- Record: found
- Abstract: found
- Article: found
Microarray profiling of lung long non-coding RNAs and mRNAs in lipopolysaccharide-induced acute lung injury mouse model
research-article
Author(s):
Juan Wang
1 ,
Yong-Chun Shen
2 ,
Zhen-Ni Chen
3 ,
Zhi-Cheng Yuan
2 ,
Hao Wang
2 ,
Da-Jiang Li
1
,
3 ,
Kai Liu
1
,
3 ,
Fu-Qiang Wen
2
Publication date (Electronic):
30 April 2019
Journal:
Bioscience Reports
Publisher:
Portland Press Ltd.
Read this article at
There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.
Abstract
Long non-coding RNAs (lncRNAs) are involved in various biological processes as well as many respiratory diseases, while the role of lncRNAs in acute lung injury (ALI) remains unclear. The present study aimed to profile the expression of lung lncRNAs and mRNAs in lipopolysaccharide (LPS)-induced ALI mouse model. C57BL/6 mice were exposed to LPS or phosphate-buffered saline for 24 h, and lncRNAs and mRNAs were profiled by Arraystar mouse LncRNA Array V3.0. Bioinformatics analysis gene ontology including (GO) and pathway analysis and cell study in vitro was used to investigate potential mechanisms. Based on the microarray results, 2632 lncRNAs and 2352 mRNAs were differentially expressed between ALI and control mice. The microarray results were confirmed by the quantitative real-time PCR (qRT-PCR) results of ten randomized selected lncRNAs. GO analysis showed that the altered mRNAs were mainly related to the processes of immune system, immune response and defense response. Pathway analysis suggests that tumor necrosis factor (TNF) signaling pathway, NOD-like receptor pathway, and cytokine–cytokine receptor interaction may be involved in ALI. LncRNA-mRNA co-expression network analysis indicated that one individual lncRNA may interact with several mRNAs, and one individual mRNA may also interact with several lncRNAs. Small interfering RNA (siRNA) for ENSMUST00000170214.1, - ENSMUST00000016031.13 significantly inhibited LPS-induced TNF-α and interleukin (IL)-1β production in murine RAW264.7 macrophages. Our results found significant changes of lncRNAs and mRNAs in the lungs of LPS-induced ALI mouse model, and intervention targeting lncRNAs may attenuate LPS-induced inflammation, which may help to elucidate the role of lncRNAs in the pathogenesis and treatment of ALI.
Related collections
Most cited references22
- Record: found
- Abstract: found
- Article: not found
An official American Thoracic Society workshop report: features and measurements of experimental acute lung injury in animals.
Acute lung injury (ALI) is well defined in humans, but there is no agreement as to the main features of acute lung injury in animal models. A Committee was organized to determine the main features that characterize ALI in animal models and to identify the most relevant methods to assess these features. We used a Delphi approach in which a series of questionnaires were distributed to a panel of experts in experimental lung injury. The Committee concluded that the main features of experimental ALI include histological evidence of tissue injury, alteration of the alveolar capillary barrier, presence of an inflammatory response, and evidence of physiological dysfunction; they recommended that, to determine if ALI has occurred, at least three of these four main features of ALI should be present. The Committee also identified key "very relevant" and "somewhat relevant" measurements for each of the main features of ALI and recommended the use of least one "very relevant" measurement and preferably one or two additional separate measurements to determine if a main feature of ALI is present. Finally, the Committee emphasized that not all of the measurements listed can or should be performed in every study, and that measurements not included in the list are by no means "irrelevant." Our list of features and measurements of ALI is intended as a guide for investigators, and ultimately investigators should choose the particular measurements that best suit the experimental questions being addressed as well as take into consideration any unique aspects of the experimental design.
- Record: found
- Abstract: found
- Article: not found
Transient expression of IL-1beta induces acute lung injury and chronic repair leading to pulmonary fibrosis.
M. Kolb, P J Margetts, D C Anthony … (2001)
IL-1beta is one of a family of proinflammatory cytokines thought to be involved in many acute and chronic diseases. Although it is considered to participate in wound repair, no major role has been attributed to IL-1beta in tissue fibrosis. We used adenoviral gene transfer to transiently overexpress IL-1beta in rat lungs after intratracheal administration. The high expression of IL-1beta in the first week after injection was accompanied by local increase of the proinflammatory cytokines IL-6 and TNF-alpha and a vigorous acute inflammatory tissue response with evidence of tissue injury. The profibrotic cytokines PDGF and TGF-beta1 were increased in lung fluid samples 1 week after peak expression of IL-1beta. Although PDGF returned to baseline in the third week, TGF-beta1 showed increased concentrations in bronchoalveolar lavage fluid for up to 60 days. This was associated with severe progressive tissue fibrosis in the lung, as shown by the presence of myofibroblasts, fibroblast foci, and significant extracellular accumulations of collagen and fibronectin. These data directly demonstrate how acute tissue injury in the lung, initiated by a highly proinflammatory cytokine, IL-1beta, converts to progressive fibrotic changes. IL-1beta should be considered a valid target for therapeutic intervention in diseases associated with fibrosis and tissue remodeling.
