Adipogenic miRNA and meta-signature miRNAs involved in human adipocyte differentiation and obesity

MicroRNAs (miRNAs) have been identified as a new class of regulatory molecules that influence many biological functions, including metabolism, adipocyte differentiation. To determine the role of adipogenic miRNAs in the adipocyte differentiation process, we used microarray technology to monitor miRNA levels in human adipose-derived mesenchymal stem cells (hMSCs-Ad), human stromal vascular cells (SVCs) and differentiated adipocytes. 79 miRNAs were found to be differentially expressed, most of which are located in obesity related chromosomal regions but have not been previously linked to adipocyte differentiation process. A systematic search was made for relevant studies in academic data bases, involving the Gene Expression Omnibus (GEO) ArrayExpress, Pubmed and Embase database. Eight studies on human adipocyte differentiation or obesity were included in the final analysis. After combining our microarray data with meta-analysis of published microarray data, we detected 42 differently expressed miRNAs (meta-signature miRNAs) in mature adipocytes compared to SVCs or hMSCs-Ad. Our study shows meta-signature miRNAs specific for adipogenesis, several of which are correlated with key gene targets demonstrating functional relationships to pathways in BMP signaling pathway, Cell differentiation, Wnt signaling, insulin receptor signaling pathway, MAPK signaling, Cell cycle and lipid metabolic process. Our study shows that the first evidence of hsa-let-7 family, hsa-miR-15a-5p, hsa-miR-27a-3p, hsa-miR-106b-5p, hsa-miR-148a-3p and hsa-miR-26b-5p got a great weight in adipogenesis. We concluded that meta-signature miRNAs involved in adipocyte differentiation and provided pathophysiological roles and novel insight into obesity and its related metabolic diseases.


INTRODUCTION
Obesity has become a major worldwide health problem associated with metabolic and cardiovascular disorders. Obesity is a progressive and multifactorial pathology characterized by excessive expansion of white adipose tissue, which due to an increase in adipocyte hyperplasia and hypertrophy [1]. Since obesity-related hyperplasia and hypertrophy are often associated with various metabolic disorders, proper understanding of the molecular events regulating adipogenesis and lipid metabolism can provide valuable information for exploiting comprehensive and effective therapeutic strategies against obesity.
MicroRNAs (miRNAs), small non-coding RNAs of 18-25 nucleotides in length, which were seen to control gene expression in virtually obesity or other metabolism syndrome, were abundantly investigated relating to proliferation and differentiation of adipogenesis. Indeed, the role of some miRNAs has been described in obesity. For example, miR-103, miR-107, and miR-143 regulate liver insulin sensitivity in diet-induced obesity through different mechanisms [2,3]. The involvement of miRNAs in obesity pathogenesis is well established, as they can Review Oncotarget 40831 www.impactjournals.com/oncotarget behave as pro-adipogenic or adipogenic suppressor genes depending on the cellular function of their targets [4].
Emerging evidence shows that miRNAs may directly or indirectly modulate adipocyte differentiation. In our previous study, we found that miR-148a, as a biomarker of obesity, promoted adipogenesis by inhibiting Wnt1 [5]. miR-146b as a positive regulator of accelerated adipocyte differentiation through modulation of KLF7 [6]. Recent findings [7] have suggested that miR-26b inhibits adipogenic differentiation and overexpression of miR-1908 inhibited adipogenic differentiation [8]. However, to date, there is no meta-analysis to interpret the biology of miRNA and its contribution to obesity development which may provide early promising therapeutic targets and effective control of obesity.
In the present study, using a miRNA array strategy, we identified 79 differentially expressed miRNAs, most of which located in obesity related chromosomal regions but have not been previously linked to adipocyte differentiation process. Additionally, to overcome the limitations in current researches, we performed metaanalysis applying the robust rank aggregation method [9], followed by pathway analysis, to identify miRNA regulation in adipogenesis and the pathways that key miRNAs may impact. Identification of miRNA metasignature and involved pathways would provide the potential target and novel insight into obesity and its related metabolic diseases.

