Integrative analysis of long non-coding RNA (lncRNA) and mRNA expression in TLR4-primed MSCs of ankylosing spondylitis

Background: Ankylosing spondylitis (AS) is a chronic autoimmune disease, and the precise pathogenesis is largely unknown at present. Our previous study found that the expression of toll-like receptor 4 (TLR4) of mesenchymal stem cells from AS patients (AS-MSCs) was reduced and activation of TLR4 by lipopolysaccharide (LPS) could enhance the immunoregulatory ability of AS-MSCs. However, the potential mechanism by which TLR4 affect the immunoregulatory function of AS-MSCs remains unclear. Objective: The goal of this study was to explore the expression proles and functional networks of lncRNAs and mRNAs in TLR4-primed AS-MSCs and to clarify the potential mechanisms by which TLR4-primed AS-MSCs exert immunoregulatory effects. Methods: Immunoregulatory effects of MSCs were determined after TLR4 activation. Then, the differentially expressed (DE) long non-coding RNAs (lncRNA) and messenger RNAs (mRNA) between AS-MSCs and TLR4-primed AS-MSCs (stimulated by LPS) were identied through high-throughput sequencing followed by qRT-PCR conrmation. Finally, bioinformatic analyses were performed to identify the critical biological functions, signalling pathways and associated functional networks involved in the TLR4-primed immunoregulatory function of AS-MSCs. Results: A total of 147 DE lncRNAs and 698 DE mRNAs were identied between TLR4-primed AS-MSCs and unstimulated AS-MSCs. Of total, 107 lncRNAs were upregulated and 40 were downregulated (fold change ≥ 2, P <0.05), while 504 mRNAs upregulated and 194 downregulated (fold change ≥ 2, P <0.05). 5 lncRNAs and 5 mRNAs with largest fold changes were respectively veried by qRT-PCR. GO and KEGG analysis demonstrated that the DE mRNAs and lncRNAs were highly associated with the inammatory response, such as NOD-like receptor (NLR) signalling pathway, the TNF signalling pathway and the NF-kappa B signalling pathway. Cis-regulation prediction revealed 8 novel lncRNAs while trans-regulation prediction revealed 15 lncRNAs, respectively. 8 core pairs of LncRNA and target mRNA in the lncRNA-TF-mRNA network were: PACERR-PTGS2, LOC105378085-SOD2, LOC107986655-HIVEP2, MICB-DT-MICB, LOC105373925-SP140L, LOC107984251-IFIT5, LOC112268267-GBP2 and LOC101926887-IFIT3, respectively. Conclusion: In AS, TLR4 activation can enhance the immunoregulation ability of MSCs. Eight core pairs of lncRNA and target mRNA have been found in TLR4-primed AS-MSCs, which could contribute to elucidate the potential mechanism of immunoregulatory dysfunction of AS-MSCs. The current study presents an integrative analysis of lncRNA-mRNA expression proles and functional networks involved in the TLR4-primed immunoregulation of AS-MSCs. These results improve our understanding of the roles of lncRNAs in the immunoregulatory ability of AS-MSCs and could provide potential targets to improve the curative effect of MSCs on DE PCC the co-expression between 147 DE lncRNAs and 698 DE mRNAs total of 1,072 DE lncRNA-DE mRNA co-expression pairs obtained an absolute PCC value ≥ 0.85 and P < 0.05. Among these pairs, 706 lncRNA-mRNA pairs were identied as being co-expressed, whereas 366 lncRNA-mRNA pairs were found to be negatively co-expressed high-throughput and expression and functional of TLR4-primed AS-MSCs. ndings by qRT-PCR. KEGG pathway that some key pathways, such as the NF-kappa B and TLR signalling pathways, might contribute to the immunoregulatory function of TLR4-primed AS-MSCs. In addition, we obtained novel ndings by bioinformatic analyses of DE transcripts, including identication of the most signicantly altered GO categories, construction of a co-expression network for lncRNA function prediction, and cis- and trans-regulation predictions of lncRNAs. Our results provide a model that can be used to explore the roles of lncRNAs and mRNAs in the immunoregulatory mechanisms of TLR4-primed AS-MSCs. expression proles reect behaviours of pathway signicant TLR4-primed AS-MSCs unstimulated AS-MSCs. the TNF the NF-kappa B signalling pathway, cytokine-cytokine receptor interaction and the TLR activation of TLRs activate NF-κB and MAPK signalling pathways secretion of pro-inammatory cytokines, IL-6, IL-12, TNF-α and type I IFNs, drive inammation in [32]. ndings, TNF-α, IL-6 IL1β in AS-MSCs after TLR4 and MAPK signalling pathways identied by microarray analysis. In addition, as pathogen recognition receptors, both TLR and NLR activate pathways mediated through different adaptor proteins that are commonly found to activate NF-κB [33]. The NF-κB-mediated activation of MSCs leads to the secretion of TNF-α and other cytokines. Elevated pro-inammatory cytokine levels are one of the main manifestations of AS, as conrmed by previous research. Our results further conrmed the signicant role of TLR4 in the pathology of AS. The selected top DE mRNAs and lncRNAs included in the NF-κB pathway might be possible upstream targets of pathological inammation in AS. with of the DE lncRNA in TLR4-primed AS-MSCs. co-expressed DE lncRNA-DE mRNA networks visualized using Cytoscape 3.0