- Record: found
- Abstract: found
- Article: found
Long non-coding RNAs and enhancer RNAs regulate the lipopolysaccharide-induced inflammatory response in human monocytes
Nicholas E. IIott, James A. Heward, Benoit Roux … (2014)
Genome-scale transcriptional responses of protein-coding genes following lipopolysaccharide (LPS)-induced activation of the innate immune response in human monocytes and monocytic cell lines have been extensively studied using both microarrays1 and serial analysis of gene expression2 3. These studies have clearly demonstrated the induction of many members of proinflammatory cytokines (for example, IL1β, CXCL8, IL6 and tumor necrosis factor alpha (TNFα)), chemokines and cell surface markers in response to Toll-like receptor 4 (TLR4) signalling3. Significantly, the advent of next-generation sequencing technology has allowed for an unbiased investigation of the full repertoire of transcription from both coding and non-coding regions of the genome. Evidence for pervasive transcription across the genome now exists in the form of large sequenced cDNA libraries4, histone modification chromatin immunoprecipitation (ChIP)-seq data sets5 6 and messenger RNA (mRNA) sequencing (RNA-seq) data sets7 8 as part of multi-centre collaborative projects such as FANTOM, ENCODE and the HumanBodyMap. These data sets have revealed the presence of a large number of non-coding RNAs, which display both tissue- and developmental time point-specific transcription6 7 9. Several classes of non-coding RNAs have been characterized, and some, such as microRNAs (miRNA), have been well studied. Indeed, miRNAs have been shown to play a role in the cellular response to LPS10 11. However, the regulatory potential of long non-coding RNAs (lncRNAs) has emerged only recently (reviewed in ref. 12). While the function of the majority of these RNA species is yet to be elucidated, there is increasing evidence to support their diverse mechanistic roles. Some act as structural molecules in the recruitment of histone-modifying enzymes either in cis or trans (reviewed in ref. 12), while others are transcribed from enhancer regions (eRNAs13), facilitating transcription of protein-coding genes via RNA-dependent or RNA-independent mechanisms14. Studies in mouse macrophages and bone marrow-derived dendritic cells have revealed the regulation of intergenic transcription in response to proinflammatory signals (LPS and IFNγ)6 15 16. Interestingly, transcription from intergenic loci is not always coincident with canonical promoter histone modification signatures. Indeed, a large proportion is derived from regions of the genome that contain a high H3K4me1/H3K4me3 ratio, suggesting that lncRNAs are commonly derived from enhancer elements (eRNAs)15 17. Transcriptional complexity is further increased when we consider the presence of non-polyadenylated, bidirectional transcripts produced from additional sets of enhancers18. While the function of lncRNAs in the human innate immune response is not well described, recent studies have begun to elucidate the mechanism of action of some mouse lncRNAs. Thus, Rapicavoli et al.19 identified a pseudogene for ribosomal protein S15a (Rsp15a), renamed Lethe, that negatively regulates the inflammatory response through an interaction with RelA (p65), a component of the NF-κB complex19. Similarly, Carpenter et al.20 showed rapid induction in the expression of a lincRNA localized ~50 Kb downstream of COX2 (named lincRNA-Cox2) that mediates both the activation and repression of inflammatory response through interaction with the nuclear ribonucleoprotein A/B and A2/B1. A single publication in human macrophages derived from human monocytic THP-1 cells has identified a lincRNA named TNFα and heterogeneous nuclear ribonucleoprotein L-related immunoregulatory LincRNA from a microarray screen that regulated the inflammatory response21. While these studies have been valuable in determining lncRNA regulation in response to LPS in cells of the mouse innate immune system, sequencing studies have not yet been performed in humans. In contrast to the observed global human–mouse transcriptional conservation of protein-coding genes22, rapid transcriptional turnover of a vast number of lncRNAs in rodent species suggests the evolution of lineage-specific transcription23. This provides the impetus to use human cells to relate lncRNA transcription to human immune function. In this study, we aim to characterize the unexplored aspects of non-coding transcription in cells of the human innate immune system. Specifically, using ab initio transcript assembly of a deeply sequenced RNA data set, we examine the response of lncRNAs to TLR4 signalling through LPS stimulation. By integration of publically available H3K4me1 and H3K4me3 ChIP-seq data from the ENCODE project, we estimate the proportion of LPS-regulated lncRNAs that are derived from enhancer elements (eRNAs) and determine how changes in expression correlate with changes in proximal protein-coding mRNA expression. We also characterize the expression of bidirectional transcripts, a class of elements that have not yet been examined in the innate immune response and use knockdown experiments to determine the functional role of lncRNAs. Our data demonstrate that many lncRNAs in human monocytes show differential expression in response to LPS stimulation—a subset of which are derived from uni- and bidirectional enhancer regions that coexpress with neighbouring inflammatory mediators. Interestingly, knockdown of an eRNA (IL1β-eRNA) and a region of bidirectional transcription (IL1β-RBT46) selectively attenuates LPS-induced expression and release of the proinflammatory mediators, IL1β and CXCL8. This suggests a general role for lncRNAs in the regulation of