Foam Cell Specific LXRα Ligand

Objective The liver X receptor α (LXRα) is a ligand-dependent nuclear receptor and the major regulator of reverse cholesterol transport in macrophages. This makes it an interesting target for mechanistic study and treatment of atherosclerosis. Methods and Results We optimized a promising stilbenoid structure (STX4) in order to reach nanomolar effective concentrations in LXRα reporter-gene assays. STX4 displayed the unique property to activate LXRα effectively but not its subtype LXRβ. The potential of STX4 to increase transcriptional activity as an LXRα ligand was tested with gene expression analyses in THP1-derived human macrophages and oxLDL-loaded human foam cells. Only in foam cells but not in macrophage cells STX4 treatment showed athero-protective effects with similar potency as the synthetic LXR ligand T0901317 (T09). Surprisingly, combinatorial treatment with STX4 and T09 resulted in an additive effect on reporter-gene activation and target gene expression. In physiological tests the cellular content of total and esterified cholesterol was significantly reduced by STX4 without the undesirable increase in triglyceride levels as observed for T09. Conclusions STX4 is a new LXRα-ligand to study transcriptional regulation of anti-atherogenic processes in cell or ex vivo models, and provides a promising lead structure for pharmaceutical development.


Introduction
Atherosclerosis and cardiovascular diseases have become enormous health problems during the last decades. Macrophages play a pivotal role in the development and progression of atherosclerosis [1]. A key event of atherosclerosis involves the uncontrolled uptake of oxidized low density lipoproteins (oxLDL) in macrophages, which agglomerate at the subendothelial space of blood vessel walls [2]. When macrophages fail to restore their cellular cholesterol homeostasis via regulate reverse cholesterol transport (RCT) they form diseased foam cells, main components of fatty streaks [3]. The ligand-depend transcription factors LXRa and LXRb act as cholesterol sensors and respond to elevated oxysterol levels by activating target genes with impact on cholesterol metabolism and atherosclerosis [4]. Many approaches have been pursued to identify a promising LXR ligand with exclusively beneficial properties [5,6]. Unfortunately, LXRs are also implicated in other counteracting physiological processes, such as triglyceride synthesis [7]. A current challenge is to develop powerful and selective LXR modulators as biochemical tools for mechanistic studies on LXR biology, and to provide pharmaceutically exploitable compounds without mentioned adverse sideeffects [8].
While in the past selective agonists for the ubiquitously expressed subtype LXRb showed in general low efficiency to counteract atherogenic processes, recent evidence suggests that rather the LXRa subtype plays a crucial role for regulating antiatherogenic gene expression profiles [9]. In human macrophages LXRa is regulated via a feed-forward autocatalytic loop leading to significantly increased levels by LXRa ligand-treatment [10]. In mouse macrophages this effect was not observed, indicating that activation in human foam cells differs strongly from mouse foam cells.
Here we introduce an LXRa ligand that specifically activates target genes in diseased human foam cells but not in macrophages. This activation resulted in significant reduction of total and esterified cholesterol with similar potency as for the synthetic LXRa/b ligand T0901317 (T09) -without the observed counteracting increases of triglyceride levels known from previously published LXRa/b ligands.

LXRa Ligand Screening
We used the LXR alpha coactivator kit (Invitrogen) and screened a biologically diverse library of .7,000 compounds for nanomolar binders according to manufacturer's instructions. T0901317 (T09) was obtained from Sigma-Aldrich.

STX4 Ligand Synthesis
All chemicals were obtained from commercial suppliers and used without further purification. 1 13

Cell Culture
We used the human THP1 cell line, which is widely applied as a macrophage and foam cell model with similar properties as primary cells [11]. THP-1, a human acute monocytic leukemia cell line, was obtained from Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ) and cultivated at 37uC (5% CO 2 , 95% humidity) in RPMI 1640 (Invitrogen) supplemented with 10% FBS Superior (Biochrom). For generating macrophages differentiation was induced by 10 28 M Phorbol 12-myristate 13acetate (PMA) (Sigma-Aldrich) for 48 h. Ligand treatment of macrophages was performed for further 24 h. To induce foam cell formation macrophages were treated for additional 48 h with 100 mg/mL modified human oxidized low density lipoprotein (oxLDL) obtained from Source BioScience (RP-048). Ligand treatment was performed simultaneously with foam cell formation. Cholesterol loading and treatment was controlled with Oil Red O staining (Alfa Aeser). To investigate ligand induced effects on cell proliferation of THP-1 macrophages CellTiter-Glo Luminescent Cell Viability Assay (Promega) was utilized.

