Long noncoding RNA expression profile in fibroblast-like synoviocytes from patients with rheumatoid arthritis

RA synovial tissue specimens were obtained from the knee joints of 10 patients with RA, who were undergoing total knee arthroplasty (two men and eight women, age range 38–65 years). In addition, normal synovial tissue specimens were obtained from the knee joints of 10 patients with trauma, who were undergoing arthroscopic surgery for knee ligament injury or meniscus injury (three men and seven women, age range 35–61 years). The clinical characteristics of the patients are shown in Additional file 1: Table S1. All patients with RA included in the study fulfilled the 2010 American College of Rheumatology/European league Against Rheumatism (ACR/EULAR) classification criteria for RA [11]. Written informed consent was obtained from each participant prior to sample collection. The study was approved by the Human Ethics Committee of Jinling Hospital.

Tissue specimens were minced into small pieces and digested for 2 hours with 2 mg/ml of type II collagenase (Invitrogen, Carlsbad, CA, USA) in high-glucose Dulbecco’s modified Eagle’s medium (DMEM) at 37 °C [12]. After centrifugation at 210 × g for 5 minutes, the precipitate was resuspended with 1 ml of high-glucose DMEM containing 10 % fetal bovine serum (FBS), 100 units/ml penicillin, and 100 units/ml streptomycin, and then cultured in 25-cm² cell culture flasks (Corning) in a humidified 5 % CO₂ incubator. After 10 hours, 4 ml of high-glucose DMEM containing 10 % FBS was added to the cell culture flask. All experiments were conducted using cells at passage 3.

FLSs at passage 3 were identified by flow cytometry based on the expression of CD68 (a macrophage marker) and CD90 (a fibroblast marker) [13]. Cells were washed three times with phosphate-buffered saline (PBS) and were then incubated with fluorescein isothiocyanate (FITC)-conjugated anti-CD68 antibody, phycoerythrin (PE)-conjugated anti-CD90 antibody, FITC-conjugated mouse IgG2b, or PE-conjugated mouse IgG1 (Miltenyi Biotec, Germany) for 20 minutes in the dark. Cells were washed with PBS and then analyzed on a FACSCalibur flow cytometer (BD Biosciences, San Diego, CA, USA).

Sample labeling and array hybridization were performed according to the Agilent One-Color Microarray-Based Gene Expression Analysis protocol (Agilent Technology). Briefly, RNA was purified using the RNeasy Mini Kit (Qiagen, Germany). Each sample was then amplified and labeled with cyanine-3-CTP. The labeled cRNAs were purified again with the RNeasy Mini Kit. The production of cRNAs needed to reach 1.65 μg to meet the requirements of the microarray. The specific activity of the labeled cRNAs needed to reach 9.0 pmol Cy3/μg cRNA. RNA quantity and quality were measured according to the A260 nm/A280 nm ratio using a NanoDrop ND-1000 spectrometer. RNA integrity was detected by standard denaturing agarose gel electrophoresis. For each microarray, 0.6 μg cRNA, 5 μl of 10× blocking agent, 1 μl of 25× fragmentation buffer, and nuclease-free water were added to reach a total volume of 25 μl: 25 μl of 2× GE Hybridization Buffer was then added to stop the fragmentation reaction. The hybridization solution and Arraystar Human LncRNA Microarray V3.0 were incubated at 65 °C for 17 hours in an Agilent Hybridization Oven. Approximately 30,586 lncRNAs and 26,109 coding transcripts can be detected using the third-generation lncRNA microarray. After washing the chip, a microarray scanner (Agilent DNA Microarray Scanner) was used to measure the fluorescence intensity. Agilent Feature Extraction Software was used to analyze the raw data.

The microarray data were log-transformed and normalized using quantile normalization. After filtering to remove unreliable transcripts, the remaining data were statistically analyzed to identify lncRNAs and mRNAs with significantly differential normalization. Volcano plots are useful tools for visualizing genes expressed differentially between two groups. Transcripts were distributed according to statistical significance (y-axis) and the magnitude of change (log2 ratio of RA FLSs/normal FLSs) (x-axis). Hierarchical cluster analysis was used to identify distinguishable RNA expression profiles between different samples.