miRNA microarray results
miRNA microarray analysis was performed to identify differentially expressed miRNAs in mature adipocytes and then compared to SVCs and hMSCs-Ad. A larger number of microRNAs (79 miRNAs, 45 up-regulated miRNAs and 34 down-regulated miRNAs) were significantly up-or down-regulated in mature adipocytes compared to SVCs and hMSCs-Ad (Figure1, Supplementary Table 1).Our data showed that differentially expressed miRNAs were overlapping either in mature adipocytes versus SVCs or in mature adipocytes versus hMSCs-Ad (Supplementary Table 1, P< 0.05). Between mature adipocytes and SVCs/hMSCs-Ad, miR-146b-5p and miR-335 were found to be differentially expressed with a fold change >10 (Supplementary Table  1). Whereas, miR-424 was differentially expressed with a fold change of more than 10 in mature adipocytes when compared to hMSCs-Ad, and the expression level was about 9.2-fold when compared to SVCs. Of all differentially expressed miRNAs, miR-1275, miR-155 and miR-1268 were down-regulated in mature adipocytes over 10-fold when compared with SVCs and hMSCs-Ad. We identified that miR-148a, miR-26b, miR-132, miR-365 and miR-1908 were highly expressed in mature adipocytes with over 5-fold compared to SVCs/hMSCs-Ad. Additionally, miR-93 and miR-720 were lowly expressed with 5-fold in mature adipocytes compared to SVCs/hMSCs-Ad. Among these differentially expressed miRNAs, our group did further research for the adipogenic miRNAs in vivo and in vitro, including miR-148a [5], miR-26b [10], miR-146b [6] and miR-1275.

PCR validation of significant differentially expressed miRNAs expression
To confirm the microarray hybridization results, qRT-PCR was performed on 8 up-regulated miRNAs (miR-335, miR-146b-5p, miR-26b, miR-30b, miR-21, miR-378, miR-143 and miR-148a) [10] and 3 downregulated miRNA (miR-155,miR-221, and miR-1275), chosen on the basis of their levels of expression on the microarray and their biological significance. qRT-PCR was performed in triplicate on the diluted cDNA and the experiments were conducted twice on differentiated hMSCs-Ad/mature adipocytes and hMSCs-Ad/SVCs. The qRT-PCR results were closely mirrored with the microarray analysis showing P-value <0.05 ( Figure 2).

The location of differentially expressed miRNAs in obesity gene map
To elucidate the correlation pattern between miRNA and obesity, the chromosomal regions of differentially expressed miRNAs were performed using Genomic map view analysis. We compared the chromosomal regions of differentially expressed miRNAs with obesity gene map published in 2005 [11]. As shown in Table 1, there were about 65% (51/79) miRNAs located in obesity related chromosomal regions. Our results indicated that these differentially expressed miRNAs have a close relationship with obesity.

Study selection and data extraction
We retrieved publications by the term of "adipocyte OR fat cell OR adipogenesis OR MSC (mesenchymal stem cells) OR obesity OR adipocyte differentiation" AND "miRNA OR microRNA" AND human". Database searches initially yielded a total of 526 publications and 8 studies met the inclusion criteria ( Figure 3, Table 2).
Our study found 79 miRNAs (45 up-regulated miRNAs and 34 down-regulated miRNAs) were significantly up-or down-regulated in mature adipocytes compared to SVCs and hMSCs-Ad (Figure1, Supplementary Table 1). We detected 42 meta-signature miRNAs from these miRNA array data, which were differently expressed miRNAs in mature adipocytes compared to SVCs or MSC (Table 3). According to the sources of cells, we detected the meta-signature miRNAs from human adipocytes and MSC, respectively. We also identified 17 meta-signature differently expressed miRNAs in obese and lean subjects (Table 4).

GO and pathway analysis
To determine which GO and pathways of metasignature miRNAs are involved in adipogenesis, the predicted gene targets were compared against pathways in the Go and Kyoto Encyclopedia of Genes and Genomes (KEGG) database. The predicted targets were found to be significantly over represented in cell differentiation, cell cycle, cell proliferation and Wnt signaling pathway of GO terms (Figure4, Supplementary Table 3). The KEGG enrichment pathway analysis of our microarray data revealed that almost all of the meta-signature miRNAs were particularly involved in Wnt signaling, MAPK signaling, mTOR signaling and ECM-receptor interaction. Several genes in the Wnt signaling set (WNT4; LRP6; DAAM2; FRAT2;SKP1; PRKX; FZD3; FZD9; NFATC1; NFAT5; CHP; PPP3CA) were also targeted by multiple miRNAs as part of a complex regulatory network ( Figure  5, Supplementary Table 4).

