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

SUMMARY 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.


INTRODUCTION
Obesity is a state of chronic low-grade inflammation that increases the risk for cardiometabolic diseases (CMDs), a cluster of conditions that include cardiovascular disease linc-ADAIN KD had increased adipocyte size and macrophage infiltration compared to control, mimicking hallmark features of adipose tissue remodeling seen in obesity.

Adipose-enriched linc-ADAIN expression is modulated by obesity in humans
linc-ADAIN (previously named linc-DMRT2) was identified as an inflammatory modulated lincRNA by our laboratory through deep RNA sequencing (RNA-seq) of gluteal subcutaneous white adipose tissue (sWAT) from healthy, non-obese volunteers in the Genetics and Evoked Responses to Niacin and Endotoxemia (GENE) study. 17Adipose expression of lincRNAs, including linc-ADAIN, was validated in lean healthy subjects from GENE (N = 25) and obese subjects from the Penn Human Adipose Resource (HAR) (N = 39).Here, we show that linc-ADAIN expression is reduced in sWAT of obese compared to lean humans (Figure 1A), with similar expression in sWAT and visceral adipose tissue (VAT) (Figure 1B) and slightly higher expression in males than in females (Figure 1C).linc-ADAIN expression negatively correlates with percentage of total body fat (Figure 1D).During endotoxemia, linc-ADAIN expression was significantly downregulated in human sWAT (Figure 1E), demonstrating that its adipose expression is suppressed in this wellstudied model of acute systemic inflammation.

Human linc-ADAIN expression is induced during adipocyte differentiation and is modulated by canonical adipocyte transcription factors such as PPARγ
The linc-ADAIN gene is located upstream of the protein-coding gene DMRT2 on human chromosome 9. RNA-seq, 3′ RACE (rapid amplification of cDNA ends) (Figure S1A), and the 5′ transcription start site (TSS) identified by cap analysis of gene expression (CAGE) 25 in human adipocytes suggest that linc-ADAIN has three alternatively spliced isoforms, with isoforms 1 and 2 sharing the same promoter (Figures 2A and S2A).Isoform 3 is the most abundant and therefore is the predominant focus for follow-on studies.Analysis of the phylogenetic information-based codon substitution frequency (PhyloCSF/Expasy) tracks on the UCSC genome browser 26 suggested that the three linc-ADAIN transcripts had low probabilities of containing open reading frames for coding sequences.In vitro transcription/ translation assays failed to detect any peptide products, supporting its predicted non-coding status (Figure S1B).Tissue expression of linc-ADAIN using RNA-seq data from GTEx (Genotype-Tissue Expression project) shows that linc-ADAIN is predominantly expressed in adipose and kidney tissues (Figure 2B).Previously, we showed that linc-ADAIN was detected in adipocytes, but not in monocytes, unpolarized macrophages, or macrophages polarized to M1 or M2 by qPCR. 17Using the Broad Institute online single-cell RNA-seq in adipose tissue dataset from Emont et al., 27 expression of linc-ADAIN, listed as linc01230, is almost exclusively limited to adipocytes and not other cell types within human WAT (Figure 2G).Human adipocyte chromatin immunoprecipitation (ChIP)-seq data revealed high occupancy of H3K4me3 at the linc-ADAIN locus, with multiple PPARγ and C/EBPα binding peaks, and H3K4me and H3K27ac histone enhancer marks upstream of its TSS (Figure 2A).In mature adipocytes, linc-ADAIN expression was reduced by GW9662, a PPARγ antagonist (Figure S1C).Conversely, the PPARγ agonist rosiglitazone induced its expression (Figure 2C).linc-ADAIN was slightly expressed in ASC pre-adipocytes but was markedly induced during ASC differentiation into adipocytes (Figure 2D).The linc-ADAIN locus is syntenic in mouse as defined by homologous neighboring protein-coding genes.However, unlike human linc-ADAIN, the 2610016A17Rik lincRNA transcript upstream of DMRT2 in mouse is on the antisense strand, is not regulated by PPARγ in the same manner as human linc-ADAIN, and shares a promoter region with mouse DMRT2 (Figure S1D).Overall, these data show that human linc-ADAIN is exclusively expressed in adipocytes in WAT, is markedly enriched in WAT over other tissues, is directly regulated by the canonical adipose transcription factor PPARγ, and most likely is not conserved.

linc-ADAIN is located predominantly in the cytoplasm
Following fractionation of mature primary human ASC adipocytes, linc-ADAIN transcripts are found mostly in the cytoplasm (Figures 2E and S2B).RNA scope confirms linc-ADAIN isoforms to be mostly cytoplasmic, often in perinuclear locations, particularly the most abundant isoform, 3 (Figure 2F).Using lentiviral delivery of an shRNA to target all three isoforms in human hTERT adipocytes (Figure S2D), ~85% KD of linc-ADAIN isoforms was achieved compared to the scramble shRNA (Figure S1E).linc-ADAIN KD had no effect on the expression of nearby protein-coding genes ± 500 kb of the linc-ADAIN locus, e.g., DMRT2, DMRT1, or DMRT3 (Figure S1F) or DMRT2 protein (Figure S2C).

linc-ADAIN KD increases lipid accumulation, the adipogenic program, and cytokine secretion in human adipocytes
Due to linc-ADAIN's association with human obesity, we tested whether linc-ADAIN regulated human adipocyte differentiation and lipid storage.KD of linc-ADAIN via shRNA increased cell size and lipid storage, measured by Nile red staining and cellular triglyceride levels, in ASC hTERT adipocytes (Figures 3A, 3B, and 3C) at 14 and 21 days post differentiation.KD of linc-ADAIN also increased mRNA (Figure 3D) and protein expression (Figure 3E) of adipogenic markers PPARγ, FASN, ATGL (Pnpla2), perilipin 1, FABP4, GLUT4, C/EBPα, and adiponectin and increased adipocyte secretion of adiponectin at day 14 of differentiation (Figure S3A).Levels of multiple RNAs that regulate adipocyte lipolysis and de novo lipogenesis, including ATGL, HSL, MGL, and MTP, were increased by KD of linc-ADAIN at day 14 of adipocyte differentiation (Figure S3C).
Because adipose tissue inflammation modulated its expression, we tested whether linc-ADAIN regulated inflammatory pathways in human adipocytes.Using a human cytokine array (C5 Ray Biotech, 80 cytokines) as an initial screen, we found that, compared to scramble control, linc-ADAIN KD tended to increase adipocyte secretion of several cytokines, including IL-8, MCP-1, IL-6, M-CSF, and MIP-1 (Figures S3E and S3F).Using a human adipocyte panel Luminex assay, we confirmed that linc-ADAIN KD increased adipocyte secretion of IL-8, MCP-1, and IL-6 (Figure 3F) and validated this by ELISA (Figures S4A and S4B).Suggestive of post-translational regulation, of these, only IL-8 had significant increases at the mRNA level, and this was to a much lesser extent (4fold) than increases in secreted IL-8 protein (20-fold) (Figures 3G and S4B).KD of linc-ADAIN tended to increase IL-1β and TNF-α mRNA transcripts but did not increase IL-1β or TNF-α protein secretion (Figures 3F and 3G), known to be very low at baseline in adipocytes. 28Cytokine findings were consistent in independent experiments using an antisense oligonucleotide (ASO) against linc-ADAIN and a negative control (Figure S3F to S4F).Ingenuity Pathway Analysis (IPA) and gene ontology (GO) enrichment analysis of RNA-seq data from linc-ADAIN KD vs. scramble control 14-day adipocytes showed cellular movement as a top molecular function (Figure S5A) and inflammatory response and angiogenesis as top enriched GO biological processes (Figures S5B, S6A, and S6B) for genes enriched in linc-ADAIN KD vs. scramble control.

