Poly(ADP-ribose) Polymerase 1 Represses Liver X Receptor-mediated ABCA1 Expression and Cholesterol Efflux in Macrophages*

Liver X receptors (LXR) are oxysterol-activated nuclear receptors that play a central role in reverse cholesterol transport through up-regulation of ATP-binding cassette transporters (ABCA1 and ABCG1) that mediate cellular cholesterol efflux. Mouse models of atherosclerosis exhibit reduced atherosclerosis and enhanced regression of established plaques upon LXR activation. However, the coregulatory factors that affect LXR-dependent gene activation in macrophages remain to be elucidated. To identify novel regulators of LXR that modulate its activity, we used affinity purification and mass spectrometry to analyze nuclear LXRα complexes and identified poly(ADP-ribose) polymerase-1 (PARP-1) as an LXR-associated factor. In fact, PARP-1 interacted with both LXRα and LXRβ. Both depletion of PARP-1 and inhibition of PARP-1 activity augmented LXR ligand-induced ABCA1 expression in the RAW 264.7 macrophage line and primary bone marrow-derived macrophages but did not affect LXR-dependent expression of other target genes, ABCG1 and SREBP-1c. Chromatin immunoprecipitation experiments confirmed PARP-1 recruitment at the LXR response element in the promoter of the ABCA1 gene. Further, we demonstrated that LXR is poly(ADP-ribosyl)ated by PARP-1, a potential mechanism by which PARP-1 influences LXR function. Importantly, the PARP inhibitor 3-aminobenzamide enhanced macrophage ABCA1-mediated cholesterol efflux to the lipid-poor apolipoprotein AI. These findings shed light on the important role of PARP-1 on LXR-regulated lipid homeostasis. Understanding the interplay between PARP-1 and LXR may provide insights into developing novel therapeutics for treating atherosclerosis.

Atherosclerosis, characterized by the buildup of cholesterolladen macrophages in the arterial wall, is the most common cause of cardiovascular disease (1). Normally, arterial macrophages ingest lipoprotein-derived cholesterol and transfer it to HDL for delivery to the liver and excretion from the body through bile. This process is known as reverse cholesterol transport and has an atheroprotective function because it promotes clearance of lipid particles from the body (2,3). However, persistent hyperlipidemia promotes atherogenesis by overwhelming the efflux process and leading to the formation of lipid-laden macrophage foam cells that are retained within the arterial wall (4,5).
LXRs 2 belong to the nuclear receptor family of transcription factors and are activated physiologically by oxysterol cholesterol metabolites and cholesterol precursors (6 -8). The LXRs, LXR␣ and LXR␤, act as key regulators of lipid homeostasis by controlling the expression of a number of genes involved in cholesterol absorption, transport, and elimination (9). Although the expression of LXR␣ is restricted to macrophages and tissues involved in lipid metabolism, LXR␤ is ubiquitously expressed (10). The transcriptional activity of LXR is dependent on formation of a heterodimer, with retinoid X receptors, that binds to a DNA motif termed LXR response element (LXRE) present in the promoter of LXR target genes (6). The heterodimer in its unliganded state inhibits gene expression by forming a complex with corepressors such as silencing mediator of retinoic acid and thyroid hormone receptor and nuclear receptor corepressor (NCoR) (11). Ligand binding causes a conformational change in the receptor that releases the corepressors and recruits coactivators such as E1A-associated protein p300 (Ep300) and activating signal cointegrator 2 (ASC2) to induce target gene expression (12,13).
LDL receptor Ϫ/Ϫ and ApoE Ϫ/Ϫ mice that are predisposed to hypercholesterolemia and atherosclerosis exhibit regression of plaques and reduced atherosclerosis upon systemic administration of an LXR agonist (20 -22). Moreover, ApoE Ϫ/Ϫ mice deficient in both LXR␣ and LXR␤ die within 10 weeks of age because of extensive accumulation of cholesterol in peripheral tissue macrophages (9). Recently, ABCA1 was shown to be a critical mediator of the anti-inflammatory effects of LXR (23). Given the beneficial roles of LXRs in modulating ABCA1 gene expression, there remains a lack of understanding of the cellular factors controlling LXR activity in a gene-specific manner.
