LncRNA-PCAT1 targeting miR-145-5p promotes TLR4-associated osteogenic differentiation of adipose-derived stem cells

2.1 Clinical samples

The study group consisted of eight osteoporosis patients who had undergone an iliac bone graft procedure during surgery at Peking Union Medical College Hospital, Peking Union Medical College and Chinese Academy of Medical Science in 2016. All patients provided informed consent and the present study was approved by the ethics committee of the Peking Union Medical College Hospital, Peking Union Medical College and Chinese Academy of Medical Science. Eight normal donors were used as controls and none of them had previously experienced bone fractures before. A standardized clinical evaluation was performed to exclude possible comorbidities.

2.2 Microarray analysis

Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/) provided the gene expression data derived from four different donors (GSE89330). To ascertain those genes that had higher chances of being differentially expressed in one group compared with another, we used the statistical test in the R Limma package. Those genes with at least an absolute value of 1 for log₂ (FC) and adj. P < 0.05 cut-off were identified. The overall profile was normalized for each array to correct for systematic data bias and remove nonbiological impact.

2.3 Pathway analysis

Gene set enrichment analysis (GSEA) tested whether a set of predefined genes, usually united by shared association, showed enriched in expression levels. In our study, the uploaded gene set consisted of normalized expression data of mRNAs expression data and was sorted by the mean log₂ signal ratios. The aggregated distribution of gene expression levels in pathways was then determined with a normalized enrichment score, which represented the statistical significance through enrichment analysis. Pathways that were significantly biased in the undifferentiated or osteogenesis group were identified. The GSEA enrichment plot offered us with visualized results.

2.4 DEG correlation network analysis

By evaluating the correlation between gene expression patterns, we hoped to discover potential drug targets and candidate biomarkers. The “Pearson” approach in the R package Psych was harnessed to identify linear correlations among DEGs in a specific pathway, this was adjusted by the “BH” method. Then, the networks were visualized using Cytoscape software. Nodes in the network represented DEGs while the edges indicated the presence of coexpression.

2.5 Cell cultivation

Human adipose-derived stem cells were purchased from the ScienCell Company (Carlsbad, CA, USA). The cells were cultured in growth medium consisting of Dulbecco's Modified Eagle's Medium (DMEM; Gibco, New York, NY, USA) supplemented with 10% foetal bovine serum (FBS) and 1% antibiotics (Gibco). All cell-based in vitro experiments were performed in triplicate.

2.6 Osteoblast differentiation induction

To induce osteoblast differentiation, the media was replaced with osteoblasts-specific induction medium upon reaching 80%-90% confluence (cells covered area percentage). Low glucose of DMEM with 10% FBS, 10 mM β-glycerophosphate, 0.1 μM dexamethasone, and 0.2 mM antiscorbic acid (Sigma-Aldrich, Louis, MO, USA) was useful for inducing osteoblast differentiation for 0, 7, 14, and 21 days.

2.7 Cell proliferation dynamics analysis

Passage 4 cells were seeded into 96-well plates at 6.25 × 10⁴, 1.25 × 10⁵, 2.5 × 10⁵, 5 × 10⁵, 1 × 10⁶ and 2 × 10⁶ cells/well cell concentration, respectively. Taking the cell density as the X-axis, the absorbance as the Y-axis, draw the standard curve. Cell proliferation was detected by using the Cell Counting Kit-8 kit (Beyotime, Shanghai, China). Cells were seeded into 96-well plates at 2 × 10³ cells/well cell concentration. After cell culture for 1-21 days, 10 μL CCK-8 solution was added to each well and incubated at 37°C for 2 hours. The absorbance was measured at a wavelength of 570 nm for each well. The population doubling was calculated according to the standard and growth curves. Population doubling T_d = T × [lg2/lg(N_t/N₀)]. T, proliferation time; N₀, cell inoculum density; N_t, late cell density.

2.8 hADSC phenotypic characterization

hADSC-specific surface antigens were stained with PE-conjugated antibodies of anti-human CD29 (BD, Franklin Lakes, NJ, USA), anti-human CD34 (BD), anti-human CD44 (BD), and anti-human CD45 (BD) or their corresponding isotype control. Then, the stained cells were analysed by fluorescence-activated cell sorting (FACS). In brief, hADSCs were separately by trypsin-EDTA and gently blown into single cells, which were fixed and permeated with a Cytofix/Cytoperm Fixation/Permeabilization Solution Kit (BD) following the manufacturer s instructions. After stained with PE-conjugated antibodies for 30 minutes at 4°C, these stained cells were analysed on a fluorescence-activated cell sorter (Beckman, Miami, FL, USA). The experiments were repeated three times, the results are presented as fold change ± SD, and P < 0.05 was considered significantly different.

