Analysis of differentially expressed microRNA of TNF-α-stimulated mesenchymal stem cells and exosomes from their culture supernatant.

INTRODUCTION: To analyze the microRNA expression of tumor necrosi factor α (TNF-α) stimulated mesenchymal stem cells (MSCs) and exosomes from their culture supernatant.
MATERIAL AND METHODS: TNF-α (20 ng/ml) was used to stimulate MSCs, which were then regarded as TNF-α cells (TC), while unstimulated cells were the normal control cells (NCC). MSCs and their culture supernatant were harvested after 48 h. Subsequently, exosomes were isolated from culture supernatants with ExoQuick-TC and were divided into two groups, TNF-α exosomes (TE) and normal control exosomes (NCE). Then, the microRNAs were measured by high-throughput sequencing and the results were differentially analyzed. Finally, the correlation of the target genes corresponding to differently expressed microRNAs was analyzed by gene ontology (GO) and KEGG pathway analysis.
RESULTS: High-throughput sequencing showed that the cellular compartment (TC vs. NCC) had 280 microRNAs. miR-146a-5p was a uniquely up-regulated microRNA (p < 0.001) and the most significantly down-regulated microRNA among the 279 microRNAs included was miR-150-5p (p < 0.001). There were 180 differentially expressed microRNAs in the exosome compartment (TE vs. NCE), where miR-146-5p (p < 0.001) was one of 176 upregulated microRNAs and miR-203b-5p (p < 0.001) was one of 4 downregulated microRNAs. Coincidentally, bioinformatics analysis showed that IRAK1 was a critical target gene of miR-146-5p related to the Toll-like receptor (TLR) signaling pathway.
CONCLUSIONS: In contrast with the control group, there were significantly differentially expressed microRNAs in both MSCs and exosomes. Interestingly, miR-146a-5p was up-regulated in both comparative groups, and its target gene IRAK1 plays a crucial part in the TLR signaling pathway. These investigations demonstrate a new direction for subsequent inflammation mechanistic studies.

Introduction

Mesenchymal stem cells (MSCs) have the potential to differentiate and impact important immune regulation, tissue repair, etc. [1-5]. Some of these approaches have been applied in clinical practice. However, there have been some problems, such as limited differentiation potential, and the fact that MSCs may transform into tumor cells or improve tumor progression [6]. Moreover, the survival rate of MSCs in injured tissues was low. It is believed that MSCs can repair damaged tissue mainly by paracrine signaling [7]. However, the specific mechanism needs further investigation. In 2010, Lai et al. [8] isolated nutritional factors that had reparative effects and first confirmed exosomes from mesenchymal stem cells (MSC exosomes). MSC exosomes are homologous vesicles secreted in physiological or pathological conditions, with a diameter of approximately 30–100 nm. Exosomes treat diseases by delivering specific proteins, mRNA, and microRNA to target tissue. Mesenchymal stem cells can secrete more exosomes than other sources [9]. An increasing number of studies have reported that the protective effects of MSCs after pretreatment, such as with conditioned media, ischemia-reperfusion [10], hypoxia induction [11], and inflammatory factors, became more pronounced. Because MSCs secreted more bioactive substances after pretreatment [12], the conditioned medium can also simulate MSCs to protect cells and tissues and regulate immunity and inflammation [13].

Tumor necrosis factor α is an activator of a variety of immune cells. Tumor necrosis factor α can increase the immune regulation of MSCs in vitro and in vivo [14]. The culture medium can reduce the inflammation after stimulating MSCs with TNF-α [15]. The culture medium can simulate the biological function of MSCs after being pretreated by TNF-α and may transfer biological information through exosomes as a carrier, but its mechanism has not been elucidated.

In this study, we used TNF-α as an inducible factor in human mesenchymal stem cells. Cells and exosomes were collected, and RNA of cells and exosomes was extracted to detect microRNAs by high-throughput sequencing. The differential expression of microRNAs in the cells and exosomes was analyzed, and subsequent analysis with an elaborate bioinformatics strategy, including gene ontology (GO) analysis and pathway analysis, was performed.