Nuclear receptors serve as the best examples of transcriptional regulation through the targeted recruitment of protein complexes that reversibly alter chromatin to activate or repress gene expression (24). The nature of nuclear receptor interactions with transcriptional cofactors can be influenced by conformational changes in the receptor induced by ligands, posttranslation modifications, and DNA elements to selectively affect target gene expression. We hypothesized that LXR activity could be modulated in a gene-specific manner by altering the function of receptor-associated transcriptional regulators that selectively act at certain target genes but not others. Here we have performed a mass spectrometry-based proteomics study and identified poly(ADP-ribose) polymerase-1 (PARP-1) as a novel LXR interacting protein that selectively regulates LXR-dependent ABCA1 expression in macrophages. PARP-1, the founding member of the PARP superfamily, is a highly expressed and ubiquitous nuclear protein that regulates many nuclear processes (25). It is an enzyme responsible for the majority of cellular ADP-ribosylation (26), which is a reversible post-translational modification that occurs via covalent transfer of poly-ADP-ribose units from NAD ϩ to glutamine, asparagine, lysine, and/or arginine amino acids of target proteins (27). PARP-1 catalyzes the polymerization of ADP-ribose units to form linear or branched ADP-ribose chains on target proteins, thus modifying them to various masses. Although PARP-1 was originally known for its roles in DNA repair pathways, it is now well established that PARP-1 also regulates gene transcription both under basal and signal-activated conditions (25). In this study, we demonstrate that in macrophages LXR transcriptional activity at the ABCA1 gene can be increased by reducing PARP-1 levels or by inhibiting the catalytic activity of PARP-1, thereby facilitating cholesterol efflux.

Experimental Procedures
Cell Culture-HEK293T cells obtained from the ATCC were infected with recombinant retroviruses that were produced by transfecting LZRSpBMN-GFP or LZRSpBMN-GFP/LXR␣ into 293GP cells to create stable lines expressing vector only (293T-Vo) or FLAG-tagged human LXR␣ (293T-LXR␣), respectively. After infection, the cells positive for green fluorescence protein expression were sorted by fluorescence-activated cell sorting and cultured in DMEM (Corning) containing 10% FBS and 1 unit/ml penicillin and 1 g/ml streptomycin. HEK293 and RAW 264.7 cells from ATCC and RAW 264.7 cells stably expressing FLAG-tagged human LXR␣ previously described (28) were also maintained under the same culture conditions. The cells were routinely tested for mycoplasma and were myco-plasma-free. Bone marrow-derived macrophages were prepared from monocytes harvested from the tibia and femur of 6 -8-week-old C57BL/6 male mice. Bone marrow cells were incubated in red blood cell lysis buffer (Sigma) to remove red blood cells. The cells were then resuspended and maintained for a week in DMEM containing 20% FBS and 10 ng/ml macrophage colony-stimulating factor (PeproTech Inc., Rocky Hill, NJ) to differentiate them into unactivated (M0) macrophages.
Affinity Purification of FLAG-LXR␣ Protein Complexes-HEK293T-LXR␣ or HEK 293T-Vo cells were lysed by incubating cells in hypotonic buffer (10 mM HEPES, 1.5 mM MgCl 2 , 10 mM KCl, pH 7.9) and passing them 10 times through a 25-gauge syringe needle. The crude nuclear pellet obtained after centrifugation was resuspended in 10 mM HEPES, 25% glycerol, 420 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA solution at pH 7.9. Supernatant containing the nuclear proteins was collected after centrifugation. We then brought the final concentration of the solution to 300 mM salt and 1% Triton X-100 by adding HEMG0 buffer containing 25 mM HEPES, 12.5 mM MgCl 2 , 10% glycerol, 1 mM EDTA, and Triton X-100. FLAG-LXR␣ was immunoprecipitated from the nuclear extracts using agarose beads conjugated to FLAG antibody (Sigma). The beads were incubated with the samples for 6 h and were washed three times in HEMG0 buffer with 300 mM KCl and then in HEMG0 buffer with 150 mM KCl and two times in TBS. The beads were incubated in 50 l of TBS with 0.5 mg/ml FLAG peptide (Sigma F3165) for 1 h, with gentle mixing of the solution every 10 min (29,30). The proteins in the supernatant were precipitated with TCA overnight. The TCA precipitate was processed and then subjected to analysis by Multidimensional Protein Identification Technology and LTQ and LTQ orbitrap mass spectrometry as described previously (31).
Mass Spectrometry-TCA precipitate was resuspended in 8 M urea, and the extracts were processed with ProteasMAX (Promega, Madison, WI) following the manufacturer's instructions. The samples were subsequently reduced by incubation with 5 mM tris-(2 carboxyethyl) phosphine at room temperature for 20 min and alkylated in the dark by treatment with 10 mM iodoacetamide for an additional 20 min. The proteins were digested overnight at 37°C with sequencing grade modified trypsin (Promega). The reaction was stopped by acidification with formic acid.