Introduction
Ankylosing spondylitis (AS) is a chronic in ammatory rheumatic disease characterized by in ammatory back pain and asymmetrical peripheral arthritis [1]. Previous studies have shown that AS is closely related to immune dysfunction. However, the pathogenesis of AS is largely unknown.
Mesenchymal stem cells (MSCs) are a group of self-renewing cells that have a signi cant immunomodulatory ability that allows them to regulate T cell proliferation and differentiation and inhibit dendritic cell (DC) maturation [2,3]. According to recent studies, abnormal immunoregulation by MSCs can lead to several autoimmune diseases, such as immune thrombocytopenia and systemic lupus erythematosus (SLE) [4][5][6]. In our previous study, we found the reduced immunoregulatory ability of MSCs from AS patients may participate a signi cant role in the pathogenesis of this disease [7]. Furthermore, our clinical trial demonstrated that intravenous infusion of MSCs from healthy donors (HDs) was a feasible, safe, and effective approach for the treatment of AS [8]. Our research results indicate that it is essential to elucidate the mechanism of abnormal immunoregulatory function of AS-MSCs. However, our understanding of the immunoregulatory function of AS-MSCs is still in its infancy, and further characterization and identi cation of key factors regulating these properties are still needed.
Toll-like receptors (TLRs), the most characterized pattern recognition receptors (PRRs), which are capable of potently activating different cell types, could be highly expressed on most immune cells, as well as other cell types, including chondrocytes, endothelial cells, and broblasts [9]. Their downstream signalling pathways lead to the production of a wide range of immune stimulatory cytokines and chemokines [10]. Aberrant activation of TLRs may result in unrestricted in ammatory responses, suggesting that the family of TLRs may play a pivotal role in the development of autoimmune diseases [11]. TLRs have also been demonstrated to play important roles in regulating the immunomodulatory properties of MSCs [12]. Our previous study found that the expression of TLR4 in MSCs was downregulated in AS patients compared to healthy donors. The inhibitory effects of MSCs on CD4 + T cell proliferation in AS were enhanced by stimulation by lipopolysaccharide (LPS), which is the speci c ligand of TLR4. It has been suggested that TLR4 plays a signi cant role in regulating the immunomodulatory ability of MSCs in AS [13]. Liotta et al. found that ligation of TLR4 suppressed the inhibitory effects of human bone marrow (BM)-MSCs on T cell proliferation by downregulating Jagged-1 expression [14]. Waterman et al. demonstrated that MSCs could be primed towards a pro-in ammatory phenotype after TLR4 activation [15]. In contrast, Opitz et al. reported that TLR4 enhanced the immunosuppressive properties of human BM-MSCs by directly inducing indoleamine 2,3-dioxygenase 1 (IDO1) [16]. Inconsistent results have been reported in recent years, the speci c immunoregulatory mechanisms by which TLR4 controls MSC immunoregulation remains unclear and must be addressed.
Noncoding RNAs (ncRNAs) are a class of non-protein-coding RNA and functional RNA molecules that include microRNA (miRNA), long noncoding RNAs (lncRNA) and circular RNAs (circRNA) [17]. Overwhelming evidence has indicated that various ncRNAs are implicated in human disease process. For example, miRNAs are the important molecules of various diseases [18]. In recent years, lncRNAs (> 200nt RNA molecules) have been found and studied to play a role in development, evolution and disease. LncRNAs are important epigenetic regulators and thus participate in crucial roles in various cell biology behaviours [19]. For example, LncRNA-PCAT1 negatively regulated miR-145-5p, which promoted TLR4 expression to promote osteogenic differentiation by activating the TLR signalling pathway in human adipose-derived stem cells [20]. Speci cally, lncRNAs are widely involved in the regulation of immune system homeostasis. LncRNA NEAT1 promotes in ammatory response in sepsis-induced liver injury via the Let-7a/TLR4 axis [21]. LPS-induced lncRNA Mirt2 can function as a checkpoint to prevent aberrant activation of in ammation, and is a potential regulator of macrophage polarization [22]. LncRNA MALAT1 regulates in ammatory cytokine production in lipopolysaccharidestimulated human gingival broblasts through sponging miR-20a and activating TLR4 pathway [23]. However, the molecular mechanism by which lncRNAs participate in the immunoregulatory function of MSCs, particularly in AS, is still unclear.
The current study presents an integrative analysis of lncRNA-mRNA expression pro les and functional networks involved in the TLR4-primed immunoregulation of AS-MSCs. These results improve our understanding of the roles of lncRNAs in the immunoregulatory ability of AS-MSCs and could provide potential targets to improve the curative effect of MSCs on AS.

