Main

It is well established that the cholesterol biosynthetic pathway is downregulated under conditions in which sterols are abundant through the inhibition of sterol regulatory element-binding protein (SREBP) processing4. Notably, however, under conditions in which hepatic cholesterol content was not enriched, activation of LXRs with the selective synthetic agonist GW3965 also acutely suppressed the expression of sterol synthesis genes in mouse liver (Fig. 1a and Extended Data Fig. 1a). The effect could not be explained by changes in intracellular cholesterol levels, as LXR activation has been shown to lower hepatic cholesterol content5, which would lead to upregulation of the SREBP-2 pathway.

Figure 1: LXR activation inhibits cholesterol biosynthesis and induces LeXis expression.
figure 1

a, qPCR analysis of gene expression in livers from C57BL/6 mice treated by oral gavage with 40 mg kg−1 GW3965 for the indicated time (n = 6 mice per group). All curves are statistically different from baseline expression (P < 0.05, one way analysis of variance (ANOVA)). b, Schematic representation of the LeXis gene locus on an Integrative Genome Viewer (IGV) showing histone marks from LICR ENCODE data. c, qPCR analysis of gene expression in primary mouse hepatocytes treated with GW3965 (GW; 1 μM) and/or the RXR ligand LG268 (LG; 50 nM). DKO, double knockout (LXRα−/− and LXRβ−/−). Results are representative of four independent experiments. d, qPCR analysis of gene expression in livers from male C57BL/6 mice gavaged with 40 mg kg−1 GW3965 before collection at the indicated time (n = 6 per group). e, Gene expression in livers obtained from mice maintained on chow (n = 2 per group) or a Western diet (n = 5 per group). f, Gene expression in primary mouse peritoneal macrophages treated with 1 μM GW3965 and/or 50 nM LG268 for 16 h. Results are representative of four independent experiments. Values are mean ± s.d. (c, f), or mean ± s.e.m. (a, d, e). *P < 0.05; **P < 0.01 (analysis of variance (ANOVA) with multi-group comparison in a, d and e).

PowerPoint slide

Full size image

To investigate the mechanism by which LXRs suppress cholesterol biosynthesis, we performed genome-wide transcriptional profiling on primary mouse hepatocytes treated with vehicle or GW3965 (Extended Data Fig. 1b). The most robustly induced gene in our RNA-sequencing (RNA-seq) analysis was a predicted non-coding RNA annotated as 4930412L05Rik (Extended Data Fig. 1c). Parallel profiling of non-coding and protein-coding transcripts using microarrays also identified 4930412L05Rik as the highest induced transcript (Extended Data Fig. 1d). We named this transcript LeXis (liver-expressed LXR-induced sequence). Notably, the LeXis gene locus lies in close proximity to the canonical LXR target gene Abca1 in mouse. Analysis of chromatin structure from The ENCODE Project6,7 indicated that LeXis and Abca1 were distinct genes with separate promoters (Fig. 1b). We defined the transcripts produced from the LeXis gene using rapid amplification of complementary DNA ends (RACE) (Extended Data Fig. 2). LeXis and Abca1 were induced by LXR and retinoid X receptor (RXR) agonists (LG268 and GW3965, respectively) in primary hepatocytes in an LXR-dependent manner (Fig. 1c and Extended Data Fig. 3a). LeXis was induced in LXRα−/− and LXRβ−/− (also known as Nr1h3−/− and Nr1h2−/−, respectively) hepatocytes, indicating that both LXR isotypes are capable of regulating LeXis (Extended Data Fig. 3b). Induction of LeXis was not sensitive to the protein synthesis inhibitor cycloheximide, and was not dependent on SREBPs, since 25-hydroxycholesterol (which blocks SREBP processing) also induced LeXis (Extended Data Fig. 3c, d).

Administration of GW3965 to mice induced the expression of LeXis in several metabolically active tissues (Fig. 1d and Extended Data Fig. 3e). We also observed a prominent, LXR-dependent induction of LeXis expression in response to Western diet feeding, consistent with a potential role for LeXis in the response to cholesterol excess (Fig. 1e). Despite being physically adjacent, the LeXis and Abca1 loci are regulated independently. LeXis was neither expressed at baseline nor induced by LXR in mouse peritoneal macrophages, a cell type in which Abca1 expression is prominent (Fig. 1f). A luciferase reporter containing the LeXis promoter was induced by LXR and RXR in co-transfection assays (Extended Data Fig. 3f), and we identified an LXR-response element within the LeXis promoter region that was bound by LXRα in chromatin immunoprecipitation and quantitative PCR (ChIP-qPCR) assays (Extended Data Fig. 3g). The coding potential calculator and coding-non-coding index algorithms predict low coding potential of LeXis (Extended Data Fig. 3h, i). In addition, we found no evidence of production of a protein product from LeXis using in vitro transcription–translation assays (Extended Data Fig. 3j).

