Integrated analysis of mRNA and miRNA expression in response to interleukin-6 in hepatocytes

Integrated analysis of mRNA and miRNA expression in response to interleukin-6 in hepatocytes

YGENO-08734; No. of pages: 9; 4C: 6 Genomics xxx (2015) xxx–xxx Contents lists available at ScienceDirect Genomics journal homepage: www.elsevier.co...

2MB Sizes 1 Downloads 52 Views

YGENO-08734; No. of pages: 9; 4C: 6 Genomics xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Genomics journal homepage: www.elsevier.com/locate/ygeno

Integrated analysis of mRNA and miRNA expression in response to interleukin-6 in hepatocytes Samuel W. Lukowski a,b,⁎, Richard J. Fish a,b, Juliette Martin-Levilain c, Carmen Gonelle-Gispert d, Leo H. Bühler d, Pierre Maechler c, Emmanouil T. Dermitzakis a,b,e, Marguerite Neerman-Arbez a,b a

Department of Genetic Medicine and Development, University of Geneva Medical School, 1211 Geneva, Switzerland Institute of Genetics and Genomics in Geneva (iGE3), 1211 Geneva, Switzerland Department of Cell Physiology and Metabolism, University of Geneva Medical Center, Geneva, Switzerland d Surgical Research Unit, Department of Surgery, University Hospital, 1211 Geneva, Switzerland e Swiss Institute of Bioinformatics, 1211 Geneva, Switzerland b c

a r t i c l e

i n f o

Article history: Received 24 October 2014 Accepted 5 May 2015 Available online xxxx Keywords: Acute phase response Inflammation IL-6 Hepatocyte Post-transcriptional regulation miRNA Transcriptome

a b s t r a c t The expression of plasma proteins changes dramatically as a result of cytokine induction, particularly interleukin6, and their levels are used as clinical markers of inflammation. miRNAs are important regulators of gene expression and play significant roles in many inflammatory diseases and processes. The interactions between miRNAs and the genes that they regulate during the acute phase response have not been investigated. We examined the effects of IL-6 stimulation on the transcriptome and miRNome of human and mouse primary hepatocytes and the HepG2 cell line. Using an integrated analysis, we identified differentially expressed miRNAs whose seed sequences are significantly enriched in the 3′ untranslated regions of differentially expressed genes, many of which are involved in inflammation-related pathways. Our finding that certain miRNAs may de-repress critical acute phase proteins within acute timeframes has important biological and clinical implications. © 2015 Elsevier Inc. All rights reserved.

1. Introduction The hepatocyte is the primary site of acute phase protein (APP) production and therefore plays a central role in inflammation. Following the secretion of cytokines into the blood by inflammatory cells such as neutrophils and macrophages, the production of acute phase reactants by the liver drives immune and hemostatic responses. Hepatic acute phase proteins are positively or negatively regulated in response to cytokine stimulation [1]. Interleukin-6 (IL-6) is both a pro- and anti-inflammatory cytokine secreted by inflammatory cells in response to infection and inflammation [2,3]. IL-6 drives acute phase protein synthesis and increased IL-6 levels are present in a multitude of diseases, including atherosclerosis [4,5], diabetes [6,7], rheumatoid arthritis [8–10] and pulmonary hypertension [11,12]. This vital connection between IL-6 and hepatocytes leads to a complex network of gene and

Abbreviations: APR, acute phase response; APP, acute phase protein; IL-6, interleukin-6; GLM, generalized linear modeling; cpm, counts per million; RPKM, reads per kilobase of exon per million mapped reads; FDR, false discovery rate; UR, up-regulated; DR, downregulated; DEG, differentially expressed gene; DE miRNA, differentially expressed miRNA; GO, gene ontology; UTR, untranslated region. ⁎ Corresponding author at: Department of Genetic Medicine and Development, University of Geneva Medical School, 1211 Geneva, Switzerland. E-mail address: [email protected] (S.W. Lukowski).

protein expression that requires intensive scrutiny in order to discover new disease mechanisms, biomarkers or novel anti-IL-6 drug targets. A vast amount of research into the mechanisms that underlie the acute phase response (APR), particularly cytokine-stimulated pathways, has been performed in vitro and in vivo, including several microarray-based transcriptome [13–15] and proteomic approaches [16–18]. In recent times, several studies concerned with understanding gene expression in the liver have used microarray technology to examine the unfolded protein response [19], the effect of alcohol exposure on hepatoma (HepG2) cells [20] and the murine circadian clock [21]. miRNA-mediated regulation of gene expression is well-documented and there are numerous reports of single miRNA (or miRNA family)– mRNA interactions. There are also increasing numbers of reports describing miRNA signatures in acute and chronic diseases, including acute myocardial infarction [22], arterial hypertension [23], systemic lupus erythematosus [24], and cancer [25]. One recent study [26] used a combined miRNA/mRNA expression analysis to investigate miRNA expression compared to the expression of its host gene in hypoxia and found a low correlation between the two, suggesting a need for concurrent analyses in such cases. Studies investigating the regulation of gene expression by miRNAs following IL-6 induction in cellular models have revealed a number of interesting and recurring miRNAs, including miR-18a and miR-26a, and that the effects can be tissue-specific. These miRNAs influence the

http://dx.doi.org/10.1016/j.ygeno.2015.05.001 0888-7543/© 2015 Elsevier Inc. All rights reserved.