Material and methods

Ethics statement

The experiments were approved by the Ethics Committee of the Second Clinical Medical College of Jinan University.

Preparation of HuMSCs

Human umbilical cord mesenchymal stem cells (HuMSCs) were isolated from human cord tissue using a method described previously [16]. With the written consent of the parents, fresh human umbilical cords were obtained. Wharton’s jelly was diced into cubes of approximately 0.5 cm3 and poured into a centrifuge tube. The tissue was treated with collagenase type II (Sigma, USA) at 37°C for 8 h, and the resulting solution was filtered. After the mesenchymal tissue was digested and poured into a 200 mesh filter, the filtrate was diluted with PBS, and then the diluted liquid was centrifuged at 2000 rpm for 5 min. After removal of the supernatant fraction, the cells were resuspended in PBS. The resuspended liquid was centrifuged again at 2000 rpm for 5 min. After removal of the supernatant fraction again, the precipitate (mesenchymal tissue) was resuspended in Dulbecco’s modified Eagle’s medium (Gibco, Carlsbad, CA, USA) with 10% fetal bovine serum and cultured at 37°C in an incubator containing 5% CO2.

Cell identification and differentiation studies

To verify the cultured cells, the 3rd-passage cells (1 × 105) were suspended in 50 ml of PBS containing 20 ng/ml fluorescein isothiocyanate (FITC)-coupled antibodies against CD29, CD34, CD45, CD90, CD105, and HLA-DR (Biolegend, San Diego, CA, USA) and FITC-coupled nonspecific IgG as an isotype control (Biolegend, San Diego, CA, USA).

Stimulation of human mesenchymal stem cells by TNF-α

The cells were washed twice with Hank’s Balanced Salt Solution (HBSS) to remove serum when MSCs were grown to 80–90% confluence. Then, the cells were divided into the following groups: 1) normal control cells: without intervention; 2) TNF-α cells: with stimulant alone, 48 h TNF-α (20 ng/ml). After 48 h incubation, supernatants were harvested for isolation of exosomes.

Isolation of exosomes with ExoQuick-TC

MSC-derived exosomes were isolated using ExoQuick-TC according to the manufacturer’s protocol. In brief, cell culture supernatants were collected and centrifuged at 3,000 g for 15 min to remove cells and cell debris. Two ml of ExoQuick-TC Exosome Precipitation Solution was added to 10 ml of the supernatants and the mixture was refrigerated overnight (at least 12 h). The ExoQuick-TC/biofluid mixture was centrifuged at 1,500 g for 30 min and the supernatants were aspirated. Then, the residual solution was centrifuged at 1,500 g for 5 min and removed. The exosome pellet was resuspended in the appropriate buffer for protein or RNA analysis.

Characterizations of exosomes

The quality of exosomes was confirmed by dynamic light scattering using a particle and molecular size analyzer (ZetasizerNano ZS, Malvern Instruments) according to the manufacturer’s instructions. The quantity of exosomes was determined by the Micro-BCA assay (Pierce, Rockford, IL) for measurement of total protein.

RNA isolation of MSCs

Total RNAs from HuMSCs were extracted with the mirVana microRNA isolation kit (Life Technologies, Gaithersburg, MD, USA) according to the manufacturer’s protocol. RNA integrity was assessed with electrophoresis.

Isolation of RNA from exosomes

Exosome supernatants were added to 40 pM synthetic cel-miR-39 (UCACCGGGUGUAAAUCAGCUUG) to control and normalize the efficiency of RNA extraction and then transferred to RNase-free tubes for RNA isolation using an miRNeasy Mini Kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s protocol. The RNA samples were washed twice in 500 μl of RPE buffer and eluted in RNase-free water. The isolated RNA was measured using a NanoDrop 1000 ultraviolet spectrophotometer (Thermo Fisher Scientific) and analyzed by reverse transcription polymerase chain reaction (RT-PCR), followed by quantitative PCR (qPCR).