LXR Knockdown
Target specificity of ligand-dependent gene expression effects was investigated in siRNA-mediated LXRa and b-knockdown foam cells with subsequent qPCR or microarray analysis. Therefore, 2610 5 THP-1 cells/well were differentiated in 24-well plates and induced to foam cells prior to knockdown as described above. Foam cells were transfected using TransIT-TKO transfection reagent (Mirus Bio), for combined knockdown with 15 nM LXRa Silencer Validated siRNA (Ambion, ID 5458) and 15 nM LXRb Silencer Select Validated siRNA (Ambion, ID s14684) and with 30 nM LXRa or LXRb for individual knockdowns, respectively. As control we used 30 nM Silencer Select Negative Control #1 (Ambion). Transfection was carried out for 48 h and followed by treatment with 10 mM ligand or vehicle control for 24 h.

Real-time Polymerase Chain Reaction
Total RNA isolation and purification was performed by using the RNeasy Mini Kit (QIAGEN) and RNase-Free DNase Set (QIAGEN). cDNA was generated by reverse transcription using the High Capacity cDNA Reverse Transcription Kit (Life Technologies). Gene expression was quantified using the SYBR Green PCR Master Mix (Life Technologies) and ABI 7900HT Fast Real Time PCR System (Life Technologies). In each reaction 0.8 ng/mL cDNA and 0.2 mM primers were used. Gene expression was calculated with the DDCt-method. ß-Actin was used for normalisation.

Microarray Analysis
Microarray analyses were performed according to instructions of Illumina's TotalPrep RNA Amplification Kit followed by hybridisation on HumanHT-12 v3 or HumanHT-12 v4 Expression BeadChips (Illumina). Data analysis was performed with GenomeStudio V2011.1 (Illumina). Differential expression analysis was performed on background-subtracted data with cubic spline normalization and Benjamini-Hochberg FDR correction. Significant data was considered to have a detection p-value of #0.01 and differential expression p-value of #0.05 according to Illumina's t-test error model. Gene expression data were submitted in MIAME-compliant form to the NCBI Gene Expression Omnibus database (GSE39079) [12]. Bead array data were validated by quantitative real-time PCR. The Gene Distance Matrix was conducted in Euclidean space using the MeV 4.3 software tool [13]. Gene Set Enrichment Analysis (GSEA) was performed as described elsewhere [14] with the following parameters: 1000 gene set permutations, weighted enrichment **P,0.01, ***P,0.001 vs. negative siRNA. B, Gene expression in THP-1 macrophages and foam cells after treatment with T09 (10 mM) or STX4 (10 mM). Data are expressed as mean6SEM (n = 4). C, Western blot analysis of LXRa and APOE content in foam cells and STX4 treated foam cells. Bar plot displays the results of densitometry analysis (mean6SEM, n = 3). *P,0.05, **P,0.01 vs. foam cell.D, Transcriptional activation of LXRa by STX4 in the presence or absence of 200 nM T09 (see also table 1). Data are expressed as mean6SD (n = 3). E, Gene expression in THP-1 macrophages after treatment with different concentrations of T09 or STX4 in the presence or absence of 1 mM T09. Data are expressed as mean6SEM (n = 2-3). *P,0.05, **P,0.01, ***P,0.001 vs. DMSO; n.s., not significant. doi:10.1371/journal.pone.0057311.g002 statistics, minimal gene set size of 5, and signal-to-noise metric. Enrichment of pathways was tested using the Reactome (v3.0) and the KEGG (v3.0) database from the Molecular Signature Database (MSigDB). Heatmaps were carried out with Mayday 2.8 [15]. For presentation of treatment effects in the heatmap, the pathway normalized enrichment score (NES) was adjusted with the appropriate FDR as follows: adjustedNES = (12FDR)6NES.