the innate immune response and the release of inflammatory mediators in human monocytes. Results RNA sequencing shows LPS induction of innate immunity LPS stimulation produced significant increases in the release of the inflammatory mediators CXCL8 and TNFα (Fig. 1b), confirming activation of the immune response. Analysis of RNA-seq data for Ensembl (version 66)-annotated mRNAs confirmed this activation at the transcriptional level. A total of 1,621 protein-coding genes were differentially expressed on LPS stimulation at a false discovery rate (FDR) 2 (Supplementary Fig. 1a and Supplementary Table 1; 1,045 upregulated, 576 downregulated). Further cross-platform analysis revealed that fold changes determined by RNA-seq correlated well with those from microarray-based analysis of the same data set (see Supplementary Fig. 1b online; r=0.74) and both methods called similar sets of genes as differentially expressed (hypergeometric test; P=1.32 × 10−11). A total of 534 genes were called by both methods (415 upregulated, 115 downregulated, 4 incongruent), with 530 representing a robust set of inflammation-regulated genes. As expected, the upregulated genes from this set were significantly enriched (FDR 5 Kb from a protein-coding gene) and mRNA-flanking ( 1) lncRNAs were differentially expressed (see Supplementary Table 3). We observed 182 up and 39 downregulated lncRNAs (Fig. 1d), of which 76 were from the novel lncRNAs identified in primary human monocytes. The majority of differentially expressed lncRNAs were located >5 Kb from protein-coding annotations with class representation in the order of their proportion being the following: intergenic (51%), antisense (33%) and mRNA-flanking lncRNA (antisense upstream (11%) and antisense downstream (5%)) (Fig. 1e). qRT–PCR analysis of a subset of differentially expressed lncRNAs confirmed our RNA-seq results with 14/18 (78%) validated using this method (Supplementary Fig. 3). Of note, only 2 of the 221 differentially expressed lncRNAs displayed positional overlap with previously identified LPS-regulated lncRNA exons from mouse bone marrow-derived dendritic cells6 and displayed no significant homology. This result is consistent with rapid transcriptional turnover of lncRNAs across species23. Our initial differential expression analysis thus revealed a widespread programme of LPS-induced expression changes in a class of RNA molecule that has not been previously studied in primary human monocytes. LncRNAs are associated with an enhancer-like chromatin state Previous studies have reported the presence of transcription at active enhancers marked by H3K4me1 (ref. 18). While the poised/active promoter-associated mark, H3K4me3, may also be present at distal enhancer loci24, the ratio of H3K4me1/H3K4me3 is commonly used as a discriminatory mark between enhancers and promoters (reviewed in refs 14, 17). To explore whether our differentially expressed lncRNAs represent transcription from enhancer-like regions, we utilized recently available histone modification ChIP-seq data from the ENCODE consortium. We downloaded alignments for H3K4me1 and H3K4me3 ChIP-seq data in CD14+ resting monocytes and assessed the read coverage over intervals surrounding the transcription start site (TSS, ±0.5 Kb) of differentially expressed lncRNAs. An equivalent analysis using differentially expressed protein-coding genes (n=530) provided a comparison set. To eliminate confounding influences of marks associated with protein-coding genes on lncRNA marks, we removed lncRNAs that either overlapped (that is, antisense) or shared a TSS interval ( 1.2 and 1 and had>20 read counts in at least two samples. This resulted in a set of 349 intervals. To address the presence of regions of bi-directional transcription (RBT), we plotted the distribution of the forward strand/reverse strand read count ratio across these intervals. The distribution was tri-modal (Fig. 4a), demonstrating that a proportion of intervals represent genomic regions that are transcribed bi-directionally (log2(ratio)~0). This was in contrast to protein-coding genes and can-lncRNAs/eRNAs, whose distributions suggested the predominance of transcription coming from one of either the forward or reverse strand (Fig. 4a). In order to identify LPS-regulated RBT, we assessed differential expression across 69 loci that had a forward strand/reverse strand ratio 1.2). The reason for bi-directional transcription at these loci remains obscure. Nevertheless, we observed a positive correlation between LPS-induced changes in RBT expression and changes in expression of their neighbouring protein-coding gene (Fig. 4c). As with our described unidirectional eRNAs, we observed co-regulation of RBT and important regulators of the immune response. For example, TRAF1, IL37, CXCR4, IFNGR2 and CCRN4L (Supplementary Table 5) all have upstream co-regulated RBT. Of particular interest, we identified an mRNA-flanking RBT (IL1β−RBT46) and another within an enhancer region (IL1β−RBT47) located upstream of the important inflammatory mediator, IL1β. Together with our analysis of multi-exonic eRNAs, these data suggest that transcriptional regulation at this locus, and potentially many more, may involve complex regulatory networks that include transcription from both uni- and bi-directionally transcribed enhancers. LPS-induced lncRNAs are enriched for NFκB binding sites Given the importance of the inflammatory transcription factor (TF) NF-κB in regulating the transcriptional response to infection, we were interested in identifying whether our differentially expressed lncRNAs showed evidence for NF-κB binding. We used the genomic association tester27 to assess the overlap of eRNA and can-lncRNA promoters (±0.5 Kb around TSS) as well as RBT intervals