Progress and prospects of long noncoding RNAs in lipid homeostasis.

BACKGROUND: Long noncoding RNAs (lncRNAs) are a novel group of universally present, non-coding RNAs (>200 nt) that are increasingly recognized as key regulators of many physiological and pathological processes. SCOPE OF REVIEW: Recent publications have shown that lncRNAs influence lipid homeostasis by controlling lipid metabolism in the liver and by regulating adipogenesis. lncRNAs control lipid metabolism-related gene expression by either base-pairing with RNA and DNA or by binding to proteins. MAJOR
CONCLUSIONS: The recent advances and future prospects in understanding the roles of lncRNAs in lipid homeostasis are discussed.

Introduction

Obesity and type 2 diabetes (T2D) are epidemic health problems that affect millions of people worldwide. Fat accumulation is determined by the balance between anabolic (adipogenesis and lipogenesis) and catabolic (lipolysis, fatty acid β-oxidation, and thermogenesis) processes. Adipogenesis is the process by which preadipocytes develop into mature white or brown adipocytes, contributing to energy balance [1]. Lipogenesis is the process of fatty acid synthesis and subsequent triglyceride synthesis in both white adipose tissue (WAT) and liver. Lipogenesis is triggered by circulating insulin and ingestion of nutrients [2]. Conversely, other hormones and exercise induce lipolysis, the breakdown of triglycerides into glycerol and free fatty acids, in both WAT and muscle [3], [4]. Free fatty acids released into the circulation during lipolysis are subsequently taken up by liver, muscle, and brown adipose tissue (BAT) as an energy source for β-oxidation [4]. Lastly, adaptive thermogenesis in BAT is a catabolic process in which oxidative phosphorylation and fatty acid β-oxidation are uncoupled to generate heat [5], [6]. All of these processes are initiated and regulated by hormones, nutrients, and/or environmental stress, transduced by signal pathways, and controlled by transcription factors [2], [4], [6]. The balance of these processes is critical for maintaining normal adiposity and regulating systemic lipid metabolism. All too often, however, dysregulation of these processes leads to increased adiposity, dyslipidemia, and metabolic perturbations that can accelerate diabetes, cardiovascular diseases, and nonalcoholic fatty liver disease (NAFLD).

Emerging studies now suggest that non-coding RNAs are also key regulators of lipid homeostasis. More than 90% of the human genome is likely to be transcribed; yet less than 2% of the genome encodes approximately 20,000 proteins (International Human Genome Sequencing Consortium 2004). This leaves the balance of the human genome (∼98%) to be transcribed into thousands of non-coding RNAs (ncRNAs). ncRNAs are classified into two main subgroups: short ncRNAs (<200 nt) and long ncRNAs (>200 nt). Short ncRNAs include microRNAs (miRNAs), which regulate many biological processes by inducing mRNA degradation via the RNA interference pathway. miRNAs are well studied and have been implicated in human diseases including cancer [7], cardiovascular disease [8], diabetes [9], and neurodegenerative disorders [10]. Furthermore, miRNAs regulate lipid metabolism and adipogenesis [11], [12]. By contrast, long non-coding RNAs (lncRNAs) are less well-studied. The most recent release from GenCode (version 22) has annotated ∼15,900 lncRNA genes in humans. lncRNAs are categorized based on genome location into intergenic, intronic, antisense and enhancer lncRNAs [13]. Multiple studies have shown that many lncRNAs are regulated during development, exhibit cell type-specific expression patterns, localize to specific subcellular compartments, and are associated with human diseases such as cancer [14] and diabetes [15], [16], [17], [18], [19], [20]. As summarized in Table 1, there is now accumulating evidence that lncRNAs are important regulators of lipid metabolism and adipogenesis [21], [22], [23], [24], [25], [26]. Recent studies that implicate lncRNAs in the regulation of lipid metabolism in the liver and adipogenesis are reviewed here.