Results
The effect of TLR4 activation on MSCs is time and dose dependent To investigate whether the activation of TLR4 in MSCs can affect the immunoregulatory ability of these cells, MSCs from AS patients were prestimulated with LPS before being co-cultured with PBMCs. To determine the best stimulation time and concentration for the TLR4 ligand used, the level of p38 phosphorylation was examined by western blotting. AS-MSCs were rst exposed to LPS at a concentration of 1 µg/ml for the indicated times (0, 2, 4, 8, 12 and 24 hours) and then treated with three different concentrations (0, 0.1, 1 and 10 µg/ml) of stimuli for the previously selected time. The upregulation of the phospho-p38 level was highest at 4 hours with the LPS concentration of 1 µg/ml and declined thereafter (Fig. 1A).
TLR4-primed AS-MSCs demonstrate an enhanced inhibitory effect on CD4 + T cell proliferation Previous reports have shown that co-culture of unprimed MSCs with PBMCs can inhibit PBMC proliferation and/or activation [24]. Thus, we sought to assess the potential in uence of TLR4 activation on the immunoregulatory effect of MSCs derived from AS patients. To this end, MSC-PBMC co-culturing was conducted with CFSE-labelled PBMCs (responder cells) co-cultured with unprimed or TLR4-primed MSCs (effector cells) for 5 days, after which time the proliferating responders were sorted by ow cytometry for CD4 positivity and then gated on CFSE expression. As shown in Fig. 1C, 72.8% of the CD4 + T cells underwent proliferation when cultured without MSCs, but this proportion was reduced to 46.3% when cultured with AS-MSCs. Activation of TLR4 with 1 µg/ml LPS signi cantly enhanced the immunoregulatory effect of AS-MSCs, which reduced CD4 + T cell proliferation to 40.3% (p < 0.05) (Fig. 1B). qRT-PCR results suggested that the expression of several cytokines and chemokines (TNF-α, CXCL-9, PDL1, IL-1β, IL-6 and iNOS) was strengthened after stimulation with LPS (Fig. 1C).

