Long non-coding RNAs and enhancer RNAs regulate the lipopolysaccharide-induced inflammatory response in human monocytes

Abstract

Early reports indicate that long non-coding RNAs (lncRNAs) are novel regulators of biological responses. However, their role in the human innate immune response, which provides the initial defence against infection, is largely unexplored. To address this issue, here we characterize the long non-coding RNA transcriptome in primary human monocytes using RNA sequencing. We identify 76 enhancer RNAs (eRNAs), 40 canonical lncRNAs, 65 antisense lncRNAs and 35 regions of bidirectional transcription (RBT) that are differentially expressed in response to bacterial lipopolysaccharide (LPS). Crucially, we demonstrate that knockdown of nuclear-localized, NF-κB-regulated, eRNAs (IL1β-eRNA) and RBT (IL1β-RBT46) surrounding the IL1β locus, attenuates LPS-induced messenger RNA transcription and release of the proinflammatory mediators, IL1β and CXCL8. We predict that lncRNAs can be important regulators of the human innate immune response.

Figure 1. LncRNA expression is regulated by LPS in human monocytes.

(a) Pipeline for predicting lncRNAs from cufflinks assembled transfrags. HumanBodyMap=lincRNAs predicted in ref. . Can-lncRNAs=canonical lncRNAs characterized by a low H3K4me1/H3K4me3 ratio. eRNAs characterized by a high H3K4me1/H3K4me3 ratio. (b) Release of CXCL8 and TNFα measured by ELISA shows activation of the innate immune response. Data are the mean±s.e.m. of four independent experiments. Statistical significance was determined using a Student t-test where *P<0.05, **P<0.01 and ***P<0.001. (c) Overlap of lncRNAs in monocytes, Gencode v13 and lncRNAs from the HumanBodyMap catalogues. (d) Volcano plot displaying differentially expressed lncRNAs (n=4 in each group). Differential expression analysis was performed using a test based on the negative binomial distribution implemented in DESeq. LncRNAs that are significant at an FDR<0.05 are displayed in blue. (e) Breakdown of differentially expressed lncRNAs based on positional classifications.

Figure 2. LncRNAs can be distinguished by canonical promoter and enhancer chromatin signatures.

(a) H3k4me3 and H3K4me1 binding across a 1 Kb interval centred on the transcription start site of expressed protein-coding genes. Profiles are sorted based on the height of the H3K4me3 peak. Also provided is a plot of the H3K4me1/H3K4me3 log2(ratio) at each TSS (mean over interval). (b) H3k4me3 and H3K4me1 binding across a 1 Kb interval centred on the transcription start site of differentially expressed lncRNAs. Profiles are sorted on the height of the H3K4me3 peak. Also provided is a plot of the H3K4me1/H3K4me3 log2(ratio) at each TSS (mean). (c) Example of a lncRNA with a canonical promoter-like chromatin signature (can-lncRNA, left) and a lncRNA with an enhancer signature (eRNA, right). Figures were produced using the UCSC genome browser.

Figure 3. eRNA expression correlates with protein-coding gene expression.

(a) Distribution of distances from the closest expressed protein-coding gene for eRNAs and can-lncRNAs. (b) Correlation between absolute fold changes for eRNAs and can-lncRNAs with their nearest expressed protein-coding gene neighbour.

Figure 4. Bi-directional transcription defines a second set of eRNAs.

(a) Distribution of the ratio of forward strand/reverse strand read counts for monoexonic loci, lncRNAs (including eRNAs) and protein-coding genes. Dashed lines indicate a forward strand/reverse strand ratio of 2. The distribution for monoexonic loci suggests a subset of loci that are transcribed from both DNA strands (ratio<2). (b) H3K4me1 and H3K4me3 profile across differentially expressed RBT. RBT distally located from protein-coding genes display enhancer chromatin signatures. (c) Scatterplot displaying the relationship between the fold changes observed (log2) for RBT and fold changes (log2) for their nearest downstream protein-coding gene. Solid line indicates no change and dashed lines indicate fold changes>2. (d) In addition to a uni-directional eRNA downstream of IL1β, 2 upstream RBT are regulated by LPS and have evidence for NF-κB binding (lower panel).

Figure 5. Characterisation of LPS-induced lncRNA and mRNA expression.

Following exposure of human monocytic THP-1 cells to buffer or 1 μg ml⁻¹ LPS, we measured the time course of expression of (a) IL1β-eRNA, (b) IL1β-RBT46(+), (c) IL1β mRNA, CXCL8 mRNA and IL6 mRNA and (d) release of IL1β, CXCL8 and IL6 protein. Subsequent studies determine the expression of IL1β-eRNA, IL1β-RBT46(+), IL1β mRNA, CXCL8 mRNA and IL6 mRNA following 2 h exposure to a range of inflammatory agonists (e). Data are the mean±s.e.m. of three independent experiments. Statistical significance was determined using a one-way analysis of variance with a Dunnett’s post test versus untreated control, where *P<0.05, **P<0.01 and ***P<0.001.

Figure 6. NF-κB dependency and subcellular distribution of IL1β-eRNA and IL1β-RBT46(+).

NF-κB dependency in human monocytic THP-1 cells was evaluated using TPCA-1, an inhibitor of IKK2. (a) Cells were incubated with TCPA-1 (10 μM) for 30 min prior to stimulation with LPS (1 μg ml⁻¹) for 60 min. ChIP in combination with qRT–PCR was used to detect NF-κB binding within the promoter regions of IL1β-eRNA, IL1β-RBT46(+), CXCL8 and IL1β. Data were expressed as the fold enrichment of negative control primers, amplifying regions of the genome not transcribed or bound by NF-κB in human monocytes. (b) Cells were incubated with TCPA-1 (indicated concentration) for 30 min prior to stimulation with LPS (1 μg ml⁻¹) for 2 h and the expression of IL1β-eRNA, IL1β-RBT46(+), IL1β and CXCL8 mRNA was measured by qRT–PCR. (c) The subcellular distribution was determined by fractionation into total (T), nuclear (N) or cytoplasmic (C) fractions following exposure to buffer or 1 μg ml⁻¹ LPS for 2 h. Data are the mean±s.e.m. of three independent experiments. Statistical significance was determined using a one-way analysis of variance with Tukey’s multiple comparisons test, where **P<0.01 and ***P<0.001 versus unstimulated vehicle and ^#P<0.05 and ^##P<0.01 versus LPS-stimulated vehicle.

Figure 7. IL1β-eRNA and IL1β-RBT46(+) regulate LPS-induced IL1β and CXCL8 expression and release.

Human monocytic THP-1 cells (Anti-RBT46+) were transfected with two negative control LNAs or an LNA antisense against either IL1β-eRNA (Anti-eRNA) or IL1(β)-RBT46(+) (Anti-RBT46+) at a final concentration of 30 nM. Cells were then treated with either buffer (non-stimulated) or LPS prior to quantification of (a) IL1β-eRNA1 and IL1β-RBT46(+) expression at 2 h (b) IL1β mRNA, CXCL8 mRNA and IL6 mRNA at 24 h, (c) IL1β, CXCL8 and IL6 protein release at 24 h and (d) IL1α mRNA and IL1RN mRNA at 24 h. Data are the mean±s.e.m. of nine independent experiments. Statistical significance was determined using a one-way analysis of variance with a Dunnett’s post test, where *P<0.05, **P<0.01 and ***P<0.001.