microRNA deep sequencing analysis

To prepare for deep sequencing, four small RNA libraries were constructed using the Small RNA Expression Kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturer’s protocol. Briefly, RNA samples were hybridized and ligated with an adapter mix. Then, reverse transcription was performed, followed by RNase H digestion and PCR amplification. The PCR products were then size-selected and purified. Each pool of libraries included three biological repeats. Then, the SOLID V2 sequencing system (Applied Biosystems, Foster City, CA, USA) was performed to generate read counts of nucleotide sequences. The sequencing results were analyzed by the SOLID system small RNA analysis pipeline tool (RNA2MAP). The raw data were first filtered to remove low-quality reads. Then the data were mapped to GenBank (http://www.ncbi.nlm.nih.gov/genbank), and sequences matching rRNAs, tRNAs, snoRNAs and scRNAs were discarded. Finally, by mapping to the miRBase database (http://www.mirbase.org), known and novel microRNAs were identified. The total read counts of each sample were normalized to reads per million (RPM). To perform differential expression analysis, a DEGseq package tool [17] was applied and the significantly differential microRNAs were identified according to the p-value (< 0.05).

Bioinformatics analysis

GO analysis and pathway analysis were used a method that we described previously [18].

Target gene analysis of microRNA expression profiles

To analyze target genes repressed by micro-RNAs, we used a previously developed method. Repression scores for individual target genes were calculated according to the formula below. Briefly, the TargetScan Database provides an individual microRNA’s repression rate for each of its target genes. By combining the repression rates of all related microRNAs, we can calculate the effect of the entire microRNA profile based on the repression rate of one specific target. Importantly, this method takes into consideration the abundance of each microRNA. Therefore, the repression score represents the effect of the microRNA profiles on a specific target. The significantly differential targets were further identified using DEGseq (p < 0.05). To identify microRNAs corresponding to each pathway, target genes were analyzed by the miRWalk tool [19]. Only significantly expressed microRNAs were identified.

Results

Characterization of MSCs

HuMSCs were generated by standard procedures in ex vivo culture. HuMSCs were morphologically defined by a fibroblast-like appearance (Figure 1). Flow cytometric analysis of passage 3 cells confirmed that cells were negative for CD34, CD45 and HLA-DR and that cells were positive for CD29, CD90 and CD105.

Figure 1

Image of HuMSCs observed by electron microscopy (original magnification × 100). A – Control-HuMSCs growth state at 24 h, B – TNF-α HuMSCs growth state at 24 h, C – control HuMSCs growth state at 48 h, D – TNF-α HuMSCs growth state at 48 h

Characterization of MSC-Exos

Transmission electron microscopy clearly revealed that MSC-Exos exhibited a cup-shaped or round-shaped morphology with a diameter of 30–100 nm. The detection of origin immunoblotting indicated that the MSC-Exos expressed exosomal markers such as CD81 and CD63 (Figure 2).

Figure 2

Characterization of exosomes from culture supernatant of MSCs. A – Exosomes at the measurement of 30 nm to 100 nm in an electron microscope; B – surface marker CD63/CD81 of positive exosomes and HuMSCs

microRNA profiles of MSCs

Using a microRNA deep sequencing method, we investigated the microRNA expression changes of MSCs in the TNF-α cell group and control cell group. High-throughput raw sequencing reads were generated by the SOLiD sequencing system. As a result, 36,842,471 (TC group) and 15,483,812 (NC group) reads were obtained. Compared with the GenBank and Rfam databases, we found 13,205,775 and 4,389,656 microRNA fragments, respectively. We then sought to investigate the TNF-α stimulation-related changes in microRNAs. The DEGseq package tool identified differentially expressed genes based on the Poisson distribution. Using DEGSeq, we performed a differential comparison of the microRNA expression patterns of the TC group MSCs and NC MSCs. To filter out the non-significantly altered microRNAs, we limited the p-value to p < 0.01. As a result, 280 microRNAs were identified as the most differentially expressed microRNAs in the TNF-α cell group, including 1 up-regulated microRNA and 279 down-regulated microRNAs (Figure 3, Table I), of which only miR-146a-5p was up-regulated, 1.92 times greater than in the control group, while hsa-miR-150-5p, hsa-miR-1273-c, hsa-miR-1247-5p and hsa-miR-1208 were down-regulated.

