Small RNA sequencing reveals miR-642a-3p as a novel adipocyte-specific microRNA and miR-30 as a key regulator of human adipogenesis

Abstract

Background: In severe obesity, as well as in normal development, the growth of adipose tissue is the result of an increase in adipocyte size and numbers, which is underlain by the stimulation of adipogenic differentiation of precursor cells. A better knowledge of the pathways that regulate adipogenesis is therefore essential for an improved understanding of adipose tissue expansion. As microRNAs (miRNAs) have a critical role in many differentiation processes, our study aimed to identify the role of miRNA-mediated gene silencing in the regulation of adipogenic differentiation.

Results: We used deep sequencing to identify small RNAs that are differentially expressed during adipogenesis of adipose tissue-derived stem cells. This approach revealed the un-annotated miR-642a-3p as a highly adipocyte-specific miRNA. We then focused our study on the miR-30 family, which was also up-regulated during adipogenic differentiation and for which the role in adipogenesis had not yet been elucidated. Inhibition of the miR-30 family blocked adipogenesis, whilst over-expression of miR-30a and miR-30d stimulated this process. We additionally showed that both miR-30a and miR-30d target the transcription factor RUNX2, and stimulate adipogenesis via the modulation of this major regulator of osteogenesis.

Conclusions: Overall, our data suggest that the miR-30 family plays a central role in adipocyte development. Moreover, as adipose tissue-derived stem cells can differentiate into either adipocytes or osteoblasts, the down-regulation of the osteogenesis regulator RUNX2 represents a plausible mechanism by which miR-30 miRNAs may contribute to adipogenic differentiation of adipose tissue-derived stem cells.

Figures

Figure 1

Distribution of deep-sequenced small RNAs across non-coding RNA categories. (a) Reads were matched versus the hg19 genome build and then distributed in an exclusive manner to human miRNAs, as well as miRNAs of species other than human (mirBase 16), to UCSC annotated sequences (UCSC Refflat file) and finally to non-coding RNA classes (fRNAdb, database of ncRNA.org): piwi-interacting RNA (piRNA), tRNA, rRNA, small nucleolar RNA (snoRNA) and other non-coding RNA (ncRNA). Reads that did not match any of those non-coding RNA classes were labeled as 'non-annotated'. Data are the average of read sequencing frequency (percentage) for each experimental condition. ND, undifferentiated cells; AD3, adipogenesis day 3; AD8, adipogenesis day 8. (b) Relative abundance of reads corresponding to the 30 most expressed miRNAs in undifferentiated hMADS cells. Read counts are normalized to 106 total miRNA reads per sample. Data are the average of sequencing of samples from two independent experiments, each with two technical replicates with opposite sequencing directions (error bars represent ± standard error).

Figure 2

miRNA expression data in differentiated versus undifferentiated human adipose tissue-derived stem cells. (a) Heatmap of the fold-change (log2 transformed) of miRNA expression in differentiated versus undifferentiated hMADS cells. The top 26 regulated miRNAs are represented (P-value < 0.05). Two independent experiments are displayed. (b) Relative abundance of the miR-30 family over total detected miRNAs in undifferentiated and adipocyte-differentiated hMADS cells. Data are the average of sequencing of samples from two independent experiments, each with two technical replicates with opposite sequencing directions. ND, undifferentiated cells; AD3.1, adipogenesis day 3, replicate 1; AD3.2, adipogenesis day 3, replicate 2; AD8.1, adipogenesis day 8, replicate 1; AD8.2, adipogenesis day 8, replicate 2.

Figure 3

Abundance of each base along the miR-642a pre-miR. Each experimental condition is pictured using the color code in the insert. Light grey shading highlights miR-642-5p (bases in orange) and miR-642-3p (bases in blue). Represented samples were sequenced in the 3' to 5' direction.

Figure 4

The miR-30 family positively regulates hMADS cell adipocyte differentiation. (a) Sub-confluent hMADS cells were transfected with one of anti-miR control (anti-neg), anti-miR 30, pre-miR control (pre-miR-neg), pre-miR-30a, or pre-miR-30d and were induced to undergo adipocyte differentiation 3 days later. Differentiation was assessed at the indicated time points (D4 or D10) by photomicrographic recording (top row) and Oil red O plus crystal violet counter-staining (lower row). (b) Assessment of adipogenesis by GPDH enzymatic activity. Results are means of three culture wells (24-well plates). Error bars represent mean ± standard error of the mean (N = 3). *P < 0.05. (c) Expression of adipogenesis-induced genes (CEBPβ, PPARγ and FABP4) by qPCR. The level of expression of each gene in control cells (anti-miR-neg or pre-miR-neg) was taken as 1.

Figure 5

RUNX2 mRNA is a primary target for miR-30a and miR-30d. (a) Predicted interaction between miR-30a and miR-30d and their putative binding sites in the 3' UTR of RUNX2. The representation is limited to the region around the miR-30a and miR-30d complementary sites. In bold is the 'seed' region with a conserved anchoring adenosine that is complementary to the first nucleotide of miR-30a and miR-30d (underlined). (b) Schematic representation of the construct used in the luciferase assay: a 300-bp (reporter 1) and 319-bp (reporter 2) region of the 3' UTR of human RUNX2 containing the putative miR-30a and/or miR-30d target sites (black boxes) were cloned into the pSi-CHECK™-2 vector. (c) Normalized luciferase activity 48 hours after co-transfection of human pre-miR-30a, pre-miR-30d, pre-miR-378 or pre-miR-control (neg) together with pSi-CHECK™-2 constructs in HEK 293 cells. Data were obtained from four independent experiments (error bars represent average ± standard error); n.s., not significant compared to pre-miR-control; *significant compared to pre-miR-control (P < 0.05); **significant compared to pre-miR-control (P < 0.01). (d) Undifferentiated hMADS cells were transfected with either pre-miR control (pre-miR-neg) or pre-miR-30a or pre-miR-30d. Four days later, cell lysates were prepared and expression of RUNX2 was investigated by western blotting. Tubulin was used as a loading control. The integrated density of each band was quantified with Image J. Densities obtained for RUNX2 signals were divided by the corresponding tubulin densities. Numbers below the blot are the density fold changes compared to the control condition. (e) Undifferentiated hMADS cells were transfected with control target site blocker (TSB-neg) or with RUNX2 target site blocker (TSB-RUNX2) as well as with pre-miR-30a. Adipogenic differentiation was evaluated by analyzing adiponectin and PPARγ expression by qPCR. The level of expression of each gene in the pre-miR-30a condition was taken as 1.