Analyzing the genomic context of lncRNAs can help to predict their functional roles. According to the sequence and relative position between lncRNAs and their associated protein-coding genes, the lncRNAs detected by microarray were characterized as natural antisense, intronic antisense, exon sense overlapping, intron sense overlapping, and bidirectional, and intergenic, among others [14]. Natural antisense lncRNAs are RNA molecules that are transcribed from the antisense strand and overlap with coding transcripts. Intronic antisense lncRNAs are RNA molecules that are transcribed from the antisense strand, but do not share overlapping exons. Exon sense overlapping lncRNAs can be considered as transcript variants of protein-coding mRNAs, as they overlap with coding transcripts from the same genomic strand. Intron sense overlapping lncRNAs overlap with the introns of annotated coding genes on the same genomic strand. A bidirectional lncRNA is oriented head-to-head with a protein-coding gene within 1000 bp. Intergenic lncRNAs involve no overlapping or bidirectional coding transcripts near the lncRNAs. The “others” category mainly included lncRNAs for which the associated genes were pseudogenes.

Pathway analysis and Gene Ontology (GO) analysis were applied to explore the potential roles that the differentially expressed mRNAs play in a biological pathway or GO function, including three categories: biological process, cellular component, and molecular function.

A gene co-expression network was built according to the normalized signal intensity of specifically expressed lncRNAs and mRNAs using the Cytoscape program. Pearson correlation analysis was used to evaluate the significance of the correlation between the expression levels between each pair of genes. Pearson correlation coefficients were selected for inclusion in the network when they were above 0.95. By analyzing the function of these mRNAs, we could link lncRNAs with particular functions and signaling pathways, which could facilitate the prediction of their biological functions and mechanisms of action. If the area of the node increased with increasing connections between mRNAs and lncRNAs, this indicated that the lncRNA may play an important role.

The primers used for qPCR were synthesized by Kangcheng (Shanghai, China). Each PCR contained 2 μl template cDNA, 10 μl 2× SYBR Green mix (TaKaRa, Japan), 1 μl forward primer, 1 μl reverse primer, and nuclease-free water to a total volume of 20 μl. The PCR conditions were as follows: 95 °C for 10 minutes, followed by 40 cycles of 95 °C for 10 seconds and 60 °C for 60 seconds. GAPDH mRNA was used as an endogenous control to normalize lncRNA expression levels using the 2^-△△Ct method. The primer sequences used in this study were as follows: for GAPDH, 5′-GGGAAACTGTGGCGTGAT-3′ (forward) and 5′-GAGTGGGTGTCGCTGTTGA-3′ (reverse); for ENST00000483588, 5′-CACGTGAAAGGGGGAGAAA-3′ (forward) and 5′-CCAACAGCACAGAAGGCGT-3′ (reverse); for NR_073012, 5′-ATTCCCACGTATGCGGAGTG3′ (forward) and 5′-ACGGGTGTAGTAGGCGTTTC-3′ (reverse); for ENST00000567753, 5′-CGTGGCAGGACTTTGCTTTC-3′ (forward) and 5′-GGGGTCTTACTGTGTGGGGT-3′ (reverse); for uc003xhp.3, 5′-AGGTATCAGTCTGGGGGACC-3′ (forward) and 5′- ACGCACAGTGGAGGAATGAG-3′ (reverse); for ENST00000557804, 5′- CAGAACGATCAAGCGACCTC-3′ (forward) and 5′- CGTGGAAGGAAGATGACCC-3′ (reverse); for ENST00000438399, 5′- AGAAAGTCAAGGGAAGATAAGG-3′ (forward) and 5′- TAGTCAACCAGGGAAGCAGT-3′ (reverse); for uc004afb.1, 5′- AAACTATGCCTGATTTGTGGTC-3′ (forward) and 5′-CTGGTGTAAGAGCATGGGGT-3′ (reverse); for ENST00000412143, 5′- ATTACTCTTTCCCAGCCCAGC-3′ (forward) and 5′- GCCCTCCTTGCAGCATCAT-3′ (reverse); for ENST00000452247, 5′- GACTCCTCCTCCTGCTTTCAC-3′ (forward) and 5′-GGCAATAGAGCTGGATCTTGT-3′ (reverse).

SPSS software 19.0 and GraphPad Prism 6.0 were used to analyze the data. The two-tailed Student t test and rank-sum test were used as appropriate to analyze the expression levels between two groups. The relationships between the expression levels of lncRNAs and clinical characteristics were analyzed by Pearson’s correlation coefficient. P values below 0.05 were regarded as statistically significant. The diagnostic value was evaluated with a receiver operating characteristic (ROC) curve.

The purity of FLSs at passage 3 was determined by flow cytometry. The purity of the third generation of FLSs of the primary culture reached 99.0 % (Fig. 1).