Figure 3

TNF-α cell vs. control cell microRNAs in a volcano plot. The differentially expressed microRNAs of TE compared to NCE group in a volcano plot; each dot represents one microRNA; the blue dots indicate the upregulated microRNAs; the red dots indicate the downregulated microRNAs; the green ones have no significant difference

Table I

The most significantly expressed microRNAs: 1 up-regulated and 19 down-regulated microRNAs

microRNA ID	Relative count (TPM) and group		Fold change	P-value
	TC group	NC group
hsa-miR-146a-5p	19.9	5.25	1.92	< 0.001
hsa-miR-150-5p	4.65	477.64	–6.68	< 0.001
hsa-miR-1273c	1.03	94.41	–6.51	< 0.001
hsa-miR-1247-5p	2.91	255.07	–6.45	< 0.001
hsa-miR-1208	8.17	703.83	–6.43	< 0.001
hsa-miR-378h	2.78	201.65	–6.18	< 0.001
hsa-miR-5693	4.9	338.22	–6.11	< 0.001
hsa-miR-767-5p	4.41	302.93	–6.11	< 0.001
hsa-miR-4792	1.06	65.47	–5.95	< 0.001
hsa-miR-2392	2.89	167.52	–5.86	< 0.001
hsa-miR-4651	2.23	122.32	–5.78	< 0.001
hsa-miR-718	1.36	73.24	–5.75	< 0.001
hsa-miR-4315	1.06	55.5	–5.71	< 0.001
hsa-miR-8069	1.06	53.75	–5.66	< 0.001
hsa-miR-6075	1.47	73.56	–5.65	< 0.001
hsa-miR-208b-3p	4.49	211.36	–5.56	< 0.001
hsa-miR-4668-5p	4.41	207.41	–5.56	< 0.001
hsa-miR-4540	2.91	134.5	–5.53	< 0.001
hsa-miR-4294	10.72	494.09	–5.53	< 0.001
hsa-miR-4518	2.34	107.75	–5.52	< 0.001

TPM – transcripts per million.

microRNA profiles of exosomes

Using a microRNA deep sequencing method, we investigated the microRNA expression changes of MSCs in the TE group and in the NE group. High-throughput raw sequencing reads were generated by the SOLiD sequencing system. As a result, 10,569,238 (TE group) and 15,454,239 (NE group) reads were obtained. Compared with the GenBank and Rfam databases, we found 14,739,30 and 631,271 microRNA fragments, respectively. We then sought to investigate the TNF-α-related changes in microRNAs. The DEGseq package tool identified differentially expressed genes based on the Poisson distribution. Using DEGSeq, we performed a differential comparison of the microRNA expression patterns of the TE group exosomes and NE group exosomes. To filter out the non-significantly altered microRNAs, we limited the p-value to p < 0.01. As a result, 180 microRNAs were identified as the most differentially expressed microRNAs in TE group exosomes, including 176 upregulated microRNAs and 4 downregulated microRNAs (Figure 4, Table II). The expression of miR-146-5p, miR-193b-5p, miR-193a-5p and miR-382-5p was up-regulated the most. The expression of miR-203b-5p, miR-203a-3p, miR-1273h-3p, and miR-4540 was down-regulated.