Table 1

Functional lncRNAs involved in lipid metabolism and adipogenesis.

Name	Tissue/cell type	Loss-of-function phenotype	Gain-of-function phenotype	Assays	References
lncLSTR	Liver	Reduce plasma TG level		Knockdown	[23]
HULC	Hepatoma cell		Increase triglyceride and cholesterol levels	Overexpression	[30]
APOA1-AS	Liver	Increase APOA1 expression both in vitro and in vivo		Knockdown	[31]
DYNLRB2-2	Macrophage		Decrease cellular cholesterol level, increase APOA1 mediated cholesterol efflux	Overexpression	[34]
SRA	Fat	Reduce adipose size and liver TG level		Knockout	[37], [38], [39]
ADINR	3T3-L1 cell	Reduce adipogenesis		Knockdown	[41]
NEAT1	3T3-L1 cell		Increase adipogenesis	Overexpression	[42], [43]
HOTAIR	Gluteal adipose tissue		Increase adipogenesis	Overexpression	[44]
lnc-BATE1	Brown adipose tissue	Reduce BAT activation		knockdown	[35]
Blnc1	Brown adipose tissue/epididymal white adipose tissue	Reduce brown adipogenesis	Stimulate brown adipogenesis	Knockdown/overexpression	[26]

Tissue specific lncRNAs

Unlike mRNAs, lncRNAs are poorly conserved. lncRNAs are expressed in a species-, cell-, tissue-, and developmental stage-specific manner. There are approximately 11,000 primate-specific lncRNAs but only about 425 highly conserved lncRNAs. Conserved lncRNAs appear to predominantly regulate embryonic development [27]. Notably, each tissue generates unique lncRNAs during development [27]. For example, in mice, about 1109 polyadenylated lncRNAs are expressed in erythroblasts, megakaryocytes, and megakaryocyte-erythroid precursors, whereas 594 lncRNAs are expressed in human erythroblasts [28]. Collectively, 53.6% of lncRNAs in megakaryocyte-erythroid precursors are conserved between mouse and human, whereas only 15% of mouse erythroid lncRNAs are expressed in human erythroblasts [28], indicating that the conservation of lncRNAs is highly dependent on both species and development stage.

Each tissue has its own catalog of specific lncRNAs, which may contribute to the unique function of each tissue. Liver, skeletal muscle, and adipose tissue are three major metabolic tissues controlling lipid metabolism. An analysis of a dataset of multi-tissue gene expression profiles identified 30 lncRNAs that are enriched in liver, muscle, or adipose tissues [23]. These lncRNAs may regulate lipid metabolism in these tissues.

lncRNAs regulate lipid metabolism in the liver

Analysis of liver-enriched lncRNAs identifies lncLSTR as a putative regulator of plasma triglyceride (TG) levels [23]. Liver-specific knockdown of lncLSTR increases ApoC2 expression and LPL activities, enhances plasma TG clearance, and ultimately results in decreased plasma TG [23]. However, knockdown of lncLSTR in primary hepatocytes fails to increase ApoC2 expression, suggesting that another mediator exists in the liver cells. Liver-specific knockdown of lncLSTR decreases the expression of Cyp8b1, increases the ratio of muricholic acid (MA) and cholic acid (CA) in bile acid, and enhances FXR activity, leading to increased ApoC2 expression [23]. Mechanistically, lncLSTR is shown to directly bind to TDP43 and inhibit Cyp8b1 expression [23]. Another lncRNA, HULC, which is abnormally overexpressed in hepatocellular carcinoma (HCC) [29], has been shown to increase triglyceride and cholesterol levels by activating PPARα and ACSL1 in hepatoma cells [30]. Together, these studies demonstrate that liver-enriched lncRNAs regulate lipid metabolism in the liver.