Please cite this article as: S.W. Lukowski, et al., Genomics (2015), http://dx.doi.org/10.1016/j.ygeno.2015.05.001

2

S.W. Lukowski et al. / Genomics xxx (2015) xxx–xxx

production of acute phase response genes such as fibrinogen and haptoglobin [27], regulate transcription factors such as c-Myc [28], or are themselves regulated in response to IL-6 stimulation [28]. To date, the global interactions between hepatic miRNAs and hepatic mRNA expression during the APR have not been investigated. In this study, we used mRNA- and small RNA-seq to compare the transcriptomes of several hepatocyte models following stimulation with IL-6 at different time points. Our data provides an important resource for understanding the global effects of the APR on gene expression in mouse and human hepatocytes. 2. Materials and methods

depending on the cell origin, was added to the media at a concentration of 50 ng/mL. Cells were collected at 0 h (untreated), 6 h and 24 h following exposure to human or mouse IL-6. 2.6. RNA extraction and cDNA synthesis At the appropriate time point, cells were washed in PBS, scraped and centrifuged before RNA was extracted using the standard Qiagen miRNeasy (Qiagen) protocol as per the manufacturer's instructions, which retains the total RNA, including the small RNA fraction. cDNA from mRNA was synthesized using SuperScript II (Invitrogen), and from miRNA using the Exiqon Universal cDNA Synthesis kit II, according to the manufacturer's protocols.

2.1. Tissue collection 2.7. Quantitative real-time PCR Authorization for human hepatocyte isolation from surgical liver biopsies was obtained from the Institutional Ethics Committee of the Department of Surgery. This study was conducted under experimental protocols and consent procedures approved by the Ethical Committee of the Geneva University Medical School. All patients gave written consent. Animal studies were approved by the Ethical Committee for Animal Experimentation of the Geneva University Medical School and the Canton of Geneva Animal Experimentation Veterinary authority.

mRNA and miRNA expression was measured using an Applied Biosystems StepOne qPCR instrument and KAPA Biosystems SYBR Fast Mastermix (ABI Prism) following the manufacturer's suggested thermal cycling parameters. Relative gene expression was quantified using the Applied Biosystems Expression Suite software. The reference genes used were SCLY for mRNA, and U6 for miRNA. miRcury LNA miRNAspecific qPCR primers were obtained from Exiqon.

2.2. Human primary hepatocyte collection and culture

2.8. Library preparation and sequencing

Human hepatocytes were isolated from surgical liver biopsies of patients undergoing segmental hepatectomies. At the start of the intervention, a wedge of macroscopically normal tissue (15–30 g) located within the part of the liver to be resected was excised and immediately used thereafter. Human hepatocyte isolations were performed using a two-step collagenase perfusion method as previously described [29].

For mRNA-seq of 6 conditions (IL-6 (+/−) at 0, 6, 24 h), two biological replicates for HepG2 cells (12 samples) and three biological replicates each for human and mouse primary hepatocytes (18 samples each) were prepared using the standard Illumina TruSeq RNA v2 protocol. Libraries were randomized and sequenced as 49 bp pairedend reactions using 6 samples per lane. For small RNA-seq, four biological replicates for HepG2 cells (24 samples) and three biological replicates each for human and mouse primary hepatocytes (18 samples) were prepared using the standard Illumina TruSeq small RNA protocol. The libraries were randomized and sequenced as 36 bp single-end reactions using 15 samples per lane. All library preparations were checked for quality using an Agilent Bioanalyzer (Agilent) and sample concentration was measured with a Qubit spectrophotometer (Invitrogen). All sequencing was performed on an Illumina HiSeq 2000.