Multidimensional Protein Identification Technology and LTQ Mass Spectrometry-The protein digest was pressureloaded onto a 250-m inner diameter capillary packed with 2.5 cm of 10-m Jupiter C18 resin (Phenomenex, Torrance, CA) followed by an additional 2.5 cm of 5-m Partisphere strong cation exchanger (Whatman, Clifton, NJ). The column was washed with buffer solution containing 5% acetonitrile, 0.1% formic acid, and 95% water. Next a 100-m inner diameter capillary with a 5-m pulled tip packed with 15 cm of 4-m Jupiter C18 resin (Phenomenex) was attached to the filter union, and the entire split column (desalting column-filter union-analytical column) was placed in line with an Agilent 1100 quaternary HPLC (Palo Alto, CA) and analyzed using a modified five-step separation as described previously (63). The buffer solutions used were buffer A (5% acetonitrile, 0.1% formic acid), buffer B (80% acetonitrile, 0.1% formic acid), and buffer C (500 mM ammonium acetate, 5% acetonitrile, 0.1% formic acid).
Step 1 consisted of a 75-min gradient from 0 -100% buffer B, and steps 2-5 had a similar profile except 100% buffer A for 3 min, X% buffer C for 5 min, gradient from 0 to 15% buffer B for 10 min, and a 105-min gradient from 10 to 55% buffer B (except for step 5, which %B was increased from 10% to 100%). The 5-min buffer C percentages (X) were 10, 40, 60, and 100%, respectively, for the five-step analysis. As peptides eluted from the microcapillary column, they were electrosprayed directly into an LTQ mass spectrometer (Ther-moFinnigan, Palo Alto, CA). For LTQ analysis, as peptides eluted from the microcapillary column, they were electrosprayed directly into an LTQ two-dimensional ion trap mass spectrometer (ThermoFinnigan) with the application of a distal 2.4-kV spray voltage. A cycle of one full scan mass spectrum (400 -2000 m/z) followed by seven data-dependent MS/MS spectra at a 35% normalized collision energy was repeated continuously throughout each step of the multidimensional separation. Application of mass spectrometer scan functions and HPLC solvent gradients were controlled by the Xcalibur data system.
Analysis of Tandem Mass Spectra-Protein identification and quantification analyses were done with Integrated Proteomics Pipeline (IP2, Integrated Proteomics Applications, Inc., San Diego, CA) using ProLuCID, DTASelect2, and Census. Tandem mass spectra were extracted into ms1 and ms2 files from raw files using Raw Extract 1.9.9 and were searched against IPI human protein database (version 3_57_01, released on January 1, 2009; plus sequences of known contaminants such as keratin and porcine trypsin concatenated to a decoy database in which the sequence for each entry in the original database was reversed using ProLuCID/Sequest. LTQ data were searched with 3000.0 milli-amu precursor tolerance, and the fragment ions were restricted to a 600.0-ppm tolerance. All searches were parallelized and performed on the Garibaldi 64-bit LINUX cluster with 2848 cores at the Scripps Research Institute. Search space included all fully and half-tryptic peptide candidates with no missed cleavage restrictions. Carbamidomethylation (ϩ57.02146) of cysteine was considered a static modification, and we require two peptides per protein and at least one tryptic terminus for each peptide identification. The ProLuCID search results were assembled and filtered using the DTASelect program (version 2.0), with a false discovery rate of 0.05; under such filtering conditions, the estimated false discovery rate was less than 1% at the protein level in all analyses.
Preparation of Whole Cell Extracts and Immunoprecipitation-The cells were washed twice in PBS and lysed in Triton lysis buffer (50 mM HEPES, pH 7.6, 150 mM NaCl, 1 mM EDTA, pH 8.0, 1 mM EGTA, pH 8.0, 1 mM NaF, 1% Triton X-100, 10% glycerol) with protease inhibitor mixture (Cell Signaling). Protein concentrations were measured by using a Bradford assay (Bio-Rad). For immunoprecipitation, the lysates were incubated with FLAG antibody-conjugated agarose beads or protein G magnetic beads (Invitrogen) cross-linked to either 5 g of LXR␣ antibody (Abcam ab41902) or mouse IgG (Sigma-Aldrich) overnight at 4°C. The beads were washed once in the lysis buffer and three times in TBS. Proteins associated with the antibody were eluted in TBS by competition with FLAG peptides as described above for FLAG beads and by boiling at 95°C in the presence of 5ϫ Laemmli sample buffer and ␤-mercaptoethanol for 5 min.