Figure 4

TE vs. NCE microRNAs in a volcano plot. The differentially expressed microRNAs of TE compared to NCE group in a volcano plot; each dot represents one microRNA; the blue dots indicate the upregulated microRNAs; the red dots indicate the downregulated microRNAs; the green ones have no significant difference

Table II

The most significantly expressed microRNAs: 4 up-regulated and 4 down-regulated microRNAs

microRNA ID	Relative count (TPM) and group		Fold change	P-value
	TE group	NE group
hsa-miR-146a-5p	836.23	43	4.28	< 0.001
hsa-miR-382-5p	321	83.54	2.34	< 0.001
hsa-miR-193a-5p	63.04	13.78	2.20	< 0.001
hsa-miR-193b-5p	9.18	1.82	2.19	< 0.001
hsa-miR-203b-5p	1.89	21.72	–3.52	< 0.001
hsa-miR-2036a-5p	2.27	22.3	–3.29	< 0.001
hsa-miR-1273h-3p	5.87	13.5	–1.20	< 0.001
hsa-miR-4540	8.71	18.56	–1.09	< 0.001

GO annotation and analysis of the differences

The cell and exosome genes characterized in the study were evaluated based on their molecular function, biological process and cellular component annotations. As shown in Figure 5 A, the most enriched cellular components were organelle parts and organelles in cells. In contrast, the cell part is the largest in the exosome genes, as shown in Figure 6 A. The largest group in the distribution of the various functional cell categories is proteins with roles in catalytic activity, as shown in Figure 5 B. The molecular function of exosomes shown in Figure 6 B. Proteins related to cellular and multicellular organismal processes, which comprise the largest group in the distribution of biological processes of cells and exosomes, are shown in Figures 5 C and 6 C, respectively.

Figure 5

Distribution of significantly differentially expressed microRNAs of cells in the cellular component, molecular function and biological process with high-throughput sequencing

Figure 6

Distribution of significantly differentially expressed microRNAs of exosomes in the cellular component, molecular function and biological process with high-throughput sequencing

Signaling pathway analyses

As described in detail previously [18], we mapped differentially modified genes to the KEGG pathway database using GenMAPP v2.1 and then performed a statistical test to identify enriched metabolic pathways. miR-146a-5p was significantly up-regulated in both the cell group and exosome group. There were 53 target genes in the database that had a corresponding metabolic pathway. There were seven target genes that were closely related to inflammation immunity in nine signaling pathways (Table III), while the Toll-like receptor (TLR) signaling pathway was the most widely distributed (Figure 7).

Table III

Distributions of the main target gene of miR-146a-5p in the KEGG pathway

Pathway	Target gene
TOLL_LIKE_RECEPTOR_SIGNALING_PATHWAY	TRAF6, IRAK1, CCL5
NF-kB_SIGNALING_PATHWAY	TRAF6, IRAK1
WNT_SIGNALING_PATHWAY	CXXC4, WNT3
CYTOKINE_RECEPTOR_INTERACTION	CNTF
JAK_STAT_SIGNALING_PATHWAY	CNTF
CHEMOKINE_SIGNALING_PATHWAY	CCL5
NOD_LIKE_RECEPTOR_SIGNALING_PATHWAY	CCL5
TGF_BETA_SIGNALING_PATHWAY	RBL1
HEDGEHOG_SIGNALING_PATHWAY	WNT3

Figure 7

Gene network of miR-146a-5p in the Toll-like receptor signaling pathway

Discussion

Many studies have shown that MSCs and MSC exosomes significantly affect kidney diseases [20, 21], cardio and pulmonary diseases [22, 23], limb ischemic injury [24] and other diseases. The use of LPS [13], TNF-α [25-28], hypoxia injury [10] and other pretreatments can regulate cell phenotype, secretion of soluble factors and exosomes [29] to play protective roles. Later studies found that exosomes had corresponding target microRNAs that protect the heart [30], inhibit tumor growth [31], enhance tumor chemosensitivity [32], etc. Exosomes may even become a marker to diagnose disease and assess the prognosis of diseases [33]. However, cultured inflammatory-stimulated stem cells can enhance the anti-inflammatory effects [15], which is in contrast with our traditional understanding that inflammatory factors can only cause tissue damage.