Another potential group of lncRNAs that regulates lipid metabolism is natural antisense transcripts (NATs). About 70% of lncRNAs are NATs, which regulate sense gene expression in a positive or negative manner [31]. APOA1-AS has been shown to negatively regulate APOA1 expression both in vitro and in vivo [31]. Moreover, APOA1-AS regulates different histone methylation patterns that activate or suppress gene expression. Knockdown of APOA1-AS increases the level of H3K4-me3, decreases the level of H3K27-me3, but does not alter the level of H3K9-me3 at APO gene cluster, leading to increased expression of APOA1, APOA4 and APOC3 [31]. APOA1 is a major component of high-density lipoprotein (HDL), protecting against cardiovascular disease [32], making APOA1-AS a potential therapeutic target for treating cardiovascular disease.

The expression of many lncRNAs can be induced by hormones [33], ligands [26], or lipoprotein [34], and these lncRNAs could regulate lipid metabolism. For example, oxidized LDL (Ox-LDL) significantly induces long intervening noncoding RNA (lincRNA)-DYNLRB2-2 expression, resulting in the upregulation of GPR119 and ABCA1 expression through the glucagon-like peptide 1 receptor signaling pathway in THP-1 macrophage-derived foam cells [34]. As a negative feedback, GPR119 significantly decreases cellular cholesterol content and increases APOA1-mediated cholesterol efflux in the liver, reducing atherosclerosis in APOE knockout mice [34]. This study shows that inducible lncRNAs regulate lipid metabolism.

lncRNAs regulate adipogenesis

Adipocytes, including white, brown, and beige, play important roles in lipid storage or clearance. White adipocytes are the major constituent of white adipose tissue (WAT), controlling the storage of triacylglycerol [1]. Brown adipose tissue (BAT) and beige fat are responsible for thermogenesis. BAT is an important tissue controlling plasma lipid clearance in response to cold stimulation [5]. Both white and brown adipogenesis are tightly controlled by signal pathways, transcription factors, miRNA, and lncRNAs [1], [11], [26], [35], [36].

The first evidence of a potential role for lncRNAs in adipogenesis was reported by Xu et al [37]. The non-coding RNA, steroid receptor RNA activator (SRA), promotes adipogenesis in vitro through regulation of PPARγ and P38/JNK phosphorylation [37], [38]. Genetic deletion of SRA protects high fat diet induced obesity and fatty liver disease, reduces the size of adipocytes, and improves glucose tolerance [39].

To identify adipogenesis associated lncRNAs, transcriptomic analyses of primary brown and white adipocytes, preadipocytes, and cultured adipocytes have been performed [24], [26], [35], [40]. A total of 175 lncRNAs were found to be significantly up- or down-regulated, by more than two-fold, during differentiation of both brown and white adipocytes [40]. Many lncRNAs are adipose-enriched and strongly induced during adipogenesis. Key transcription factors such as PPARγ and C/EBPα not only control mRNA expression related to adipogenesis but also regulate lncRNA expression during adipogenesis [40]. PPARγ is physically bound within the promoter region of 23 (13%) of the 175 lncRNAs while C/EBPα is bound upstream of 34 up-regulated lncRNAs (19%) during adipogenesis [40].

Some lncRNAs have been shown to regulate adipogenesis by loss- and gain-of-function assays [40], [41]. lncRNA ADINR specifically binds to PA1 and recruits MLL3/4 histone methyl-transferase complexes to increase H3K4me3 and decrease H3K27me3 histone modification in the C/EBPα locus, leading to transcriptional activation of C/EBPα and increased adipogenesis [41]. lncRNA NEAT1 regulates PPARγ2 splicing during adipogenesis [42]. It also mediates miR-140 induced adipogenesis [43]. In human, lncRNA HOTAIR is highly expressed in gluteal but not in abdominal adipose tissue [44]. Ectopic expression of HOTAIR in abdominal preadipocytes increases cell differentiation by inducing expression of PPARγ and LPL [44]. These data demonstrate that lncRNAs regulate white adipogenesis through lncRNA-protein and lncRNA-miRNA interactions.

