linc-ADAIN, a human adipose lincRNA, regulates adipogenesis by modulating KLF5 and IL-8 mRNA stability

Abstract

Adipose tissue remodeling and dysfunction, characterized by elevated inflammation and insulin resistance, play a central role in obesity-related development of type 2 diabetes (T2D) and cardiovascular diseases. Long intergenic non-coding RNAs (lincRNAs) are important regulators of cellular functions. Here, we describe the functions of linc-ADAIN (adipose anti-inflammatory), an adipose lincRNA that is downregulated in white adipose tissue of obese humans. We demonstrate that linc-ADAIN knockdown (KD) increases KLF5 and interleukin-8 (IL-8) mRNA stability and translation by interacting with IGF2BP2. Upregulation of KLF5 and IL-8, via linc-ADAIN KD, leads to an enhanced adipogenic program and adipose tissue inflammation, mirroring the obese state, in vitro and in vivo. KD of linc-ADAIN in human adipose stromal cell (ASC) hTERT adipocytes implanted into mice increases adipocyte size and macrophage infiltration compared to implanted control adipocytes, mimicking hallmark features of obesity-induced adipose tissue remodeling. linc-ADAIN is an anti-inflammatory lincRNA that limits adipose tissue expansion and lipid storage.

Keywords: CP: Metabolism; CP: Molecular biology; HuR; IGF2BP2; IL-8; KLF5; adipogenesis; linc-ADAIN; linc-DMRT2; linc01230; long non-coding RNA; obesity.

Figure 1.. Adipose-enriched linc-ADAIN expression is modulated by obesity in humans

(A) Gluteal sWAT expression of linc-ADAIN is reduced in obese humans (N = 29 [lean], N = 39 [obese]); ****p < 0.0001 with relation to lean by Mann-Whitney U test. (B) linc-ADAIN expression in human subcutaneous or visceral adipose tissue (N = 39 [obese]); p = 0.2119 by Mann-Whitney U test. (C) linc-ADAIN expression in subcutaneous adipose tissue in males (N = 11) vs. females (N = 14); *p < 0.05 with relation to male by Mann-Whitney U test. (D) linc-ADAIN expression correlates negatively with total percentage of body fat; Pearson correlations. (E) Experimental human endotoxemia suppresses linc-ADAIN expression in human sWAT (N = 12); *p < 0.05 with relation to pre-lipopolysaccharide (LPS) by one-way ANOVA. Data are presented as the mean ± SEM.

Figure 2.. Human linc-ADAIN expression is induced during adipocyte differentiation and is modulated by the canonical adipocyte transcription factor PPARγ

(A) Regulatory features at the linc-ADAIN locus, RNA-seq coverage (human adipocytes), transcription factor binding, and active histone modification markers. (B) Tissue expression of linc-ADAIN from GTEx (gene mean transcripts per million [TPM]). (C and D) Induction of linc-ADAIN by the PPARγ agonist rosiglitazone (10 μM) (C) and during adipocyte differentiation in vitro (D) (N = 3 in triplicate). (E) Cellular fractionation of primary ASC adipocytes. qPCR of MALAT1, GAPDH, and linc-ADAIN of nuclear and cytoplasmic fractions. Data were normalized by averaging of loading controls GAPDH, β-ACTIN, MALAT1, U6, and HPRT and then subtracting the nucleus value and getting a fold change of gene expression compared to nucleus (N = 3). (F) RNA scope assay showing the spatial expression of linc-ADAIN (red) and nuclei (DAPI/blue) in scramble and linc-ADAIN shRNA hTERT ASC adipocytes (scale bar, 20 μm). (G) UMAP (uniform manifold approximation and projection) projection of linc-ADAIN (linc01230) expression in single-cell RNA-seq of human subcutaneous WAT (Broad Institute). Data are presented as the mean ± SEM.

Figure 3.. linc-ADAIN knockdown increases lipid deposition, the adipogenic program, and cytokine secretion

(A–C) Nile red staining of ASC hTERT adipocytes expressing shRNA against linc-ADAIN or scramble control (A and B) (scale bar, 20 μm) and triglyceride accumulation at days 14 and 21 post differentiation (C). (D) qPCR of adipogenic gene expression upon linc-ADAIN KD with shRNA vs. scramble control at 14 days (N = 3 in duplicate). (E) Western blot of adipogenic markers in scramble and linc-ADAIN shRNA ASC adipocytes at day 14: PPARγ, FASN, ATGL, perilipin 1, FABP4, GLUT4, C/EBPα, and adiponectin. (F) Luminex protein panel in adipocyte medium at 14 days (N = 3 in duplicate). (G) Fold change of transcripts per million (TPM) of cytokine mRNAs via RNA-seq upon linc-ADAIN KD at 14 days (N = 1 in triplicate). All data, **p < 0.01, ***p < 0.001 with respect to scramble shR by two-way ANOVA. Data are presented as the mean ± SEM.

