhibits transcriptional activity of LXRα and LXRβ by antagonizing its interaction with nuclear coactivators [31]. Polyunsaturated fatty acids (PUFAs) of the n-3 and n-6 family, such as arachidonic acid, eicosapentaenoic acid, docosahexaenoic acid, Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd 301

6 Liver X Receptors Bełtowski and linoleic acid, are competitive antagonists of the interaction between LXR and their ligands [32]. Role of LXR in the Regulation of Cholesterol Metabolism LXR sense excess cholesterol and trigger various adaptive mechanisms protecting the cells from cholesterol overload. Activation of LXR results in: (1) stimulation of cholesterol removal from the cell, its transport to the liver and biliary excretion; processes together referred to as reverse cholesterol transport (RCT); (2) inhibition of intestinal cholesterol absorption; and (3) inhibition of cholesterol synthesis and uptake by the cells. Reverse Cholesterol Transport RCT is cholesterol traffic from peripheral tissues to the liver, where it is excreted in the bile as unchanged cholesterol or after conversion to bile acids. RCT is initiated by cholesterol removal from the cell to HDL or to lipid-free apolipoproteins such as apoa-i or apoe. Within HDL, most of free cholesterol is esterified by lecithin:cholesterol acyltransferase (LCAT), and cholesterol esters are then transported to the liver either in HDL or in apob-containing lipoproteins (VLDL and LDL), to which they are first transferred by cholesterol ester transfer protein (CETP). Naik et al. [33] have demonstrated that if mice are injected with macrophages previously loaded with titrated cholesterol and then are treated with GW3965 for 10 days, fecal excretion of cholesterol tracer is markedly higher and excretion of titrated bile acids tends to be higher than in vehicle-treated animals. These data clearly indicate that LXR agonist stimulates cholesterol transport to the liver for subsequent excretion and/or metabolism. Most, if not all, steps of RCT are stimulated by LXR. First, LXR upregulate the expression of transporters involved in cholesterol removal from plasma membrane to extracellular acceptors, the ATP-binding cassette transporters A1 (ABCA1) and G1 (ABCG1). ABCA1 and ABCG1 are abundant in macrophages where they prevent foam cell formation by removing excess cholesterol, but are also expressed in most other cell types including hepatocytes, enterocytes, adipocytes, and skeletal muscle cells. ABCA1 is a full transporter, that is, it operates as a single molecule, and transfers both cholesterol and phospholipids from plasma membranes to small discoid-shape pre-hdl or to lipid-free apoa-i. Thus, apart from RCT, ABCA1 is also involved in the formation of nascent HDL particles in the liver and small intestine. Deficiency of ABCA1 is responsible for Tangier disease, a rare inherited disorder characterized by very low HDL level, overaccumulation of cholesterol in peripheral tissues, and premature atherosclerosis. In contrast, ABCG1 is a half-transporter, that is, it operates as a homodimer, and transfers cholesterol to HDL but not to lipid-free apoa-i [34]. Multiple studies have demonstrated that LXR agonists increase the expression of ABCA1 and ABCG1, which is accompanied by enhanced cholesterol efflux from various cell types both in vitro and in vivo [35,36]. In particular, the expression of ABCG1 may be increased up to 1000-fold by natural and synthetic LXR agonists [37]. Interestingly, various ABC transporters may be involved in cholesterol efflux from different intracellular pools. In particular, cholesterol delivered by intact LDL through the LDL receptor-dependent pathway is subsequently removed by ABCG1/ABCG4 heterodimer, whereas cholesterolderivedfrommodifiedldl,whichenteredthecell through the scavenger receptor is exported by ABCA1 [38]. Cholesterol loading of the cell results in the increased formation of oxysterols, which stimulate LXR and enhance cholesterol efflux through these transporters, thus providing the regulatory feedback mechanism maintaining constant cholesterol content [39]. LXR are also involved in the regulation of intracellular cholesterol traffic. Before efflux, cholesterol has to be transported from the endosomal compartment to the plasma membrane. This process is mediated by two carriers, Niemann Pick C1 (NPC1) and C2 (NPC2) proteins. LXR agonists increase the expression of NPC1 and NPC2 and thus stimulate redistribution of cholesterol from endosomal compartment to the plasma membrane where it becomes available for efflux to extracellular acceptors [40]. Scavenger receptor type B1 (SR-B1) is a specific HDL receptor expressed in the liver and, to a lesser extent, in other tissues. Hepatic SR-B1 is essential for delivery of cholesterol esters from HDL to hepatocytes. 22(R)-HC increased SR-B1 expression in human hepatoma cells by binding to the LXRE within the promoter region of SR-B1 gene [41]. However, the effect of LXR agonists on SR-B1 in vivo has not been studied so far. The first and rate-limiting enzyme of hepatic bile acid synthesis is cholesterol 7α-hydroxylase (CYP7A1). Promoter regions of mouse and rat CYP7A1 genes contain LXRE, and their transcription is potently upregulated by natural and synthetic LXR agonists [42]. LXRE of CYP7A1 gene binds LXRα more potently than LXRβ, and therefore LXRα / mice but not LXRβ / animals accumulate large amounts of cholesterol in the liver in response to high cholesterol diet. This is accompanied by hepatomegaly and severe liver damage. LXRα / mice do not exhibit any increase in CYP7A1 expression when switched to 302 Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd

7 Bełtowski Liver X Receptors Figure 3 Intestinal absorption of cholesterol (A) and plant sterols (B). CHE, cholesterol ester; PE, phytosterol ester; ACAT-2, acyl-coa:cholesterol acyltransferase-2; CHM, chylomicrons. cholesterol-rich diet, nor are they able to increase bile acid synthesis and fecal excretion [43]. In contrast to rat and mice, human CYP7A1 gene does not contain the LXRE and its transcription is not upregulated by LXR agonists [44]. This species difference is responsible, at least in part, for much greater susceptibility of humans to dietaryinduced hypercholesterolemia [42]. Indeed, cholesterol 7α-hydroxylase is not upregulated in response to cholesterol feeding in mice expressing human CYP7A1, and these animals easily develop hypercholesterolemia when challenged with high-cholesterol diet [45]. Intestinal Cholesterol Absorption An important role in intestinal cholesterol absorption is played by the Niemann Pick C1-like 1 protein (NPC1L1) contained in the apical membranes of enterocytes. Next, the majority of cholesterol is esterified inside the enterocyte by acyl-coa:cholesterol acyltransferase-2 (ACAT-2) and is incorporated into chylomicrons. Some cholesterol is exported through the ABCA1 contained in the basolateral membrane to apoa-i; the process essential for intestinal HDL formation (Fig. 3A). In addition, a substantial part of intracellular cholesterol is back extruded from enterocyte to intestinal lumen through the ABCG5/ABCG8 heterodimer contained in the apical membrane [46]. Repa et al. [47] demonstrated that T reduced intestinal absorption of dietary cholesterol. Because this effect was accompanied by increase in ABCA1 expression in the small intestine, it was suggested that this transporter drives cholesterol removal from the enterocytes to the intestinal lumen and thus reduces net cholesterol absorption. However, this possibility is unlikely since ABCA1 is contained mainly in the basolateral membrane Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd 303

8 Liver X Receptors Bełtowski of enterocytes and the effect of LXR agonists on fecal cholesterol excretion is preserved in ABCA1 knockout mice [48]. Indeed, LXR agonists stimulate cholesterol recycling from the enterocyte to the intestinal lumen by upregulating ABCG5 and ABCG8 [49]. Apart from the intestine, ABCG5 and ABCG8 are abundantly expressed in the canalicular membrane of hepatocytes where they drive cholesterol transport to the bile. Hepatic ABCG5 and ABCG8 are also stimulated by LXR agonists, which results in the enhanced biliary cholesterol excretion [50]. In contrast to wild-type animals, T does not stimulate biliary cholesterol excretion and fails to reduce fractional cholesterol absorption in ABCG5/ABCG8 double knockout mice [50]. However, increase in fecal cholesterol by LXR agonists is not solely dependent on biliary cholesterol excretion. Indeed, GW3965 does not stimulate biliary cholesterol content but increases fecal cholesterol loss in mice lacking the multidrug resistance protein-2 (Mdr-2, ABCB4) involved in the export of phospholipids from hepatocytes to bile canaliculi [51]. It is suggested that apart from the liver, the intestine is an alternative organ involved in RCT and excretion. According to this hypothesis, HDL may deliver cholesterol from peripheral tissues directly to the intestine, where it is removed to the gut lumen. Finally, LXR agonists decrease the expression of NPC1L1 protein in the apical membrane of enterocytes, which is another mechanism leading to impaired cholesterol absorption [52]. Effect of LXR on Intestinal HDL Formation Apart from the liver, small intestine is a major HDLgenerating organ in the body. Although LXR agonists inhibit cholesterol absorption in ABCA1-independent manner, they do stimulate this transporter in enterocytes and enhance intestinal HDL formation. Indeed, LXR agonists stimulate apoa-i mediated cholesterol efflux from the basolateral but not from the apical membrane of enterocytes, while having no effect on the formation of chylomicrons [53]. Intestine-specific deletion of ABCA1 results in about 30% reduction of plasma HDL, whereas tissue-specific stimulation of intestinal ABCA1 by LXR agonists raises plasma HDL level [54,55]. Indeed, GW3965 increases plasma HDL in wild-type mice as well as in animals lacking hepatic ABCA1, but fails to do so in mice lacking ABCA1 selectively in the small intestine. In addition, the effect of T on plasma HDL is impaired in NPC1L1 / mice, indicating that alimentary cholesterol is involved in intestinal HDL formation in response to LXR agonists [56]. Effect of LXR on Plant Sterol Absorption Although plant sterols are almost as abundant as cholesterol in food, only a small fraction of them is absorbed in the intestine for two reasons. First, plant sterols are much more effectively exported by the ABCG5/ABCG8 heterodimer. In addition, plant sterols are very poor substrates for ACAT-2 and are poorly incorporated into chylomicrons. Thus, most of plant sterols reaching the enterocyte are rapidly removed to the intestinal lumen (Fig. 3B). Any absorbed plant sterols are also rapidly excreted to the bile by hepatic ABCG5/ABCG8 complex. Therefore, plasma concentration of plant sterols is normally very low [57]. Mutation of either ABCG5 or ABCG8 gene results in sitosterolemia (phytosterolemia), a rare inherited recessive disease characterized by very high level of plant sterols in plasma and tissues, moderate hypercholesterolemia, and accelerated atherosclerosis, presumably related