with TNFα-induced NF-κB binding in lymphoblastoid cell lines. First, we established the overlap between differentially expressed protein-coding promoters and NF-κB. Protein-coding promoters were significantly enriched for NF-κB binding (Table 1), validating the use of this cell type/stimulation procedure for the analysis of lncRNAs. Consistent with immune regulation of can-lncRNAs, we found a significant enrichment for NF-κB binding (Table 1). However, there was no significant overlap between eRNA promoters and NF-κB binding sites (Table 1). To assess whether the difference in NF-κB binding between can-lncRNAs and eRNAs was due to differences in binding location, we also assessed NF-κB binding across gene bodies (TSS-TTS). We found that both eRNAs and can-lncRNAs were enriched for NF-κB binding across their gene bodies (Table 1). We did not see a significant overlap between RBT intervals and NF-κB binding sites, although this may be due to low numbers in this group. Significantly, where we have shown uni- and bidirectional non-coding transcription at the IL1β locus, we also observed coincident NF-κB binding (Fig. 4d). This evidence suggests that eRNAs are either regulated by NF-κB themselves, or that they are transcribed from regions that contain NF-κB binding sites important for regulating the expression of nearby mRNA. Characterization of IL1β-eRNA and IL1β-RBT46 expression Having shown widespread changes in the expression of eRNAs, can-lncRNAs and RBTs following LPS stimulation, we wanted to examine whether these regulated the inflammatory response. To this end, we focused upon the production and release of IL1β as a marker of monocyte activation, since this is the second most highly expressed mediator in response to LPS (Supplementary Table 1) and is an important driver of the inflammation associated with the innate immune response28. This cytokine is also situated in a region that we have shown to display high transcriptional complexity—it has a downstream eRNA (IL1β-eRNA) and an upstream mRNA-flanking RBT (IL1β-RBT46) (Fig. 4d). Given the difficulty in transfecting primary human monocytes, these studies were performed in the monocytic THP-1 cell line. For clarity, we have only included the data for the +ve strand of RBT46, since the results with –ve strand were identical. Measurement of the time course showed rapid LPS induction in IL1β-eRNA (Fig. 5a) and IL1β-RBT46(+) (Fig. 5b) expression that peaked at ~2 h and ~6 h, respectively. This correlated with that of IL1β mRNA production as well as the generation of two additional, highly expressed inflammatory mediators, CXCL8 and IL6 (Fig. 5c). As might be expected, mRNA production preceded the extracellular release of IL1β, CXCL8 and IL6 protein (Fig. 5d). Since our studies have focused upon LPS stimulation via TLR4, we also examined the profile of IL1β-eRNA and IL1β-RBT46(+) expression in response to a range of alternative inflammatory mediators including IL1β and other TLR agonists. These studies showed that IL1β, CXCL8 and IL6 mRNA expression at 2 h was only induced in response to LPS (via TLR4) and FSL-1 (via TLR-2/6), with a small but non-significant increase following exposure to Pam3CSK4 (via TLR-1/2) and HKLM (via TLR-2) (Fig. 5e). In the case of IL6 mRNA expression, this early time point meant that the changes were not significant. This selective response is likely to reflect the profile of receptor expression on the monocytic THP-1 cell line. As might be expected if their activation was mediated via similar intracellular pathways, we found that the induction of IL1β-eRNA and IL1β-RBT46(+) mirrored that of IL1β (Fig. 5e) Having predicted the presence of NF-κB binding sites at the genomic locations of IL1β-eRNA and IL1β-RBT46, we proceeded to measure NF-κB binding by ChIP in combination with qRT–PCR, using an antibody to the p65 RelA DNA binding subunit. These studies showed a 3.2- and 4.3-fold enrichment in NF-κB binding (relative to non-treated controls) within the promoter regions of IL1β-eRNA and IL1β-RBT46(+) following exposure to LPS, respectively (Fig. 6a). This was comparable with the 3.7-fold enrichment seen within the promoter region of the NF-κB regulated chemokine, CXCL8. As predicted by the existing ChIP-seq data in ENCODE, the promoter region of IL1β did not demonstrate NF-κB binding as measured by ChIP–qPCR (Fig. 6a). To confirm the ChIP–qPCR data, we examined the action of TPCA-1, an inhibitor of IκB kinase 2 that is an upstream activator of NF-κB29. TPCA-1 attenuated the LPS-induced production of IL1β-eRNA, IL1β-RBT46(+) and CXCL8 with an IC50 of 1.0 μM, 0.9 μM and 1.3 μM, respectively (Fig. 6b). TPCA-1-mediated inhibition of NF-κB binding was confirmed using ChIP–qPCR (Fig. 6a). Interestingly, despite the absence of NF-κB binding, TPCA-1 attenuated IL1β production, giving an IC50 of, 0.7 μM (Fig. 6b) and suggests that IL1β production might be indirectly regulated through expression of IL1β-eRNA and IL1β-RBT46(+). Much of the previous functional analysis of lncRNAs and eRNAs has suggested that these operate within the nucleus to regulate the transcription of protein-coding mRNAs12 30. We therefore determined the subcellular localization of IL1β-eRNA and IL1β-RBT46(+). We assessed the effectiveness of our separation procedure by using MALAT1 lncRNA as a nuclear marker and the mitochondrial cytochrome c oxidase 1 (MT-CO1) as a cytoplasmic marker. Figure 6c demonstrates successful subcellular fractionation, with MALAT1 predominantly associated with the nucleus and MT-CO1 located in the cytoplasmic fraction. There was no significant change in the expression of these markers following exposure to LPS. Interestingly, both IL1β-eRNA and IL1β-RBT46(+) were predominantly localized in the nucleus of