A PPARγ/long noncoding RNA axis regulates adipose thermoneutral remodeling in mice

Interplay between energy-storing white adipose cells and thermogenic beige adipocytes contributes to obesity and insulin resistance. Irrespective of specialized niche, adipocytes require the activity of the nuclear receptor PPARγ for proper function. Exposure to cold or adrenergic signaling enriches thermogenic cells though multiple pathways that act synergistically with PPARγ; however, the molecular mechanisms by which PPARγ licenses white adipose tissue to preferentially adopt a thermogenic or white adipose fate in response to dietary cues or thermoneutral conditions are not fully elucidated. Here, we show that a PPARγ/long noncoding RNA (lncRNA) axis integrates canonical and noncanonical thermogenesis to restrain white adipose tissue heat dissipation during thermoneutrality and diet-induced obesity. Pharmacologic inhibition or genetic deletion of the lncRNA Lexis enhances uncoupling protein 1–dependent (UCP1-dependent) and -independent thermogenesis. Adipose-specific deletion of Lexis counteracted diet-induced obesity, improved insulin sensitivity, and enhanced energy expenditure. Single-nuclei transcriptomics revealed that Lexis regulates a distinct population of thermogenic adipocytes. We systematically map Lexis motif preferences and show that it regulates the thermogenic program through the activity of the metabolic GWAS gene and WNT modulator TCF7L2. Collectively, our studies uncover a new mode of crosstalk between PPARγ and WNT that preserves white adipose tissue plasticity.


Supplemental Material
A PPARγ-long noncoding RNA axis regulates adipose thermoneutral remodeling in mice

Supplemental Figures Page
Supplemental Figure Supplemental Table 2 Page 23 Supplemental Table 3 Page 24 Supplemental Table 3 Page 26 free conditions.Lexis global knockout mice and TCF7L2 flox mice were generated per our previous study (4) (1).Lexis flox mice were generated by Cyagen Biotechnology using gene-targeting technology.Adipose tissue specific knockout mice and littermate controls were generated by crossing with Adipoq-cre mice (Jackson Laboratory) using the strategy outlined in Supplemental Figure 3 (A) were created in our previous study Ucp1 KO mice were bought from Jackson Laboratory (003124, Jackson Laboratory).
Experiments used male or female mice as indicated except thermoneutrality studies included a mix of both.Ucp1 KO mice experiments were pooled from two smaller studies since we were not able to generate a sufficient number of Ucp1 KO in a single cohort.

Anti-sense oligonucleotide (ASO) studies
Acute Lexis antisense oligonucleotide (ASO) studies were performed as we previously described (2,4).Male mice at 9-10 weeks of age were intraperitoneally injected with control or Lexis ASO at the dose of 25 mg /kg twice a week.Body weight was measured weekly and body composition was determined by EchoMRI Body Composition Analyzer.

Diet-induced Obesity Studies
The majority of studies used mice aged 8-12 weeks old and fed a calorie-rich diet (Western Diet, D12079B, Research Diet).Although in pilot studies we tested Western and High Fat diet, we used Western Diet for most experiments because the induction of Lexis was noted to be highest.Body composition was determined by EchoMRI analysis (EchoMRI, 3-in-1).For the food intake experiment, mice were singly housed in standard housing and fed chow or western diets.The mass of diet administered and consumed was recorded daily over a period of seven days and shown as gram of diet per mouse per day.

Thermal stress challenge
For the thermoneutrality studies, mice were group housed in monitored climatecontrolled facilities maintained on a 12 h light-dark cycle at a constant temperature of 30 °C.For chronic cold exposure studies, mice were group housed and placed at 6°C with bedding and free access to food and water.The rectal temperature of mice was monitored using Bioseb rodent thermometer (BIOSEB)

Metabolic chamber studies
Energy expenditure was measured using male mice by indirect calorimetry with chow or Western Diet feeding as indicated in figure legend (Oxymax CLAMS, Columbus Instruments).We detected the body composition and body mass right before the calorimetry experiments.The indirect calorimetry experiment generated data was analyzed using CalR (5).