We performed genome-wide analysis of the expression profiles of lncRNAs and mRNAs in three pairs of FLSs samples from patients with RA and patients with trauma using Arraystar Human LncRNA Microarray V3.0. Based on the criteria of a fold change >2.0 and a P value <0.05, 135 lncRNAs and 103 mRNAs were differentially expressed between the RA FLSs and normal FLSs (Additional files 2 and 3: Tables S2 and S3). Of the 135 lncRNAs identified, 62 lncRNAs were up-regulated and 73 lncRNAs were down-regulated in the RA FLSs. Of the 103 mRNAs detected, 36 mRNAs were up-regulated and 67 mRNAs were down-regulated in the RA FLSs.

To visualize the differentially expressed lncRNAs and mRNAs, volcano plot analysis was conducted to further explore the difference between the RA FLSs and normal FLSs (Fig. 2a and b). The data showed that 135 (0.44 %) of the lncRNAs and 103 (0.39 %) of the mRNAs were significantly differentially expressed. The results of hierarchical cluster analyses showed distinguishable lncRNA and mRNA expression profiles between the RA FLSs and normal FLSs (Fig. 2c and d).

Pathway and GO analyses were applied to explore the potential roles that the differentially expressed mRNAs might play in RA. Pathway analysis showed that the differentially expressed mRNAs in RA FLSs were related with mTOR, HIF-1, Ras, proteoglycans, and apoptosis signaling pathways, among others (Additional file 4: Figure S1). These mRNAs were further characterized as being mainly involved in regulating biological processes related to cell growth, vascular permeability, and cell activation (Additional files 5 and 6: Figures S2 and S3).

Of the 62 up-regulated lncRNAs identified, we observed 4 (6.5 %) natural antisense, 10 (16.1 %) intronic antisense, 4 (6.5 %) exon sense overlapping, 1 (1.6 %) intron sense overlapping, 5 (8.1 %) bidirectional, 35 (56.5 %) intergenic, and 3 (4.8 %) other lncRNAs (Fig. 3a). Of the 73 down-regulated lncRNAs identified, we observed 8 (11.0 %) natural antisense, 8 (11.0 %) intronic antisense, 8 (11.0 %) exon sense overlapping, 3 (4.1 %) intron sense overlapping, 1 (1.4 %) bidirectional, 43 (58.9 %) intergenic, and 2 (2.7 %) other lncRNAs (Fig. 3b).

Considering the observed fold changes, the calculated P values, and the primer specificities, we selected the following nine lncRNAs for further validation of expression levels by qPCR: ENST00000483588, NR_073012, ENST00000567753, uc003xhp.3, ENST00000557804, ENST00000438399, uc004afb.1, ENST00000412143, and ENST00000452247 (Table 1). The qPCR results confirmed that ENST00000483588 was overexpressed in the RA FLSs, whereas the expression levels of ENST00000438399, uc004afb.1, and ENST00000452247 were decreased in the RA FLSs (Fig. 4). These four lncRNAs showed similar trends in qPCR, as determined by microarray. Nevertheless, some lncRNAs showed no significant difference between RA FLSs and normal FLSs in qPCR in contrast to the results determined in the microarray.

We further analyzed the correlation between the expression levels of the lncRNAs ENST00000483588, ENST00000438399, uc004afb.1, and ENST00000452247 and clinical characteristics (erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) expression, Simplified Disease Activity Index (SDAI) score, and age). The expression level of ENST00000483588 was positively correlated with the CRP level (r = 0.74, P < 0.05) and the SDAI score (r = 0.79, P < 0.01) (Fig. 5). However, none of these lncRNAs were significantly correlated with ESR or age.

The potential diagnostic value of four lncRNAs for RA was evaluated using patients with trauma patients as control subjects. The area under the ROC curve for the ENST00000483588, ENST00000438399, uc004afb.1, and ENST00000452247 lncRNAs was 0.85, 0.92, 0.97, and 0.92, respectively, indicating potential diagnostic value for RA (Fig. 6).

An lncRNA and mRNA co-expression network was constructed based on the correlation analysis between the differentially expressed lncRNAs and mRNAs. We further focused on co-expression networks centering on ENST00000483588, ENST00000438399, uc004afb.1, and ENST00000452247. The co-expression networks showed that ENST00000483588 expression was positively correlated with ATAD3A and NDUFA4L2 mRNA expression levels (Additional file 7: Figure S4), ENST00000438399 expression was positively correlated with PTPRQ mRNA expression and negatively correlated with PTHLH mRNA expression (Additional file 8: Figure S5), uc004afb.1 expression was positively correlated with TNFRSF11B mRNA expression (Additional file 9: Figure S6), and ENST00000452247 expression was positively correlated with ZNF154 mRNA expression and negatively correlated with WISP3 mRNA expression (Additional file 10: Figure S7).