To identify lncRNAs associated with adipogenesis in brown fat, RNA-seq and microarray have been performed by several independent groups [24], [26], [45], [46]. One study shows that 127 lncRNAs are BAT-specific, induced during brown adipose differentiation, and targeted by key transcription factors such as PPARγ, C/EBPα, and C/EBPβ [35]. One of these lncRNAs, lnc-BATE1, is essential to maintain BAT identity and thermogenic capacity. Knockdown of lnc-BATE1 impairs the activation of BAT. lnc-BATE1 binds to heterogeneous nuclear ribonucleoprotein U, and both are required for brown adipogenesis [35]. Another independent study revealed a cluster of 21 lncRNAs, the expression of which is enriched in BAT and induced during brown adipocyte differentiation [26]. Among these lncRNAs, Blnc1 was shown to be a key regulator of brown adipogenesis [26]. Knockdown of Blnc1 inhibits brown adipogenesis, while overexpression of Blnc1 enhances lipid accumulation in differentiated brown adipocytes [26]. Mechanistically, Blnc1 forms a ribonucleoprotein complex with transcription factor EBF2 to stimulate the thermogenic gene program, driving brown adipogenesis [26]. These studies show that lncRNAs induced during adipogenesis regulate adipogenesis and thermogenesis through lncRNA-protein interactions.

Future perspectives

Many lncRNAs associated with lipid metabolism and adipogenesis have been identified through RNA-seq and bioinformatics analyses. The roles of these lncRNAs are beginning to be classified. However, we are still at the tip of the iceberg regarding our understanding of lncRNAs in the regulation of lipid metabolism, and numerous future potential research directions should be taken if we are to fully understand how lncRNAs regulate lipid metabolism at the cellular and organismal levels (Figure 1).

Figure 1

Perspectives of lncRNAs in the regulation of lipid metabolism. Fat accumulation is determined by the balance between anabolic (adipogenesis and lipogenesis) and catabolic (lipolysis, fatty acid β-oxidation, and thermogenesis) processes that are mainly controlled by WAT, BAT, liver and muscle. These processes are initiated and regulated by hormones, nutrients, and/or environmental stress, transduced by signal pathways, controlled by transcription factors, and exerted by multiple enzymes, UCP1, and lipid droplet associated proteins. lncRNAs may be involved in all these processes. It is very important to identify lncRNAs that are induced by hormones (e.g. Insulin, Glucagon, and Epinephrine) (1), thyroid and steroid hormones or nutrients/transcription factors (2, 3, 4), and those that are associated with lipid droplets (5) and obesity/type 2 diabetes (6), and plasma lncRNAs (7). These lncRNAs may regulate lipid metabolism through well-known or other potential mechanisms, including lncRNA-protein, lncRNA-DNA, antisense-sense (8), and lncRNA-miRNA-mRNA interactions (9), and expression of small peptides (10).

First, identifying the hormonal regulation of lncRNAs during fat accumulation is very important. Lipogenesis, lipolysis, β-oxidation, adipogenesis, and thermogenesis are highly regulated processes. It is likely that many of the same hormones and signaling pathways that regulate lipid metabolism (e.g. insulin [2], [47], glucocorticoids [47], GLP-1 [48], thyroid hormone [49], adiponectin [50], and leptin [51], [52]) influence the expression and activity of lncRNAs. Similarly, nutritional inputs (glucose and fatty acids) may also regulate key lncRNAs to control lipogenesis. One could envision a scenario in which lncRNAs might be induced by key transcription factors that control metabolism (e.g. SREBP1 [53], ChREBP [54], [55], LXRs [56], [57], FXR [56], [57], PPARγ [58], [59], and PPARα [60], [61]). Subsequently, subsets of lncRNAs could either mediate lipogenesis or β-oxidation by fine-tuning gene expression programs or by providing feedback to regulate the activity of metabolic transcription factors.