linc-ADAIN interacts with HuR and IGF2BP2 in human adipocytes
We screened for candidate linc-ADAIN-protein interactions as potential functional mediators.First, RBPmap 29 was used to predict RNA binding protein (RBP) motifs within the linc-ADAIN RNA sequence.A total of 92 proteins were predicted to have binding with isoform 3, the most abundant linc-ADAIN (Table S2).Second, a biotinylated pull-down assay of linc-ADAIN coupled to mass spectrometry (MS) in primary ASC adipocytes identified 118 proteins that had a ≥5-fold total spectrum count (TSC) bound to linc-ADAIN compared to antisense control (Table S1).Since we were interested in how linc-ADAIN may regulate specific cytokine mRNAs, we used RBPmap to predict binding proteins that were common for IL-8, MCP-1, and IL-6 (Tables S3-S5) as well as for linc-ADAIN.We then overlapped the five datasets, which included (1) linc-ADAIN MS-pull-down data and the RBPmap-predicted proteins for (2) linc-ADAIN, (3) IL-8, (4) MCP-1, and (5) IL-6 (Tables S1-S5), and we found seven proteins in common: HuR, IGF2BP2, G3BP2, FXR2, MATR3, FMR1, and SFPQ (Figures 4A, 4B, and S7A).STRING analysis of the seven proteins showed regulation of translation and mRNA stability as top biological processes (Figure 4C), and two of these proteins (IGF2BP2 and HuR [ELAV1]) are found in the coding region instability determinant complex, which stabilizes mRNAs to protect them from degradation. 21,30Specific proteins involved in mRNA stability are HuR (ELAVL1), IGF2BP2, FMR1, and FXR2 (Figure 4D).HuR and IGF2BP2 had the highest Z score and number of predicted binding motifs for linc-ADAIN of the seven proteins (Figure 4B).For this reason, we chose to validate their interactions with linc-ADAIN.Using RNA immunoprecipitation (RIP) assays, we found that linc-ADAIN RNA interacted with HuR protein in mature ASC adipocytes (Figure 4E) but not IGF2BP2 (data not shown).However, consistent with previously reported interactions between IGF2BP2 and HuR proteins, 21,30 we found that IGF2BP2 and HuR interact in human hTERT ASC adipocytes, as shown by western blot of an immunoprecipitation pull-down of IGF2BP2 probed for HuR protein (Figure 4F).Combining RNA scope and immunofluorescence (IF), we observed co-localization of linc-ADAIN with HuR and IGF2BP2 in adipocyte perinuclear regions (Figure 4G).

KD of linc-ADAIN stabilizes IL-8 and KLF5 mRNA, likely through IGF2BP2
We next investigated how linc-ADAIN might modulate HuR and IGF2BP2 functions.KD of linc-ADAIN via shRNA had no effect on IGF2BP2 or HuR protein expression in mature ASC adipocytes (Figure S7B).RIP assays showed that KD of linc-ADAIN affected binding of mRNAs to both HuR and IGF2BP2 (Figure 5).For IL-8 and KLF5, linc-ADAIN KD blunted the interaction with HuR (Figures 5A and 5C), but increased binding with IGF2BP2 (Figure 5B).With linc-ADAIN KD, there was enrichment of IL-8 and KLF5 mRNA on IGF2BP2 protein (Figures 5B and 5D).The interaction of MCP-1 and HuR was blunted by linc-ADAIN KD, whereas the MCP1-IGF2BP2 interaction appeared unchanged (Figures S8A and S8B).c-myc mRNA, known to be modulated by IGF2BPs, 30 also had reduced binding to HuR and IGF2BP2 upon linc-ADAIN KD (Figures S8E and S8F) without an effect on c-Myc protein expression (Figure S8G).Negligible effects were observed for IL-6 (Figures S8C and S8D) and GAPDH, an mRNA not known to be modulated by the mRNA stabilizing coding region instability determinant complex (Figures 5E and 5F).
Given the roles of HuR and IGF2BP2 in mRNA stabilization and in adipogenesis, 22,[30][31][32] we next explored whether mRNA stabilization of specific targets, such as IL-8 and KLF5, might mediate linc-ADAIN effects on adipogenic gene expression noted above.We crossreferenced mRNAs with the greatest fold change in IGF2BP2 binding vs. IgG in HEK293T cells from public data 21 and mRNAs with the greatest differential expression in our bulk RNA-seq data from linc-ADAIN shRNA vs. scramble shRNA in 14-day ASC adipocytes (Figure S9A).The top candidate mRNA was KLF5, a transcription factor shown to be a key regulator of early adipogenesis 20 and also found in the angiogenesis GO pathway above (Figure S6B).KLF5 mRNA was upregulated by linc-ADAIN KD (Figure S9B) and showed the same pattern of binding to HuR and IGF2BP2 as IL-8 (Figures 5C and 5D), indicating that KLK5 and IL-8 mRNAs may be regulated similarly.
We investigated the effect of linc-ADAIN KD on mRNA stability in mature ASC hTERT adipocytes.Using actinomycin-D to pause transcription, linc-ADAIN KD delayed IL-8 and KLF5 mRNA decay responses (Figures 5G and 5H), suggesting that endogenous linc-ADAIN accelerates the decay of these two mRNAs in maturing adipocytes.linc-ADAIN KD did not affect GAPDH, MCP1, or IL-6 mRNA stability (Figures 5I, S8H, and S8I), but did cause a modest delay in the MYC (Figure S8J) mRNA decay response.linc-ADAIN KD enhanced IGF2BP2 binding to IL-8 and KLF5 but not to MCP-1 or IL-6, suggesting that IGF2BP2 may mediate linc-ADAIN induced differences in patterns of IL-8 and KLF5 but not MCP-1 or IL-6 mRNA and protein.Overall, because of its established role in mRNA stabilization, [33][34][35] IGF2BP2 may be a mediator of IL-8 and KLF5 stabilization observed during linc-ADAIN KD.