Identi cation of DE mRNAs and lncRNAs
A total of 698 mRNAs were DE in TLR4-primed MSCs compared to unprimed MSCs from AS patients. Among these genes, 594 mRNAs were upregulated, and 104 mRNAs were downregulated. The DE mRNAs are depicted using a clustergram ( Fig. 2A) and volcano plots (Fig. 2C). The 20 mRNAs with the largest fold changes are shown in Table 1. Several immunoregulatory cytokines and chemokines, such as CXCL10, CXCL11, IDO1, CXCL8, CXCL1, CCL20, IL6 and SOD2, which are secreted by MSCs and play important roles in regulating immunocytes, were included in this list. A total of 147 lncRNAs, including 107 upregulated and 40 downregulated lncRNAs, were differentially expressed in TLR4-primed MSCs compared to unprimed MSCs from AS patients. The DE lncRNAs are depicted in a clustergram (Fig. 2B) and volcano plots (Fig. 2D). The 10 lncRNAs with the largest fold changes are shown in Table 2. Validation of DE mRNA and lncRNA expression levels To con rm the RNA-seq results, several important DE mRNAs and lncRNAs were assessed by qRT-PCR. We found that the expression of mRNAs (CXCL1, CXCL8, CXCL10, CXCL11 and CCL20) and lncRNAs (MIR3142HG, LOC105371619, LOC105374444, PACERR and LOC105375914) was signi cantly upregulated in TLR4-primed AS-MSCs compared to unstimulated AS-MSCs (P < 0.05) ( Fig. 3A and 3B). All qRT-PCR results were consistent with the RNA-seq results, con rming the reliability of the sequencing data.

GO and KEGG analyses
We performed GO analysis of the DE mRNAs and lncRNAs. The top 10 GO terms related to biological processes, cellular components and molecular functions are provided in Fig. 4A  were identi ed as being positively co-expressed, whereas 366 lncRNA-mRNA pairs were found to be negatively co-expressed ( Fig. 5B).

Cis-regulation prediction of DE lncRNAs
Cis-regulation, which regulates the transcription of nearby genes located on the same chromosome, is vital for gene expression. A total of 8 lncRNA transcripts and their predicted cis-regulated protein-coding genes were identi ed in the top 20 cis-regulated genes (Fig. 6A).
LOC107986655 was predicted to cis-regulate HIVEP2, PACERR was predicted to cis-regulate PTGS2, MICB-DT was predicted to cis-regulate MICB, LOC105373925 was predicted to cis-regulate SP140L, LOC107984251 was predicted to cis-regulate IFIT5, LOC112268267 was predicted to cisregulate GBP2, LOC101926887 was predicted to cis-regulate IFIT3, and LOC105378085 was predicted to cis-regulate SOD2. These networks may provide valuable clues about these lncRNAs and their nearby coding genes in the development of AS.

Trans-regulation prediction of DE lncRNAs
One of the important mechanisms by which lncRNAs function is by participating in particular pathways regulated by TFs. A top 500 lncRNA-TF network, which showed that 15 lncRNAs participate in pathways regulated by TFs, was constructed to provide key data for subsequent research

Discussion
In our present research, we utilized high-throughput sequencing followed by bioinformatic analysis to analyse the mRNA and lncRNA expression pro les and functional networks of TLR4-primed AS-MSCs. These ndings were then con rmed by qRT-PCR. KEGG pathway analysis indicated that some key pathways, such as the NF-kappa B and TLR signalling pathways, might contribute to the immunoregulatory function of TLR4primed AS-MSCs. In addition, we obtained novel ndings by bioinformatic analyses of DE transcripts, including identi cation of the most signi cantly altered GO categories, construction of a co-expression network for lncRNA function prediction, and cis-and trans-regulation predictions of lncRNAs. Our results provide a model that can be used to explore the roles of lncRNAs and mRNAs in the immunoregulatory mechanisms of TLR4-primed AS-MSCs.