to deposition of plant sterols in the arterial wall. Under normal conditions, LXR agonists limit intestinal absorption of plant sterols and increase their biliary excretion by stimulating ABCG5 and ABCG8. However, in ABCG5 knockout mice [58] and in ABCG5/ABCG8 double knockout mice [50], LXR agonists paradoxically increase plasma concentration of phytosterols the effect associated with unopposed stimulation of ABCA1- driven basolateral phytosterol efflux from the enterocyte to HDL (Fig. 3B). Indeed, ABCA1 does not discriminate between cholesterol and plant sterols and T stimulates basolateral apoa-i mediated efflux of β-sitosterol from cultured intestinal epithelial cells [59]. These data suggest that LXR agonists could have a paradoxical undesirable effect on plasma plant sterols in patients with sitosterolemia, and possibly also in much more frequent asymptomatic ABCG5 or ABCG8 mutant heterozygotes. Cholesterol Synthesis and Uptake Cellular cholesterol content depends on the balance among three processes: (1) intracellular synthesis; (2) uptake from plasma lipoproteins, mainly LDL, and (3) removal from the cell to plasma lipoproteins, mainly HDL. Cholesterol synthesis is precisely regulated to maintain constant cholesterol content and thus is adjusted to the remaining two processes. Due to the potent effect of LXR agonists on cholesterol removal from the cell, it is difficult to separate any direct effect of these receptors on cholesterol synthesis from adaptor changes induced by altered uptake and/or removal. Several key enzymes in the cholesterol biosynthesis pathway including HMG-CoA reductase are 304 Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd

9 Bełtowski Liver X Receptors stimulated by the sterol regulatory element-binding protein-2 (SREBP-2). SREBPs are transcription factors synthesized as the inactive precursors which reside in the endoplasmic reticulum and are subsequently activated by proteolysis, move to the nucleus, and stimulate the expression of target genes. Three isoforms of SREBPs were identified; two of them (SREBP-1a and SREBP- 1c) are encoded by a single gene whereas the third, SREBP-2, is encoded by a separate gene. SREBP-1a stimulates the expression of genes involved in both cholesterol and fatty acid synthesis, SREBP-1c stimulates lipogenesis, and SREBP-2 stimulates cholesterol-synthesizing enzymes and LDL receptor. SREBP-2 is activated in response to cholesterol depletion and stimulates compensatory upregulation of cholesterol uptake and synthesis [60]. LXRα null mice are characterized by higher expression of SREBP-2 as well as of several target genes including HMG-CoA synthase and HMG-CoA reductase [43]. The similar although much smaller upregulation of cholesterol-synthesizing enzymes is observed in LXRβ null mice [61]. In addition, T reduces hepatic expression of HMG-CoA synthase and squalene synthase in wild-type mice [24]. These data suggest that LXR inhibit cholesterol synthesis. However, stimulation of cholesterol efflux by LXR agonists may cause a compensatory upregulation of cholesterol synthesis in certain cell types. For example, T and GW3965 enhance cholesterol synthesis as well as increase LDL-receptor expression in human hepatoma HepG2 cells, probably by stimulating cholesterol efflux and activating SREBP- 2 dependent pathway. In contrast, natural agonists have no (22(R)-HC) or slight inhibitory (24(S),25-EC) effect on cholesterol production, most likely by directly suppressing SREBP-2 processing in LXR-independent manner [62]. Finally, natural and synthetic LXR agonists inhibit foam cell formation by suppressing LDL uptake by macrophages; however, this effect is not associated with the downregulation of LDL receptors but with inhibiting fluid-phase pinocytosis [63]. LXR in the Regulation of Lipid Metabolism Lipogenesis Schultz et al. [24] first demonstrated that T markedly increased hepatic triglyceride content and plasma triglyceride concentration in mice and hamsters. Although the effect on plasma triglycerides was transient, that on hepatic triglycerides was persistent and led to severe liver steatosis and dysfunction. In addition, plasma triglycerides are about 4-fold lower in LXRα/LXRβ double knockout mice than in wild-type animals. LXR agonists markedly stimulate fatty acid synthesis (lipogenesis) in hepatocytes. This effect is partially mediated by increased expression of SREBP-1c, which subsequently binds to sterol response element (SRE) within the promoter region of genes encoding numerous lipogenic enzymes [64]. Yoshikawa et al. [65] have identified two LXREs within the SREBP-1c gene promoter, and demonstrated that LXR as well as RXR agonists increase its transcriptional activity. In addition, LXR directly (in SREBP-1c independent manner) regulates the expression of several lipogenic enzymes including acetyl-coa carboxylase (ACC) [66], fatty acid synthase (FAS) [67], and stearoyl-coa desaturase-1 (SCD-1), which introduces the double bond at n-9 position of saturated fatty acids and forms their monounsaturated counterparts [68]. T increases SREBP-1c, ACC, FAS and SCD-1 gene expression in the liver of wild-type mice [24]. This effect is preserved in LXRβ / but not in LXRα / mice. In addition, basal level of SREBP-1c, ACC, FAS, and SCD- 1 is reduced in LXRα / mice. These data suggest that LXRα is a major isoform responsible for lipogenic effect. LXR agonists stimulate production of triglyceride-rich, large VLDL particles in the liver [69]. Although the number of VLDL particles does not change, their diameter increases due to higher amount of triglycerides in each