KD of linc-ADAIN prevents the normal reduction of KLF5 expression during late adipocyte differentiation
As KLF5 is an important transcription factor in early adipogenesis, 20 we investigated its expression during adipocyte differentiation in the presence and absence of linc-ADAIN.Normally, linc-ADAIN is induced gradually as adipocytes differentiate (Figure 2D), whereas KLF5 expression increases during early differentiation but falls as adipocytes mature (Figures 6B and 6E, scramble). 20linc-ADAIN KD prevented the drop in KLF5 mRNA (Figures 6A and 6B) and protein (Figure 6E) during later adipocyte differentiation, with marked upregulated expression at day 14 compared to absent KLF5 expression in controls at these later time points. 20This linc-ADAIN-KD-induced persistence of KLF5 expression was associated with increased expression of PPARγ1 and 2 (Figure 6E) and CEBPα at day 14 (Figure 6D).Consistent with published data, 36 IL-8 mRNA expression is high at baseline in pre-adipocytes, and levels decline as adipocytes differentiate (Figure 6F, scramble).Silencing of linc-ADAIN before differentiation led to increased IL-8 mRNA at baseline (Figure 6F) and a sustained increase in IL-8 secretion throughout differentiation (Figure 6G).To test if KLF5 and IL-8 could mediate the adipogenic and inflammatory marker phenotypes observed with linc-ADAIN KD, we performed ASO-induced KD of KLF5 and IL-8 in ASC hTERTs, achieving 81% and 89% KD, respectively (Figures S9C  and S9D).We observed significant reductions in PPARγ and IL-6 with KLF5 KD (Figure S9E).Indeed, KD of KLF5 had the opposite effect on PPARγ and IL-6 compared to linc-ADAIN KD (Figures S3D and S3F).These findings support the concept that linc-ADAIN may control adipogenesis through KLF5.IL-8 KD showed no changes in adipogenic genes such as KLF5 or PPARγ-this was expected because IL-8 is not known to regulate these factors.However, there was a significant reduction in IL-6 gene expression with IL-8 KD.Again, this is in the opposite direction of the linc-ADAIN KD effect on IL-6 (see Figure 3F).
Overall, our data suggest that the sustained increase in KLF5 and IL-8 mRNA induced by linc-ADAIN KD during adipocyte differentiation is mediated by IGF2BP2 stabilization of their mRNA.Sustained expression of KLF5 is likely to drive increased expression of adipogenic transcription factors, PPARγ and CEBPα, as has been reported, 20,37 and increased adipocyte lipid accumulation.

Adipocytes with reduced linc-ADAIN have increased size and lipid storage and greater macrophage infiltration in vivo
Next, we sought to determine the impact of the depletion of linc-ADAIN on adipocyte maturation, lipid storage, and KLF5 and IL-8 expression in vivo.Although, the human linc-ADAIN locus is syntenic in mice, our data (Figure S1) suggest that linc-ADAIN is unlikely to be a conservation of the mouse lincRNA (2610016A17Rik) at this syntenic locus.Further 2610016A17Rik is reversed, on the antisense strand, and shares a common promoter region with the nearest protein-coding gene, DMRT2.Thus, it is not appropriate or feasible to generate a 2610016A17Rik-knockout mouse model to study linc-ADAIN loss of function in vivo (Figure S1C).For this reason, we chose to adapt a transplant model, whereby cultured human adipocytes are implanted with Matrigel into NSG mice 23,24 to generate a functional human fat pad.We implanted day 14 hTERT ASC human adipocytes overexpressing shRNA against linc-ADAIN or scramble control into the flank regions of female NSG mice.We extracted these human adipose transplants after 16 weeks in vivo.The mice had no significant changes in blood cell counts, weight gain, or organ weight during the protocol (Figures S10C-S10E).Compared to controls, linc-ADAIN-KD explants had significantly fewer small adipocytes (<20 μm diameter) and more large adipocytes (>50 μm diameter) (Figures 7A and 7B), and the linc-ADAIN-KD explants weighed significantly more than control explants, and the mice with linc-ADAIN-KD implants also had increased plasma human adiponectin compared to control mice (Figures S10A and S10B).linc-ADAIN-KD explants also had increased infiltration of macrophages, as measured by CD68 and F4/80 (Adgre1) by immunohistochemistry (IHC) (Figures 7A, 7C, and 7D) and via qPCR (Figures 7E and 7F).Although not all were statistically significant, compared to control explants, mRNA levels of MCP-1, IL-8, IL-6, and KLF5 tended to be higher in linc-ADAIN-KD explants (Figures 7G, 7H, 7I, and 7J).Expression of adipogenic genes, like PPARγ and CEBPα, that was increased by linc-ADAIN KD in vitro was decreased in vivo (Figure 7K).This may be due to increased infiltration of macrophages and their secreted inflammatory