Second, given that lipid droplet formation in adipocytes and at ectopic sites (muscle and liver) plays an important role in physiological and pathological conditions of obesity [62], [63], [64], it is important to identify lncRNAs that regulate the formation of lipid droplets. Regulation of this process by lncRNAs may be achieved by several means. For example, previous studies have indicated that lncRNAs might express small peptides [65], [66], which could alter lipid droplet formation. miRNAs also are important to lipid metabolism [11], [12], and many publications show that lncRNAs can reduce miRNA levels by acting either as sponges or through base-pairing with primary miRNAs to block their processing into mature miRNAs. Therefore, lncRNAs and miRNAs may interact to regulate or dysregulate genes that encode key lipid-binding proteins and/or other regulators of lipid droplet formation. Along these same lines, it will be important to identify more natural antisense transcripts (NATs) associated with lipid metabolism. About 70% of lncRNAs are NATs that regulate gene expression in a positive or negative manner. NATs are good therapeutic targets for human diseases, including immune diseases [31], [67] and cardiovascular diseases [31]. However, only one NAT, APOA1-AS, has been identified as a negative regulator of APOA1 expression both in vitro and in vivo [31].

Third, it will be very important to demonstrate that aberrant lncRNA expression is linked to the development of lipid accumulation that accompanies obesity, diabetes, and nonalcoholic fatty liver disease (NAFLD) in human populations. Clinically, circulating lncRNAs serve as biomarkers for cancer [68], [69], heart failure [70], and kidney injury [71]. So it is possible to monitor biologically relevant lncRNAs non-invasively. However, using lncRNAs as biomarkers for obesity is not as critical. Recent data show that miRNA and other noncoding RNA in plasma exosomes regulate the immune response [72], [73]. It would be very interesting to test whether plasma lncRNAs regulate lipid metabolism in liver, skeletal muscle, and adipose tissue. Moreover, it would be interesting to determine whether lncRNAs that regulate lipogenesis, lipolysis, β-oxidation, adipogenesis, and thermogenesis in cells can serve as meaningful biomarkers to monitor the efficacy of anti-obesity treatments and/or therapies that target dyslipidemias.

Finally, experimental tools are essential for progress in any of the future directions. As shown in Figure 2, lncRNAs associated with lipid metabolism could be further identified through RNA-seq and bioinformatics analysis. For functional assays of these identified lncRNAs, both loss- and gain-of-function assays can be performed in cultured cells [23], [26], [31], mice [23], and monkeys [31]. Traditional RNAi/antisense oligonucleotides [26], [31], knockout [74], and newly developed CRISPR/Cas9 gene-editing tools [75] can be used to suppress lncRNA expression, while adenovirus, lentivirus [26], [44], and transgene [76] mediated overexpression of lncRNAs can be used in gain-of-function assays. For mechanistic studies of lncRNAs, it is very important to identify the interacting proteins or nuclei acids (RNA or DNA) by lncRNA-protein (e.g. crosslinking immunoprecipitation, CLIP [77]), lncRNA-RNA (crosslinking analysis of synthetic hybrids, CLASH [78]) or lncRNA-DNA (capture hybridization analysis of RNA targets, CHART [79]; and chromatin isolation by RNA purification, ChIRP [78]) interaction assays. Moreover, this field will be greatly advanced by the development of additional technologies that effectively determine the function and mechanisms of lncRNAs in the regulation of lipid metabolism under normal and obese conditions.

Figure 2

Workflow for identification, functional assay and mechanistic study of lncRNAs in the regulation of lipid metabolism.

Funding

This study was supported by Changbai Mountain Scholars Program of The People's Government of Jilin Province 2013046 (to Z. C.), the National Natural Science Foundation of China Grant 31500957 (to Z. C.), the Jilin Talent Development Foundation 111860000 (to Z. C.), and the startup funds from Northeast Normal University Grant 120401204 (to Z. C.).