Western Blotting
Whole cell extracts were harvested from three biological replicates and denaturized for SDS-PAGE (NuPAGE Novex 4-12% Bis-Tris Gels). Proteins were blotted on nitrocellulose Hybond ECL membrane and incubated with strictly validated published antibodies against LXRa (Abcam, ab 41902), APOE (Abcam ab1906), and as housekeeping protein control b-Actin (Santa Cruz Biotechnology, sc-47778, C4). Secondary antibodies were HRP labeled anti-mouse and anti-rabbit (Santa Cruz Biotechnology). After detection with Western Lightning Plus-ECL solution membranes were stripped with Restore Plus Western Blot Stripping Buffer. Densitometry was performed in ImageQuant TL (GE Healthcare). This tool measures quantitatively the optical density and provides more accurate data than simple eye-observations. The calculation of log (protein of interest/b-Actin) corrects data for loading imbalances.

Cholesterol and Triglyceride Analyses
THP-1 monocytes were plated out into 24-well plates (Nunc) at a density of 2610 5 cells/well. After differentiation and treatment cells were lysed with 100 mL lysis buffer (PBS, 0.25 M NaCl, 1% Triton X-100). Total and free cholesterol were determined using the colorimetric enzymatic Amplex Red Cholesterol Assay Kit (Invitrogen). Triglycerides were quantified by Triglyceride Assay Kit (BioVision). For normalisation protein content was determined with Pierce 660 nm Protein Assay Reagent (Thermo Scientific).

Statistical Analyses
Statistical significance was determined by unpaired two-tailed Student's t-test for single comparisons and one-way ANOVA with Dunnett's post-hoc test for multiple comparisons. Statistical analyses were carried out using GraphPad Prism 5.0. A p-value #0.05 was defined as statistically significant.

Results
During the screening of a natural products library [16], we identified the stilbenoid backbone as a suitable candidate structure for activating LXR. Thereby, we observed an increased activation potential of stilbenoids with an epoxide (STX4, Figure 1A). To analyse binding and activation ability of the new ligand STX4 we applied dual luciferase reporter-gene assays with LXRa and LXRb ligand binding domains (LBD) in HEK-293 cells ( Figure 1B and Figure 1C). Whereas the control ligand T09 bound to LXRa and LXRb with similar affinity, STX4 showed characteristic specificity for LXRa (EC 50 = 35 nM) but less efficient transcriptional activation (6% vs. T09). We additionally measured the LXRa activation potential of the natural ligand 22-R-Hydro-xycholesterol (EC 50 = 13.3 mM, Figure 1B) and observed even lower transcriptional activation (4% vs. T09).
As for T09, viability assays revealed no cytotoxic effects of STX4 up to 50 mM ( Figure 1D), making this compound suitable for cell culture studies.
To test for ligand specificity we performed LXRa and LXRb knockdown experiments (Figure 2A). For the T09 ligand we detected in absence of both subtypes a significant reduction of LXRa and ABCA1 expression, whereas STX4 treatment showed a higher dependency on the LXRa subtype. Notably, the STX4 knockdown study was less robust than with T09 due to apparent cross-reactivity of STX4 with the transfection reagent. However, consistent with the above described reporter gene assays we observed in foam cells higher LXRa than LXRb specificity of STX4.
For testing the activation of LXRa in human macrophages and in diseased human foam cells we performed gene expression analyses of central LXRa target genes (namely LXRa itself, SREBF1, ABCA1, and APOE in Figure 2B). In macrophages we observed increased expression of LXRa target genes upon T09 treatment, but not with STX4. Strikingly, in foam cells STX4 treatment resulted in potent target gene expression similar to T09, which was consistent with increased protein expression levels of LXRa target genes ( Figure 2C).
In a competitive reporter-gene assay with a fixed concentration of T09 and increasing levels of STX4, we detected additive effects on reporter gene activation (efficacy range 64%-111%, Figure 2D, Table 1) and ABCA1 expression in macrophages ( Figure 2E), indicating conditional activation potency of STX4 in foam cells. These data suggest that activation of LXRa upon STX4 treatment occurred selectively under the diseased condition of the foam cell.
To study the genome-wide effects of STX4 treatment we performed microarray-based expression analyses on STX4-, T09or DMSO-treated macrophages and foam cells ( Figure 3A). Consistent with qPCR data (Figure 3B), expression of genes involved in lipid metabolism processes was activated in T09treated but not in STX4-treated macrophages. However, in foam cells we found a high similarity between STX4 and T09 treatment ( Figure 3C). Lipid metabolism processes were comparably upregulated by T09 and STX4 in foam cells ( Figure 3A), which consequently resulted in a significant reduction of total and esterified cholesterol content in contrast to macrophage treatments ( Figure 4A and Figure 4B).
Interestingly, we noted a difference between T09 and STX4 treatments in terms of triacylglyceride biosynthesis. While this process was transcriptionally induced by T09 treatment, it was not regulated by STX4 treatment in foam cells ( Figure 3A). Consistently, treatment with T09 but not STX4 led to increased triglyceride levels in foam cells ( Figure 4C). In macrophages the triglyceride level stayed rather unchanged for both treatments ( Figure 4D).
Notably, despite the observed up-regulation of SREBF1 mRNA in STX4-treated cells ( Figure 2B), which is a common marker gene for lipogenesis, triglyceride levels were not increased. This example illustrated the strength of the applied gene set enrichment analysis method to decipher functional pathways and explain complex effects of nuclear receptor ligands compared to limited information from conventional single gene centered assessments. Further studies are needed to explore the observed effects. Anyway, the LXRa agonist STX4 featured a high potential to reduce excess cholesterol without undesirable increase in triglycerides in foam cells.
In summary, STX4 is a foam cell specific gene regulating molecule, which showed no significant activity in normal macrophages. This new LXRa ligand can be applied for investigating mechanistic aspects of macrophage homeostasis during (anti-) atherogenic processes, and can be used as a lead structure for pharmaceutical development.