CRISPR-Cas9-mediated PARP-1 Deletion-A guide sequence 5Ј-CGA GTC GAG TAC GCC AAG AGC GG-3Ј targeting exon 1 of human PARP-1 gene was designed by using CRISPR design tool. The sequence was cloned into the expression plasmid pSpCas9 (BB)-2A-GFP (32), which bears a guide RNA backbone, Cas9, and GFP. The resulting plasmid was transfected into HEK293 cells. After 24 h, single cells were sorted for the expression of GFP into 96-well plates by FACS. Clones deficient in PARP-1 were identified by Western blot using PARP-1 antibody.
Transfection of Plasmids and siRNA and Pharmacological Inhibition of PARP-1-RAW-LXR␣ cells were transfected with constructs that express human PARP-1 (pCMV6-PARP-1; Ori-Gene Technologies, Rockville, MD) or the vector construct (pCMV6-Entry) as a control using Lipofectamine LTX (Life Technologies) following the manufacturer's instructions. HEK293 cells were transfected with pCMV6-FLAG-mLXR␣, pCMV6-FLAG-mLXR␤, or their respective vector control with Lipofectamine 2000. For knockdown experiments, RAW-LXR␣ cells were transfected with 50 nM On-Targetplus siControl nontargeting pool or the On-Targetplus Smartpool for mouse PARP-1 (GE Dharmacon, Lafayette, CO) with Hiperfect transfection reagent (Qiagen) for 72 h as per the manufacturer's instructions. The cells were cultured overnight in 1% FBS prior to treatment with vehicle or agonists. For PARP inhibition studies, the cells were pretreated with 3-aminobenzamide (3-AB) for 16 h in 1% FBS. The cells were then treated with ligands for 1, 4, and 8 h before harvesting them for ChIP assay, RNA, and protein analysis, respectively, unless otherwise noted. Bodipy

PARP-1 Interacts with LXRs-
We performed shotgun proteomic mass spectrometry (35) to identify LXR␣ interacting proteins from nuclear extracts of HEK293T cells expressing FLAG-tagged human LXR␣ (293T-LXR␣). The number of peptides identified in the FLAG immunoprecipitate was compared between HEK293 control cells expressing the empty vector (293T-Vo) and 293T-hLXR␣ cells. A protein was considered a specific LXR␣ interactor if it was identified in the immunoprecipitated material from LXR␣ expressing cells but not control cells in replicate analyses.
We identified a number of established LXR␣-associated proteins such as retinoid X receptor, the obligatory heterodimeric partner of LXR, and NCoR, thus validating the approach (Table  1). Importantly, a number of proteins that have roles as transcriptional regulators were also revealed as potential LXR␣ interactors. These included TAF15, a TATA-binding proteinassociated factor (36); NAT10, an acetyltransferase that is known to acetylate histones (37,38); SMARCB1 (39); SMARCE1 (40); components of chromatin remodeling SWI/ SNF complex; and PARP-1. Given that PARP-1 impacts the transcriptional activity of several nuclear receptors including estrogen receptor ␣ (41), retinoic acid receptor (42), and farnesoid X receptor (43), we focused on the relationship between PARP-1 and LXR␣ in controlling transcription of target genes important for cholesterol homeostasis.
PARP-1 Represses LXR-mediated ABCA1 Expression-We next investigated the function of PARP-1 in LXR-mediated transcriptional regulation by using HEK293 cells deficient in PARP-1 (293-PARP-1-KO) generated by the CRISPR/Cas9 system ( Fig. 2A). LXR␣ or LXR␤ was ectopically expressed in cell lines generated from two independent clones each of wild-type HEK293 cells (293-WT) and 293-PARP-1-KO cells, and expression levels of the LXR target gene ABCA1 were examined. Interestingly, PARP-1-deficient cells expressing LXR␣ showed significantly higher ABCA1 than WT cells (Fig. 2B). Similarly, we observed increased ABCA1 expression in LXR␤ expressing 293-PARP-1-KO cells compared with 293-WT cells ( Fig. 2C; data not shown for the second clone). The effect of PARP-1 deletion was also reflected in the newly formed nascent mRNA levels, thereby confirming that the change in ABCA1 expression is a result of a transcriptional event (Fig. 2D).