particle. In animals treated with LXR agonists, plasma triglyceride concentration increases only transiently, because VLDL metabolism is simultaneously stimulated, presumably due to the upregulation of lipoprotein lipase (LPL). In addition, T stimulates peroxisomal β- oxidation of very long chain fatty acids in the liver [70]. Apart from direct (SREBP-1c-independent) and indirect (SREBP-1c-dependent) stimulation of lipogenic enzymes, several other mechanisms of lipogenic effect of LXR agonists have been described. Carbohydrate response element binding protein (ChREBP) is a glucosesensitive transcription factor which stimulates the expression of lipogenic enzymes when glucose concentration is high. LXR agonists stimulate the expression of ChREBP in the liver both in vitro and in vivo, and stimulatory effects of T on FAS, ACC, and SCD-1 are attenuated in ChREBP-knockout mice [71]. Adiponectin is secreted in large amounts by adipocytes and exerts many beneficial metabolic effects such as increase in insulin sensitivity and stimulation of mitochondrial fatty acid oxidation. Adiponectin inhibits hepatic lipogenesis and prevents hepatic steatosis by reducing SREBP-1c expression. Interestingly, 25-HC reduces the expression of adiponectin receptor, AdipoR1, in primary hepatocytes [72]. Zhang et al. [73] have demonstrated that LXR agonists stimulate the expression of liver-specific uridine phosphorylase, Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd 305

10 Liver X Receptors Bełtowski the rate-limiting enzyme in uridine catabolism. A major metabolic product of uridine catabolism in the liver is β- alanine, which serves as a precursor for fatty acid synthesis. Thus, the effect on uridine phosphorylase could be involved in LXR-induced lipogenesis. Kaplan et al. [74] have shown that the natural and synthetic LXR agonists stimulate the expression of angiopoietin-like protein 3 (Angptl3) in the liver as well as its plasma concentration. Angptl3 may mediate, at least in part, the effect of LXR ligands on plasma lipid profile. Indeed, Angptl3-deficient KK/San mice are characterized by low plasma triglycerides and cholesterol levels despite being obese and diabetic, whereas overexpression of Angptl3 gene or injection of Angptl3 protein increase plasma triglycerides [75]. Angptl3 increases plasma VLDL mainly by inhibiting LPL [76], and the hypertriglyceridemic effect of LXR agonists is markedly reduced in Angptl3 knockout mice [77]. Effect of LXR Agonists on Apolipoproteins LXR agonists stimulate synthesis of apolipoprotein-e in macrophages and in adipose tissue both in vitro and in vivo [78]. In contrast, apoe expression in the liver is not regulated by LXR. In fact, apoe was the first identified LXRresponsive gene which is regulated in a tissue-specific manner. The stimulatory effect on apoe is greater in human than in murine macrophages. In addition to apoa- I, apoe is an alternative extracellular cholesterol acceptor and thus is involved in LXR-stimulated cholesterol efflux. Apart from apoe, LXR agonists stimulate the expression of apolipoproteins C-I, C-II, and C-IV, which also operate as cholesterol acceptors [79]. LXR agonists increase the expression of apolipoprotein D in adipocytes [80]. ApoD is a 29-kDa glycoprotein that is primarily associated with plasma HDL. It is synthesized in many tissues including spleen, testes, and brain. Physiological function of apod is unclear and it is not known if apod is regulated by LXR also outside the adipose tissue. The next apolipoprotein regulated by LXR is apoa- IV. ApoA-IV is synthesized in the intestine and, to a lesser extent, in the liver, and is detected mainly in chylomicrons and in lower amounts in HDL, as well as a free plasma protein. In HDL, apoa-iv stimulates LCAT, whereas free plasma apoa-iv facilitates cholesterol efflux from the cells and exerts antioxidant activity. Transgenic overexpression of apoa-iv reduces atherosclerosis in the mouse models, and plasma apoa-iv is inversely correlated with atherosclerosis in humans [81]. Although apoa-iv gene contains the LXRE, LXR agonists stimulate its expression only in hepatocytes but not in the intestine. In vivo, T increases HDL-associated apoa-iv, which is presumably of hepatic origin, but has no effect on intestine-derived chylomicron-associated apoprotein. Surprisingly, LXR agonists inhibit apoa-i production in human hepatocytes [82]. Apolipoprotein A-V is the only known apolipoprotein downregulated by LXR agonists [83]. ApoA-V is synthesized in the liver and is incorporated into the HDL and VLDL fractions. Overexpression of apoa-v reduces plasma triglycerides in mice, whereas deletion of apoa-v gene induces hypertriglyceridemia. Hypolipidemic effect of apoa-v is mediated by its stimulatory effect on LPL. Plasma Lipoprotein-Remodeling Transporters LXR agonists stimulate the expression of two plasma lipoprotein-remodeling proteins, CETP and phospholipid transfer protein (PLTP). CETP is synthesized in the liver and circulates in the HDL fraction. CETP transfers cholesterol esters from HDL to VLDL and LDL in the exchange for triglycerides. Thus, CETP may facilitate cholesterol transport to the liver. CETP contributes to high LDL in humans and rabbits, whereas species lacking CETP such as rat and mice have very low level of LDL [84]. LXR agonists increase hepatic CETP synthesis and plasma CETP concentration [85]. Therefore, LXR agonists markedly elevate plasma HDL-cholesterol in species lacking CETP but have only a weak or no effect on HDL in CETP-positive animals. In addition, in two CETP-positive species, Syrian hamster and