2.9 QRT-PCR

After osteoblast differentiation was induced in cells for 14 days, the cells were dissociated with 0.25% trypsin with EDTA and washed twice with cold PBS. The cell suspension was centrifuged at 111.9 × g for 5 minutes to get the precipitation. RNA extraction was performed using the Takara Minibest universal RNA extraction kit (Takara, Katsushika, Tokyo) according to the manufacturer's protocol. In total 500 ng of total RNA was reverse-transcribed to cDNA using the Takara PrimeScript RT master mix kit (Takara) according to the manufacturer's instructions. MiRNAs were reverse-transcribed with an All-in-OneTM miRNA first-strand cDNA synthesis kit (GeneCopoeia, Rockville, MD, USA). Equivalent amounts of cDNA were used for real-time PCR in a 20 μL reaction mixture using SYBR Premix Ex TaqII kit (catalog number: RR820A; Takara). QRT-PCR reactions were performed with 40 cycles of amplification on an ABI Prism 7300 Real-Time PCR system (Applied Biosystems, Waltham, MA, USA). The primers (Invitrogen, Waltham, MA, USA) used are listed in Table S1. The results were evaluated by the 2^−ΔΔCt method.

2.10 Western blot

After osteoblast differentiation was induced in cells for 14 days, the cells were washed twice with PBS and harvested in 200 μL lysis buffer (100 μM HEPES, 10 mM KCl, 2 mM MgCl₂, and 1 mM DTT) containing complete protease inhibitor (Roche Diagnostics Deutschland GmbH, Mannheim, Germany). Five microliters 10% Nonidet-P40 was added, followed by centrifugation at 18911.1 × g for 30 seconds at 4°C. The Bradford protein assay (Bio-Rad, Hercules, CA, USA) was used to quantify the protein concentration. Samples were separated by SDS polyacrylamide gel electrophoresis and then transferred to a PVDF membrane using the iBlot Dry Blotting Transfer System (Life Technologies Corporation, Gaithersburg, MD, USA). Membranes were blocked with 2.5% nonfat milk powder in TTBS buffer (0.1 M Tris, 150 mM NaCl, and 0.1% Tween-20) and incubated with primary antibodies against TLR4 (1/500, ab13556; Abcam, Invitrogen, OR, USA), ERK1/2 (1/1000, ab17942; Abcam), p-ERK1/2 (1/1000, ab214362; Abcam), JNK (1/1000, ab179461; Abcam), p- JNK (1/1000, ab124956; Abcam), RUNX2 (1/1000, ab23981; Abcam), OPN (1/1000, ab8448; Abcam), OCN (1/500, ab93876; Abcam), primary antibodies overnight at 4°C. Membranes were washed three times for 15 minutes with TTBS followed by incubation for 2 hours at room temperature with secondary antibodies. After three more washes for 15 minutes with TBST, specific staining was detected using the chemiluminescence (ECL) system (VWR International GmbH, Darmstadt, Germany). All bands were densitometrically analysed with ImageJ.

2.11 Cell transfection

The miR-145-5p mimics, si-PACT1 and si-TLR4 were synthesized by Genomeditech (Shanghai, China). Oligonucleotide transfection was conducted using Lipofectamine 2000 transfection reagent (Invitrogen) following manufacturer's recommendations. Forty-eight hours after transfection, the cells were collected for further investigations.

2.12 Alkaline phosphatase staining

The expression of alkaline phosphatase (ALP) in the cell layers was assessed using a kit according to the manufacturer's instructions (Yeasen, Shanghai, China). Fourteen days after osteogenic induction, the hADSCs were rinsed three times with PBS three times and fixed with 70% alcohol for 30 minutes. The fixed cells were soaked in BCIP/NBT solution, washed with ddH2O, and then observed with a scanner (ImageScanner III, GE Healthcare Bio-Sciences Corp., Piscataway, NJ, USA). Each assay condition was repeated in triplicate. All the results were repeated in three independent experiments.

2.13 Alizarin red S staining

Alizarin red S staining was used to detect matrix mineralization. Fourteen days after osteogenic differentiation, the cultured cells were treated with 75% ethanol for 20 minutes, and then stained with 1% Alizarin Red S pH 4.2 (ShangHai Haoran Biological Technology, Shanghai, China) for 1 hour at 37°C for the semi-quantitative evaluation of the degree of mineralization.