both unstimulated and LPS-stimulated cells (Fig. 6c). These data show that LPS-induced IL1β-eRNA and IL1β-RBT46(+) expression follows a similar time course to that of IL1β, is dependent on the activation of the inflammatory transcription factor NF-κB and if functional, are acting in the nucleus to regulate gene transcription. IL1β eRNA and IL1β RBT46 regulate IL1β expression Given their genomic position (Fig. 4d) and nuclear localization (Fig. 6c), we speculated that IL1β-eRNA and IL1β-RBT46 might regulate the transcription of IL1β. To examine this hypothesis, we designed a panel of five locked nucleic acid (LNA)-based antisense inhibitors against IL1β-eRNA and IL1β-RBT46(+) and transfected them into the monocytic THP-1 cells. Following LPS stimulation, we found that in the case of both IL1β-eRNA and IL1β-RBT46(+), only one (of the five) attenuated lncRNA production (Fig. 7a). However, these LNA antisense inhibitors, but not two negative controls, reduced LPS-induced IL1β-eRNA and IL1β-RBT46(+) generation by 85±9% and 53±9%, respectively (Fig. 7a). Of relevance, we also failed to attenuate LPS-induced IL1β-eRNA production using a panel of four siRNAs (Supplementary Fig. 4a) despite showing a 64±4% reduction in LPS-induced IL6 mRNA production using a positive control siRNA (Supplementary Fig. S4b). In contrast, we were able to show knockdown using all five LNA antisense inhibitors targeted against the constitutively expressed lncRNA, OIP5-mf-lncRNA (Supplementary Fig. S4c). These studies suggested that unlike previous reports that have successfully employed both LNA antisense and siRNA for the knockout of constitutively expressed lncRNAs and eRNAs31 32 33, this approach is more problematic when applied to lncRNAs that are located within the nucleus and rapidly induced upon exposure to LPS. Despite these limitations, we demonstrated that knockdown of IL1β-eRNA and IL1β-RBT46(+) attenuated LPS-induced IL1β mRNA expression and protein release at 24 h (Fig. 7b/c). Thus, we observed a 40 and 35% reduction in IL1β mRNA (Fig. 7b) and a 66 and 67% reduction in IL1β protein release (Fig. 7c) following inhibition of IL1β-eRNA and IL1β-RBT46(+), respectively (Fig. 7c). The effect of IL1β-eRNA and IL1β-RBT46(+) knockdown did not appear to be non-specific since this had no effect upon LPS-induced expression of IL1α and IL1RN (Fig. 7d), that are located in the same genomic region or upon expression of the distally located IL6 (Fig. 7c). Significantly, we showed a 47 and 52% reduction in CXCL8 mRNA (Fig. 7b) that translated into a small but significant reduction of 35 and 33% in CXCL8 protein release (Fig. 7c) following inhibition of IL1β-eRNA and IL1β-RBT46(+), respectively (Fig. 7c). Overall, this indicates that these LPS-induced IL1β-eRNA and IL1β-RBT46(+) regulate mRNA expression and downstream release of IL1β and CXCL8 following activation of the innate immune response. Discussion To further our knowledge on the potential role of lncRNAs during the activation of the innate immune response, we have used RNA sequencing of ribosomal RNA (rRNA)-depleted, stranded RNA libraries to study the human monocytic response to LPS stimulation. We identified 2,607 lncRNAs using an ab initio transcript assembly. Interestingly, our data support the notion that lncRNA expression is tissue-restricted34 as we observe 1,318 lncRNAs that have not been previously identified in Gencode8 or the HumanBodyMap7. In addition to these multi-exonic lncRNAs, we uncovered a set of novel bi-directionally transcribed genomic loci (RBT, n=69), a feature of our data that was made possible through stranded RNA sequencing. Previous studies have shown that exposure to LPS induces changes in multiple miRNAs that regulate the innate immune response through targeting the translation of key signalling proteins10 35. Significantly, we found 221 lncRNAs and 35 RBTs to be differentially expressed in response to LPS and thus may be important in monocyte activation. Examination of the position and homology between these lncRNAs and those previous identified in mouse bone marrow-derived dendritic cells6 showed virtually no overlap between humans and mice, which is consistent with rapid transcriptional turnover across species23. Furthermore, this implies that we must be cautious when extrapolating functional and mechanistic observations between species. As an example, a BLAST search failed to identify the presence in the human genome of the mouse lincRNA-Cox2 that has been reported to regulate the inflammatory response in mouse bone marrow-derived macrophages20. Recent investigations in mouse macrophages have indicated that many monocyte lncRNAs are transcribed from enhancer regions14 15 36. In support of this observation, we found that many of the monocyte expressed lncRNAs (58%) are transcribed from regions of the genome that are marked by a high H3K4me1/H3K4me3 ratio, a marker of the existence of enhancer transcription (eRNAs)37. In addition, we found that changes in expression of these eRNAs correlated with those of neighbouring protein-coding mRNAs. Mechanistically, this would imply that these eRNAs might act in cis to regulate the expression of their most proximal coding gene, which is consistent with previous data in mouse macrophages15. Bi-directional transcription has been shown to be a defining feature of a subset of active enhancers in mouse cortical neurons and human fetal lung fibroblasts18 38. These non-polyadenylated transcripts are regulated by neuronal activity, a feature that is correlated with activity, regulation of their neighbouring protein-coding gene18. We have shown for the first time, the phenomenon of stimulus-dependent bi-directional transcription from both enhancer and promoter regions of activated human