Glucose tolerance tests (GTT) and insulin tolerance tests (ITT)
GTT and ITT were performed using our previous described method (2).Briefly, for glucose tolerance tests, we fasted the mice for 6 hours and then challenged them with an IP injection of glucose with the amount of 1 g kg -1 (according to lean mass or total body mass as indicated in figure legends).For insulin tolerance tests, we fasted the mice for 6 hours and i.p. injected insulin with the amount of 1 U kg -1 .Blood glucose levels were detected at different time points as shown in the figures using the ACCUCHEK active glucometer (Roche).

Cellular and Mitochondrial Respiration assay
Tissue OCR was measured through the Mitochondrial and Metabolism Core at UCLA using mitochondrial isolated from iWAT samples according to previous publications with modification (6,7).Briefly, frozen iWAT samples were stored in -80 °C until use in the Seahorse experiments.Frozen tissues were thawed on ice and homogenized in 1 mL MAS buffer (70 mM sucrose, 220 mM mannitol, 5 mM KH2PO4, 5 mM MgCl2, 1 mM EGTA, 2 mM HEPES pH 7.2) with 15-20 strokes in a Dounce tissue homogenizer.
Homogenates were centrifuged at 20,000 × g for 10 minutes at 4 °C, the supernatant was removed, and the pellet was resuspended in 75 µL MAS buffer.The sample was then centrifuged at 500x g and the supernatant was collected.

Gene expression analysis and immunoblot analysis
TRIzol reagent (Invitrogen) was used for total RNA isolation.Random Hexamer Primers and Olig-dT primers were used for reverse transcription.Gene expression was quantified using cDNA by qRT-PCR using SYBR Green Master Mix (Bio-Rad) on Bio-Rad 384 Real-time PCR instrument.Gene expression levels were determined by using a standard curve or 2(-Delta-Delta C(T)) method.Each gene was normalized to the housekeeping gene 36B4.For immunoblot analysis, whole cell lysate or tissue lysate was extracted using RIPA lysis buffer (Boston Bioproducts) supplemented with complete protease inhibitor cocktail (Roche).Proteins were diluted in Nupage loading dye (Invitrogen), heated at 95 °C for 5 min, and run on 4-12% NuPAGE Bis-Tris Gel (Invitrogen).Proteins were transferred to PVDF blotting membrane (0.45 μm) blocked with 5% milk to quench nonspecific protein binding and blotted with the indicated primary antibody.For complete listing of antibodies please see Supplemental Table 2.

Haematoxylin & Eosin (H&E) staining
All paraffin embedment and H&E staining in this paper was performed by the Translational Pathology Core Laboratory (TPCL) in UCLA.The iWAT tissues of mice were fixed in 10% formaldehyde overnight, washed with tap water for 15 minutes and stored in 70% ethanol before being mounted in paraffin.After Paraffin embedment, 5 μm sections were sliced and stained with haematoxylin and eosin.Images were taken with a fluorescence microscope (ZEISS).

Subcellular fractionation
A total of 20 million preadipocytes were fractionated into cytosolic and nuclear compartments using detergent lysis according to published protocol with modification (8).Briefly, after collection and washing with cold PBS, cells were resuspended in cold cytoplasmic buffer (0.15% NP-40, 10 mM Tris pH 7.5 and 150 mM NaCl) and incubated on ice for 10 min.Lysates were layered onto 2.5 volumes of a chilled sucrose buffer (10 mM Tris 7.5, 150 mM NaCl, 24% sucrose) and spined at 14,000 rpm× 10 min× 4 °C.