cynomolgus monkey, LXR agonists significantly elevate LDL-cholesterol [86]. In addition, transgenic mice expressing human CETP demonstrate the elevation of LDL and no change in HDL in response to LXR agonists [87,88]. PLTP mediates transport of phospholipids from VLDL and chylomicron remnants to HDL or to lipid-poor apoa-i. In addition, PLTP mediates phospholipid transfer between different HDL subfractions, thus generating lipid-poor small pre-β HDL which are better cholesterol acceptors than large HDL particles. T increases PLTP activity in plasma as well as PLTP mrna in the liver, adipose tissue, and intestine, which is accompanied by the increase in phospholipid content in the HDL fraction [89]. Stimulation of PLTP may contribute to LXR-induced VLDL production because PLTP is involved in VLDL assembly in hepatocytes. Indeed, hypertriglyceridemic effect of T is markedly reduced in PLTP deficient mice [90]. Enzymes Involved in Plasma Lipoprotein Metabolism Triglycerides and/or phospholipids contained in plasma lipoproteins are hydrolyzed by three lipases expressed on the surface of endothelial cells: LPL, hepatic lipase (HPL), and endothelial lipase (EL). LPL has a high affinity for 306 Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd

11 Bełtowski Liver X Receptors triglycerides and hydrolyzes triglyceride-rich lipoproteins such as VLDL and chylomicrons. Mice and human LPL genes contain LXRE and are direct targets for LXRα but not for LXRβ [91]. The stimulatory effect of LXR on LPL expression is restricted to the liver and, to a lesser extent, to macrophages. In contrast, LXR agonists have no effect on LPL in adipose tissue. EL hydrolyzes plasma phospholipids, including those associated with HDL. Sovic et al. [92] have demonstrated that 24(S)-HC reduces EL mrna and protein levels in brain capillary endothelial cells. This effect may contribute to LXR-induced increase in HDL fraction. However, Norata et al. [93] observed increase in EL expression in human umbilical vein endothelial cells, human fibroblasts, and cultured hepatocytes in response to T The role of LXR in the regulation of HPL, the third enzyme involved in intravascular lipolysis which is active toward both triglycerides and phospholipids, has not been studied. Effect of LXR Agonists on Atherosclerosis in Experimental Studies Several studies have shown that LXR agonists reduce atherosclerotic lesions in animal models. Joseph et al. [94] have demonstrated that GW3965 reduces size of aortic atherosclerotic lesions in apoe / and in male LDL-R / mice by about 50%, and in female LDL-R null animals by 35%. Interestingly, GW3965 had differential effects on plasma lipids in both models; that is, it slightly reduced total cholesterol and had no effect on triglycerides or HDL in LDL-R / mice, but had no effect on total or HDL-cholesterol and increased triglycerides in apoe / mice. These data suggest that GW3965 protects against atherosclerosis independently of changes in plasma lipids. Similarly, T administered to LDL-R / mice for 8 weeks dose-dependently reduced atherosclerotic lesions by up to 70% [95]. In that study, no changes in total plasma cholesterol were observed, but T transiently increased plasma triglycerides and to some extent corrected the fall in HDL induced by atherogenic diet. Importantly, T given to LDL-R / mice with developed atherosclerosis induced the regression of established aortic plaques indicating that LXR agonists may be effective not only during lesion development but also in established atherosclerosis. In addition, treatment with T beneficially affected structure of aortic plaques, that is, reduced macrophage and increased collagen content, thus increasing plaque stability and reducing the risk of rupture [96]. Administration of GW3965 reduced atherosclerotic lesions in apoe / mice and in apoe / LXRα / double knockout mice, indicating that activation of LXRβ is sufficient to ameliorate atherogenesis [97]. LXR Agonists as Potential Drugs for Dyslipidemia Currently, several groups of drugs are used in the treatment of dyslipidemia including HMG-CoA reductase inhibitors (statins), peroxisome proliferatoractivated receptor-α (PPAR-α) agonists, fibrates, bile acid-sequestering resins, cholesterol absorption inhibitors (ezetimibe), etc. Given the potent effect of LXR agonists on HDL formation, cholesterol efflux and intestinal cholesterol absorption, these compounds could be considered as a novel therapeutic option for patients with hypercholesterolemia and/or low HDL level. However, the possible side effects should also be kept in mind; the most important of them being liver steatosis and/or hypertriglyceridemia. Thus, specific LXR modulators with little or no effect on lipogenesis would be greatly desirable. Several strategies may be applied to generate LXR ligands devoid of lipogenic potential. First, most data suggest that lipogenic effect of LXR agonists in the liver is mediated by LXRα. In contrast, macrophages contain both isoforms and both are involved in cholesterol efflux. Therefore, specific LXRβ agonists might specifically target cholesterol balance. Quinet et al. [98] have demonstrated that synthetic LXR agonists stimulate ABCA1 expression and cholesterol efflux from peritoneal macrophages obtained from wild-type, LXRα /,andlxrβ / mice, confirming that both receptors may mediate cholesterol efflux. In vivo, LXR agonists were less effective in increasing hepatic SREBP-1c gene expression in LXRα / than in wild-type mice and failed to increase plasma triglycerides in LXRα / animals, whereas increase in HDL-cholesterol was comparable