To explore the function of LeXis in vivo, we transduced mice with adenoviral vectors encoding green fluorescent protein (GFP) control or LeXis (Fig. 2a and Extended Data Fig. 4a). Remarkably, LeXis expression decreased serum cholesterol, but not triglycerides, in chow-fed C57BL/6 mice (Fig. 2a, b). No differences in liver function tests were observed between the two groups, and there was no evidence of ER stress or inflammation (Fig. 2b and Extended Data Fig. 4b, c). Fractionation of lipoproteins revealed reduced cholesterol in both the low-density lipoprotein (LDL) and high-density lipoprotein (HDL) fractions in LeXis-expressing mice (Fig. 2c). The effects of LeXis were distinct from the consequences of hepatic expression of other LXR target genes, such as Abca1 and Idol (also known as Mylip), which raise serum cholesterol8,9.

Figure 2: LeXis expression reduces serum cholesterol and sterol synthesis through a pathway requiring intact SREBP signalling.
figure 2

a, Total serum cholesterol levels in 10-week-old chow-fed male C57BL/6 mice transduced with adenoviral vectors encoding GFP control (Ad-GFP) or LeXis (Ad-LeXis) for 6 days (n = 24 per group). b, Total serum triglycerides levels in the mice shown in a (n = 12–16 per group). c, Cholesterol levels in pooled fractionated serum from mice treated with Ad-GFP or Ad-LeXis. VLDL, very-low-density lipoprotein. d, Analysis of gene expression in livers obtained after 6 days of transduction with Ad-GFP or Ad-LeXis (n = 8 per group). e, Total serum cholesterol levels in chow-fed male Ldlr−/− mice (10 weeks old) transduced with Ad-GFP or Ad-LeXis for 6 days (n = 8 per group). f, Serum cholesterol levels in chow-fed wild-type (WT) or liver-specific SCAP knockout (Scap−/−) mice gavaged with 40 mg kg−1 GW3965 for 2 days. g, Cholesterol levels in pooled plasma fractions from mice shown in f. h, Total serum cholesterol levels in chow-fed Scap−/− mice transduced with Ad-GFP or Ad-LeXis for 6 days. All values are mean ± s.e.m. NS, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 (unpaired two-tailed t-test).

PowerPoint slide

Full size image

Unbiased pathway analysis of global gene expression revealed that the cholesterol biosynthetic pathway was strongly downregulated in LeXis-transduced livers (Extended Data Fig. 4d). These results were validated by qPCR (Fig. 2d). These results suggested that the cholesterol lowering effects of LeXis were due, at least in part, to suppression of cholesterol biosynthesis. Consistent with this interpretation, we observed a strong trend towards lower cholesterol content in the livers of mice overexpressing LeXis (Extended Data Fig. 4e). For reasons that are not yet clear, treatment of isolated primary hepatocytes did not reflect the effects of either LXR agonist treatment or LeXis expression on genes linked to sterol synthesis (Extended Data Fig. 4f, g).

A reduction in plasma cholesterol suggests an increase in lipoprotein clearance or a decrease in sterol production10. To assess the contribution of the low-density lipoprotein receptor (LDLR) to the actions of LeXis, we transduced Ldlr−/− mice with control or LeXis-expressing adenovirus. We observed decreases in plasma cholesterol levels and hepatic cholesterol content in response to LeXis in Ldlr−/− mice, suggesting that the LDLR is not required for LeXis effects (Fig. 2e and Extended Data Fig. 4h, i). To assess the contribution of SREBP-2 signalling to LXR-mediated inhibition of cholesterologenesis, we administered GW3965 to control or liver-specific SCAP (L-SCAP) knockout mice11. Consistent with previous studies12, GW3965 treatment did not alter serum cholesterol levels in control mice (Fig. 2f). Notably, however, GW3965 increased serum cholesterol levels in L-SCAP knockout mice, suggesting the loss of a suppressive effect (Fig. 2f, g). LXR target genes, including LeXis itself, were induced by GW3965 in both groups; however, the suppression of steroidogenic genes was abrogated in L-SCAP knockout mice (Extended Data Fig. 4j). Furthermore, expression of LeXis also failed to lower serum cholesterol or suppress cholesterogenic gene expression in L-SCAP knockout mice (Fig. 2h and Extended Data Fig. 4k).