Discussion
Specific ligands for LXRs are important tools to combat atherosclerosis and cardiovascular diseases [17]. In this study we introduced the new LXRa-specific ligand STX4, which has a stilbenoide structure that is derived from the natural LXR ligand oxysterol or sterols in general.
Our data suggest that the unique conditional activation potential of STX4 in diseased foam cells depends on a partner ligand. As we could show with reporter-gene assays and gene expression analyses, STX4 cooperates with the synthetic ligand T09 as well as with natural ligands derived from oxLDL in an additive manner. An explanation for this behaviour could be allosteric binding of STX4 to the ligand binding domain. In addition we assume that a changed set of transcriptional LXR cofactors in foam cells could be recruited by STX4 and thereby drive its specific actions. Due to its autoregulatory activation potential, LXRa has highly increased protein content in foam cells compared to macrophages [10]. Strikingly, STX4 was not efficient in activating LXR-target genes in macrophages but only in foam cells. Lipid loaded foam cells may provide a chemically favourable hydrophobic, fatty acids rich environment for the STX4 compound. Moreover, the level of LXRa in foam cells in our experiments was at least 10-fold higher than for LXRb (data not shown), suggesting that the observed effects of STX4 were mainly driven by the alpha-subtype. This hypothesis could be confirmed in reporter gene assays and by trend in LXR knockdown experiments in cell culture ( Figures 1B-C and 2A).
The discussed molecular interaction of STX4 with LXRa could eventually contribute to the observed conditional activation of LXRa pathways in diseased foam cells. The STX4 molecule can be readily applied in cell culture and ex vivo models to study gene regulation processes and resulting metabolic effects.
Most LXRa/b ligands failed as therapeutic targets due to a lack of target gene activation potency or adverse side-effects such as lipogenesis. Due to its more specific anti-atherogenic spectrum the foam cell specific ligand STX4 holds promise for further pharmaceutical development. This work will comprise chemical optimization and modification of the presented STX4 lead structure, testing of chemical stability and potential side effects including for example binding tests with nuclear receptors and other regulating proteins, and applying test panels including absorption, distribution, metabolism and excretion (ADME), prior in vivo analyses in model animals. However, the autoregulation of ligand-activated LXRa in human foam cells varies significantly from mouse foam cells [10], showing one of the severe limitations of using mouse models for studying the physiological effects of drug candidates for treating atherosclerosis [2].
As exemplified in this study, the development of potent compounds for specifically activating key metabolic nuclear receptors in diseased but not in normal target cells provides new avenues for mechanistic studies and for innovative treatment strategies.