monocytes. There is now accumulating evidence that lncRNAs regulate mRNA expression at the level of transcription and translation12 39. Similarly, recent reports have indicated that eRNAs regulate in cis local mRNA expression in multiple cell types31 40 including mouse macrophages32. Having established that LPS induced widespread changes in the expression of lncRNAs and eRNAs in human monocytes that were located close to differentially expressed inflammatory genes, it was important to determine whether these were of functional relevance. Of particular interest for regulation of the innate immune response, was our identification of multiple non-coding transcripts that are situated close to the IL1β gene, an important cytokine that stimulates inflammatory responses in multiple cell types and whose overproduction has been implicated in autoimmune diseases28 41. These included a downstream eRNA (eRNA-IL1β-eRNA) and an upstream mRNA-flanking RBT (IL1β-RBT46). Significantly, we demonstrated that expression of both IL1β-eRNA and IL1β-RBT46 was mediated by the classical proinflammatory transcription factor, NF-κB while knockdown of the IL1β-eRNA and IL1β-RBT46 was shown to attenuate LPS-induced IL1β transcription and protein release. This therefore implies that expression of these lncRNAs regulates the release of this inflammatory mediator and given the genomic position, it might be speculated that they regulate IL1β transcription in cis. However, IL1β-eRNA and IL1β-RBT46 also appear to act in trans since their knockdown inhibited the transcription and release of CXCL8, albeit to a lesser extent. It is unlikely that this effect was via a non-specific action of the LNA inhibitors since we observed no effect upon LPS-induced expression of IL6, IL1α and IL1RN. Nevertheless, we cannot rule out the possibility that the effect of CXCL8 is secondary to the inhibition of IL1β, although this is also unlikely as CXCL8 release precedes that of IL1β (Fig. 5d). In conclusion, we have shown for the first time that LPS stimulation of primary human monocytes causes widespread changes in lncRNA expression. Crucially, we have demonstrated that the nuclear-located transcripts, IL1β-eRNA and IL1β-RBT46(+) regulate the transcription and release of the key proinflammatory cytokines, IL1β and CXCL8 although the mechanism is currently unknown. As with miRNAs, we speculate that many of these eRNAs and lncRNAs are important regulators of the innate immune response and future studies will need to focus upon the identification of those that are functionally relevant and the elucidation of their mechanism of action. Methods Treatment of human primary monocytes All human volunteers gave informed written consent as approved by the London–Chelsea NRES ethics committee. Human blood (60 ml) was collected into tubes containing 2% (w/v) EDTA and red blood cells removed by dextran sedimentation. The leukocyte-rich layer was centrifuged at 400 g for 10 min at 4 °C, and the granulocytes were separated from the peripheral blood mononuclear cells (PBMC) fraction using discontinuous Percoll gradients. Percoll fractions of 81, 68 and 55% (v/v) in Dulbecco’s phosphate-buffered saline were prepared and the cell pellet from above was resuspended in 3 ml of 55% (v/v) Percoll and then overlaid onto the pre-prepared gradient. The cells were then separated according to density by centrifugation at 750 g for 25 min at 4 °C. The PBMC were harvested from the 55%/68% interface and then washed with Dulbecco’s phosphate-buffered saline. Monocytes were then isolated from the PBMC fraction using a Miltenyi Monocyte Isolation Kit II according to manufacturer’s instructions. Monocytes were centrifuged and resuspended in RPMI 1640 containing 10% (v/v) fetal calf serum, 10 mg ml−1 (1% (v/v)) penicillin/streptomycin, 2 mM L-glutamine at 1 × 106 cells ml−1 and incubated for the times indicated in the absence or presence of 10 ng ml−1 LPS42. Experimental samples were treated with 10 ng ml−1 LPS for the indicated time and the controls were left untreated (n=4 per group). The media was then removed for measurement of CXCL8 and TNFα by ELISA (R&D Systems) and the cells lysed prior to RNA extraction. Culture of human THP-1 cells Monocytic THP-1 cells were obtained from ATCC and cultured in RPMI 1640, supplemented with 10% (v/v) FBS, 1% (w/v) L-glutamine, 1% (w/v) Pen-Strep and 0.1% (v/v) β-mercaptoethanol (Invitrogen Gibco) and incubated in a 37 °C, 5% (v/v) CO2 humidified incubator. For the time courses, THP-1 cells were stimulated with 1 μg ml−1 LPS (Escherichia coli 055:B5, Sigma-Aldrich) for the indicated period of time (n=3 per group per time point). For all other experiments, THP-1 cells were stimulated with LPS at 1 μg ml−1 for the length of time indicated. Stimulation of THP-1 with TLR agonists THP-1 cells were treated with the following TLR agonists from InvivoGen for 2 h (n=3 per group) at the concentrations listed; PAM3CSK4 (100 ng ml−1), HKLM (108 cells ml−1), Poly(I:C) (10 mg ml−1), Poly(I:C) LMW (10 mg ml−1), Flagellin (100 ng ml−1), FSL-1 (100 ng ml−1), Imiquimod (5 μg ml−1), ssRNA40 (1 μg ml−1) and ODN2006 (2 μM). THP-1 cells were also treated with buffer as a control, LPS (1 μg ml−1, E. coli 055:B5, Sigma-Aldrich) and IL-1β (10 ng ml−1, recombinant, E. coli, Sigma-Aldrich). RNA isolation Total RNA was extracted using the Qiagen RNeasy kit and included an on-column DNase treatment. RNA used was of high quality (Agilent Bioanalyser (RIN>9.5)). RNA library preparation and sequencing rRNA was depleted using an early-access version of the Ribo-Zero Gold (human/mouse) rRNA Removal Kit (Epicentre). Strand-specific Illumina GA-IIx cDNA libraries were prepared using an early-access version of the ScriptSeq v2 library preparation kit (Epicentre). Two