Using an antibody that recognizes the poly(ADP-ribose) modification, we found that in the absence of PARP-1, global poly(ADP-ribosyl)ation (PARylation) was markedly reduced in whole cell extracts (Fig. 2E). PARP activity can be inhibited by the NAD ϩ analog 3-AB that competes with NAD ϩ for the catalytic site of PARP (45). To examine whether PARP-1 catalytic activity was involved in modulating LXR␣ transcriptional activity at the ABCA1 gene, we incubated 293T-LXR␣ cells with 3-AB. Treatment with 3-AB led to a significant reduction in the global PARylation levels (Fig. 2F). Interestingly, inhibition of PARP also resulted in increased ABCA1 mRNA expression, suggesting a role for PARP-1 catalytic activity in LXR-dependent ABCA1 expression (Fig. 2G).

TABLE 1 Identity of LXR␣-associated proteins by mass spectrometry
Listed are nuclear LXR␣-associated proteins recovered by FLAG immunoprecipitations from LXR␣-expressing (LXR␣) and control cells not expressing LXR␣ (Control) identified by mass spectroscopy. Proteins are listed with their international protein index (Accession); spectral counts, which are a reflection of protein abundance; and description.   MAY 20, 2016 • VOLUME 291 • NUMBER 21

JOURNAL OF BIOLOGICAL CHEMISTRY 11177
We next examined whether the regulation of ABCA1 gene expression by PARP-1 was also found in a functionally relevant macrophage cell line. We utilized RAW 264.7 macrophages stably expressing LXR␣ (RAW-LXR␣) and modulated PARP-1 levels in these cells. PARP-1 was depleted using siRNA in RAW-LXR␣ cells, and the expression of LXR target genes including ABCA1, ABCG1, and SREBP-1c was examined. As in the HEK293 cell model, depletion of PARP-1 increased ABCA1 expression upon treatment with the synthetic LXR ligand T0901317 (Fig. 3A). In contrast, ABCG1 and SREBP-1c mRNA (T), 1 M 9-cis-RA (9-cis), or a combination of T0901317 and 9-cis-retinoic acid (Tϩ9-cis) as indicated for 4 h (C, E, and F) or 24 h (D). ABCA1, SREBP1c, and/or ABCG1 mRNA expression levels were measured by qPCR as indicated similarly as above. Experiments were performed three times, and the values were averaged. The error bars represent S.E. Significance was determined using the two-tailed Student's t test. *, p Ͻ 0.05; **, p Ͻ 0.005; ***, p Ͻ 0.0005. levels remained unchanged upon PARP-1 depletion (Fig. 3A). Consistent with this observation, PARP-1 overexpression in RAW-LXR␣ significantly reduced ABCA1 mRNA expression induced by T0901317 (Fig. 3B). Moreover, inhibition of PARP by 3-AB in the RAW-LXR␣ also showed increased LXR-dependent ABCA1 mRNA expression (Fig. 3C). However, the inhibitor treatment did not affect ABCG1 and SREBP-1c mRNA expression (Fig. 3C). In addition to LXR ligands, we treated cells with 9-cis-retinoic acid, which transactivates LXR via its heterodimeric partner retinoid X receptor or with 9-cis in combination with T0901317. 3-AB diminished expression of ABCA1 under all of these ligand conditions (Fig. 3D). Importantly, primary BMDMs treated with 3-AB also showed similar effects on the expression of LXR target genes. Although ABCA1 mRNA levels were significantly enhanced with 3-AB, SREBP1c and ABCG1 levels did not change (Fig. 3E). Moreover, BMDMs cultured from mice deficient in LXR␣ and LXR␤ (LXR␣ Ϫ/Ϫ LXR␤ Ϫ/Ϫ ) were insensitive to the inhibitor treatment with regards to ABCA1 expression (Fig. 3F), suggesting that 3-ABinduced increase in ABCA1 expression is dependent on LXRs. Thus, our results indicate that PARP-1, via its catalytic activity, negatively regulates LXR-dependent transcription in a genespecific manner.