2.3. Mouse primary hepatocytes Mice with a mixed 129Sv/C57Bl6J genetic background were maintained in our animal facility. Mice were anesthetized by intraperitoneal injection with Pentothal (50 μg/mL, 50 μL/25 g of mouse body weight). Mouse hepatocytes were isolated using the conventional limited collagenase digestion method [30]. After isolation, cells were maintained in 10 mL of culture medium (Williams E medium, 5% fetal calf serum, 10 − 9 M insulin, 10 − 6 M Dexamethasone, penicillin (100 U/mL), streptomycin (100 μg/mL) and 1% Glutamax). Cell viability was determined using Trypan Blue exclusion. Hepatocytes were seeded at 0.75 × 10 5 cells/cm 2 on collagen plates and incubated at 37 °C and 5% CO2 with culture medium. 2.4. HepG2 cell culture and transfection Low-passage HepG2 cells (~15) were cultured in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum, 100 U/mL of penicillin, and 100 μg/mL streptomycin (Invitrogen). HepG2 cells were seeded at a density of 5 × 105 in 6-well plates and left overnight to increase adherence at 37 °C and 5% CO2 with culture. Pre-miR miRNA precursors and a scrambled negative control precursor were obtained from Life Technologies (hsa-miR-19b-3p, hsa-miR-181a-5p, hsa-miR-449c-5p, negative control #2) and co-transfected [30 nM] with pcDNA3.1-EGFP [1 μg] for 48 h, prior to fluorescence-activated cell sorting and subsequent RNA extraction. 2.5. Interleukin-6 treatment Human and mouse hepatocytes were cultured for 48 h post-plating (24 h for HepG2). Cells were then washed twice in PBS then serumstarved for 24 h. Recombinant human or mouse interleukin-6 (Gibco),

2.9. Read mapping and differential expression analysis We obtained a median of 58 million (HepG2), 19 million (human primary hepatocytes) and 62 million (mouse primary hepatocytes) total reads per sample for the paired-end (49 bp) experiment, and a median of 9.9 million (HepG2), 9.4 million (human primary hepatocytes) and 9.2 million (mouse primary hepatocytes) total reads per sample for the single-end (36 bp) experiment. FASTQ files containing pairedend reads (mRNA only) were mapped to the human (Homo sapiens; UCSC hg19) and mouse (Mus musculus; UCSC mm10) reference genomes using TopHat2 (v2.06) [31] with default parameters. Reads were then mapped to known exons annotated by GENCODE v12 (human). For mouse annotations, we used the UCSC Table Browser to derive a table of known mouse genes. Read counts for each gene were obtained using the HTSeq python script [32]. Differential expression analysis was performed using the generalized linear modeling (GLM) and exact test algorithms in the edgeR package (v.3.8.5) [33] with a threshold of 5 counts per million (cpm) in all samples. The cutoff for significant differential expression was log2 fold change ≥1 with a false discovery rate (FDR) of 0.1. RPKM values were obtained using the rpkm() function in the edgeR package. Single-end reads (small RNA-seq) were filtered prior to mapping using the FASTX-Toolkit (http://hannonlab. cshl.edu/fastx_toolkit) to (i) remove adapter-only reads, (ii) retain reads with a quality score ≥ 10 for 50% of bases/read, (iii) discard Nonly reads and (iv) retain reads between 16 and 27 nt. The remaining

Please cite this article as: S.W. Lukowski, et al., Genomics (2015), http://dx.doi.org/10.1016/j.ygeno.2015.05.001

S.W. Lukowski et al. / Genomics xxx (2015) xxx–xxx

reads were mapped to the reference genomes and annotated as for paired-end reads. Differential expression analysis was performed as described for paired-end reads. 2.10. Functional annotation and enriched pathways analysis Functional analyses of differentially expressed genes (DEGs) were performed using the ConsensusPath database (CPdB) [34,35]. Lists of DEGs from each time point were separated into up- and down-regulated (UR/DR) categories and analyzed using the over-representation analysis module. Gene ontology categories were limited to levels 2–4 of cellular component (C), biological process (B) and molecular function (M) with an FDR cutoff of 0.1. To identify enriched pathways in the lists of DEGs, gene lists were analyzed against the default collection of pathway databases, which includes KEGG, Reactome and BioCarta. 2.11. miRNA integration and statistical analysis mRNA–miRNA integration analysis was performed using the miRvestigator web server (http://mirvestigator.systemsbiology.net) (36) using default parameters. DEGs for each time point were combined into two lists per cell type (UR/DR) and submitted for analysis. Hypergeometric analysis was performed using the list_overlap_p Python script (https://github.com/brentp/bio-playground/blob/master/ utils/list_overlap_p.py). Gene lists required for the Python script were as follows: (i) list of UR or DR DE mRNAs, (ii) list of TargetScan-predicted, conserved mRNA targets for a given miRNA, and (iii) background list of all human or mouse genes. All other statistical analyses were performed in R v3.1.2 or GraphPad Prism 6.0d. All error bars represent standard error of the mean (SEM). 3. Results 3.1. Temporal effects of interleukin-6 stimulation on the hepatic transcriptome The liver is the site of production of a large number of acute phase proteins, which are released during an inflammatory response. Stimulation of hepatocytes with a cytokine such as interleukin-6 (IL-6) induces an acute phase response through activation of the JAK/STAT pathway [37]. Although onset is relatively rapid, the hepatic expression of acute phase proteins is not immediate or simultaneous following cytokine induction, and is variable between individuals and species. Changes in temporal expression of APPs can be detected between 4 and 5 h in