To address the role of LeXis in the setting of dietary cholesterol challenge, we used adenoviral vectors to express short hairpin RNA (shRNA) constructs targeting LeXis in mouse liver13,14. Knockdown of LeXis with either of two different shRNA constructs increased serum HDL cholesterol levels in mice fed a Western diet (Extended Data Fig. 5a–d). There was also an increase in liver cholesterol content in shLeXis-transduced mice (Extended Data Fig. 5e). Gene expression analysis revealed increased expression of cholesterol biosynthetic genes in response to LeXis knockdown (Extended Data Fig. 5f). Similar effects of LeXis knockdown were observed in mice treated with GW3965 (Extended Data Fig. 5g, h). There was no consistent evidence of ER stress or inflammation in these experiments (Extended Data Fig. 5i–k).

As a complementary acute loss-of-function approach, we used antisense oligonucleotides (ASOs) to target LeXis expression15,16. Three different ASOs that potently blocked hepatic LeXis, but not saline or non-targeting ASO controls, increased serum cholesterol levels in the setting of LXR activation, with no evidence of hepatotoxicity (Fig. 3a, b and Extended Data Fig. 5l, m). Furthermore, LeXis ASO administration increased cholesterogenic gene expression (Fig. 3c).

Figure 3: Acute and chronic inactivation of LeXis alters hepatic lipid metabolism.
figure 3

a, LeXis gene expression (normalized to 36B4, also known as Rplp0) in livers from C57BL/6 mice on a chow diet administered 25 mg kg−1 ASOs intraperitoneally on days 1, 4 and 7, and gavaged with 40 mg kg−1 GW3965 on days 4, 7 and 8 (n = 5 per group). Ctrl, control. b, Total serum cholesterol from mice in a. c, Gene expression from C57BL/6 mice on a chow diet administered 25 mg kg−1 ASOs intraperitoneally on days 1, 3 and 5, and gavaged with 40 mg kg−1 GW3965 on days 5 and 6 (n = 8 per group). d, Total serum cholesterol levels in chow-fed wild-type or LeXis−/− mice gavaged with 40 mg kg−1 GW3965 for 2 days (n = 8–10 per group). e, Gene expression from C57BL/6 wild-type or LeXis−/− mice on a chow diet gavaged with 40 mg kg−1 GW3965 for 2 days (n = 8–10 per group). f, Hepatic cholesterol content was normalized to liver mass from C57BL/6 wild-type or LeXis−/− mice fed a Western diet for 3 weeks (n = 7–11 per group). g, Representative (of three images per group) gross appearance of livers from wild-type and LeXis−/− mice after 3 weeks on a Western diet. h, Histological sections of liver from wild-type and LeXis−/−mice after 3 weeks on a Western diet (haematoxylin and eosin stain representative of three images per group). Original magnifications, ×40 (g) and ×1 (h). All values (a–f) are mean ± s.e.m. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 (ANOVA (b, c) and unpaired two-tailed t-test (e, f)).

PowerPoint slide

Full size image

We generated LeXis-deficient mice to determine the consequences of chronic loss of LeXis function (Extended Data Fig. 6a–c). Although serum cholesterol levels in LeXis-deficient mice in the setting of LXR activation were not different from controls (Fig. 3d), the expression of sterol synthesis genes in the liver was increased (Fig. 3e). Furthermore, LeXis-null mice had increased hepatic cholesterol content when challenged with a Western diet (Fig. 3f). Gross and histological examination of livers from LeXis-deficient null mice showed changes consistent with lipid accumulation (Fig. 3g, h). In contrast to the acute LXR agonist studies above, gene expression analysis of LeXis−/− mice maintained on Western diet showed a trend towards decreased sterol synthetic gene expression, probably reflecting the marked increased hepatic cholesterol content in this setting (Extended Data Fig. 6d).