PARP-1 Occupies the ABCA1 Promoter-Because PARP-1 interacts with LXRs and impacts LXR-dependent gene expression, we investigated whether PARP-1 occupies the ABCA1 gene regulatory region containing the LXR binding site. We used primers spanning the LXRE on the ABCA1 gene 85 bp upstream of the transcription start site and a control non-LXR binding site 4 kb upstream of the transcription start site. We found that PARP-1 occupancy was significantly greater at the ABCA1 LXRE compared with that at the far upstream site (Fig.  4). Interestingly, PARP-1 occupancy at the region spanning the LXRE site in the ABCA1 regulatory region was significantly greater upon T0901317 stimulation than that under the basal condition, suggesting ligand-mediated recruitment of PARP-1. PARP-1 occupancy was also observed at the LXRE sites in the promoters of ABCG1 and SREBP-1c, genes that were insensitive to changes in PARP-1 expression. However, the ligandmediated recruitment of PARP-1 was greatest at the ABCA1 locus. Thus, our results demonstrate occupancy of PARP-1 at the promoter proximal ABCA1 LXRE, which is further enhanced upon LXR ligand treatment. We also investigated whether PARP-1 depletion altered LXR␣ binding to the ABCA1 promoter through ChIP assay using an LXR␣ antibody. Depletion of PARP-1 did not change LXR␣ occupancy at the  MAY 20, 2016 • VOLUME 291 • NUMBER 21

JOURNAL OF BIOLOGICAL CHEMISTRY 11179
ABCA1 promoter LXRE, suggesting that the mechanism by which PARP-1 affects LXR-dependent gene regulation does not involve changes in LXR␣ binding to the ABCA1 promoter (Fig. 5).
PARP-1 Poly(ADP-ribosyl)ates LXR␣-Our results strongly indicate that the catalytic activity of PARP-1 controls LXRdependent ABCA1 expression. Given that PARP-1 modifies its target proteins through PARylation, we assessed whether the enzymatic activity of PARP-1 mediated ADP-ribosylation of LXR␣. Recombinant human LXR␣ was incubated with recombinant human PARP-1 in the presence or absence of the donor molecule NAD ϩ in a cell-free system. In the presence of PARP-1 and NAD ϩ , poly(ADP-ribosyl)ated LXR␣ was detected as a higher molecular mass smear that originated at 50 kDa, the molecular mass of LXR␣, indicative of the addition of the poly (ADP-ribose) polymers onto LXR␣ (Fig. 6A). Next, using immunoprecipitation we examined whether LXR␣ is a target for ADP-ribosylation by PARP-1 in mammalian 293-WT and 293-PARP-1-KO cells expressing FLAG-LXR␣. LXR␣ ADP-ribosylation was evident in 293-WT cells expressing PARP-1, whereas it was not detected in PARP-1-deficient 293-PARP-1-KO cells (Fig. 6B). LXR␣ poly(ADP-ribosyl)ation was also detected in primary bone marrow-derived macrophages, which was reduced upon inhibition of PARP-1 by 3-AB (Fig.  6C). LXR␣ activation via the T0901317 ligand did not change the ADP-ribosylation status of the receptor (Fig. 6D). Thus, our results strongly suggest that LXR␣ is poly(ADP-ribosyl)ated by PARP-1.
Inhibition of PARP-1 Activity Increases Macrophage Cholesterol Efflux-Having established that PARP-1 represses LXRdependent expression of ABCA1 mRNA, we examined whether the changes in the mRNA translated into changes in ABCA1 protein expression. We incubated RAW-LXR␣ cells with either vehicle or PARP inhibitor in the absence or presence of ligands T0901317 or T0901317 together with 9-cis-retinoic acid and measured ABCA1 protein levels. We found that macrophages treated with the PARP inhibitor, 3-AB, showed markedly enhanced ABCA1 protein expression upon ligand treatment (Fig. 7A).