DISCUSSION
Adipose tissue can store excess calories through two mechanisms, hypertrophy and hyperplasia.][42] This dysfunction contributes to CMDs. 43Here we uncover how linc-ADAIN, an adipose lincRNA downregulated in human sWAT during obesity, regulates adipocyte lipid storage and secretion of cytokines.Although the publicly annotated linc-ADAIN, known as linc01230, has been shown to have modulated expression in endometrial cancer, 44 renal cell carcinoma, 45 and lung adenocarcinoma, 46 our data suggest that in the context of noncancer human pathophysiology, linc-ADAIN is preferentially expressed in adipose over other tissues and is restricted to adipocytes within human WAT.We found that linc-ADAIN interacts and modulates the RBPs HuR and IGF2BP2 in adipocytes to reduce the stability of KLF5 and IL-8 mRNAs.linc-ADAIN disrupts the binding of IGF2BP2 to these mRNAs, increasing their decay and reducing protein translation.When linc-ADAIN is depleted, KLF5 and IL-8 mRNA binding to IGF2BP2 is enhanced, stabilizing their mRNAs and thus increasing their protein expression and secretion.Increased KLF5 enhances later adipocyte differentiation and lipid storage in vitro with higher PPARγ1 and 2 expression compared to when linc-ADAIN is present.In vivo, adipocyte implants with linc-ADAIN depletion were much larger and had evidence for increased macrophage infiltration.linc-ADAIN is induced by PPARγ early during adipocyte differentiation, whereas the increases in PPARγ induced by linc-ADAIN KD during later adipocyte differentiation are likely mediated indirectly by stabilization of KLF5 and the induction of PPARγ by sustained KLF5 expression.In vivo, adipocyte implants with linc-ADAIN depletion were much larger and had evidence for increased macrophage infiltration.While the increase in PPARγ coincident with increased adipose inflammation with linc-ADAIN KD may seems counterintuitive, this combination has often been observed in adipose remodeling in both experimental models and clinical obesity.47,48 Thus, linc-ADAIN is an adipose anti-inflammatory lincRNA that can modulate lipid storage and inflammation, two key features of adipose tissue remodeling in obesity.
RBPs play a central role in post-transcriptional gene regulation 49 and fundamental roles in cellular activities such as proliferation and differentiation.There is increasing evidence for the interplay of lncRNAs and RBPs in physiological and pathological contexts such as cancer. 50Yet, few RBPs have been studied in the context of adipogenesis. 51HuR (also known as ELAV1) is a ubiquitously expressed member of the Hu family of RNAbinding proteins that selectively binds adenylate uridylate-rich elements (AREs) and thereby stabilizes its RNA targets. 52HuR regulates the expression of mRNAs and proteins involved in cellular processes such as inflammation and apoptosis. 53,54Transgenic overexpression of HuR in myeloid cells modulated TNF-α, IL-1β, and TGF-β1 levels in the serum of LPS-challenged mice, and the effect was attributed to enhanced mRNA stability. 53HuR acts in a tissue-specific manner through cell-specific targets to exert distinctive functions on different adipose depots. 31Adipose knockout (AKO) of HuR in mice fed a high-fat diet led to increased WAT fat mass, adipocyte hypertrophy, increased circulating total cholesterol, increased triglycerides, insulin resistance, and glucose intolerance. 32Adipocyte hypertrophy in HuR AKO was attributed to reduced ATGL expression and ATGL mRNA stability, leading to reduced lipolysis. 32HuR and ATGL expression is reduced in subcutaneous adipose tissue of obese humans. 32Of specific relevance to the actions of linc-ADAIN, AKO of HuR can enhance the adipogenic gene program and an inflammatory program in epidydimal white fat, by regulating the stability of hundreds of adipocyte mRNAs, including Insig1.31 In the presence of linc-ADAIN, we observe increased binding of cytokines IL-8, MCP-1, and IL-6 to HuR compared to when linc-ADAIN is depleted.Indeed HuR has been shown to enhance the stabilization of IL-8, 55 IL-6, 56 and MCP-1 57 in other cell types.HuR complexes with IGF2BP family proteins are part of the coding region instability determinant complex, which stabilize mRNAs to protect them from degradation, with c-myc mRNA being a prime example. 21,30e RBP IGF2BP2 has been shown to interact with lncRNAs in adipocytes by us and others, regulating adipocyte differentiation and lipogenesis via linc-ADAL 16 or lipolysis via lncRAP. 58In genome-wide association studies (GWASs), the IGF2BP2 locus has been repeatedly associated with T2D. 59,60Similar to HuR, IGF2BP2 functions are tissue specific due to IGF2BP2 binding different RNAs across tissues. 51Global deletion of IGF2BP2 in mice leads to resistance to high-fat-diet-induced obesity, fatty liver, and glucose tolerance, due to increased protein but not mRNA of UCP1, which promotes increased energy expenditure.IGF2BP2 binds to UCP1 mRNA and inhibits its translation in brown adipose tissue (BAT). 61In contrast, hepatocyte-specific IGF2BP2 knockout results in diet-induced fatty liver by impairing fatty acid oxidation. 62Like linc-ADAIN, many lincRNAs are tissue specific in their expression, 63 providing one molecular mechanism for how RBPs have tissue-specific functions, e.g., for HuR. 32,53Through direct interactions with lincRNAs, HuR and IGF2BP2 can stabilize and increase the translation of certain mRNAs in cancer. 64or example, circEIF3H acts as a scaffold for IGF2BP2 and HuR to regulate mRNA stability of HSPD1, RB8A, and G3BP1 in triple-negative breast cancer. 65We found HuR bound to IGF2BP2 in human adipocytes, and when linc-ADAIN was depleted, this enhanced the binding of HuR and IGF2BP2 to IL-8 and KLF5 mRNA, increasing their mRNA stability and protein levels.IGF2BP2 is known to recruit HuR and MATR3 as stabilizers, 21 and we also found that MATR3 was present in our linc-ADAIN pull-down (Figure 4B).lincRNAs can regulate cytokines in a transcriptional and post-transcriptional manner. 66inc-ADAIN depletion increased IL-8 mRNA stability, mRNA levels, and protein secretion, likely through increased binding of IGF2BP2 to IL-8 mRNAs.IL-8 mRNA stabilization in macrophages is distinct from other cytokines (TNF-α and IL-6) and is ARE independent. 67L-8 mRNA is unstable in undifferentiated THP-1 monocytes but becomes constitutively elevated and stabilized in differentiated THP-1 macrophage cells. 67IL-8 stabilization is cell-type dependent.We found high expression of IL-8 mRNA in pre-adipocytes, which then dropped during adipocyte differentiation coincident with an increase in linc-ADAIN.IL-8 is a chemokine for neutrophils, and in early stages of obesity, neutrophils infiltrate adipose tissue, producing chemokines that promote macrophage infiltration. 68Indeed, with linc-ADAIN depletion we observed increased macrophage infiltration in 16-week-old adipose explants, which may be mediated by increased IL-8.
KLF5 is an early transcription factor for adipocyte differentiation and adipogenesis 20 and is needed to activate the PPARγ2 promoter.KLF5 is induced early (within 1 h) during adipocyte differentiation and gradually declines thereafter. 20KLF5 +/− mice have delayed adipose tissue development, have reduced lipid droplets, and are resistant to high-fat-diet-induced obesity. 20linc-ADAIN depletion during adipocyte differentiation elevated KLF5 in pre-adipocytes and resulted in sustained KLF5 levels and increased PPARγ1 and 2 expression, as well as levels of adipogenic genes.Like IL-8, we observed enhanced stability of KLF5 mRNA with linc-ADAIN depletion, likely due to its increased binding to IGF2BP2 protein.Notably, both KLF5 and IL-8 mRNA levels are high in pre-adipocytes and decline during differentiation, coincident with increased linc-ADAIN expression.When linc-ADAIN is depleted, both KLF5 and IL-8 remain high throughout differentiation, suggesting that linc-ADAIN is an important regulator of their expression and functions during physiological adipocyte differentiation, adipogenesis, and lipid storage in humans.Furthermore, consistent with KLF5 and IL-8 being potential mediators of linc-ADAIN effects, we found that KD of IL-8 and KLF5 produced the phenotypes opposite that observed with linc-ADAIN KD.Our findings suggest a potential negative feedback regulation by linc-ADAIN of KLF5 under normal physiological circumstances.With reduction of linc-ADAIN expression experimentally or in the context of obesity, this regulation may be lost, permitting adipose remodeling and expansion.To our knowledge, linc-ADAIN is the only lincRNA to be shown to regulate KLF5 or IL-8 expression and functions within adipose tissue.KLF5 expression can promote some cancers and tumor angiogenesis. 69For example, KLF5 promoted angiogenesis in bladder cancer by directly increasing vascular endothelial growth factor A (VEGFA) transcription. 70KD of lncRNA MCM3AP-AS1 increases miR-211, which targets KLF5 and inhibits its expression, affecting angiogenesis in glioblastoma. 71nterestingly, IL-8 is also known as a pro-angiogenic factor, particularly in cancer. 72In our pathway analyses of adipocyte RNA-seq data from linc-ADAIN KD vs. scramble control, angiogenesis was one of the top biological processes identified.Angiogenesis is necessary for normal adipose tissue development to provide nutrients and oxygen.In obesity, angiogenesis tends to precede hyperplastic expansion, but follows hypertrophic expansion, leading to hypoxia in hypertrophic expansion. 73The mechanisms by which adipose tissue growth is coordinated with capillary networks are still not fully understood and may underlie aspects of adipose tissue dysfunction in obesity and metabolic disease. 73linc-ADAIN increases during adipocyte differentiation and regulates IL-8 and KLF5 expression and thus may contribute to adipose tissue remodeling through adipocyte expansion, lipid storage, angiogenesis, and immune cell recruitment during normal adipose tissue development and physiology.
Most lincRNAs are not conserved across species; thus, creating knockout or overexpression mouse models is challenging.linc-ADAIN has a syntenic locus in mouse, yet linc-ADAIN is probably not a conservation of the mouse 2610016A17Rik lincRNA at this locus.In addition, mouse 2610016A17Rik is on the antisense strand and shares a promoter region with DMRT2.Thus, deleting 2610016A17Rik would likely affect DMRT2 expression in the mouse and confound studies of 2610016A17Rik.Therefore, we adapted an implant model to generate a functional human adipose fat pad in mice. 23,24Human ASC hTERT adipocytes with a lentiviral expression of an shRNA targeting linc-ADAIN or a scramble control after 14 days of differentiation in vitro were implanted into the flanks of NSG mice.We hypothesized that linc-ADAIN KD adipocytes would have increased differentiation and lipid storage over time in vivo and that increased adipocyte cytokine secretion from linc-ADAIN KD would lead to increased macrophage infiltration in vivo.By 16 weeks after implantation, linc-ADAIN-KD adipocytes produced larger adipose depots and had increased macrophage infiltration compared to scramble shRNA adipocytes.These results mirror many hallmark features of adipose tissue remodeling in obesity and provide proof of principle that linc-ADAIN regulates adipogenesis, lipid storage, and tissue inflammation in vivo.