Figure 4.. linc-ADAIN interacts with HuR and IGF2BP2 in human adipocytes

(A and B) Venn diagram overlapping linc-ADAIN biotinylated pull-down-MS with ≥5-fold total spectrum count (TSC) with linc-ADAIN RNA compared to antisense control and RBPmap-predicted binding proteins for linc-ADAIN, IL-8, MCP-1, and IL-6 mRNAs (A) and the table of six proteins that overlapped (B). (C) STRING analysis of the six proteins’ gene ontology. (D) Proteins involved in mRNA stability (IGF2BP2, ELAVL1 [HuR], FMR1, and FXR2). (E) RNA immunoprecipitation assay (RIP) measuring interaction between HuR and linc-ADAIN in ASC hTERT adipocytes. (F) Western blot of co-immunoprecipitation assay with IGF2BP2 in ASC hTERT adipocytes and HuR. (G) RNA SCOPE (linc-ADAIN) immunofluorescence (IF) of HuR and IGF2BP2 by confocal microscopy in ASC hTERT adipocytes (white arrows point to co-localization [orange] of linc-ADAIN and proteins HuR or IGF2BP2; scale bar, 10 μm). Data are presented as the mean ± SEM.

Figure 5.. Knockdown of linc-ADAIN stabilizes IL-8 and KLF5 mRNA, likely through IGF2BP2

(A–F) RNA immunoprecipitation (RIP) assays measure the interaction of IL-8, KLF5, and GAPDH with HuR (A, C, E) (N = 3) and IGF2PBP2 (B, D, F) (N = 3–4), upon linc-ADAIN shRNA knockdown compared to scramble shRNA in day 14 ASC hTERT adipocytes. (G–I) Scramble and linc-ADAIN shRNA-expressing ASC adipocytes were treated with actinomycin-D to halt transcription and fraction of mRNA measured at 0.5, 1, 2, and 4 h post treatment of IL-8 (normalized to GAPDH) (G), KLF5 (normalized to GAPDH) (H), and GAPDH (I) (N = 3). *p < 0.05, **p < 0.01 ****p < 0.0001 with respect to scramble shRNA by two-way ANOVA. Data are presented as the mean ± SEM.

Figure 6.. KLF5 and IL-8 have sustained upregulation in linc-ADAIN shRNA KD during differentiation

(A and B) Gene expression of KLF5 at day 14 of adipocyte differentiation (A) (N = 3 in duplicate); ***p < 0.001 with respect to scramble, and during adipocyte differentiation (B). (C and D) PPARγ and CEBPα gene expression during adipocyte differentiation. (E) Protein expression of KLF5, PPARγ, and CEBPα in ASC hTERT adipocytes expressing scramble or linc-ADAIN shRNA. (F and G) IL-8 gene expression during adipocyte differentiation (F) and IL-8 secretion during adipocyte differentiation (N = 3) (G); *p < 0.05, **p < 0.01, ****p < 0.0001 with respect to scramble shRNA control by two-way ANOVA. Data are presented as the mean ± SEM.

Figure 7.. Adipocytes with reduced linc-ADAIN show increased size and lipid storage and macrophage infiltration in vivo

(A and B) ASC adipocytes with scramble or linc-ADAIN shRNA were implanted into the flanks of NSG mice and removed 16 weeks later. H&E staining of scramble and linc-ADAIN-KD implants after 16 weeks (scale bar, 60 μm) (A) and size distribution of adipocytes, quantified with Adiposoft in a 300 × 300 μm section (B). (C and D) CD68 (C) and F4/80 (D) IHC staining area quantified.

(E–K) Gene expression of mouse Cd68 (E) and Adgre1 (F) and human expression of IL-8 (G), MCP-1 (H), IL-6 (I), KLF5 (J), and PPARγ and CEBPα (K) in the scramble and linc-ADAIN shRNA-expressing explants after 16 weeks or the ASC hTERT adipocytes prior to implantation (N = 4–7); *p < 0.05, **p < 0.01, ***p < 0.001 with relation to scramble shRNA explant; ^##p < 0.01 with relation to linc-ADAIN shRNA explant; ^{$ $ $}p < 0.01 with reslation to scramble shRNA adipocytes by two-way ANOVA. Data are presented as the mean ± SEM.