In macrophages, ABCA1 facilitates the clearance of excess cellular free cholesterol by pumping out the cholesterol to acceptor proteins, primarily apoA-I. Given the increase in ABCA1 levels by PARP inhibition, we tested whether PARP-1 inhibition would enhance the cholesterol efflux capacity of macrophages. To test this, we used a BODIPY-cholesterol efflux assay, which utilizes fluorescently labeled cholesterol to measure efflux to extracellular acceptors. We found that ABCA1-mediated cholesterol efflux to apoA-I was significantly greater in RAW cells treated with the PARP inhibitor compared with the efflux in vehicle-treated cells (Fig. 7B). We also measured the ability of primary BMDMs treated with vehicle or PARP inhibitor to efflux cholesterol to apoA-I by cholesterol efflux assay using [ 3 H]cholesterol. In the presence of the inhibitor, apoAI-directed cholesterol efflux was greatly enhanced (Fig. 7C). We found that both ABCA1 and LXRs were indispensable for the inhibitor-mediated increase in cholesterol efflux because macrophages from mice lacking LXR␣ and LXR␤ (Lxr␣ Ϫ/Ϫ Lxr␤ Ϫ/Ϫ ) and ABCA1 (Abca1 Ϫ/Ϫ ) did not exhibit changes in cholesterol efflux with the addition of the inhibitor. Our data suggest that in macrophages, PARP-1 is an LXR␣ regulator that selectively represses ABCA1 expression, such that when PARP activity is reduced, cholesterol efflux is enhanced. Chromatin immunoprecipitation was performed using PARP-1 antibody or rabbit IgG. Precipitated DNA was quantified by qPCR using the primers spanning a non-LXRE site 4 kb upstream of the transcription start site of the ABCA1 gene, LXRE site at the ABCA1 promoter 85 bp upstream of the transcription start site, and ABCG1 promoter LXRE and SREBP-1c LXRE sites; normalized to total input chromatin levels; and measured as a percentage of input. The experiment was performed three times, and the values were averaged. The error bars represent S.E. Significance was determined using the two-tailed Student's t test. *, p Ͻ 0.05; **, p Ͻ 0.005; ***, p Ͻ 0.0005. Chromatin immunoprecipitation was performed using LXR␣ antibody or mouse IgG. Precipitated DNA was quantified by qPCR using primers spanning ABCA1 LXRE or a control non-LXRE site, normalized to total input chromatin levels, and measured as a percentage of input. The experiment was performed three times, and the values were averaged. The error bars represent S.E. The two-tailed Student's t test showed no significant (ns) differences in LXR␣ occupancy between si control and si PARP1 groups.

Discussion
PARP-1 has been reported to have roles as a promoter-specific coactivator or corepressor for several DNA binding transcriptional regulators independent of its role in the DNA damage response (25). In fact, PARP-1 has been shown to be a transcriptional coregulator for a number of nuclear receptors including thyroid hormone receptor, estrogen receptor ␣, retinoic acid receptor, and farnesoid X receptor (42,43,46). Here we report that PARP-1 interacts with both LXR␣ and LXR␤, and functions as a gene-specific corepressor of LXR-mediated gene expression. Whereas ABCG1 and SREBP-1C were unaffected by changes in PARP-1 expression or activity, the cholesterol transporter ABCA1 was up-regulated by either PARP-1 depletion or inhibition of its catalytic activity. Consistent with this finding, inhibition of PARP activity significantly increased ABCA1 protein levels and cholesterol efflux to apoA-I, an ABCA1-specific cholesterol acceptor in an ABCA1-and LXRdependent manner. Thus, our study suggests that PARP inhibition in vivo would be anti-atherogenic by virtue of enhanced cholesterol efflux from macrophages.
Indeed, previous reports indicated that PARP-1 activation was pro-atherogenic, whereas PARP inhibition was anti-atherogenic. These conclusions were drawn from studies where factors that promoted atherosclerosis, such as oxidized LDL, hyperglycemia, and H 2 O 2 -stimulated PARP activity (47)(48)(49)(50). PARP-1 was activated within atherosclerotic plaques of ApoE Ϫ/Ϫ mice fed a high fat diet (51). PARP-1 deletion, on the other hand, reduced atherosclerotic lesion formation in high fat diet-fed ApoE Ϫ/Ϫ mice (52), and PARP catalytic inhibitors reduced atherosclerotic plaque burden in mouse models of atherosclerosis (53). Thus, PARP inhibition in vivo is atheroprotective and is consistent with our data that PARP-1 inhibition, through increasing lipid efflux, impeded the accumulation of cholesterol in macrophages, a process that is central to the pathogenesis of atherosclerosis.