Limitations of the study
In our work, the role of linc-ADAIN in adipogenesis has been characterized in much more detail than its role in adipose tissue inflammation.To understand fully its role in modulating adipose tissue inflammation will require more work, including a closer examination in vitro and in vivo of the effects of adipose linc-ADAIN on recruitment of neutrophils as well as of macrophages via linc-ADAIN-regulated chemokines and cytokines, especially IL-8.Our in vivo adipocyte transplant model is innovative but challenging and has limitations for assessing whole-body metabolic effects of this non-conserved lincRNA.Future in vivo studies using high-fat-diet feeding can examine the effects of adipose linc-ADAIN expression on systemic measures of glucose tolerance, insulin sensitivity, and adipocyte-tomacrophage crosstalk under pathophysiological stress.We could not address directly why linc-ADAIN is downregulated in obesity.We hypothesize this may be due to reduced PPARγ activity caused by inflammatory macrophage-secreted factors in obese adipose. 39ur data showing PPARγ ChIP-seq peaks upstream of linc-ADAIN and PPARγ agonist, rosiglitazone, induction of linc-ADAIN suggest that PPARγ is a transcriptional activator of linc-ADAIN.Previously, the publicly annotated linc-ADAIN, known as linc01230, had been shown to be transcriptionally activated by PPARγ in endothelial cells. 19Overexpression of linc-ADAIN in ASC-hTERTs using a tetracycline-inducible lentivirus (pSILKNeo) system did not significantly change lipid storage or inflammatory gene expression.However, viral vector overexpression of lincRNAs is associated with several technical challenges, including introduction of additional sequence elements that may have non-specific effects and failure of ectopic expression to reproduce physiological subcellular localization.For lincRNAs where endogenous expression is physiologically "protective," overexpression systems fail to produce the phenotype opposite to the loss-of-function models. 74,75Furthermore, we emphasize that our data show significant, consistent and reproducible effects of independent experimental KD approaches using shRNA and ASO in cultured adipocytes in vitro as well as in our adipocyte transplant model in vivo.In humans, linc-ADAIN is downregulated in obese human adipose tissue and is suppressed by acute inflammation during endotoxemia, and its expression is negatively correlated with total body fat percentage. 17Larger human population studies are required to determine if linc-ADAIN is a causal gene in CMD traits.Future work can address if induction of linc-ADAIN (e.g., through RNA activation) in obese adipose in vivo can provide anti-inflammatory and adipose remodeling effects that may attenuate the cardiometabolic complications of obesity.Because cytoplasmic lincRNAs can act as competing endogenous RNAs for microRNA (miRNA), we performed in silico analyses using DIANA LncBase 76 and identified predicted linc-ADAIN interactions with hsa-let-7d-5p, hsa-miR-34a-5p, and hsa-miR-423-5p.Although Ago2 was not identified in our protein-binding analyses, future studies of these predicted miRNA interactions may be warranted.
In summary, we provide a functional study of human adipose linc-ADAIN and its roles in adipose remodeling.We demonstrate that linc-ADAIN is an anti-inflammatory adipose lincRNA that limits adipose expansion and lipid storage and is downregulated in obese white adipose tissue.linc-ADAIN fine-tunes adipogenesis and lipid storage via its regulation of IL-8 and KLF5 mRNA stability and expression, two genes of known importance in adipose tissue remodeling and adipogenesis.linc-ADAIN inhibits the interaction between IL-8 and KLF5 mRNAs and IGF2BP2, an RBP that stabilizes mRNA as part of the coding region instability determinant complex.When adipocyte linc-ADAIN is depleted, IL-8 and KLF5 mRNAs remain stable, and adipocytes have increased mRNA and protein expression throughout adipogenesis, leading to increased expression of IL-8, other chemokines, PPARγ, and the adipogenic program.This results in hypertrophied and inflamed adipocytes.In vivo, linc-ADAIN KD in adipocytes led to adipocyte hypertrophy and larger fat pads as well as increased macrophage infiltration of fat pads implanted into NSG mice-key features of obesity-induced adipose tissue remodeling in obesity and CMD.