Stromal vascular fraction (SVF) isolation and immortalization
SVFs were isolated from the iWAT of 4-6 weeks old mice and induced for differentiation based on previous published protocol (1,9).SVF were immortalized by retroviral expression of the SV40 large T antigen according to our previous publication (10).
Genetic deletion by the CRISPR-Cas9 system or pharmacological inhibition of Atp2a2 by thaipsigargin (2 µM) as previously described (11).We used the exact guide RNA sequences published by Ikeda et al., (11) The data were sequenced on Illumina HiSeq 3000.RNA-seq reads were aligned with HISAT2 to the mouse genome (mm9) (12).Differential gene expression analysis was performed using R package DEseq2 (13).Enrichment analysis for Gene Ontology terms among the differential expressed genes was performed using Metascape (14).For analysis of RNA-seq data download from public dataset under GEO number GSE94654 (2) or GSE151324 (15), reads were aligned to mouse genome (mm39) or human genome (GRCh38) with HISAT2 (12).FeatureCounts ( 16) was applied to call read counts of lncRNAs for each sample with the mice gene annotation of GENCODE (Release M31, GRCm39) or human gene annotation of GENCODE (Release 42, GRCh38.p13)(17).Differential gene expression analysis was performed using R package DEseq2 (13).

Chromatin-affinity assays
Chromatin Isolation by RNA Purification (ChIRP) qPCR in preadipocytes was performed according to previous published protocol (18).Cells were treated with GW1929 for 24 hours before ChIRP steps.A total number of 50 million preadipocytes were used for each replicate and 3 replicates were performed for each group.Cells are crosslinked in 1% fresh made Glutaraldehyde for 12 min.After cell lysis, sonication was performed using Covaris E220 according to the manufacturer's protocol.The sequence of primers used for ChIRP-qPCR were provided in Supplemental Table 3. HiChIRP assay was performed according to the published protocol with modification (19).Briefly, a total of 100 million preadipocytes (pre-treated with GW1929 for 24 hours) were used for each replicate and 2 replicates were performed for each group.For negative control, RNase Center (BSCRC) Sequencing Core at UCLA.For the data analysis, raw reads were uniquely mapped to reference mouse genome (NCBI37/mm9) using Bowtie2 (20).
MACS2 was used for peak calling (21,22).Motif analysis was performed based on Lexis genomic binding sites and visualized by PscanChIP and MotifStack (23,24).

Lexis binding sites prediction
Software "LongTarget" was used to predict DNA binding motifs and binding sites of Lexis in a genomic region based on potential base pairing rules between an RNA sequence and a DNA duplex according to published methods (25,26).
Chromatin was immunoprecipitated with 5 µg antibodies against TCF7L2 antibody (Cell Signaling Technology) or IgG (Millipore) overnight at 4°C.For iWAT ChIP, adipocyte nuclei isolation from 500 mg of iWAT from 10-week-old Lexis-AdWT or Lexis-AdKO mice were performed as previously described (9).After reversal of cross-linking, DNA was isolated and DNA enrichment was quantified by qRT-PCR.Axin2 was used as positive control of TCF7L2 ChIP-qPCR according to previous publication (29).The sequence of primers used for ChIP-qPCR is provided in Supplemental Table 3.

RNA fluorescence in situ hybridization (RNA FISH)
Custom design RNA FISH probes against Lexis were designed using Stellaris FISH probe designer and ordered from Stellaris (LGC Biosearch Technologies).Sequences of RNA FISH Probes were provided in Supplemental

Electrophoretic Mobility Shifting Assay (EMSA)
EMSA was performed to determine the binding of TCF7L2 to the Atp2a2 promoter region bearing one putative TCF7L2 recognition site (Sequences in Supplemental Table 3) in preadipocyte.The corresponding 3' end biotin labeled oligonucleotides were synthesized by Integrated DNA Technology (IDT).Nuclear extracts were prepared using the NE-PER Nuclear Extraction Reagent Kit (ThermoFisher Scientific).The LightShift EMSA Kit (ThermoFisher Scientific) was used for binding reaction according to the manufacturer's profocol.20 fM concentration of labeled duplex DNA and 4 mg of nuclear protein extract were used in the reaction.For super-shift, we added 1 µl of TCF7L2-specific ChIP grade antibody (Cell Signaling Technology) to nuclear extract proteins for 30 min at room temperature before adding labeled probes.The electrophoresis was performed by running reaction mix on 5% native agarose gel in 0.5X TBE buffer.After transfer onto a positively charged nylon membrane (ThermoFisher Scientific), DNA-protein complex was crosslinked by UV and visualized with Horseradish Peroxidase-Conjugated Streptavidin (ThermoFisher Scientific).