MSCs are one of the most important immunoregulatory cell types and regulate the functions of many immune cells, including T cells, B cells and
DCs [25][26][27]. Abnormal immunoregulation by MSCs can lead to several autoimmune diseases [4]. Moreover, MSCs exert considerable therapeutic effects on several autoimmune diseases owing to their multilineage differentiation potential and highly immunoregulatory properties [28,29]. In our previous study, we found that impairment in the immunoregulatory functions of MSCs played a key role in the pathogenesis of AS [7].
Additionally, our clinical trial study demonstrated that infusion of MSCs isolated from healthy individuals is a safe and e cient method for the treatment of AS [8].
Accumulating evidence suggests that TLR activation can modulate the immunoregulatory functions of MSCs [14][15][16]. In addition, emerging evidence further suggests a role for TLRs in the pathogenesis of spondyloarthropathies, including AS [30]. According to our previous study, the expression of TLR4 was downregulated in MSCs derived from AS patients, and compared with MSCs from healthy donors, TLR4-primed AS-MSCs possessed enhanced immunoregulatory effects limiting the proliferation of naive CD4 + T cells [31]. However, the precise mechanism underlying the enhanced immunoregulatory ability of TLR4-primed AS-MSCs remains unclear. Therefore, we measured the differential expression pro les of lncRNAs and mRNAs in AS-MSCs after TLR4 activation to identify the regulatory network of lncRNAs and mRNAs in these cells. The results showed that there were 698 DE mRNAs and 147 DE lncRNAs in TLR4-primed AS-MSCs compared with unstimulated cells. The top 5 mRNAs and lncRNAs, which may be involved in the regulatory dysfunction of AS-MSCs, were veri ed by qRT-PCR.
The mRNA expression pro les re ect the biological behaviours and functions of cells. In this study, KEGG pathway analysis revealed that 75 signalling pathways exhibited signi cant differences between TLR4-primed AS-MSCs and unstimulated AS-MSCs. Among these pathways, the NLR signalling pathway, the TNF signalling pathway, the NF-kappa B signalling pathway, cytokine-cytokine receptor interaction and the TLR signalling pathway were prominent. Recent studies indicate that the activation of TLRs can activate NF-κB and MAPK signalling pathways to promote the secretion of pro-in ammatory cytokines, such as IL-6, IL-12, TNF-α and type I IFNs, which drive in ammation in AS [32]. Consistent with these ndings, we found that the expression of TNF-α, IL-6 and IL1β in AS-MSCs was signi cantly increased after TLR4 activation, which supported the crucial roles for TLR4 in the NF-κB and MAPK signalling pathways identi ed by microarray analysis. In addition, as pathogen recognition receptors, both TLR and NLR activate pathways mediated through different adaptor proteins that are commonly found to activate NF-κB [33]. The mRNA expression pro les are under the control of a series of epigenetic regulators, of which lncRNAs are an important component [19,34]. In recent years, an increasing number of lncRNAs have been reported to perform key roles in the pathogenesis and development of AS. For example, lncRNA-AK001085 expression was found to be downregulated in AS patients, which served as a potential diagnostic indicator; thus, this lncRNA is considered a potential suppressor of AS [35]. Our previous microarray study identi ed four lncRNAs (lnc-ZNF354A-1, lnc-LIN54-1, lnc-FRG2C-3 and lnc-USP50-2) that are involved in the abnormal osteogenic differentiation of AS-MSCs [36]. However, the immunoregulatory function of AS-MSCs regulated by lncRNAs has not been explored. Our research identi es the lncRNA expression pro le in an in ammatory environment based on previous studies, which provides a possible way to further explore the regulatory function of lncRNAs in AS. To the best of our knowledge, this study is the rst to use microarray analyses to examine the roles of lncRNAs in TLR4-primed AS-MSCs.