To begin to understand how LeXis was influencing hepatic metabolism, we analysed its subcellular localization. LeXis was almost exclusively located in the insoluble nuclear pellet in fractionation studies, along with the known nuclear long non-coding RNAs (lncRNAs) XIST and histone H3 (Fig. 4a). Single molecule RNA fluorescence in-situ hybridization with LeXis-specific probes further confirmed its nuclear localization (Fig. 4b).

Figure 4: LeXis interacts with RALY to regulate metabolic gene expression.
figure 4

a, Hepa1-6 cells were transfected with LeXis, and 24 h later cellular content was separated into cytoplasmic soluble (cyt), nuclear soluble (nuc) and insoluble pellet (pel) fractions. Transcripts in each fraction were analysed by qPCR, and fraction purity was validated by western blotting with the indicated compartment markers (n = 3 per group). b, Representative (of three) micrograph showing LeXis subcellular localization in primary mouse hepatocytes by single molecule fluorescence in situ hybridization using anti-sense probes to LeXis (red). Nuclei were counterstained with DAPI (blue). Original magnification, ×63. c, Recruitment of RNA polymerase II (RNPII) to promoter regions as determined by ChIP-qPCR analysis in livers transduced with control (Ad-GFP) or LeXis-expressing (Ad-LeXis) adenoviruses. Data are expressed as percentage input retrieved normalized to an upstream site (region 1) (n = 3 per group). d, Total serum cholesterol levels in 14-week-old chow-fed male C57BL/6 mice transduced with control (shCtrl) or adenoviral vectors expressing Raly shRNA (shRaly) (n = 8 per group). e, Gene expression in livers of the mice shown in f. f, Total serum cholesterol in chow-fed male C57BL/6 mice transduced with control (Ad-GFP) or Ad-LeXis (1.0 × 109 plaque-forming units, p.f.u.) and shCtrl or shRaly (2.0 × 109 p.f.u.) (n = 7–8 per group). Values are mean ± s.d. (a, c) or mean ± s.e.m. (d). *P < 0.05; **P < 0.01; ***P < 0.001 (unpaired two-tailed t-test (d, e) and ANOVA with multi-group comparison (f)).

PowerPoint slide

Full size image

Owing to the presence of LeXis in the nucleus, we tested its ability to affect RNA polymerase II-dependent transcription. Expression of LeXis in mouse liver reduced RNA polymerase II engagement at the promoters of Srebf2 and its target genes (Fig. 4c). Previous work has shown that nuclear lncRNAs can affect transcription by modifying the recruitment of proteins to chromatin17. We used an unbiased lncRNA–chromatin affinity capture technique to pull-down LeXis from mouse liver and identify interacting proteins18 (Extended Data Fig. 7a). Analysis of the LeXis interactome by mass spectrometry identified the heterogeneous ribonucleoprotein RALY19 as a binding partner. Similar to LeXis, RALY was located in the nuclear pellet (chromatin) fraction of hepatocytes (Extended Data Fig. 7b). Moreover, an antibody to RALY retrieved LeXis in co-immunoprecipitation studies (Extended Data Fig. 7c, d).

RALY contains both an RNA-binding domain and a leucine-zipper coiled domain, suggesting it may act as a regulatory factor20. Notably, previous unbiased analysis of gene coexpression networks has identified Srebf2 as one of the top genes positively coregulated with Raly21. Other studies have shown direct binding of SREBP-2 at the Raly promoter22. Unbiased protein homology analysis revealed extensive structural conservation between RALY and RNA binding motif protein 14 (RBM14, also known as CoAA)23 (Extended Data Fig. 7e), a known steroid receptor coactivator24. This led us to hypothesize that RALY may act as transcriptional cofactor for genes involved in cholesterol biosynthesis. In line with this idea, adenovirus-mediated knockdown of RALY in mouse liver reduced serum cholesterol, mimicking the effect of LeXis expression (Fig. 4d and Extended Data Fig. 7f). This effect was correlated with reduced expression of Srebf2 and its target genes (Fig. 4e and Extended Data Fig. 7g). Unbiased gene expression profiling of liver revealed that RALY knockdown preferentially affected cholesterol biosynthetic pathways (Extended Data Fig. 8a, b). The effects of RALY were independent of LDLR expression, since they were preserved in Ldlr-null mice (Extended Data Fig. 9a, b). The actions of LeXis in vivo were dependent on RALY, since the ability of LeXis to alter serum cholesterol levels and hepatic gene expression was impaired in the setting of RALY knockdown (Fig. 4f and Extended Data Fig. 9c). Finally, ChIP–qPCR analysis of mouse liver revealed that RALY associated with cholesterol biosynthetic gene promoters, and that RALY occupancy was reduced in the setting of LeXis expression (Extended Data Fig. 9d).