3

rats [38], but are generally measured between 6 and 8 h and again at 24 and 72 h [39,40]. The early time points reflect a balance between detectable protein levels and inter-individual and -animal variabilities, while the later time points allow the measurement of later-expressed APPs, as well as accounting for those with longer mRNA half-lives. To investigate the dynamics of hepatocyte-specific transcription during different stages of the APR, we stimulated three hepatocyte cell types with IL-6 and collected RNA from each at 0 h (untreated), 6 h and 24 h and performed RNA-seq on the mRNA and small RNA fractions. We tested for differential gene expression in three time windows (IL-6 (+) 0–6 h, 0–24 h, and 6–24 h), and at two static time points, with or without IL-6 stimulation (IL-6 (+/−) 6 h and 24 h). Time-zero samples were not exposed to IL-6. Over the 24 h time period, we identified a total of 582 DEGs in HepG2, 892 in human primary hepatocytes and 1716 in mouse primary hepatocytes (FDR 10%, ≥ 5 reads per treatment in all samples, log2 FC (fold-change) ≥ 1). Venn diagram analysis shows the total number, as well as the intersection, of DEGs for each cell type at specific time points in IL-6stimulated cells (Fig. 1). For IL-6 (+) cells, a greater number of genes were differentially expressed in the earlier time point (0–6 h) in HepG2 (H) and human primary hepatocytes (h) than in mouse (m) (17%, 99/582 (H); 25.1%, 224/892 (h); 5%, 86/1716 (m)), whereas a higher proportion of DEGs appeared between 6 and 24 h in mouse than human (19.8% (H); 6.3% (h); 41.7% (m)), suggesting a prolonged or delayed response to the IL-6 stimulation. This was also evident in the static time point (treated vs. untreated) data for IL-6-treated cells at 6 h (26.8% (H); 38.4% (h); 8.6% (m)) and 24 h (73.2% (H); 61.6% (h); 91.4% (m)). Next, we assembled lists of the top 20 up- and down-regulated genes per time point and condition, sorted by log2 fold change. Differentially expressed genes at representative time points (0–6 h and 0–24 h) are shown in Fig. 2 while the remaining time points (6–24 h; IL-6 (+) 6 h; IL-6 (+) 24 h) are available in Fig. 1, Ref. [41]. In terms of gene IDs, there seemed to be little similarity between the cell types, therefore, we conducted a deeper analysis of enriched biological pathways and gene ontology (GO) terms using the ConsensusPath database [34,35]. Gene ontology analysis (FDR ≤ 0.1) revealed significantly enriched terms associated with cytokine stimulation and the activation of coagulation and immune response pathways in the IL-6 (+) 0–6 h and static IL-6 (+/−) 6 h time points, modeling the acute phase response. In HepG2 and mouse hepatocytes, this response was even more pronounced in the IL-6 (+) 0–24 h and static IL-6 (+/−) 24 h time points. In human primary hepatocytes, there was a shift toward hypoxia response pathways and cytokine production at 24 h. This was also

Fig. 1. Venn diagram of differentially expressed genes in HepG2 and human and mouse hepatocytes. Overlap of DEGs between cell types at different time points (0–6 h, 0–24 h and 6–24 h) in hepatocytes stimulated with IL-6. Time zero = untreated cells. Genes located in the intersection of all cell types at times 0–6 h and 0–24 h, and between any two cell types at 6–24 h are listed below each diagram. DEGs are genes with a read count ≥5 in all samples, a fold-change ≥1 or ≤1 and below the false discovery rate (FDR) threshold of 0.1.

Please cite this article as: S.W. Lukowski, et al., Genomics (2015), http://dx.doi.org/10.1016/j.ygeno.2015.05.001

4

S.W. Lukowski et al. / Genomics xxx (2015) xxx–xxx

Fig. 2. Top 20 up- and down-regulated genes in IL-6 stimulated hepatocytes. Top 20 up- and down-regulated genes in HepG2, human primary hepatocytes and mouse primary hepatocytes between 0 and 6 h (panels A–C, respectively), and between 0 and 24 h (panels D–F, respectively). Plotted values are the log2 fold-change of gene expression.

evident in the IL-6 (+) 6–24 h window, suggesting that the intensity and duration of the IL-6 response in human primary hepatocytes are reduced compared to the other cell types (see Figs. 2–4 in Ref. [41]). Enriched pathway analysis of the DEGs from the three cell types at each time point revealed IL-6-mediated signaling in the form of JAK– STAT pathway activation, in addition to other cytokine-mediated pathways such as IL-2, -4, -7 and -17. In HepG2 and mouse primary hepatocytes, numerous pathways including those of the complement system, coagulation and fibrinolysis, were significantly enriched at 0–6 and 0– 24 h. These pathways were only strongly enriched at 0–6 h in human primary hepatocytes, further supporting a shorter, human hepatocytespecific, IL-6-induced response. To validate our RNA-seq data, we measured the mRNA expression of several intersecting genes in HepG2 cells (CP, OSMR, IL1R1, FGA, FGG,