In this study, several lncRNAs with the largest fold changes among DE lncRNAs were studied. For example, GBP1P1 is a lncRNA that acts as a prognostic biomarker for hepatocellular carcinoma [37]. MIR3142HG can regulate the in ammatory response following IL-1β-mediated activation of human lung broblasts, which is a positive regulator of IL-8 and CCL2 release [38]. The in ammatory response regulation by MIR3142HG indicates that it may contribute to the enhanced immunoregulatory ability of TLR4-primed AS-MSCs and the immunoregulatory dysfunction seen in AS. Unfortunately, most of the DE lncRNAs, such as LOC101926887, LOC105378410, and LOC107984251 have not been studied yet. Further exploration is needed in the future.
Regulatory lncRNAs act in a cis and/or trans manner to in uence or interact with nearby or distant genes [39]. In our study, 8 lncRNA transcripts were predicted to cis-regulate nearby protein-coding genes. In addition, we predicted the functions of trans-regulatory lncRNAs using TFs that Antisense NFKB1 Complex-Mediated Expression Regulator RNA, also known as PACER) is a novel long noncoding RNA that has been found to interact with NF-kB transcriptional regulators to promote expression of prostaglandin-endoperoxide synthase 2 (PTGS2, cyclooxygenase-2, also known as COX2) [40]. COX2 could produce prostaglandin E2 (PGE2) and amplify the Th17 mediated autoimmune process through COX2-PGE2-EP2/EP4-NF-κB loop [41]. Targeting COX2 activity with non-steroidal anti-in ammatory drugs (NSAIDs) or agents designed to speci cally block COX-2 activity have been approved for therapeutic use [42]. Of note, NSAIDs are recommended as rst-line treatment for AS patients for their high e ency in reducing back pain and stiffness in patients with axial spondyloarthritis [43]. The pair of PACERR and PTGS2 may participate a key role in the immunuregulation in TLR4-primed AS-MSCs, which maybe an important potential biomark for diagnosis and target for drug therapy. SOD2 is a component of antioxidant defence systems, which are crucial in defending cells against oxidative stress. SOD2-overexpressing BM-MSCs have an enhanced therapeutic effect on brain injury treatment in traumatic brain injury mice [44]. Moreover, MSCs could enhance the expression of the SOD2 antioxidant gene to adapt to the oxidative environment and exert their therapeutic effect [45]. Exposure to LPS induces oxidative stress in AS-MSCs, LOC105378085 may regulate the expression of SOD2 in AS-MSCs to defend cells against oxidative stress and to exert the immunoregulatory effects of the AS-MSCs. Finally, the above mentioned 23 DE lncRNA and 8 lncRNA-target mRNA pairs identi ed in the present study provide novel information for understanding the biological functions of lncRNAs in AS-MSCs, but the underlying mechanisms of how "lncRNA-TF-target gene" networks affect TLR4-induced immunoregulation remain to be clari ed.
Our study has several limitations. First, RNA-seq is an important method to screen possible lncRNAs and mRNAs associated with speci c diseases, but the results of big-data analyses may include false positives. Therefore, we performed qRT-PCR to further verify differential expression. Second, we predicted lncRNA functions only indirectly using bioinformatic analysis and validated several DE lncRNAs. Further functional studies on the mechanism are warranted to clarify the roles of lncRNAs.