This work identifies the non-coding RNA LeXis as an additional mediator of the complex effects of LXR signalling on hepatic lipid metabolism. Our data suggest that LeXis contributes to the ability of LXRs to inhibit cholesterol synthesis. It is important to acknowledge, however, that the involvement of additional pathways in this crosstalk is not excluded by the present work. The demonstration that LeXis expression is responsive to dietary cues and can modulate physiological pathways with links to common diseases expands our understanding of the regulatory potential of non-coding RNA. Notably, the consequences of acute and chronic loss of LeXis expression are only partially overlapping, perhaps reflecting compensation in the setting of developmental deletion25.

Although the rapid sequence evolution of lncRNAs presents a challenge to identifying functional counterparts between species26, batch coordinate conversion between mouse and human assemblies revealed moderate conservation of the LeXis genomic sequence in a region adjacent to the human ABCA1 gene. An annotated putative lncRNA (TCONS_00016452) in this region was robustly induced by LXR activation in human hepatocyte cell lines (Extended Data Fig. 10). In the future it will be of interest to assess whether this sequence or an as yet to be identified lncRNA is a functional orthologue of LeXis.

Methods

Reagents, plasmids and gene expression

GW3965 was synthesized as previously described27. LG268 was from Ligand Pharmaceuticals. Oxysterols were purchased from Sigma and used as described28. Simvastatin sodium salt was from Calbiochem. Ligands were dissolved in dimethyl sulfoxide before use in cell culture. LeXis was amplified from GW3695-treated primary mouse hepatocytes using KOD polymerase (Millipore) and primers designed to provide flanking attB sequences and a SacI site at the immediate 3′ end. The fragments were then cloned into pDONR221 using the Gateway system and the minimal SV40 polyadenylation sequence was inserted at the SacI site. For transient transfections and viral vector production the entry clone was transferred into the pAd/CMV/V5-DEST Gateway vector by LR recombination. We estimate transcription from this vector to append 109 nucleotides at the 5′ end, and 29 nucleotides at the 3′ end of the cloned LeXis sequence. To obtain the shLeXis adenovirus, we used BLOCK-iT kit as described (Invitrogen)3. In brief, Invitrogen based software was used for original nucleotide generation targeting the LEXIS fragment and cloned into pENTR/U6. The resulting pENTR/U6-LEXIS shRNA plasmids were tested for their ability to inhibit overexpressed LEXIS in transient transfection experiments in HEK293T cells and then transferred by Gateway recombination into the pAd/BLOCK-iT-DEST destination vector for viral particle generation. Viruses were amplified, purified and titred by Viraquest. For gene expression analysis, RNA was isolated using TRIzol reagent (Invitrogen) and analysed by qPCR using an Applied Biosystems 7900HT sequence detector or Applied Biosystems Quant Studio 6 Flex. Results are normalized to 36B4 or cyclophilin (also known as Ppia). Immunohistochemical staining of paraffin-embedded livers were done by the UCLA Translational Pathology Core Laboratory.

Animals and diets

All animals (C57BL/6, greater than 10 generations backcrossed) were housed in a temperature-controlled room under a 12-h light/12-h dark cycle and pathogen-free conditions. For adenovirus experiments, aged-matched mice were purchased from Jackson Laboratories. Littermates were manually randomized to different treatment groups. Investigators were blinded to group allocation for some but not all studies. LXRα−/−, LXRβ−/− and LXRαβ−/− mice were originally provided by D. Mangelsdorf. Floxed Scap−/− mice were previously described29. LeXis global knockout mice were generated at UCDavis KOMP using strategy outlined in Extended Data Fig. 6. Mice were fed a chow diet except as indicated, where mice were placed on a Western diet (21% fat, 0.21% cholesterol; D12079B; Research Diets Inc.) or were gavaged with either vehicle or 40 mg kg−1 GW3965. Livers were obtained 4 h after the last gavage. We measured cholesterol and triglycerides as previously described30. For adenoviral infections, age-matched (9–11 weeks old) male mice were injected with 2.0 × 109 p.f.u. by tail-vein injection unless otherwise specified. Mice were euthanized 6 days later after a 6-h fast. At the time of euthanization, liver tissue and blood was collected by cardiac puncture and immediately frozen in liquid nitrogen and stored at −80 °C. Liver tissue was processed for isolation of RNA and protein as above. Generation 2.5 constrained ethyl ASOs, synthesized as described previously31, were administered by three 25 mg kg−1 intraperitoneal doses together with 40 mg kg−1 GW3965. Animals were euthanized on day 6 or 8 as indicated in figure legends. Most experiments were performed using male mice. All animal experiments were approved by the UCLA Institutional Animal Care and Research Advisory Committee.