FGB) using qPCR (Fig. 3; also see Fig. 5 in Ref. [41]). Since they are strongly influenced by IL-6 stimulation, we also measured changes in fibrinogen mRNA compared to the RNA-seq data (RPKM) in the other two cell lines, as well as fibrinogen protein synthesis in the conditioned media of human primary hepatocytes by western blot (see Fig. 6 in Ref. [41]). 3.2. Differentially expressed miRNAs are integrated with differentially expressed mRNAs miRNAs are becoming increasingly important as biomarkers in many types of cancer and, specifically, as circulating markers of both acute and chronic diseases. We decided to investigate the changes in hepatic miRNA expression during an IL-6-induced inflammatory

Please cite this article as: S.W. Lukowski, et al., Genomics (2015), http://dx.doi.org/10.1016/j.ygeno.2015.05.001

S.W. Lukowski et al. / Genomics xxx (2015) xxx–xxx

5

Fig. 3. Expression profiles of differentially expressed genes determined by RNA-seq at 0, 6 and 24 h in HepG2 cells treated with IL-6 and validated by qPCR. The expression profiles of differentially expressed genes identified in the intersection of multiple cell types treated with IL-6 at different time points were validated in HepG2 cells by qPCR analysis. The black bars show the baseline expression in untreated cells over 24 h, whereas the white bars show the gene expression in cells treated with IL-6 [50 ng/mL] at 6 h and 24 h. Error bars represent the standard error of the mean.

response using small RNA-seq. Using both pairwise and generalized linear modeling (GLM) analyses (filtered as for mRNA analysis, except ≥1 read per sample), we identified a combined total of 20 differentially expressed miRNAs (DE miRNAs) from three cell types (H = 4; h = 4; m = 12) (Fig. 4). Given the high level of miRNA conservation between human and mouse, we analyzed their expression profiles together to provide more power for integration with mRNA data, with the exception of 3 human-specific and 1 mouse-specific miRNA, which were analyzed in a species-specific manner. Of the 20 DE miRNAs, we observed 5 that were down-regulated at 6 h post-IL-6 treatment but returned to normal, unstimulated levels at 24 h (miR-19ab, -26ab, -181ab, -211a, -455), 9 that were significantly up-regulated (UR) at 24 h (miR-34c, -92b, -126, -211, 500b, -1286, -3177, -3662 and mmumiR-1199) and 2 that were significantly down-regulated (DR) at 24 h (miR-449a, -455) (Fig. 4). To validate the small RNA-seq expression profiles, we selected several miRNA genes with altered expression in HepG2 cells following IL-6 stimulation (MIR19B, MIR181A, MIR455, MIR126 and MIR21), and measured their expression by qPCR (Fig. 5). Comparing these data to the small RNA-seq data, we found that the expression changes for each miRNA over the IL-6 stimulation time were largely consistent between both methods. We hypothesized that the time points in which these DE miRNAs showed altered expression would correlate with a set of differentially and inversely expressed mRNAs in each cell type, with the potential for a delayed change in mRNA levels. To test whether a common regulatory thread existed between the differentially expressed mRNAs, we created gene lists that combined all up- or down-regulated DEGs, irrespective of time point, to increase the likelihood of finding

a common binding motif in the 3′UTRs of DE mRNAs. We then used the miRvestigator algorithm [36] to predict (i) whether a common miRNA binding motif existed in the 3′UTRs of the differentially expressed mRNAs, and (ii) which miRNAs were likely to bind to each identified motif (Table 1; also see Table 2 in [41]). For those miRvestigator-identified miRNAs, we used TargetScan (v6.2) [42] to generate a list of the conserved targets for each and applied a hypergeometric test to determine the probability of the DE mRNAs present in each list of conserved targets being regulated by a given miRNA (see Table 3 in [41]). We found that DE miRNAs were more likely to target differentially expressed, up-regulated genes, compared to differentially expressed, down-regulated genes, in response to IL-6 stimulation. This suggests a mechanism whereby those miRNAs with reduced expression at 6 h post-IL-6 stimulation may derepress the production of certain APR proteins that are their predicted regulatory targets. The 3′UTRs of up-regulated DE mRNAs were mainly enriched for potential binding motifs complementary to DE miRNAs at 0–6 h and 0–24 h in HepG2; 0–6 h, 6–24 h and 6 h in human primary hepatocytes; and all time points for mouse primary hepatocytes with a higher proportion of enriched DE mRNAs at 0–24 h and 6–24 h. For down-regulated DE mRNAs, the only significant miRNA seed sequence enrichment occurred at 0–6 h, in which HepG2 cells were enriched for miR-18ab; human primary hepatocytes for miR-17, miR-18ab, miR92ab and miR-3177; and mouse primary hepatocytes for mmu-miR19ab. Many of the significant miRNAs identified as potential regulators of genes with altered expression in response to IL-6 stimulation arise from the miR-17–92 cluster (whose members include miR-17-5p, miR-18a, miR-19a, miR-19b, miR-20a, and miR-92). This data supports

