Cell culture and adipogenesis
hMSCs (HUXMA-01001, Cyagen Biosciences, China; cell lot no. of three donors: 150724I31, 160202I31, and 161125R41) were phenotyped via flow cytometry and were confirmed to be ≥ 95% CD73-, CD90-, and CD105-positive, as well as CD11b-, CD19-, CD45-, CD34-, and CD HLA-DR-negative (≤ 5%) as shown in additional File S1 (Supplementary Files). hMSCs were cultured at 5 × 104 cells/cm2 in OriCell hMSC Growth Media (HUXMA-90011; Cyagen Biosciences), supplemented with L-glutamine, 10% FBS, and penicillin/streptomycin. Cells were harvested using trypsin-EDTA (Invitrogen, CA, USA) and were passaged until reaching the sixth passage at which time they were used for experimentation.
Adipogenesis was induced by replacing normal hMSC growth media with DMEM, containing 10% FBS, 10% FBS, 10% g/mL insulin, 0.5 mM 3-isobutyl-1-methylxanthine, and 0.5 mM dexamethasone after cells had reached confluence[26]. These cells were then cultured in this media for 0, 7, 14, or 21 days after which they underwent high-throughput sequencing analyses.
Library construction and small RNA-seq
Cellular RNA was extracted using Trizol (Invitrogen), with RNA quality confirmed using an Agilent 2200 machine. Samples that had an RNA integrity (RIN) score>7.0 were then used for cDNA library preparation with the NEBNext Small RNA Library Prep Kit for Illumina. This RNA was first ligated to provided 3’ and 5’ adapters, after which first strand cDNA synthesis was performed. Index sequences and Illumina sequence adapters were then applied through index PCR, after which the library was purified, assessed for quality control with a Bioanalyzer 2200 instrument (Agilent, CA, USA), and sequences using a HiSeq X-ten platform (Illumina, CA, USA) with 150 bp paired-end reads.
Small RNA analysis
Low quality reads and adapter sequences were removed from raw data using Trim Galore, after which all remaining sequences from12 - 50 nucleotides long were subjected to sequence alignment with those sequences found in miRBase (http://www.mirbase.org/). Identification of known miRNAs was conducted using BWA after which unmapped reads were aligned to rRNA sequences (https://rnacentral.org/). Next, internally designed tRNA sequence database (using sequences from http://gtrnadb.ucsc.edu/ and https://cm.jefferson.edu/MINTbase/) was used to analyze any remaining unmapped reads, First, intronic sequences were removed after which CCA was added to the end of each tRNA sequence. A total of 50 genomic nucleotides were then added behind these CCA residues, with the resultant mapped reads being identified as potential tsRNAs that were then classified using tRFdb (http://genome.bioch.virginia.edu/trfdb/) and MINTBase (https://cm.jefferson.edu/MINTbase/).
Differential small RNA expression and target gene prediction
The EB-Seq algorithm[27] was used to identify those tsRNAs that were differentially regulated during adipogenesis based upon P-value and FDR significance analyses[28], with the cutoff criteria for differentially regulated sncRNAs being: i) Fold Change>2 or <0.5; ii) P<0.05, FDR<0.05. Next, putative tsRNA and miRNA target genes were identified using the miRanda[29] and RNAhybrid tools[30]. A sequencing data analysis flowchart was shown in Fig. S1 (Supplementary Files).
Lentiviral transduction and hMSC screening
Knockdown of tsRNA-16902 and overexpression of RARγ was conducted using lentiviral particles from Shanghai Genechem Co., Ltd. A short hairpin RNA (shRNA) targeting tsRNA-16902 was designed (target sequence,5′-TGGTGTCCTTGGAAAAAGGTTTTCATCTCCGGTTTACAA-3′). In addition, a negative control (NC) shRNA was used (sequence, 5′-TTCTCCGAACGTGTCACGT-3′). Lentiviral transduction reactions were conducted by plating 5×104 hMSCs/cm2 in 6-well plates until 20-30% confluent, after which they were infected using 10 uL of the appropriate lentivirus (1×108 infectious units/ml) in complete media supplemented with 5 µg/ml polybrene. After 10 h, this transduction media was removed and fresh media was added, after which the cells were grown for an additional 72 h. Cells were then cultured in 0.5µg/ml puromycin for 48 h, after which they were screened over a 6 day period with fresh media added every 1-2 days.
At various time points during adipogenesis, Oil Red O staining was used to confirm the presence of lipid droplets within cells. In addition, cells were collected to assess the expression of adipogenic markers and RARγ.
Oil red O staining
Cells were washed with PBS and were then fixed with 10% formalin for 30 min, after which they were washed using 60% isopropanol prior to staining for 10 minutes with Oil red O (0.3%; Sigma-Aldrich) while shaking gently. Cells were then washed using distilled water to remove free dye. In addition, free dye was eluted from these cells using 100% isopropanol, after which absorbance at 490 nm was assessed via spectrophotometry[31].
qRT‑PCR
TRIzol (Invitrogen) was used to extract cellular RNA as above, and a cDNA Reverse Transcription Kit (Thermo, CA, USA) was then utilized for cDNA preparation. All qRT-PCR reactions were conducted using a SYBR Premix Ex Taq kit (Toyobo, Osaka, Japan) with an ABI Prism 7500 instrument (Applied Biosystems). The primers were used in this study are compiled in Table S1 (Supplementary Files). Relative gene expression was quantified using the 2−ΔΔCt method[32-33].
Western blotting
Cells were lysed using RIPA buffer, after which 15 ug of protein per sample was boiled in 5×SDS sample buffer for 5 minutes, separated via 10% SDS-PAGE, and transferred to a PVDF membrane (Millipore). Blots were blocked with 5% non-fat milk for 2 h, and were probed overnight with the following antibodies: rabbit anti-PPARγ (abcam 191407), rabbit anti- C/EBPα (abcam 40764), rabbit anti-FABP4 (abcam 92501), rabbit anti-RARγ (abcam 191368), rabbit anti-Smad2/3(abcam 63672), rabbit anti-p-Smad2/3 (abcam 63399) and mouse anti-β-actin (1:2,000; abcam 173838). All rabbit antibodies were diluted 1:1,000. Blots were then probed for 1 h with an appropriate secondary HRP-linked antibody (1:5,000; CST), after which enhanced chemiluminescence (BeyoECL Plus; Beyotime Institute of Biotechnology) was used for protein visualization.
Luciferase reporter assay
Briefly, a WT or mutated version of the RARγ 3’-UTR sequence containing this potential tsRNA binding site was cloned into the pGL3 vector downstream of luciferase, after which DNA sequencing was used to confirm construct identity. This vector was then co-transfected into 293T cells with or without a tsRNA-16902 mimic. After 48 h, the luciferase activity in these cells was quantified with Renilla luciferase activity being used for normalization purposes.
Statistical analysis
SPSS v16.0 (SPSS, IL, USA) was used for statistical testing. Data are given as means ± standard deviation (SD). Data were compared using Student’s t-tests and one-way ANOVAs, as appropriate, with a two-tailed P <0.05 as the significance threshold.