Cell culture

Primary peritoneal macrophages were isolated 4 days after thioglycollate injection and prepared as described32. Mouse primary hepatocytes were isolated as previously described and cultured in William’s E medium with 5% FBS28. Peritoneal cells were incubated in 0.5% FBS in DMEM, with 5 μM simvastatin and 100 μM mevalonic acid. Five to eight hours later, cells were pretreated with dimethylsulfoxide (DMSO) or appropriate ligand overnight. In vitro translation assay was performed using TnT Coupled Transcription/Translation System (PROMEGA) according to the manufacturer’s protocol. The cell lines HEK293T, HEK293A and Hepa1-6 were originally obtained from ATCC. All cells were tested for mycoplasma contamination.

RACE

The 5′ and 3′ ends of the LeXis transcript were defined using mouse liver RNA and the FirstChoice RLM-RACE kit (Ambion) according to manufacturer’s protocol, with modifications. In brief, for the 5′ RACE, degraded messenger RNA 5′ ends were dephosphorylated with CIP, and then full-length mRNA was decapped with TAP. Following 5′ RACE adaptor ligation, reverse transcription was performed using SuperScriptIII First-Strand Synthesis system (Invitrogen) and LeXis-specific primers. For the 3′ RACE, RNA was reverse transcribed using SuperScriptIII First-Strand Synthesis system (Invitrogen) and the adaptor-linked oligo dTs. The resulting cDNA was amplified by nested PCR across a 55–65 °C melting temperature gradient using KOD polymerase (Millipore), with the inner primers containing attB sequences. Aliquots of reactions were inspected on 1% agarose gels for product size and abundance. Products of select PCR reactions were purified using NucleoSpin Gel and PCR Cleanup kit (Clontech) and were inserted into pDONR221 by Gateway cloning. Cloned fragments were sequenced and then aligned to the mouse genome with the BLAST analysis tool.

RNA fractionation

The PAd/CMV-LeXis vector was transfected into Hepa1-6 cells using BioT reagent (Bioland Scientific LLC) and 24 h later subcellular RNA fractions were obtained according to the protocol described previously33. Lysate aliquots were inspected for fractional purity by western blotting with antibodies against α-tubulin, SNRP70 and histone H3 as cytoplasmic, nucleoplasmic and chromatin bound markers, respectively.

RNA-seq

RNA-seq libraries, starting with 500 ng total RNA, were constructed with the TruSeq RNA Sample Prep Kits from Illumina on RNA isolated from primary hepatocytes treated with or without GW3965. Samples were indexed with adapters and submitted for paired-end 2 × 100-bp sequencing in Illumina HiSeq2000. RNA-seq reads were aligned with TopHatv2.0.2 to the mouse genome, version mm9 (ref. 34). The TopHat alignment rate was 85%, resulting in an average of 65 million reads per sample. Transcripts were assessed and quantities were determined by Cufflinks v2.0.2, using a GTF file based on Ensembl mouse NCBI37. Comparison expression levels were made using fragments per kilobase of exon per million fragments mapped (FPKM) values using Cuffdiff from the Cufflinks package35. Data analysis was performed by UCLA DNA Microarray Core.

Lipid analysis

Tissue lipid was obtained using a Folch extraction. In brief, chloroform extracts were dried under nitrogen and solubilized in water. Tissue and serum cholesterol and triglycerides were determined using a commercially available enzymatic kit (Wako). Hepatic cholesterol content was normalized to liver weight and protein concentration. Mice were fasted for at least 6 h before blood collection and euthanization. Plasma lipoprotein fractions were analysed by FPLC.