STAR★METHODS RESOURCE AVAILABILITY
Lead contact-Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Muredach P. Reilly (mpr2144@cumc.columbia.edu).

Materials availability-Requests
for unique resources should be directed to the lead contact.A completed Materials Transfer Agreement may be required.

•
All data reported in this paper will be shared by the lead contact upon request.
• RNA-seq data and Mass Spec data have been deposited at https:// www.ncbi.nlm.nih.gov/geo/ and https://www.proteomexchange.org/and are publically available as of this publication.Accession numbers are listed in the key resources table.

•
This paper does not report original code.
• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental model and study participant details
Human participants: Human gluteal adipose tissue samples were collected in The Genetics of Evoked-responses to Niacin and Endotoxemia (GENE) study (N = 284, 33% African Americans, age 18-45), a National Institute of Health-sponsored protocol enrolled at University of Pennsylvania (Penn) (NIH clinical trial NCT00953667) 77 .Gluteal subcutaneous fat tissue samples were obtained at baseline, 4, 12 and 24 hours following intravenous LPS bolus (1ng/kg).As described in 16,17 , RNA isolated from a subset of gluteal subcutaneous fat tissue samples underwent deep RNA-sequencing to identify adipose lincRNAs expressed and differentially modulated by LPS in vivo.Linc-DMRT2 (renamed to Linc-ADipose Anti-INflammatory) was identified as a top LPS-modulated lincRNA and its LPS-modulation was validated by qPCR in gluteal adipose tissue of an independent sample of healthy lean GENE subjects (Figure 1E).For clinical translation, qPCR was performed of gluteal subcutaneous adipose tissue from an independent sample of healthy lean GENE subjects (N=29, 48% female; 49% European ancestry; BMI ~21), as well as in abdominal subcutaneous and visceral omental adipose biopsies of obese subjects from the Penn Human Adipose Resources (https://www.med.upenn.edu/idom/adipose.html;N=39, BMI ~48)17 (Figures 1A-1C).All clinical studies were performed with the approval of the Penn Institutional Review Board and written informed consent was obtained from all research participants.Linc-ADAIN was chosen for functional study because 1) it showed high tissue specificity for adipose tissue and 2) it is modulated in obesity and during endotoxemia in human white adipose tissue.RNA-seq library preparation, sequencing and data processing, along with de novo assembly of human adipose transcriptomes for lincRNA discovery have all been previously described in detail in 16,17 .Linc-ADAIN genomic location is chr9 1044999-1048641 (isoform1), chr9 1044999-1048641 (isoform 2), chr9 1045640-1048641 (isoform 3).

Screening of human adipose proteome profiles as coding transcript filter:
As previously described in 16 .Briefly to further exclude coding transcripts from annotated lincRNA catalogs and our de novo assembled putative lincRNAs, we screened human adipose proteome profiles for predicted peptides derived from these human lincRNA transcripts as previously described 78 .Published human adipose mass spectrometry (MS) shotgun profiling of deep coverage was used for our analyses 79 .Briefly, the linc-ADAIN transcripts (annotated and de novo assembled) were in silico translated into all possible peptides using Sixpack 80 .MS/MS spectra from human adipose MS profiling were searched against human UniProt database (20,205 reviewed canonical entries, version July 2017), MS common contaminants database and the custom in silico translated peptides from lincRNAs.MS raw files were analyzed with MaxQuant version 1.6.0.1 with a false discovery rate (FDR) of 1%.Variable modifications were indicated as methionine oxidation and protein N-terminal acetylation, while carbamidomethylation of cysteines was designated as a fixed modification.Trypsin, which cleaves after lysine and arginine residues, was selected as the digestion enzyme and two missed cleavages were permitted.The main search peptide tolerance was set to 4.5 ppm and the first search tolerance was set to 20 ppm.iBAQ (intensity-based absolute quantification) was implemented for label-free quantification to accurately identify protein copy number.We identified over 28,075 high confidence peptides which mapped into 3,226 human proteins, including proteins whose mRNA abundances in adipose, e.g.KRT7 (FPKM 0.50) and SFN (FPKM 0.43), are low and similar to lincRNAs.In this context of robust profiling of protein coding genes, we found only 2 peptides which correspond to in silico translated ORFs from 2 lincRNAs (at FDR Q-value <0.05).
Synteny and conservation analysis of human adipose lincRNAs: As previously described in 16 .Briefly conserved lincRNAs are known to have limited sequence similarity with their orthologs but typically display genomic position conservation across species 81 .We examined the synteny of human lincRNA loci relative to mouse.For linc-ADAIN, we identified the two-neighboring protein-coding genes within ±500,000bp and searched for homologous genes in mouse using HomoloGene (http://www.ncbi.nlm.nih.gov/homologene).If two neighboring genes were homologous in mouse, the lincRNA locus was classified as syntenic.Next, we looked for evidence of mouse lincRNA transcription at syntenic regions in the mouse genome.We took a published mouse adipose lincRNAs transcriptome 78 and mapped those mouse lincRNAs to syntenic human regions.A human lincRNA was considered a conserved lincRNA if the mouse adipose lincRNAs was actually expressed in the syntenic region.
Transcription factor binding analysis and histone modification analysis: As previously described in 16 .Briefly we downloaded PPARγ and C/EBPα peaks identified in adipocytes derived from human Simpson-Golabi-Behmel syndrome (SGBS) pre-adipocytes (GSE27450) 82 .Peaks from two replicates for each transcription factor were merged.We then mapped the merged peaks to ±1,500bp of the TSS for three groups of human adipose lincRNAs based on their conservation with mouse: (1) lincRNAs with mouse adipose lincRNA transcripts expressed in the syntenic loci, (2) lincRNA with syntenic loci but no mouse lincRNA expression in the region and (3) lincRNAs non-syntenic in mouse.Mapping was done at the transcript level.We then counted the number of lincRNAs with PPARγ and C/EBPα binding as well as the total number of binding events for adipose lincRNAs.For Spearman correlation analyses between lincRNA adipose abundance/fractional expression and the distance of nearby transcription factor binding sites to lincRNA TSS, we identified the midpoints of transcription factor ChIPseq peaks for lincRNAs with transcription factor binding within ±1500bp and calculated the distances between lincRNA TSS and midpoint of transcription factor binding peak.For lincRNA genes with multiple TF binding, the smallest transcription factor binding site-TSS distances were used.