2.14 Dual-luciferase reporter assay

The PCAT1 and TRL4 sequences containing the predicted potential miR-145-5p binding sites were amplified. The PCAT1-WT and TRL4-WT plasmids were constructed using a PCR method. PCAT1-MUT and TLR4-MUT were generated by site-directed mutagenesis, replacing the first six ribonucleotides of the miR-145-5p complementary sequence. HEK293T cells (BeNa Culture Collection, China) were grown in 24-well plates and cotransfected with miR-145-5p mimics and PCAT1-WT, miR-145-5p mimics and PCAT1-MUT, control mimics and PCAT1-WT, control mimics and PCAT1-MUT, miR-145-5p mimics and TLR4-WT, miR-145-5p mimics and TLR4-MUT, control mimics and TLR4-WT, control mimics and TLR4-MUT. The media was replaced after 6 hours; 24 hours after cotransfection, the Dual-Luciferase Reporter Gene Test Kit (Promega, Fitchburg, WI, USA) was used to detect the luciferase activities in different groups. The firefly and the renal luciferase reagent were added to detect the luciferase activity of each group.

2.15 Statistical analysis

Numerical data were expressed as the mean ± SD. Student's t test rendered statistical comparisons. The criterion for significance was set to as P < 0.05.

3.1 The profiling of dysregulated mRNAs, lncRNAs, and activated toll like receptor signalling pathway during osteogenic differentiation

Figure 1A and B showed 20 selected mRNAs and lncRNAs, respectively, that demonstrated the most significant differences in their expression profiles during osteogenic differentiation, with 10 ranked at the top and 10 at the bottom sorted by their fold-change value. As shown in Figure 1C and D, the pathways which were significantly biased were demonstrated by ridgeplot and dotplot based on the GSEA analysis results. The Gseaplot result showcased that the TLR signalling pathway was activated (Figure 1F). Furthermore, TLR4 was highly expressed in TLR signalling pathway, which was shown in Figure 1E. All results demonstrated that TLR signalling pathway was activated in osteogenic differentiation. Therefore, these results provided us with insights that led us to focus on TLR signalling pathway in the next study.

3.2 The putative targeted relationship between lncRNA PCAT1, miR-145-5p, and TLR4

An enrichment map was implemented to indicate those pathways containing cross-referencing genes. The TLR signalling pathway was marked with dark red, which denoted it was highly enriched. The connections showed overlapping parts, indicating that the TLR signalling pathway contained genes that cross-referenced with four other pathways (Figure 2A). The gene coexpression network in Figure 2B displays the correlation relationship between lncRNA-PCAT1 and TLR4. Constructed using the Pearson correlation coefficient, mRNAs contained in TLR signalling pathway and lncRNAs that were linearly related to mRNAs were screened as the nodes in the network depending on their expression level vairations. The results were visualized via Cytoscape 3.6.5. Considering all the miRNAs that targeted TLR4, PCAT1, and the TLR signalling pathway together, 13 miRNAs were identified by the Veen diagram. We choose miR-145-5p, which was included among these thirteen miRNAs, for further study (Figure 2C). The PCAT1, miR-145 and TLR4 binding sites as well as the putative binding relationship were illustrated in Figure 2D.

3.3 The morphology, propagation, and characterization of passage 4hADSCs

The seeded hADSCs began to grew by static adherence in 24 hours. The primary cells were relatively single, short shuttle-like, or round. While after three times of passage, cell grew into spindle-shaped, fibroblast-like appearance cells with spherical or orbicular-ovate nucleus under high power lens (×100), with rapid proliferation and swirl distribution (Figure 3A). A linear relationship between absorbance and density of live cells in growth period was displayed (Figure 3B). Furthermore, CCK-8 assay was performed to detect the proliferation of hADSCs, and the growth curve was draw for population doublings calculation. As shown in Figure 3C, 4-7 days were logarithmic phase, the population doubling was 25.38 hours calculated according to the standard curve and growth curve. In addition, FACS was used to characterize the hADSCs. Our results showed that the cell surface markers CD29 and CD44 were highly expressed and that the surface markers CD34 and CD45 were minimally expressed in hADSCs (Figure 3D).

3.4 The expression of lncRNA-PCAT1, miR-145-5p, and TLR4 were changed in hADSCs after Osteoblast differentiation induction

The changes in lncRNA PCAT1, miR-145-5p, and TLR4 mRNA expression in hADSCs were detected by qRT-PCR, after the induction of osteoblast differentiation induction for 0, 7, 14, and 21 days. As shown in Figure 4A, following the induction of osteoblast differentiation, lncRNA PCAT1 expression was up-regulated compared with 0 day, exhibiting a significant increase until 14 and 21 days after the induction of osteogenic differentiation. However, miR-145-5p expression showed the opposite tendency (Figure 4B); after induced, its expression decreased. Additionally, after induction for 14 and 21 days, the decrease was notable. The results for TLR4 mRNA were similar to lncRNA-PCAT1. Until 14 and 21 days after induced osteogenic, the expression of TLR4 was remarkable (Figure 4C). TLR4 protein expression was also detected by Western blot (Figure 4D). TLR4 protein expression was positively correlated with mRNA expression.