Isolation of nuclei from iWAT of western diet fed mice
We isolated the nuclei of iWAT from western diet fed male mice using a protocol slightly modified from recent publications (30,31).Briefly, iWAT from 3 mice of Lexis-AdWT group or Lexis-AdWT group were isolated (without lymph node) pooled (according to group) and minced. 2 ml NIB buffer (250 mM Sucrose, 10 mM HEPES, 1.5 mM MgCl2, 10 mM KCl, 0.001% IGEPAL CA-630, 0.2 mM DTT, 0.5 U/µL RNase inhibitor) were added to 400 mg pooled iWAT, then homogenized with 2 ml glass dounce homogenizer.After filtering through 70 µm cell strainer, another 2 ml NIB was added to the filtered homogenate and centrifuged at 1000 g× 10 min× 4 °C.The lipid layer was aspirated, pellet was resuspended in the remaining supernatant and then transferred to a new pre-cooled 5 ml DNA low binding tube on ice and added 2 ml NIB.Nuclei were pelleted by centrifuging at 500 g× 10 min× 4 °C, resuspended in 100 µL NRB (1% BSA in PBS, 0.04 U/ µL RNase inhibitor) and filtered through a 40 µm tip strainer.15,000 nuclei were taken out and resuspended in NRB as previous protocol described (30,31).

10x Genomics single-nucleus RNA sequencing
We performed 10x Genomics single-nucleus RNA sequencing at USC Norris Molecular Genomics Core immediately after nuclei isolation.Single cell RNA sequencing was prepared using 10x Genomics 3' v3.1 (Cat number: 1000092) following manufacturer's protocol.Samples were parsed into single nuclei using 10x Genomics Chromium Controller and libraries were simultaneously prepared.Prepared single cell RNA sequencing libraries were sequenced on the Illumina Novaseq6000 platform at a read length of 28x 90 and read depth of 50,000 reads/ cell for 5000-7000 cells.

Single-nucleus RNA-sequencing analysis
Sequencing data was package and analyzed using 10x Genomics Cell Ranger analysis tool.The R package Seurat (V4) was used to cluster the cells in the merged matrix (32).

A
and RNase H were added as the amount of 2 μg per million cells.Covaris E220 was used for sonication with the following parameters: Fill Level = 10, Duty Cycle = 5, PIP = 140, Cycles/Burst = 200, Time = 45 min.For sequencing, libraries were paired-end sequenced by NovaSeq S4 with read lengths of 150 bp at Broad Stem Cell Research

Table 3
. Probes were labeled with Quasar® 570 Dye and single molecule RNA FISH was performed according to Stellaris® RNA FISH protocol.For nuclei staining, slides were mounted with ProLong Diamond Antifade Mountant with DAPI (ThermoFisher Scientific).Fluorescent signals were captured using a laser scanning confocal microscope (SP80, Leica).

Table 2 . Key resources table
Gene expression by qRT-PCR in iWAT of the mice with rosiglitazone treatment (n= 5 per group).Rosiglitazone IP injected at the dose of 5 mg/ kg for 7 days.(B)qPCR in iWAT of Lexis WT or KO mice from Figure3(K) (n= 7 for AdWT, n= 8 for AdKO).Data was shown as mean ± SEM. (C) Ca 2+ level induced by NE was detected in SVF isolated from iWAT of Lexis-AdWT or Lexis-