Tissue expression and single-cell expression of linc-ADAIN in adipose tissue:
Gene expression in TPM and the associated sample attributes were downloaded from GTEx v8.Mean expression was calculated for each tissue and transformed by log(TPM + 1).TPMs were visualized on heatmap in a subset of 28 tissues from GTEx.Single cell UMAP data for linc01230 was taken from the Single Cell Portal -Broad Institute (https:// singlecell.broadinstitute.org)using single cell sequencing data from human white adipose tissue by Emont et al., 2022 27 .
Mouse studies-All animal experiments were approved by the Institutional Animal Care and Use Committee of Columbia University (protocol AC-AABR1551).NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice were obtained from The Jackson Laboratory (#005557) and bred in our animal facility.T method was adapted from previously published 23,24 .Female mice were used at 12 to 14 weeks of age.Under anesthesia, mice were injected subcutaneously into the flank regions with 50:50 (vol/vol) Matrigel (Corning 356231) and ASC hTERT adipocytes with lentiviral expression of shRNA targeting linc-ADAIN or scramble control as described above.ASC hTERT are differentiation for 14 days in vitro before injection, as described above.Mice were on standard chow diet and after 16 weeks, the mice were euthanized, and tissues and human adipose implants were removed for further study.Fourteen weeks after implantation, mice underwent a glucose tolerance test (GTT) or insulin tolerance test (ITT).(N= 6 for scramble shR, N=4 for Linc-ADAIN shR).Mice were fasted overnight then received an i.p injection of 1.5g/kg glucose and blood samples collected at times indicated to measure blood glucose using a/ glucometer.For ITT mice were fasted for 5 hours then received an i.p injection of 0.5U/kg of insulin and blood samples collected at times indicated to measure blood glucose.
shRNA sequences: The estimated counts and transcripts per million (TPM) values for each gene were obtained from a pseudoalignment to a kallisto index derived from human transcriptomes (GRCh38) using kallisto version 0.44.0 83 .Differential gene expression analysis was performed using DESeq2 version 1.32.0.17,174 genes with at least 10 reads total in the estimated counts were used to compare the expressions in RNA-seq data from scramble vs linc-ADAIN shRNA KD in ASC adipocytes.DESeq2 independent filtering was also applied to exclude count outliers and low mean count across samples.1,403 genes with at least 2-fold difference between two conditions (scramble vs linc-ADAIN shRNA KD) and FDR adjusted P-value of less than 0.05 were considered differentially expressed genes.Heatmap was generated using Morpheus (Broad Institute).
The supernatant was collected and centrifuged and this supernatant was the cytoplasmic fraction.The pellet was resuspended in 1x Dautry Buffer + enzymes (nuclear fraction).Total lysate, cytoplasmic fraction and nuclear fraction was used for downstream RNA extraction or western blot.
In vitro transcription/translation assay-Linc-ADAIN pcDNA3.1 expression vector T7 promoter Life-Technologies and TNT ® Quick Coupled Transcription/Translation System (Promega) were used to in vitro transcribe and translate the full-length lincRNA.The T7 luciferase expression control plasmid supplied with the kit was used as the positive control.Protein products from transcription/translation reactions were labeled with biotinylated lysine. 2 ul of the reaction products were added with 15 ul SDS sample buffer, heat denatured and resolved in NuPage SDS-polyacrylamide gel (Life Technologies).Biotinlabeled proteins were subsequently detected using Transcend ® nonradioactive translation detection system (Promega).
RNA pulldown assay and mass spectrometry-Linc-ADAIN or Anti-sense linc-ADAIN were cloned into pcDNA3.1 vector downstream of a T7 promoter (Life Technologies).The expression plasmid was linearized and transcribed in vitro using AmpliScribe T7 Flash Biotin RNA Transcription Kit (Lucigen ASB71110) following the manufacturer's instructions.Biotinylated RNAs were subsequently purified with RNeasy spin columns (Qiagen).For the pulldown assay, RNA (5000ng in 20ul) was refolded for proper secondary structure formation by heating for 2 min at 90°C and add RNA structure buffer (20mM trish ph7, 0.2M KCL, 20mmM MgCL2), incubate at RT for 30mins.Adipocyte lysates were collected from ~1×107 mature ASC-derived adipocytes in 2 ml RNA immunoprecipitation (RIP) buffer (150 mM KCl, 25 mM Tris at pH 7.4, 5 mM EDTA, 0.5mM DTT, 0.5% NP40, 1X protease inhibitor cocktail, 100 U/ml RNAseOUT).Shear cells on ice with dounce homogenizer.The lysate was cleared from cell debris by centrifugation at 15,000g for 15 min at 4°C.Protein concentration in the lysate was determined using the BCA protein assay kit.RNA pulldown was carried out as previously described in 16 .Proteins were separated by 4-12% gradient Bis-Tris gels followed by coomassie blue staining (Thermo Fisher).
For the mass spectrometry (MS) analysis, gel band was excised and in-gel digestion was performed as previously described 84 .Peptides were dissolved in 3% acetonitrile/0.1% formic acid.