Although the mechanisms by which PARP-1 imparts its gene-specific effects on LXR-dependent gene expression are not understood, it was clear that the enzymatic activity of PARP-1 was required. This is distinct from PARP-1 coactivator function for NFB, which requires PARP-1 cleavage and is independent of its catalytic activity (54). PARP-1 has been shown to modify proteins by transferring an ADP-ribose moiety to target substrates. Here, we demonstrated that LXR␣ was PARylated by PARP-1. Several post-translational modifications on LXR␣  MAY 20, 2016 • VOLUME 291 • NUMBER 21 have been previously described including phosphorylation (28), O-linked ␤-N-acetylglucosamine (O-GlcNAcylation) (55), acetylation (37), and sumoylation (56). These modifications on LXR␣ were shown to impact target gene expression through changes in LXR␣ stability, transactivation, and/or recruitment of transcriptional regulatory factors at specific target genes. By analogy, we suggest that PARylation of LXR by PARP-1 could affect coactivator-corepressor interactions at genes such as ABCA1. For example, whereas LXR occupancy was observed at the ABCA1 promoter in the absence and presence of ligand, the coregulator G-protein pathway suppressor 2 (GPS2), a component of the NCoR repression complex, was shown to occupy the ABCA1 promoter in the absence of ligand and was released upon ligand binding. At ABCG1, the opposite was true because both LXR and GPS2 were recruited by ligand (57). Thus, GPS2 in the context of ABCG1 was associated with transcriptional activation, whereas at ABCA1 GPS2 was correlated with transcriptional repression. It is conceivable that PARP-1 maintains the GPS2 containing NCoR repression complex at ABCA1 to modulate expression. Additional experiments will be required to identify the residue(s) of LXR that are PARylated and to interrogate cofactor occupancy at ABCA1 versus ABCG1 to elucidate the functional consequence of PARP-1 in LXR-dependent transcriptional regulation.

PARP-1 Affects ABCA1 Expression and Function
Elegant work from the Luger lab indicated that PARP-1 had histone chaperone function (58). This action of PARP-1, in principle, could modulate ABCA1 expression by facilitating a less open and more repressive chromatin state that precluded LXR binding. However, we did not observe any change in LXR occupancy at the ABCA1 LXRE under basal and ligand-stimulated conditions upon PARP inhibition (Fig. 4). Thus, PARP-1dependent reduction of ABCA1 expression was not a result of altered LXR occupancy at the ABCA1 promoter.
Although PARP-1 binding was observed at LXR target genes that were sensitive and insensitive to PARP-1 activity, there was significantly greater PARP-1 recruitment upon ligand stimulation at the ABCA1 compared with the ABCG1 or SREBP-1c promoters. We suggest that LXR is being modified in an allosteric manner by DNA binding at the ABCA1 LXRE that resulted in a conformation favorable for PARP-1-LXR interaction to affect the regulation for this gene (59,60). Because PARP-1 has a repressive effect on LXR-dependent gene expression, we speculate that enhanced PARP-1 recruitment acts as a negative feedback mechanism to control ABCA1 gene expression in response to ligand. Higher protein abundance of PARP-1 could potentially override such a feedback mechanism and reduce LXR occupancy at ABCA1.
In summary, our data revealed a novel interaction between LXR and PARP-1 and a role for PARP-1 in LXR-dependent expression of ABCA1 in cholesterol efflux. PARP inhibitors have garnered much attention as therapeutic agents in cancer (61). Several reports of PARP activation being critical in atherosclerotic plaque formation and destabilization support the notion that PARP inhibitors may exhibit therapeutic utility in the clinical management of atherosclerosis (62). In fact, the use of PARP inhibitors in mouse models of atherosclerosis has proven beneficial not only in preventing atherogenesis but also in promoting regression of preexisting plaques. It will be crucial . Two exposures (short and long) are shown. Hsp90 levels were measured as a loading control. The Western blot was quantitated using ImageJ software, and the ratio of ABCA1 to HSP90 is shown with DMSO and non-3-AB-treated sample set to 1. B, RAW-LXR␣ cells were labeled with medium containing BODIPY cholesterol, methyl-␤-cyclodextrin, and unlabeled cholesterol for 1 h. Then the cells were equilibrated with vehicle or 3-AB in combination with DMSO or 5 M T0901317 for 18 h before incubating for 4 h with the efflux medium containing extracellular acceptor apoA-I (10 g/ml). apoA-I-mediated efflux was calculated as a percentage of total cholesterol cleared from the cells after accounting for the cholesterol taken out from the cells in the absence of acceptors. C, WT, Lxr␣ Ϫ/Ϫ Lxr␤ Ϫ/Ϫ , and Abca1 Ϫ/Ϫ primary BMDMs were loaded with [ 3 H]cholesterol, and specific efflux to apoA-I was measured in the presence or absence of 5 mM 3-AB. The experiments in B and C were performed three times with similar results. Each efflux experiment was performed in triplicate, and a representative experiment is shown. The error bars represent S.D. Significance was determined using the two-tailed Student's t test. *, p Ͻ 0.05; **, p Ͻ 0.005.
to address what cell types and protein targets are affected by PARP-1 in atherosclerosis. In the context of this study, macrophage-specific PARP-1 knock-out mice will be essential in elucidating the contribution of PARP-1 in atherogenesis by modulating macrophage LXR signaling.