Microarray

For cDNA microarray analysis, primary hepatocytes cells were treated as indicated above with either DMSO or GW3695. These samples were from an independent cohort from those submitted for RNA-seq. For each condition, two independent samples were processed. Transcriptional profiling was performed at the University of California, Los Angeles, microarray core facility by using Agilent SurePrint G3 Gene Expression array. Data were analysed using GeneSpring software (Agilent Technologies) and David36.

ChIP

ChIP studies were performed as described elsewhere37. In brief, mouse livers were cross-linked using a final formaldehyde concentration of 1% at room temperature for 10 min. The reaction was quenched with the addition of glycine. For sonication, 0.3 ml (1/3) of nuclear lysate was sonicated for 25–30 cycles, 30 s on 30 s off at 4 °C, with BioRuptor twin sonicator (Diagenode). Sonicated Chromatin was incubated overnight at 4 °C with control IgG or 25 μg of anti-LXRα antibody (PPZ0412, ChIP Ggade, Abcam), anti-RALY antibody (EPR10121, Abcam), or Pol II antibody (N-20, Santa Cruz Biotechnology). Protein A dynabeads (50 μl per immunoprecipitation sample) were added for 4 h. After incubation beads were washed with wash buffer A (50 mM HEPES, pH 7.9, 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% Na-deoxycholate, 0.1% SDS, 1× protease inhibitors freshly added), buffer B (50 mM HEPES, pH 7.9, 500 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% Na-deoxycholate, 0.1% SDS) and finally LiCL buffer (20 mM Tris, pH 8.0, 250 mM LiCl, 1 mM EDTA, 0.5% Na-deoxycholate, 0.5% NP-40). Reverse crosslinking was performed at 60 °C overnight, mixed at 1,000 r.p.m., and DNA was extracted using a phenol–chloroform phase lock tube (5 PRIME) or Nucleospin PCR cleanup column (Macherey-Nagel). A standard curve for PCR was generated from serial dilutions of input samples and data expressed as percentage of input.

Chromatin isolation by RNA purification

Chromatin isolation by RNA purification (ChIRP) was performed as described previously18. In brief, mouse livers were cross-linked using glutaraldehyde. After glycine quenching, the nuclear lysate was sonicated for 25–30 cycles, 30 s on 30 s off at 4 °C, with BioRuptor twin sonicator (Diagenode). LeXis and LacZ pulldown probes with BiotinTEG at 3′ were designed by Biosearch Tecnologies (see Supplementary Information) and allowed to hybridize overnight with sonicated chromatin at 37 °C (100 pmol probe per 1 ml chromatin). After hybridization, C1 Dynabeads (Life Technologies) were added and incubated for 30 min. For protein elution for mass spectrometry analysis, washed beads were resuspended in 3× original volume of DNase buffer (100 mM NaCl and 0.1% NP-40), and protein was eluted with a cocktail of 50 mM triethyl ammonium bicarbonate, 12 mM sodium lauryl sarcosine, and 0.5% sodium deoxycholate supplemented with 100 μg ml−1 RNase A (Sigma-Aldrich) and 0.1 U μl−1 RNase H (Epicentre), and 100 U ml−1 DNase I (Invitrogen). For RNA isolation, beads were resuspended in protensase K buffer (100 mM NaCl, 10 mM TrisCl, pH 7.0, 1 mM EDTA, 0.5% SDS, 5% by volume proteainse K (AM2546, Ambion) 20 mg ml−1) and incubated at 50 °C followed by Trizol isolation and DNase treatment.

Single molecule RNA FISH

Custom Stellaris FISH probes were designed against LeXis. Stellaris probe set labelled with CAL Fluor Red 610 and RNA FISH performed as described previously38. In brief, hepatocytes were fixed with 3.7% formaldehyde in PBS followed by 70% ethanol treatment to permeabilize cells. Cells were washed with 10% formamide in 2× SCC followed by treatment in humidified chamber with addition of probes (125 nM) in hybridization buffer (100 mg ml−1 dextran sulfate and 10% formamide in 2× SSC). Cells were incubated in the dark at 37 °C for 4 h. DAPI nuclear stain (5 ng ml−1) was applied after washing with 10% formamide in 2× SCC. Images obtained using a Zeiss Z1 AxioObserver fluorescent microscope.

Statistical analysis

A non-paired Student’s t-test or ANOVA was used to determine statistical significance, defined at P < 0.05. Unless otherwise noted, error bars represent s.d.. Experiments were independently performed at least twice. Group sizes were based on statistical analysis of variance and prior experience with similar in vivo studies.