Please cite this article as: S.W. Lukowski, et al., Genomics (2015), http://dx.doi.org/10.1016/j.ygeno.2015.05.001

6

S.W. Lukowski et al. / Genomics xxx (2015) xxx–xxx

Fig. 4. Differentially expressed miRNAs at 0, 6 and 24 h in hepatocytes treated with IL-6. Heatmap of mean log2-transformed cpm values for DE miRNAs in IL-6-stimulated HepG2, human primary hepatocytes and mouse primary hepatocytes (untreated (UT), 6 h and 24 h). DE miRNAs are those with a cpm ≥1 in all samples, a fold-change ≥1 or ≤1, and below the FDR threshold of 0.1. The color key indicates the genes with low (blue) to high (red) expression. Gray boxes represent the values below the accepted read count threshold (undetectable expression) or miRNAs not identified in a cell type (species-specific).

a previous study by Brock et al. [27] in which they demonstrated that IL6 alters the expression of miRNAs in the miR-17–92 cluster, specifically miR-18a, in human primary hepatocytes and HepG2 cells. Having identified the significant post-IL-6 stimulation time points at which DE miRNAs had the highest probability of regulating DE mRNAs in our data, we then looked at which mRNAs predicted by TargetScan were within our DE mRNA data. We compared the DE mRNAs at 0– 6 h and 0–24 h post-IL-6 stimulation in our data to the predicted mRNA targets of hsa-miR-17/20ab, -19ab, -181a and -455 in all three cell types (see Tables 4–6 in [41]). This revealed several up-regulated mRNAs that were differentially expressed in all three cell types (ETV6, NRP1, IL1R1, MTUS1, PNRC1, TACC1), as well as several that were only observed in two (BCL3, ELF3, LPGAT1, STAT3, CLMN, DHRS3, ERBB3). Several UR mRNAs were predicted targets of more than one DR miRNA (Nek7, ETV6, LPGAT1, ERBB3, HOXD1, RORA, SMAD7, Sox4, TMCC1, Tnrc6b) in either multiple cell types or one cell type alone. The miR-17–92 cluster miRNAs are predicted to target the interleukin-1 receptor 1 (IL1R1) mRNA, with miRNA seed sequences for miR-17, -19, -20, -92 and also miR-181 in the IL1R1 3′UTR, in addition to numerous other inflammation-related genes. Based on a study in mice specifically lacking miR-17–92 in T cells, where Il1r1 expression levels were increased [43], we asked whether IL1R1 expression would change in HepG2 cells through an interaction with miR-19 or -181. To see whether the expression of IL1R1 and other predicted mRNA targets could be altered by specific DE miRNAs, we transiently overexpressed miRNA precursors sharing predicted mRNA targets (miR-19b, -181a, -449c), and measured changes in target mRNA expression by qPCR. In total, we tested nine genes, five of which were from the intersection of DEGs at the 0–24 h post-IL-6 time point shown in Fig. 1 (IL1R1, OSMR, ETV6, FABP1 and SERPINC1), and four mRNA targets predicted by TargetScan (SOCS3, PIAS3, FGB and IL1A). Many of these genes are directly involved in the inflammatory response. Our data confirms that overexpression of miR-19b, but not miR-181a, which are both predicted to target IL1R1, leads to significantly reduced IL1R1 mRNA expression in HepG2 cells (Fig. 6A; P = 0.0025). Conversely, expression of the SOCS3 mRNA, which is also a predicted target of both miR-19b and -181a, was significantly reduced by miR-181a (P = 0.0157), but not miR-19b (Fig. 6B). Expression of IL1A, whose protein product is the ligand for the IL1R1 receptor, also exhibited reduced gene expression mediated by miR-19b, despite not being a predicted target (Fig. 6C; P = 0.0330). The IL1A 3′UTR is a predicted target of both miR-181a and -449c, and our data revealed that IL1A expression was slightly reduced, but

Fig. 5. qPCR validation of differentially expressed miRNAs. Differentially expressed miRNA profiles observed in the small RNA-seq data were validated in HepG2 cells treated with IL-6 [50 ng/mL] using qPCR. Error bars represent the standard error of the mean.