3.5 The correlation between lncRNA-PCAT1, miR-154-5p, and TLR4 in osteoporosis

LncRNA-PCAT1, miR-154-5p, and TLR4 expression were detected in eight nonosteoporotic samples and eight osteoporotic samples. PCAT1 expression was decreased in the eight osteoporotic samples compared with the nonosteoporotic samples (Figure 5A). TLR4 expression was also decreased in the eight osteoporotic samples compared with nonosteoporotic samples (Figure 5C). However, the expression of miR-145-5p revealed the opposite results. The expression of miR-145-5p was higher in nonosteoporotic samples (Figure 5B). Statistical analysis showed the correlation between lncRNA-PCAT1, miR-145-5p, and TLR4. There was a significantly negative correlation between miR-145-5p and PCAT1 expression (Figure 5D). TLR4 and miR-145-5p expression also revealed a significantly negative correlation (Figure 5F). Moreover, there is a significantly positive correlation between TLR4 and miR-145-5p expression (Figure 5E).

3.6 The targeted relationship between lncRNA-PCAT1, miR-145-5p, and TLR4

The luciferase reporter gene assay was conducted in HEK293T cells for verifying their binding relationship. The result revealed that the luciferase activity of cells cotransfected with miR-145-5p and PCAT1-WT was significantly inhibited, whereas miR-145-5p had no such impact on the PCAT1-MUT fluorescence (Figure 6A). Similar results are shown for the relationship between miR-145-5p and TLR4 mRNA (Figure 6B). The luciferase activity of cells cotransfected with miR-145-5p and TLR4-WT was significantly reduced, while the luciferase activity of cells cotransfected with miR-145-5p and TLR4-MUT was practically similar. These results indicated that lncRNA-PCAT1 and miR-145-5p as well as miR-145-5p and TLR4 mRNA directly regulated one another.

To further demonstrated the relationship between lncRNA-PCAT1, miR-145-5p and TLR4 mRNA, we transfected si-PCAT1 plasmids and miR-145-5p inhibitors into hADSCs and observed the effects on PCAT1, miR-145-5p, and TLR4 expression. The results revealed that si-PCAT1 decreased PCAT1 (Figure 6C) and TLR4 mRNA (Figure 6E) expression compared with the mock and si-NC groups,whereas miR-145-5p was up-regulated (Figure 6D). MiR-145-5p was decreased after the transfection of the miR-145-5p inhibitors (Figure 6F), while PCAT1 and TLR4 expression showed completely different results. TLR4 expression was increased (Figure 6G) and PCAT1 showed barely any change in expression (Figure 6H) after transfected miR145-5p inhibitor, compared with mock and mimics NC groups. Taken together, these results revealed that lncRNA-PCAT1 could negatively regulate miR-145-5p and positively regulate TLR4. What's more, miR145-5p also negatively regulated TLR4, but did not regulate lncRNA-PCAT1.

3.7 LncRNA-PCAT1, miR-145-5p, and mRNA TLR4 affected osteogenic differentiation

The transfection of si-PCAT1 and si-TLR4 into hADSCs led to a simultaneous in TLR4 mRNA expression level and TLR4 protein levels (Figures 7A and B). PCAT1 and TLR4 knockdown remarkably inhibited ALP levels, while miR-145-5p inhibitors increased ALP levels. Mix1 and mix2 recovered ALP levels to normal. The relative ARS levels were similar to ALP levels (Figures 7C–F), Western blot detected the expression of OCN, OPN and RUNX2, which are related to osteogenic differentiation. As shown in Figures 7G and H, compared with the noninduced group, OCN, OPN, and RUNX2 expression levels decreased significantly in the si-PCAT1 and si-TLR4 groups, but were up-regulated in the miR-145-5p inhibitor group.

3.8 LncRNA PCAT1, miR-145-5p, and mRNA TLR4 affected Toll like receptor signalling pathway

The expression level of p-ERK1/2 and p-JNK, downstream proteins in TLR signalling pathway, were observed by Western blot analysis to determine whether PCAT1, miR-145-5p, and TLR4 influence this pathway. The expression of p-ERK1/2/ERK1/2 and p-JNK/JNK were inactivated when PCAT1 or TLR4 was knockdown and were activated when miR-145-5p was knockdown (Figure 8A and B). In contrast, p-ERK1/2/ERK1/2 and p-JNK/JNK expression was barely changed in mix1 group and mix2 group (P > 0.05). Therefore, our data suggested that LncRNA-PCAT1 sponged miR-145-5p to promote TLRA associated osteogenic differentiation of hADSCs via the activation of TLR signalling pathway (Figure 9A).