hundred cycles of sequencing on the Illumina GA-IIx instrument were performed to generate 2 × 100 bp paired-end sequencing reads. Quality control of RNA sequencing Quality scores across sequenced reads were assessed using FASTQC v0.9.2 (http://www.bioinformatics.babraham.ac.uk/projects/fastqc). All samples were of high quality. The average score (mean and median) at each base across reads in each sample was Q>28. Alignment and transcript assembly Reads were mapped to the human reference genome (Hg19) using TopHat v1.4.0 (ref. 43). TopHat first maps to the transcriptome that was supplied as an additional input file. Along with TopHat discovering known splice junctions, a set of known protein-coding junctions (Ensembl 66) were supplied. The following options were used for mapping reads: --mate-inner-dist 60 --num-threads 4 --library-type fr-secondstrand raw-juncs --transcriptome index -n 2 An average of 44.6M reads were mapped (range 40.86–50.82) corresponding to an average of 73.18% (range 69.90–76.05%). Consistent with successful ribosomal RNA (rRNA) depletion, an average of 4.72% (range 1.93–7.32%) of reads mapped to rRNA. Transcripts were assembled for each sample ab initio using Cufflinks v1.3.0 (ref. 1) with the following parameterization: --upper-quartile-norm --min-frags-per-transfrag 1 --pre-mrna-fraction 0.5 --junc-alpha 0.001 --overlap-radius 100 Assemblies between samples were compared using Cuffcompare, and transcripts that were present in at least two samples were retained for downstream analysis. Prediction of lncRNAs using RNA sequencing data We utilized the ab initio assembly output from Cufflinks/Cuffcompare in conjunction with the latest human lncRNA annotations (Gencode v13 and the HumanBodyMap7 to classify transcripts as putative lncRNAs. The outline of the lncRNA prediction pipeline is provided in Fig. 1a. First we removed any transfrags that overlapped (≥1 bp on the same strand) transcripts annotated by Ensembl (build 66) as ‘protein_coding’, ‘processed_pseudogene’, ‘unprocessed_pseudogene’, ‘nonsense_mediated_decay’ or ‘retained_intron’. We also filtered any transcripts that overlapped RefSeq annotated coding (CDS) intervals. Next we employed size selection, retaining transfrags that were >200 bp in length and multi-exonic. We then merged our monocyte-derived lncRNA set with the Gencode v13 and the HumanBodyMap7 sets, producing a non-redundant set of lncRNAs. Assessment of coding potential We used the coding potential calculator (CPC version 0.9-r2)44 to assess the coding potential of discovered lncRNAs. Abundance estimation and differential expression analysis For correlation analyses, RNA abundance defined as the FPKM was estimated using Cuffdiff V2.0.2. Differential expression analysis was performed using the negative binomial distribution-based method implemented in DESeq45 on the summed exon read count per gene. Genes were assessed for differential expression if they had an FPKM>1 in either LPS or unstimulated condition (average across replicates). Genes annotated as ‘protein_coding’ in Ensembl (build 66) were used for analysis of protein-coding genes. Multiple testing corrections were performed on a total of 1,065 lncRNAs and 15,020 protein-coding genes using the Q-value method46. Profiles of monocyte histone marks across lncRNAs Aligned ChIP-seq data for H3K4me3 and H3K4me1 histone modifications in resting CD14+ monocytes were downloaded from UCSC ENCODE (http://hgdownload.cse.ucsc.edu/goldenPath/hg19/ENCODEDCC/wgEncodeBroadHistone/). Two replicate alignment files for each histone mark were merged using samtools (http://samtools.sourceforge.net/) and profiles were assessed by counting alignments across windows surrounding the transcription start sites of differentially expressed lncRNAs and protein-coding genes (−0.5 to 0.5 Kb) using custom python scripts and BEDTools (code.google.com/p/bedtools/). Heatmaps of the profiles were produced using the heatmap2 (gplots) function in R. Analysis of bi-directional transcription Monoexonic lncRNAs were obtained from our initial cufflinks assembly. Overlapping predictions were merged using BEDTools. This set was filtered, retaining loci that were expressed at an FPKM>1 and were present in at least 2/8 samples with>20 uniquely mapping reads. Bi-directional transcription was assessed by counting the number of reads mapping to the forward and reverse strands. The forward/reverse ratio was used to determine the presence of bi-directional transcription that is, a ratio of 1 would indicate a 50/50 split between sense and antisense transcription. An interval with a ratio of less than twofold was considered bi-directionally transcribed. Differential expression was performed on these loci using DESeq and loci were called differentially expressed at an FDR 1 in either LPS- or unstimulated cells). The expected overlap was computed by randomizing the locations of the query set of intervals among the reference set intervals. From 10,000 randomizations, the procedure computed the expected overlap and an empirical P-value. The reported fold enrichment is the ratio of the observed overlap and the expected overlap. Microarray preparation and data analysis Microarray data were used for validation of differentially expressed protein-coding genes from the RNA-seq analysis. mRNA expression profile was determined using the Agilent SurePrint G3 Human microarrays (v2) following the manufacturer’s instructions. Two channel microarray data were analysed using LIMMA in R-2.14.1. Raw data were processed using Agilent Feature Extraction Software and probes were retained for analysis if they were flagged as being ‘well above background’, ‘not a control probe’ and ‘not saturated’ in at least three arrays. Background correction, within-array