Please cite this article as: S.W. Lukowski, et al., Genomics (2015), http://dx.doi.org/10.1016/j.ygeno.2015.05.001

S.W. Lukowski et al. / Genomics xxx (2015) xxx–xxx

7

Table 1 Identification of common miRNA seed motifs in the 3′UTRs of differentially expressed genes. For each cell type, the percentage of 3′UTRs in differentially expressed up- or down-regulated genes that contained a common 8 nt miRNA binding motif was calculated using the miRvestigator algorithm. Due to the higher quality of miRNA seed motif annotation for human 3′UTRs, the mouse genes were analyzed using 3′UTR datasets of both human and mouse. Cells

# genes analyzed

mRNA regulation direction

% UTRs containing motif

Motif

P-value

HepG2

321 120 184 108 541 517 541 517

UR DR UR DR UR DR UR DR

13.0 11.0 8.3 11.0 5.9 3.1 14.0 9.2

UGUACAUA AUGUUUGG UGUACUGU UGUACCGU UGUACUGU UGUACUGU UUGUACCG CAAUAAAG

2.4e−04 4.9e−04 7.6e−05 2.4e−04 7.6e−05 7.6e−05 3.1e−05 1.2e−04

Human Mouse (human 3′UTR set) Mouse (mouse 3′UTR set)

significantly so, by miR-181a, and was significantly reduced by miR449c (Fig. 6C; P = 0.0430). Interestingly, overexpression of miR449c leads to the unexpected increase of five mRNAs that were tested, and four of these were predicted miR-449c targets by TargetScan (Figs. 6A, B, D, E, G; P = 0.0110, 0.0183, 0.0064; 0.0682 (ns); 0.0013 respectively). This may indicate the de-repression of an upstream regulator of these mRNAs by miR-449c, rather than a direct

miRNA–mRNA interaction. In addition, two mRNAs with unexpected expression changes were (i) SERPINC1, which encodes the coagulation protein antithrombin, was significantly reduced by miR-449c (Fig. 6H; P = 0.0007) and (ii) FGB, encoding the fibrinogen beta chain, was increased by miR-181a (Fig. 6F; P = 0.0023). Neither mRNA is a predicted target of the relevant miRNA. This data suggests that, during the early stages of an IL-6-induced acute phase response in hepatocytes,

Fig. 6. mRNA abundance of differentially expressed genes predicted to be regulated by DE miRNAs. Pre-miR precursor miRNAs were overexpressed in HepG2 cells (30 nM) to test for regulatory effects on TargetScan-predicted target mRNAs. The expression of selected mRNA targets from the intersection of differentially expressed genes in IL-6-stimulated hepatocytes, and several from known inflammation pathways was measured by qPCR. TargetScan-predicted mRNA targets with altered gene expression are marked with a filled circle, and those with no change in expression are marked with an empty circle, for the relevant miRNA. Plots without circles represent the genes that are not predicted to be targeted by the relevant miRNA. Error bars represent the standard error of the mean. Significant changes in gene expression are shown with an asterisk (*, P b 0.05; **, P b 0.01; ***, P b 0.001).

Please cite this article as: S.W. Lukowski, et al., Genomics (2015), http://dx.doi.org/10.1016/j.ygeno.2015.05.001

8

S.W. Lukowski et al. / Genomics xxx (2015) xxx–xxx

specific miRNA families with reduced expression inversely regulate the expression of important inflammation genes, likely through a derepression mechanism. 4. Discussion Our data describes, in a single study, the effect of interleukin-6 on gene expression in multiple hepatocyte cell models. To our knowledge, this is the first comprehensive, high-throughput sequencing analysis of mRNA and miRNA expression in hepatocytes stimulated with IL-6. We examined changes in mRNA and microRNA expression at multiple time points in three hepatocyte model cell types – HepG2, human primary hepatocytes and mouse primary hepatocytes – using high-throughput sequencing. We sought to detail the IL-6-induced variation in gene expression in these cell types and to determine the level of integration between APR genes and the miRNAs that might regulate their expression. We used RNA-seq to identify differentially expressed genes at 0, 6 and 24 h in the presence or absence of IL-6. By performing small RNA-seq under the same conditions, we derived a list of both conserved and species-specific, differentially expressed miRNAs that can regulate the expression of differentially expressed genes during the APR. Enriched pathway and gene ontology analysis allow visualization of differentially expressed genes interacting in complex networks. As expected, all three cell types showed strong involvement in cytokine-mediated inflammatory responses, however, human hepatocytes appeared to have a shortened duration of inflammation-related gene expression compared to the other cell types, whereas mouse hepatocytes appeared to undergo a prolonged response. When we examined differential miRNA expression, we found several that were significantly down-regulated at 6 h, but returned to normal or near-normal, unstimulated levels at 24 h, as well as some miRNAs that were strongly increased or decreased at 24 h, compared to timezero, or untreated cells. By integrating the miRNA expression levels with the differentially expressed mRNAs and the predicted and conserved targets of these miRNAs, we found that a number of downregulated miRNAs had an increased probability of modulating the expression of genes that were up-regulated following IL-6 treatment. Interestingly, many of miRNAs expressed from the miR-17–92 cluster were predicted to regulate DEGs due to their reduced expression at 6 h in response to IL-6. In contrast, very few DE miRNAs were predicted to regulate the down-regulated DE mRNAs. Notably, the most significant effect that we observed in the down-regulated mRNAs was in the human hepatocyte data and was predominantly associated with miRNAs from the miR-17–92 cluster, suggesting that miRNAs in this cluster can simultaneously up- and down-regulate important acute phase genes. Some results from the miRvestigator analysis indicated that the star arm of several miRNAs could regulate differentially expressed mRNAs. Studies of miRNA star-arm activity (the less dominant arm of the miRNA precursor) are limited and generally suggest a lack of function. However, recent research demonstrated flexible miRNA star-arm selection in different rice tissues in response to stress [44], implying that tissue- and condition-specific expression of star arm transcripts may have a role in mammalian stress or inflammatory responses. We tested the functional effects of several differentially expressed miRNAs on some of their predicted targets, confirming a number of TargetScan-predicted miRNA–mRNA interactions (Fig. 6). Intriguingly, when we overexpressed the miR-449c precursor, the expression of several mRNAs that are involved in cytokine interactions or the JAK–STAT pathway significantly increased (IL1R1, SOCS3, PIAS3, OSMR). We found that the increased expression of the IL1 receptor mRNA, IL1R1, corresponded to a reduction in the expression of its ligand interleukin 1 alpha (IL1A), but for miR-19b, which only targets IL1R1, both the receptor and ligand mRNAs were reduced. IL-6 is known to inhibit IL1 protein activity [45], and it is possible that altered miR-449c expression during the IL-6 inflammatory response directly increases