LC-MS/MS analysis:
Thermo Scientific ™ UltiMate ™ 3000 RSLCnano system and Thermo Scientific EASY Spray ™ source with Thermo Scientific ™ Acclaim ™ PepMap ™ 100 2 cm × 75 μm trap column and Thermo Scientific ™ EASY-Spray ™ PepMap ™ RSLC C18 50 cm × 75 μm ID column were used to separate desalted peptides with a 5-30% acetonitrile gradient in 0.1% formic acid over 50 min at a flow rate of 250 nL/min.The column temperature was maintained at a constant 50 °C during all experiments.Thermo Scientific ™ Orbitrap Fusion ™ Tribrid ™ mass spectrometer was used for peptide MS/MS analysis.Survey scans of peptide precursors were performed from 400 to 1500 m/z at 120K FWHM resolution (at 200 m/z) with a 2 × 10 5 ion count target and a maximum injection time of 50 ms.The instrument was set to run in top speed modewith 3 s cycles for the survey and the MS/MS scans.After a survey scan, tandem MS was performed on the most abundant precursors exhibiting a charge state from 2 to 6 of greater than 5 × 10 3 intensity by isolating them in the quadrupole at 1.6 Th.CID fragmentation was applied with 35% collision energy, and resulting fragments were detected using the rapid scan rate in the ion trap.The AGC target for MS/MS was set to 1 × 10 4 and the maximum injection time was limited to 35 ms.The dynamic exclusion was set to 45 s with a 10-ppm mass tolerance around the precursor and its isotopes.Monoisotopic precursor selection was enabled.
Data analysis: Raw mass spectrometric data were analyzed using the MaxQuant environment 85 v.1.6.1.0and employed Andromeda for database search 86 at default settings with a few modifications.The default was used for the first search tolerance and main search tolerance: 20 ppm and 6 ppm, respectively.MaxQuant was set up to search with the reference human proteome database downloaded from Uniprot.MaxQuant performed the search trypsin digestion with up to 2 missed cleavages.Peptide, Site, and protein FDR were all set to 1% with a minimum of 1 peptide needed for identification, and labelfree quantitation (LFQ) was performed with a minimum ratio count of 1.The following modifications were used as variable modifications for identifications and included for protein quantification: Oxidation of methionine (M), Acetylation of the protein N-terminus, and Deamination for asparagine or glutamine (NQ).Results obtained from MaxQuant were further uploaded on Scaffold.Accession number.
The overlapping 7 proteins, HuR, IGF2BP2, G3BP2, FXR2, MATR3, FMR1, SFPQ were uploaded to STRING 88 to search for their possible interactions and biological processes.Jackson Laboratory (#005557) and bred in our animal facility.T method was adapted from previously published 23,24 .Female mice were used at 12 to 14 weeks of age.Under anesthesia, mice were injected subcutaneously into the flank regions with 50:50 (vol/vol) Matrigel (Corning 356231) and ASC hTERT adipocytes with lentiviral expression of shRNA targeting linc-ADAIN or scramble control as described above.ASC hTERT are differentiation for 14 days in vitro before injection, as described above.Mice were on standard chow diet and after 16 weeks, the mice were euthanized, and tissues and human adipose implants were removed for further study.Fourteen weeks after implantation, mice underwent a glucose tolerance test (GTT) or insulin tolerance test (ITT).(N= 6 for scramble shR, N=4 for Linc-ADAIN shR).Mice were fasted overnight then received an i.p injection of 1.5 g/kg glucose and blood samples collected at times indicated to measure blood glucose using a glucometer.For ITT mice were fasted for 5 hours then received an i.p injection of 0.5U/kg of insulin and blood samples collected at times indicated to measure blood glucose.

RNA immunoprecipitation (RIP
Immunohistochemistry staining (IHC)-Adipose explants were placed in 4% paraformaldehyde overnight.Using the Histology Service of the Molecular Pathology Shared Resource at the Columbia University Irving Medical Center, explants were paraffinembedded and 5 μm blank slides were sectioned.Slides were stained with hematoxylin and eosin.IHC staining was performed on 5 μm-thick FFPE sections with rabbit anti Mouse CD68 (Abcam ab283654) and rabbit anti-mouse F4/80 (Cell Signaling 70076S) antibodies to detect macrophages.Antigen retrieval was performed in 10 mmol/L citrate buffers (pH 6.0) with a pressure cooker.Quenching endogenous peroxidase activity was done with 3%H 2 O 2 in PBS for 10 minutes at room temperature.Slides were incubated with primary antibody CD68 1:200 or F4/80 1:400 dilutions 2hours at room temperature, followed by anti-rabbit polymer (DAKO #K4003).Detection used HRP catalyzed chromogen "diaminobenzidine".Counterstaining was with hematoxylin followed by dehydration, clearing in xylene, and mounting with Permount.Imaging was with a Leica SCN400.Adipocytes sizes were calculated using Image J and Adiposoft plugin software 89 .The CD68 and F4/80 staining area was quantified using Fiji/Image J 90 .
Nile red staining-Day 14 or Day 21 differentiation ASC hTERT adipocytes were fixed in 2% PFA then stained with 0.5uM Nile Red (Invitrogen N1142) and Hoechst 33342 (Thermo Fisher) for 30mins.Images were acquired with epifluorescence Nikon Ti-S inverted microscope at 20x resolution.Nile Red Area/ Nuclei quantification used triangle or otsu thresholding algorithms in Fiji/ Image J. 90 RNA scope immunofluorescence (IF) assay-ASC hTERT human adipocytes are plated on 24 well dishes containing glass coverslips and differentiated according to the protocol above.After 14 days differentiation, cells are dehydrated with 50, 70 and 100% ethanol incubations (5mins), then rehydrated with 100, 70 and 50% ethanol incubations (5mins).Cells were permeabilized for 5mins using pre-treatment #3 from ACD Bio RNAscope Multiplex Fluorescent Reagent Kit v2 (ACD Bio #323100) then processed with manufacturers RNA scope protocol.ACD Bio designed custom RNA scope probes for linc-ADAIN.Using ACD Bio protocol for RNA Scope and combined IF, antibodies for IGF2BP2 (Thermo Fisher #712137) and HuR (Millipore 07-468) with secondary antibodies Opal 570 and Opal 520 (Akoya Biosciences, FP1488001KT, FP1487001KT) were used to probe adipocytes.Slides were imaged using a scanning confocal: Nikon A1 with GaAsP spectral detector and Nikon Ti Eclipse inverted microscope for Z-stack images, at 20x and 40x objective magnifications.

Ingenuity pathway (IPA) and database for annotation visualization and
Integrate discovery (DAVID) gene ontology analysis-The list of differentially expressed genes (DEGs) from RNA-sequencing was uploaded to Qiagen's IPA system for core analysis and overlaid with the global molecular network in the IPA knowledge base (https://digitalinsights.qiagen.com/citation-guidelines/).Molecular and Cellular Functions were explored to interpret the effects of linc-ADAIN KD compared to scramble control on adipocyte functions.DEGs were also uploaded to DAVID database to search for top enriched gene ontology (GO) biological pathways 91 .

QUANTIFICATION AND STATISTICAL ANALYSIS
All statistics are outlined in each figure legend.In brief, Mann Whitney U non-parametric test was applied to analysis with two groups, One-Way ANOVA was applied to three or more groups, and a Two-Way ANOVA was applied when there were two or more quantitative variables with two or more categorical variables.Pearson correlations were used to assess the correlation of human linc-ADAIN expression to other variables in the GENE study.GraphPad Prism 9 (GraphPad Software Inc, San Diego, CA) was used for statistical analysis.Data in graphs are presented as mean± standard error of the mean.

Figure 2 .
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).

Figure 3 .
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) 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.

isolation and quantitative real-time PCR analysis-Total
RNA was extracted using Trizol and Direct-zol RNA miniprep kit (Zymo Research R2052).At least 100ng of RNA was reverse transcribed using High-Capacity cDNA kit (Applied Biosystems).Expression of lincRNAs and protein-coding genes was assessed by quantitative realtime PCR using an Applied Biosystems PowerUp SYBR Green system.Custom DNA oligos targeting linc-ADAIN were designed using Genescript PCR primer design tool (genescript.com)and obtained from Integrated DNA technologies.Pre-designed PrimeTime qPCR primer assays were used and obtained from Integrated DNA technologies for all other genes.GAPDH, HPRT and 18S were used as housekeeping genes.Adipose