in both groups. Thus, selective activation of LXRβ effectively stimulates HDL formation and cholesterol efflux but not hepatic lipogenesis. However, LXR agonists failed to stimulate CYP7A1 gene expression in the liver of LXRα / mice, suggesting that metabolism of cholesterol to bile acids may be ineffective following the administration of LXRβ-specific ligands. In addition, agonists with considerable selectivity for LXRβ over LXRα are not currently available, and synthesis of such compounds will be a challenge since ligand binding domains of both isoforms show almost 80% homology of their aminoacid sequences. The second possibility is suggested by the observation that, in contrast to nonsterol agonists, steroid LXR activators like oxysterols are less potent in stimulating lipogenesis, because they also inactivate SREBP-regulated Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd 307

12 Liver X Receptors Bełtowski pathways. Unfortunately, all known oxysterols are rapidly metabolized and their in vivo effectiveness is very limited. However, it cannot be excluded that synthetic sterol LXR agonists resistant to degradation will be synthesized in the future. The third approach is to obtain gene-specific LXR modulators, which will stimulate ABCA1 but not SREBP-1c. Such strategy is theoretically possible because the cooperation among LXR, coactivators, and corepressors may regulate various genes in a different manner. For example, in the absence of ligand, LXR inhibits the expression of ABCA1 by binding corepressors, but slightly stimulates that of SREBP-1c [99]. Therefore, loss of LXR results in moderate elevation of ABCA1 and depression of SREBP-1c. In theory, an agonist which induces dissociation of corepressors but does not recruit coactivators should stimulate ABCA1 but not affect, or even reduce, SREBP-1c. Among experimentally used LXR activators, T increased plasma and liver triglycerides and exerted a potent effect on hepatic SREBP-1c and FAS, whereas GW3965 failed to elevate triglycerides in C57B1/6 mice. However, both agonists induced a comparable stimulation of ABCA1 in the liver and small intestine as well as similarly increased plasma HDL [100]. The fourth approach to obtain nonlipogenic LXR agonists is to generate tissue-selective compounds which will act on macrophages but not on hepatocytes. In particular, tissue selectivity may result from pharmacokinetic properties of a specific compound. Plant sterols have been used for a long time to suppress absorption of alimentary cholesterol. In general, it is believed that their effect results from the competition with cholesterol for intestinal absorption mechanisms. Taking into account that certain oxidized phytosterols activate LXR and that such oxidized derivatives are detected in plasma of healthy humans [101], it may be suggested that the beneficial effect of oxysterols is mediated by LXR stimulation in the intestine or even in other tissues [102]. Indeed, oxidized derivative of ergosterol, YT-32, effectively inhibits cholesterol absorption by upregulating intestinal ABCG5 and ABCG8, but does not induce hypertriglyceridemia or hepatic steatosis when given orally [30]. This selectivity results from rapid removal of YT-32 from enterocytes to the gut lumen and its limited net absorption. In contrast, nonsteroid LXR agonists such as T are not transported by ABCG5 and ABCG8 and thus are avidly absorbed. Although these results may suggest a novel strategy to search for nonlipogenic LXR agonists, it should be expected that the effect of compounds like YT-32 on macrophage cholesterol efflux will also be very limited. Cholestenoic acid derivative, hypocholamide, exerts a desirable effect on cholesterol metabolism, that is, it reduces total plasma cholesterol and hepatic cholesterol content without affecting plasma triglycerides [103]. Hypocholamide effectively stimulates RCT from peripheral tissues but its hepatic effect is negligible because hypocholamide is rapidly inactivated in the liver through glucuronidation. Interestingly, hypocholamide is a relatively specific LXRα activator which indicates that LXRβ selectivity is not mandatory for a beneficial pharmacologic profile. Quinet et al. [104] described a synthetic oxysterol, N,N-dimethyl-3β-hydroxycholamide (DMHCH), which effectively stimulated the expression of hepatic and macrophage ABCA1 and ABCG1 while having only a moderate effect on SREBP-1c. Although this geneselectivity could be accounted for by its oxysterol structure, DMHCH also antagonized the stimulatory effect of GW3965 on SREBP-1c expression, suggesting that it behaves as a mixed LXR agonist-antagonist. In vivo,dmhch raised HDL but had no effect on plasma triglycerides, not only due to its gene selectivity, but also because its absorption from the intestine is limited. Indeed, orally administered DMHCH increased intestinal ABCA1 expression and HDL formation but had little effect on hepatic ABCA1. Effect of Currently Used Hypolipidemic Drugs on LXR Signaling Statins Statins are competitive inhibitors of HMG-CoA reductase, a rate-limiting enzyme in cholesterol biosynthesis which converts HMG-CoA to mevalonate. Initially introduced as cholesterol-lowering agents, statins exhibit many pleiotropic effects associated with inhibiting the formation of other sterols as well as nonsteroid isoprenoids. Statins are in general well tolerated but in some patients severe side effects are observed such as myopathy, hepatotoxicity, etc. Statins may inhibit formation of various mevalonate derivatives including GGPP, an LXR antagonist [31], and of LXR-activating oxysterols, and thus their effect on LXR seems to be complex. Several studies have demonstrated that statins reduce 24(S)-HC concentration in plasma and/or cerebrospinal fluid [105,106]. However, in most cases the effect was modest and was associated with concomitant reduction of cholesterol, which resulted in no change or even increase in 24(S)-HC/cholesterol ratio. In addition, pravastatin [107], simvastatin, and atorvastatin [108] reduced plasma 27-HC level. These data suggest that statins have some reducing effect on plasma oxysterols. However, because oxysterols are formed intracellularly, the effect of statins on tissue oxysterol level is of greater significance. In the guinea pig, pravastatin administered for 3 weeks reduced 24(S)-HC concentration in the brain only by 308 Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd

13 Bełtowski Liver X Receptors 17%, whereas simvastatin had no effect [109]. In the rat, high dose of simvastatin (100 mg/kg for 3 days) effectively reduced brain cholesterol synthesis but had no effect on 24(S)-HC, suggesting that cholesterol availability is not rate-limiting for the formation of this oxysterol [110]. It is unclear if these minor effects of statins on endogenous oxysterols have any adverse consequences for LXR signaling. Despite reducing cholesterol synthesis, statins might have no deleterious effect on ABCA1-mediated cholesterol efflux because: (1) statins have a negligible effect on oxysterols because high cholesterol level is not rate-limiting for oxysterol formation, (2) statins reduce oxysterols slightly, but physiological level of endogenous oxysterols is high enough to be not rate-limiting for LXR signaling, (3) the negative consequences of oxysterol depletion are balanced by other opposite mechanism beneficial for LXR signaling such as reduction of geranylgeraniol. However, several in vitro studies have demonstrated a considerable impairment of LXR signaling by statins. Forman et al. [31] first demonstrated that mevastatin or lovastatin reduced transcriptional activity of LXR/RXR heterodimer in cultured cells and that this effect was reversed by mevalonate and by LXR-activating oxysterols but not by farnesol or geranylgeraniol. In rat hepatoma cells, mevastatin (compactin) reduced SREBP-1c gene expression and this effect was reversed by mevalonate, T or 22(R)-HC [111]. Gouedard et al. [112] have demonstrated that statins reduce the expression of paraoxonase-1 (PON1) an antioxidant and atheroprotective enzyme in cultured human hepatocytes, and that 22(R)-HC normalized PON1 synthesis. In murine RAW264.7 macrophages, pravastatin reduced the expression of ABCA1 and ABCG1 and their levels were restored by mevalonate and 22(R)-HC [113]. However, in the same study, pravastatin had no effect on ABCA1 in human HepG2 hepatocytes, suggesting that the sensitivity to statin-induced oxysterol depletion may be cell-specific. Consistently with this notion, Sone et al. [114] observed that atorvastatin, fluvastatin, simvastatin, and lovastatin reduced ABCA1 mrna level by almost 90% in various human and murine macrophage cell lines, had no effect on ABCA1 in Swiss 3T3 fibroblasts and human embryonic kidney HEK293 cells, and increased ABCA1 by more than 2-fold in HepG2 cells. The inhibitory effect on ABCA1 in macrophages was reversed by mevalonate or 22(R)- HC. Recently, it has been demonstrated that atorvastatin suppresses not only LXR target genes, ABCA1, and apo- E, but also CYP27 in THP-1 macrophages [115]. Zanotti et al. [116] observed that compactin and pitavastatin impaired cholesterol and phospholipid efflux from murine peritoneal macrophages, which was accompanied by reduced ABCA1 expression, and mevalonate, as well as the mixture of 22(R)-HC and 9cRA, reversed this effect. Reduction of ABCA1 by statins was also observed in cultured human keratinocytes [117]. Haq et al. [118] have demonstrated that pravastatin reduced the expression of LXR target genes, retinal dehydrogenase-1 and -2, in the liver. It should be noted that statins may downregulate ABCA1 also in LXR-independent manner. In particular, statins, by depleting intracellular cholesterol, activate SREBP-2 which is a negative regulator of ABCA1 [119]. Recently, it has been demonstrated that statins downregulate ABCA1 and ABCG1 in human macrophages when cholesterol content in these cells is low, but not when cholesterol content is high. This inhibitory effect of statins was abolished by natural and synthetic LXR agonists [120]. While the effect of statins on cholesterol-derived oxysterols such as 24(S)-HC, 25-HC, or 27-HC is controversial, a more consistent effect should be expected on 24(S),25- EC, a precholesterol product of the mevalonate cascade (Fig. 2), since the rate of synthesis of this compound closely parallels the rate of cholesterol production. Wong et al. [121] have shown that statins reduce 24(S),25-EC level in primary and cultured macrophages, which is accompanied by decrease in ABCA1 and ABCG1 mrna and protein levels and impaired apoa-i mediated cholesterol efflux. The effect of statins on cholesterol efflux pumps was reversed by mevalonate as well as by exogenous 24(S),25-EC. In contrast, Argmann et al. [122] have shown that atorvastatin stimulated apoa-i or HDL-dependent cholesterol efflux despite reducing 24(S),25-EC synthesis in human THP-1 macrophages. This effect was abolished by mevalonate, farnesyl pyrophosphate, and also by GGPP (which is not converted to cholesterol). In addition, protein prenyltransferase inhibitors mimicked the effect of atorvastatin. These data suggest that atorvastatin stimulates