robust-spline normalization and between-array quantile normalization were performed using functions in LIMMA. A total of 18,739 probes corresponding to 15,937 genes were analysed for differential expression using the empirical Bayes procedure implemented in LIMMA. Genes were called differentially expressed at an FDR 2. Quantitative PCR validation of lncRNA differential expression Eighteen lncRNAs and eRNAs were chosen for validation (n=4 per group). Expression of lncRNAs, eRNAs and 18S RNA were determined by qRT–PCR using the SYBR Green PCR mix (Applied Biosystems; primers were obtained from Sigma-Aldrich and are listed in Supplementary Table 6). The separate well, 2−(ΔΔCt) method was used to determine relative quantities of individual mRNAs and lncRNAs, which were normalized to 18S RNA. Nuclear and cytoplasmic RNA fractionation THP-1 cells were stimulated with LPS at 1 μg ml−1 for the length of time indicated. The cells were centrifuged and then split into two equal fractions. Total RNA was extracted from one fraction using the normal Qiagen RNeasy protocol while the other fraction was treated with RLN buffer on ice for 5 min, in order to lyse the plasma membrane while leaving the nuclei intact. The nuclei were then isolated by centrifugation at 300 g in a pre-chilled centrifuge. RNA was then extracted from the nuclear and cytoplasm fractions using the normal Qiagen RNeasy protocol. In order to quantify gene expression within the different fractions by qRT–PCR, the 18S values from the total RNA fraction were used to normalize gene expression across all of the fractions. Transfection of THP-1 cells with LNA GAPmers To transfect with LNA GAPmers, THP-1 cells were seeded at 5 × 105 cells per well in 24 well plates, in 200 μl of complete growth medium. Transfection mixes were prepared using 190 μl of serum-free growth medium, 10 μl of HiPerFect (Qiagen) plus LNA GAPmers to give a final concentration of 30 nM. Cells were subsequently incubated for 16 h, diluted with 800 μl of complete growth medium and stimulated with LPS. Cells were removed at 2 h and 24 h with the supernatants reserved for analysis of cytokine release. LNA and siRNA sequences are listed in Supplementary Table 7. Chromatin immunoprecipitation ChIP was performed according to the manufacturer’s guidelines (Active Motif; 53040). In brief, 3 × 107 THP-1 cells were stimulated or not with LPS (1 μg ml−1) for 60 min. Whole cells were cross-linked with a 1% formaldehyde solution for 15 min at room temperature. Cells were sonicated (Branson Sonifier 250) for two cycles (output: 1, duty cycles: 20%, time: 30 s on 30 s off). DNA concentrations were quantified, and 10 μg of chromatin DNA was used for each ChIP reaction. ChIP assays were performed with 4 μg of antibody (NFκB p65, C-20, Santa Cruz) and incubated overnight at 4 °C, precipitated with agarose beads (supplied) and washed. Bead-bound DNA was reverse cross-linked and purified with DNA Purification columns (supplied). Samples were then analysed by qPCR using the probes listed in Supplementary Table S6. Author contributions N.E.I. primarily wrote the manuscript and analysed the data; J.A.H. undertook the majority of the laboratory based studies, assisted in the experimental design and contributed to the writing of the manuscript: B.R., E.T. and P.S.F. assisted in parts of the experimental work; L.L., I.G., C.H-F. and N.H. assisted with experimental design and the RNA sequencing, L.E.D. assisted in the experimental design and contributed to the writing of the paper; A.H. assisted in the experimental design and the analysis of the data; D.S. assisted in the experimental design, the analysis of the data and contributed to the writing of the manuscript and M.A.L. conceived of the experimental design, assisted in the analysis of data and contributed to the writing of the manuscript. Additional information How to cite this article: Ilott, N. E. et al. Long non-coding RNAs and enhancer RNAs regulate the lipopolysaccharide-induced inflammatory response in human monocytes. Nat. Commun. 5:3979 doi: 10.1038/ncomms4979 (2014). Accession codes: Microarray and RNA-seq data have been deposited in the EBI database (www.ebi.ac.uk/arrayexpress) under accession numbers E-MTAB-2408 and E-MTAB-2399, respectively. Supplementary Material Supplementary Information Supplementary Figures 1-4 and Supplementary Tables 1-7
Author and article information
Journal
Journal ID (nlm-ta): Biosci Rep
Journal ID (iso-abbrev): Biosci. Rep
Journal ID (hwp): ppbioscirep
Journal ID (publisher-id): BSR
Title:
Bioscience Reports
Publisher:
Portland Press Ltd.
ISSN
(Print):
0144-8463
ISSN
(Electronic):
1573-4935
Publication date
(Electronic preprint):
12
April
2019
Publication date Collection: 30
April
2019
Publication date
(Electronic):
30
April
2019
Volume: 39
Issue: 4
Electronic Location Identifier: BSR20181634
Affiliations
Author notes
Correspondence: Kai Liu (
liubusiness@
123456163.com
) or Da-Jiang Li (
lidajiang@
123456cd120.com
,
lidajiang@
123456wchscu.cn
)
Author information
Da-Jiang Li
http://orcid.org/0000-0002-9071-2591
Article
DOI: 10.1042/BSR20181634
PMC ID: 6488857
PubMed ID: 30979832
SO-VID: fc317356-ed58-4c77-81f6-b78f77c0223e
Copyright © © 2019 The Author(s).
License:
This is an open access article published by Portland Press Limited on behalf of the Biochemical Society and distributed under the Creative Commons Attribution License 4.0 (CC BY).
History
Date
received
: 16
September
2018
Date
revision received
: 03
April
2019
Date
accepted
: 10
April
2019
Page count
Pages: 12
Categories
Subject:
Research Articles
Subject:
Research Article
Subject:
50
Subject:
60
Subject:
16
Subject:
48
ScienceOpen disciplines: Life sciences
Data availability:
ScienceOpen disciplines: Life sciences