DISCUSSION
In this study, we reported the differential miRNA expression in mature adipocyte compared with SVCs or hMSCs-Ad and most of differentially expressed miRNAs located in obesity related chromosomal regions. The expression pattern of differential miRNAs between SVCs Vs. mature adipocyte and hMSCs-Ad Vs. mature adipocyte are similar, indicating that these differential miRNAs were indeed involved in adipogenesis. Our results are in line with the study from Alajez et.al. [16], describing 36 up-regulated miRNAs and 2 down-regulated miRNAs in mature adipocytes compared to hMSCs-TERT.
miR-146b-5p, miR-335, miR-424, miR-1275, miR-155, miR-1268, miR-148a, miR-26b, miR-132, miR-365, miR-1908, miR-93 and miR-720 were identified more than 3-fold changes in mature adipocyte compared with SVCs or hMSCs-Ad. More importantly, the highest fold change was observed in miR-146b in mature adipocyte. Ahn et al. reported that miR-146b identified as a positive  [20]. Interestingly, our previous study reported that miR-146b is highly expressed in the adipose tissue of obese mouse models such as DIO, diet-induced obesity, ob/ob, and db/db. This finding has implicated miR-146b as a positive regulator of accelerated adipocyte differentiation through modulation of SIRT1 and KLF7 [6]. Indeed, our previous study found that miR-148a promotes hMSCs-Ad to differentiate to mature adipocyte by targeting Wnt1 [5]. Additionally, our group [7] has also suggested that inhibited miR-26b inhibits adipogenic differentiation in human preadipocytes, and miR-335, as a potential adipogenic miRNA, is involved in adipose tissue inflammation and is highly expressed in mature 3T3-L1 [21]. Therefore, our present and previous studies bring in valuable information with respect to human obesity pathology because we have demonstrated that miR-148a, miR-146b, miR-26b and miR-335 are dysregulated in the process of adipocyte differentiation.
Since obesity-related hyperplasia and hypertrophy Oncotarget 40840 www.impactjournals.com/oncotarget are often associated with various metabolic disorders, the detail understandings of the molecular events regulating adipogenesis and lipid metabolism are important. Ortega et al. [18] isolated fat cells from both lean (BMI<25.0Kg/ m 2 , n=3) and obese (BMI>30.0Kg/m 2 , n=3) subjects for miRNA array and then differentiated to mature adipocyte, and identified 5 up-regulated miRNAs and 7 downregulated miRNAs between lean and obese subjects. In addition, 65% (51/79) differentially expressed miRNAs from our data located in obesity related chromosomal regions which was published in 2005 (11), indicating that these differentially expressed miRNAs have a close relationship with obesity.
Although, miRNAs play an important role in the progression of obesity and adipogenesis, different microarray techniques, technical approaches, tissue or cellular models, and small sample size led to inconsistent findings in previous studies. Thus, the present observation reported that 42 differentially expressed miRNAs (metasignature miRNAs) in mature adipocytes compared to SVCs or hMSCs-Ad. The number of meta-signature miRNAs studies reported varied greatly but at least 12 deregulated miRNAs were reported in each study. Ortega et al. [18] identified that miR-484 and miR-130b were lowly expressed in human subcutaneous adipose from obese and type 2 diabetes mellitus (DM), miR-199a-5p, miR-221, miR-125b and miR-1229 were lowly expressed, indicating that these miRNAs play a key role in obesity. However, this study identified miR-143 was down regulated in mature adipocyte, which conflicted with other Oncotarget 40841 www.impactjournals.com/oncotarget studies [22]. Furthermore, this microarray data was not verified by qRT-PCR, thus, this result may be influenced by experimental error, individual difference and species variation.
The KEGG enrichment pathway analysis of our microarray data revealed that almost all of the metasignature miRNAs were particularly involved in Wnt signaling, MAPK signaling, mTOR signaling and ECMreceptor interaction. Wnt signaling was first recognized as a possible negative regulator of adipogenesis when Wnt1 expression decreased significantly during adipocyte differentiation progress [23]. In particular, our study found that miR-148a through regulating Wnt1, but not Wnt10b, promotes hMSCs-Ad differentiation [5]. ERK, p38 and JNK signaling pathway belong to MAPK signaling, which are intracellular signaling pathways that play a pivotal role in many essential cellular processes such as proliferation and adipocyte differentiation [24]. Importantly, Esau et al. [25] reported that miR-143 promoted adipocyte differentiation by inhibiting ERK5. mTOR signaling belongs to the phosphoinositide 3-kinase (PI3K)-related kinase family, and promotes adipogenesis in white adipocytes [26]. Previous study reported that miR-17-92 cluster has been reported to be upregulated during the clonal expansion stage of adipocyte differentiation, posi tively regulating adipogenesis by targeting the tumor suppressor RB2/p130 [27] and knockdown of miR-106b and miR-93 significantly induced the expression of brown fat-specific genes and promoted the accumulation of lipid-droplet in differentiating brown adipocytes. In addition, ectopic expression of miR-106b and miR-93 suppressed the mRNA level of Ucp1 [28]. miR-93 is a highly expressed PCOS patients' adipocyte tissue and downregulates GLUT4 gene expression, suggesting that it plays a role in the IR of PCOS [15]. Interestingly, let-7-mediated repression of adipogenesis was also found to promote osteogenesis [29]. In the present study, we found that hsa-miR-15a-5p, hsa-miR-106b-5p, hsa-miR-181a-5p, hsa-let-7 family, hsa-miR-27a-3p, hsa-miR-130b-3p, hsa-miR-152/148a-3p and hsa-miR-26b-5p got high degree means, which indicated that these miRNAs had a great weight in adipogenesis than others. We concluded that meta-signature miRNAs involved in adipocyte differentiation and provided pathophysiological roles and novel insight into obesity and its related metabolic diseases.