It is generally believed that TNF-α is a pro-inflammatory factor that will lead to an inflammatory response. In fact, there are reports [14] that appropriate concentrations of TNF-α do not affect the biological characteristics of cells. In this study, the final concentration of TNF-α was 20 ng/ml, which was consistent with the biological characteristics and growth of the cells. Moreover, we found that TNF-α did not cause cell necrosis, accelerated apoptosis, morphological changes, etc. MicroRNAs are endogenous non-coding small RNAs that play an important regulatory role in cells and even tissue biology processes. Therefore, the differential expression of microRNAs and their signaling pathways in cells after stimulation by inflammatory factors can provide important information about pro-inflammatory factors that can also inhibit the inflammatory response.

In this study, we investigated the mechanism of pro-inflammatory factors on MSCs during ex vivo expansion based on microRNA deep sequencing methods. We identified the significantly altered microRNAs in response to TNF-α. Among them, we found 280 microRNAs as the most differentially expressed microRNAs in the TC group, including 1 up-regulated microRNA and 279 down-regulated microRNAs. We also found 180 microRNAs as the most differentially expressed microRNAs in the TE group exosomes, including 176 upregulated microRNAs and 4 downregulated microRNAs.

A number of studies [34-36] have suggested that miR-146a-5p is a microRNA that is closely related to inflammatory immune regulation. Up-regulation of miR-146a-5p has an inhibitory effect on inflammatory immune responses. Moreover, miR-146a-5p is related to important inflammatory pathways, such as NF-kB and Toll-like receptors. Up-regulation of miR-146a-5p was 3.76 and 19.45 times higher in the TC group and TE group compared to the NC group and NE control group, respectively, suggesting that appropriate pro-inflammatory cytokines can change anti-inflammatory microRNAs such as miR-146a-5p and increase anti-inflammatory effects. Recent studies in vitro and in vivo have shown that [34, 37–40] exosomes can inhibit endotoxin-induced inflammatory immune responses via miR-146-5p. In this study, the expression of miR-146-5p in the exosomes group was significantly higher than that of the cell group, which may be due to the simple structure of exosomes and the decreased expression of microRNA. Meanwhile, we believe that the use of exosomes as a small unit may be more effective and efficient because exosomes are smaller and more flexible than cells in terms of biological characteristics. Exosomes are measured in nanometers, so they have a quantitative advantage. In addition, miR-146a-5p can inhibit the transfer of kidney clear cell carcinoma [41], improve allergic rhinitis [42], promote chondrocyte phagocytosi [43], enhance differentiation of spermatogonia [44] and delay senile dementia [35].

KEGG analyses showed that miR-146a-5p was significantly upregulated in both the cell and exosome groups. There were 53 target genes in the database that had corresponding metabolic pathways. There were seven target genes that were closely related to inflammation immunity in nine signaling pathways. Some studies reported that miR-146a-5p participates in a variety of biological processes, such as inflammatory immune responses, and that the target genes TRAF6, IRAK1, and CCL5 are involved in the biological effects of multiple metabolic pathways, while the TLR signaling pathway was the most widely distributed. We selected the TLR metabolic pathway to analyze and showed that the IRAK gene played an important role in this metabolic pathway. The miR-146a-5p target gene IRAK1 is the junction of the TLR metabolic pathway and the downstream NF-kB metabolic pathway. Because miR-146a-5p has a negative regulatory immune response, up-regulation of miR-146a-5p may inhibit the transcription of the target gene IRAK1 and inhibit the next level of the inflammatory response.

In summary, the target gene IRAK1 of miR-146a-5p is involved in the biological process of the TLR metabolic pathway, which is an important inflammatory metabolic pathway. TLR4 can be seen as a major microRNA that regulates the TLR metabolic pathway. miR-146a-5p may provide a new direction for a follow-up study and clinical prevention and treatment of inflammation-related diseases.

In conclusion, we have successfully constructed an experimental model of TNF-α to stimulate MSCs and extracted small RNAs from cells and exosomes. MicroRNA-146a-5p was up-regulated in both the cell group and exosomes group, and the expression ratio was the most obvious. According to bioinformatics analysis, we found that the most important target gene was IRAK1 for microRNA-146a-5p and was involved in the TLR metabolic pathway. miR-146a-5p and IRAK1 may provide a new direction for a follow-up study and clinical prevention and treatment of inflammation-related diseases.