cholesterol efflux by attenuating isoprenylation of proteins. Specifically, the effect of atorvastatin was abolished by inhibitors of geranylgeranylated Rho proteins and Rho-dependent protein kinase (ROCK). Interestingly, PPARγ or LXR antagonists also abolished the beneficial effect of atorvastatin. Further studies revealed that atorvastatin, by reducing GGPP availability, reduces geranylgeranylation of RhoA protein leading to decreased ROCK-dependent phosphorylation of PPARγ. Dephosphorylated PPARγ has greater activity and stimulates LXR expression, ultimately leading to the upregulation of ABCA1. These data indicate that statins may modulate LXR independently of affecting oxysterols. Interestingly, atorvastatin administered for 3 weeks prevented age-related reduction of LXRα and increased its transcriptional activity in the rat liver [123]. Various mechanisms Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd 309

14 Liver X Receptors Bełtowski Figure 4 Mechanisms through which statins can modulate ABCA1- mediated cholesterol efflux. through which statins may regulate LXR-ABCA1 pathway are presented in Figure 4. Taking into account all these considerations, it seems that a combination of statins and LXR agonists might be especially useful in atheroprotection. Statins markedly reduce intracellular cholesterol synthesis and upregulate LDL receptors leading to the reduction of plasma LDLcholesterol level. Although statins raise HDL to some extent, this effect is rather limited. On the other hand, LXR agonists markedly increase HDL-cholesterol and, at the cellular level, stimulate cholesterol efflux. Thus, such combination may have potentially very beneficial effect on the lipid profile and intracellular cholesterol balance. In addition, LXR agonists may prevent some adverse effects of statins associated with the depletion of oxysterol pool. Interestingly, although oxidosqalene:lanosterol cyclase (OSC) is involved in both the regular and shunt pathways of cholesterol synthesis (Fig. 2), it preferentially cyclizes dioxidosqalene. Therefore, at the moderate concentrations selective OSC inhibitors reduce cholesterol synthesis but increases 24(S),25-EC synthesis, consequently enhancing ABCA1- and ABCG1-mediated cholesterol efflux in LXR-dependent manner [121,124]. Thus, OSC inhibitors may have favorable pharmacologic activity, simultaneously inhibiting cholesterol synthesis and stimulating 24(S),25-EC/LXR pathway [125,126]. Fibrates Fibrates are specific agonists of PPARα and are commonly used in the treatment of hyperlipidemia. Fibrates markedly reduce plasma triglycerides, slightly decrease LDL-cholesterol and elevate HDL-cholesterol. Fibrates as well as a nonfibrate PPARα agonist, Wy14643, increase LXRα expression in the liver both in vitro and in vivo [127]. In addition, 4-month treatment with fenofibrate resulted in the upregulation of LXRα in peripheral blood monocytes [128]. These data suggest that fibrates may enhance LXR signaling by upregulating the expression of these receptors. Very interesting data about fibrate-lxr relationship were provided by Thomas et al. [129]. These authors demonstrated that fenofibrate, which is used in the form of ester, is a potent LXR antagonist. Fenofibrate binds to LBD of LXR and displaces oxysterol ligands. Fenofibrate antagonizes the stimulatory effect of T on SREBP-1c and FAS in hepatocytes, and reduces basal level of these proteins by antagonizing the effect of endogenous oxysterols. Interestingly, fenofibrate ester binds to LXR with greater affinity than to PPARα. In vivo,fenofibrate is rapidly hydrolyzed to fenofibric acid which has no LXR-antagonistic activity. However, fenofibrate ester is detected in the liver of mice treated with fenofibrate, suggesting that LXR blockade may contribute to some pharmacological effects of this drug (e.g., reducing plasma triglycerides). Interestingly, fenofibrate has no effect on LXR-induced ABCA1 expression in the liver and macrophages suggesting that its effect is gene-specific. In contrast to fenofibrate, other fibrates are used as free acids which are inactive toward LXR [129]. Two studies [130,131] have demonstrated that PPARα agonists stimulate ABCA1 expression in cultured macrophages. On the other hand, the observation that fenofibrate blocks the effect of T on SREBP-1c but not on ABCA1 [129] suggests that coadministration of fenofibrate might be useful to prevent lipogenesis induced by LXR agonists. Indeed, cotreatment with fenofibrate and T caused a synergistic elevation of HDL but attenuated T induced hypertriglyceridemia and hepatic steatosis [132]. Thus, a dual LXR/PPARα agonists may have a more beneficial pharmacologic profile than pure LXR stimulators. Thiazolidinediones Thiazolidinediones (rosiglitazone, pioglitazone, and ciglitazone) are specific agonists of PPARγ and are used to improve insulin sensitivity in patients with type 2 diabetes. Although not classified as hypolipidemic drugs, thiazolidinediones affect the lipid profile. Many studies indicate that thiazolidinediones stimulate LXR signaling and thus a part of their beneficial effects may be mediated by LXR-dependent pathway. First, PPARγ agonists increase the expression of LXRα both in vitro [133] and in vivo [134]. Consequently, pioglitazone and rosiglitazone increased ABCA1 expression and cholesterol efflux from cultured macrophages. Apart from modulating LXR expression, PPARγ agonists may influence oxysterol formation. Rosiglitazone and pioglitazone 310 Cardiovascular Therapeutics 26 (2008) c 2008 The Authors. Journal Compilation c 2008 Blackwell Publishing Ltd