Conclusion
In AS, TLR4 activation can enhance the immunoregulatory ability of MSCs in AS. This study describes the lncRNA and mRNA expression pro les and functional networks in TLR4-primed AS-MSCs. Eight core pairs of lncRNA and target mRNA in the lncRNA-TF-mRNA network have been found in TLR4-primed AS-MSCs. These results provide insight into the pathogenesis of immunoregulatory dysfunction in AS-MSCs, which may help to elucidate possible molecular mechanisms and therapeutic targets in AS.

AS-MSC isolation and cell culture
MSCs were isolated from BM aspirates taken from AS patients who provided informed consent through density gradient centrifugation, as described in our previous study [46]. After density gradient centrifugation, MSCs were isolated through plastic adherence and grown at 37 °C in an atmosphere of 5% CO2 for one week. The MSCs were trypsinized when the cultures reached 80-90% con uence. MSCs in passages 3-5 were used in subsequent experiments. MSCs were identi ed on the basis of immunological phenotypes and the triple-lineage differentiation capability, as previously described [46]. After identifying MSC immunophenotypic markers by ow cytometry, cells in passages three to ve were used for subsequent experiments.

Pre-stimulation of TLR4 on AS-MSCs
To

Expression Analysis
Feature Extraction software (version 10.7.1.1, Agilent Technologies) was used to analyse array images to obtain raw data. GeneSpring (version 14.8, Agilent Technologies) was employed to complete the basic analysis with the raw data. First, the raw data were normalized with the quantile algorithm. The probes with at least 1 condition out of 2 conditions having ags in "Detected" were chosen for further data analysis. DE genes were then identi ed through fold change data as well as the P value calculated with a t-test. The threshold set for up-and downregulated genes was a fold change ≥ 2.0 and a P value ≤ 0.05. Afterwards, gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were applied to determine the roles of these differentially expressed (DE) mRNAs. Finally, hierarchical clustering was performed to display the distinguishable gene expression patterns among samples.

qRT-PCR validation
To validate the reliability of high-throughput RNA-seq and explore the expression trends of mRNAs and lncRNAs, we performed quantitative realtime PCR (qRT-PCR) for biological validation. Total RNA was isolated from AS-MSCs with or without LPS stimulation using TRIzol according to the manufacturer's protocol. cDNA was transcribed using a PrimeScript RT reagent kit (Takara, Otsu). qRT-PCR was then performed, and the data were analysed using the 2 − ΔΔCt method. The primer sequences used in the qRT-PCR assay are provided in Supplementary Table S1. Co

Cis-and trans-regulation predictions of DE lncRNAs
It has been suggested that lncRNAs regulate gene expression through both cis-and trans-regulation. For cis-regulation prediction, we identi ed each paired lncRNA and mRNA by the following procedures: (1) the mRNA loci were within 100-kb windows upstream or downstream of the given lncRNA, and (2) the Pearson correlation of lncRNA-mRNA expression was signi cant (P ≤ 0.05). For trans-regulation prediction, we enriched the coexpressed mRNAs with DE lncRNAs that signi cantly overlapped with the host genes of transcription factors (TFs). Using the threshold P < 0.05, each lncRNA could be connected with one to more than a dozen TFs, and each pair of lncRNA-TF was the result of several gene enrichments based on the hypergeometric cumulative distribution function. Then, we constructed the lncRNA-TF-mRNA network using Cytoscape software.

Statistical analysis
Data are expressed as the mean ± standard deviation (SD) and were analysed using the statistical software package SPSS16.0. Pearson correlation was used in lncRNA-mRNA co-expression analyses. A p value < 0.05, fold enrichment > 2, and log 2 FC > 1 were considered statistically signi cant.

Availability of data and materials
All data generated or analysed during this study are included in this published article and its Additional les.

Competing interests
The authors declare that they have no competing interests.   Figure 1 The immunoregulatory function of TLR4-primed AS-MSCs. A: To determine the best duration and concentration of TLR4 ligand pre-stimulation, we examined the level of p38 phosphorylation by western blotting. AS-MSCs were exposed to LPS at four different concentrations (0 µg/ml, 0.1 µg/ml, 1 µg/ml, and 10 µg/ml) for 4 hours or treated at the concentration of 1 µg/ml for the indicated time (0 hours, 2 hours, 4 hours, 8 hours, 12 hours or 24 hours). The upregulation of the phospho-p38 level was highest at 4 hours with the concentration of 1 µg/ml LPS and declined thereafter. B: AS-MSCs were pre-stimulated with or without 1 µg/ml LPS for 4 hours and then co-cultured with PBMCs at a ratio of 1:10 (MSCs: PBMCs) for 5 days. All PBMCs were then collected for assessment by ow cytometry to determine the positive percentage of CFSE-diluted cells (gated) to evaluate proliferation. AS-MSCs inhibited the proliferation of PBMCs, and this effect was strengthened by the activation of TLR4. C: The gene expression of cytokines and chemokines in AS-MSCs after LPS stimulation was detected by qRT-PCR. The symbol "*" represents P < 0.05.