miRNA microarray
RNA was labeled and hybridized to μParaflo® human miRNA microarrays following the manufacturer's instructions by LC Sciences (MRA-1001, miRHuman_14). Each chip contains 894 specific microRNA probes (repeated four times for each probe) and 50 controlled probes (according to the purpose of experiment, each probe is repeated 4-16 times). miRNA microarray analysis was performed using RNA samples from human preadipocyte, human adipose tissue-derived mesenchymal stem cells (hMSCs-Ad) and mature adipocyte.

RNA isolation and analysis of miRNAs expression by RT-quantitative PCR (qPCR)
Total RNA was prepared from SVCs or hMSCs-Ad at varying intervals after induction of adipocyte differentiation using TRIzol (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol, followed by DNase treatment (TaKaRa, Japan). The quality and concentration of RNA was assessed by Nanodrop2.0 (Thermo Fisher Scientific, Waltham, MA). 240 ng of total RNA was reverse transcribed using the high capacity reverse transcription kit (Applied Biosystems, Life Technologies) according to the manufacturer's protocol. One microliter of cDNA was then amplified by realtime qPCR using the Taqman universal PCR master mix (Applied Biosystems) and specific primers in an Applied 7500 QPCR System (Applied Biosystems). miRNA expression levels were normalized using the snoU6 miRNA expression level as internal controls. Each sample was measured in triplicate, and the gene expression levels were calculated using the 2 -ΔΔct method.

Study selection and data extraction
A systematic literature search was performed for the identification of adipogenic or obesity miRNA expression profiling studies using a two-level search strategy. First, we undertook a web-based search in Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) using search term ("adipocyte OR adipogenesis OR fat cell OR adipocyte differentiation OR obesity"[Mesh]) AND ("MicroRNAs"[Mesh]) AND "Humans" [Mesh]. To perform a comprehensive retrieval, searching in ArrayExpress (www.ebi.ac.uk/arrayexpress), Pubmed and Embase database were also performed. Second, the reference lists of all relevant and existing studies were reviewed through a manual search for further identification of potential relevant studies.
Abstracts were screened carefully and full texts of relevant potential abstracts were evaluated. Studies with original experimental design that analyzed the miRNA expression profiling in human between adipose tissues or human adipocytes were included. Meanwhile, studies were not eligible for meta-analysis if they met the following selection criteria: 1) using human preadipocytes or SVCs or hMSCs-Ad or hMSC cell, 2) preselected candidate genes research, 3) differentiating to mature adipocyte.
The lists of statistically significant expressed miRNAs were extracted from publications. Authors were contacted when the lists could not be obtained. All miRNA names were standardized through miRBase version 20.0. miRNAs that cannot be related to either -3p or -5p in miRBase were designated with hsa-miR-*, such as hsa-miR-210. Those identified as dead entry in miRBase retained their names mentioned in the literatures.

Integrative identification of miRNA targets
The meta-signature miRNAs were selected for target prediction by using TargetScan database version 6.2 [30] and miRanda (http://www.microrna.org) were used to predict the target consensus sequences for meta-miRNA.

Enrichment analysis
To identify the pathways of predicted miRNA targets, Kyoto Encyclopedia of Genes and Genomes (KEGG), Panther pathways and Gene Ontology terms were carried out with GeneCodis web tool (http:// genecodis.dacya.ucm.es/) [31]. Use the KEGG database to build the network of genes in accordance with the relationship among the genes, proteins and compounds in the database. GO analysis was applied to analyze the main function of the target genes of differential expression miRNA according to the Gene Ontology which is the key functional classification of NCBI [32]. Enrichment provides a measure of the significance of the function: as the enrichment increases, the corresponding function is more specific, which helps us to find those GOs with more concrete function description in adipogenesis.

miRNA-GO network
MiRNA-GO network is built according to the relationship of miRNAs and significant GOs. In the miRNA-GO network, the circle represents GO and the shape of red square represents miRNAs, and their relationship is represented by one edge. The center of the network is represented by degree. Degree of miRNA means the number of links one miRNA regulates GOs. The core miRNAs in the network always have the highest degrees.

Statistical analysis
The lists of miRNAs were extracted based on statistical test P-values (<0.05 was considered significant). The representatives of replicate experiments were shown in the figures and the data were shown as the means ± SEM. The statistical analyses were performed using independent Student's t-tests. In case of false positive results, Bonferroni correction was performed.