the expression of several genes, or inhibits the expression of an upstream regulator. In the case of miR-19b, the targeted inhibition of IL1R1 mRNA led to a coincident reduction of IL1A, which may represent a negative feedback effect. In addition, our data showed that SOCS3 is regulated by miR-181a in IL-6-stimulated HepG2 cells, suggesting that an IL-6-mediated reduction of miR-181a may underlie increased SOCS3 expression during the APR. A number of miRNAs highlighted in this study (miR-17/20ab, -19ab, -181a and -455) have been previously shown to regulate inflammatory mediators and cytokines such as TGF-beta, IL-1a, IL-1b, IL-6 and TNFalpha in several cell lines and animal models [27,46–51]. Our study reinforces these previous findings and adds a new perspective to the early miRNA-mediated changes in inflammation-related gene expression. The expression of many of the mRNAs in our data may be derepressed due to reduced miRNA expression at 6 h post-IL-6 stimulation. Indeed, in a physiological event such as injury or inflammation, controlled modulation of gene expression is crucial. Fine-tuning of gene expression by temporarily reduced miRNA expression may promote, or de-repress, the production of certain proteins during IL-6 stimulation in hepatocytes. In several cases, up-regulated genes that we identified in our data as being potential targets of multiple downregulated miRNAs may be the targets of miRNAs sharing similar seed sequences or arising from a single genomic cluster (e.g. miR-17–92). Depending on the expression profiles of specific miRNA family members, this could buffer or perhaps, conversely, enhance the regulatory effects on target gene expression. Based on these findings, we propose a mechanism of transiently reduced miRNA expression that could de-repress the production of acute phase proteins shortly after IL-6 stimulation, as demonstrated by the interactions between miR-19b and IL1R1, and miR-181a and SOCS3 [43]. We have identified a short time window in which several important miRNAs are down-regulated after 6 h of IL-6 stimulation. This rapid and temporary change in miRNA expression may have a substantial, and potentially long-lasting, impact on the production of a large number of acute phase proteins by de-repressing their expression. These earlystage miRNAs likely play a crucial functional role in the post-transcriptional regulation of APR genes and could be strong candidates for biomarkers or novel drug targets. Author disclosure statement No competing financial interests exist. Acknowledgments This study was supported by a grant from the Dr Henri DuboisFerrière Dinu Lipatti Foundation (http://www.dfdl.org), Swiss National Science Foundation (http://www.snf.ch/en) grant 31003A_134967 (M.N-A) and Sinergia CRSII3_147637 (P.M). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. We thank Stylianos E. Antonarakis and Karim Bouzakri for helpful discussions, and Corinne Di Sanza, Alexandra Planchon, Deborah Bielser, Luciana Romano and Christian Vesin for expert technical assistance. The computations were performed at the Vital-IT (http://www.vital-it.ch) Center for high-performance computing of the SIB Swiss Institute of Bioinformatics. References [1] H. Baumann, K.R. Prowse, S. Marinković, K.A. Won, G.P. Jahreis, Stimulation of hepatic acute phase response by cytokines and glucocorticoids, Ann. N. Y. Acad. Sci. 557 (1989) 280–296. [2] C. Gabay, I. Kushner, Acute-phase proteins and other systemic responses to inflammation, N. Engl. J. Med. 340 (1999) 448–454. [3] C. Gabay, I. Kushner, Acute phase proteins, eLS, John Wiley & Sons, Ltd, Chichester, 2001.

Please cite this article as: S.W. Lukowski, et al., Genomics (2015), http://dx.doi.org/10.1016/j.ygeno.2015.05.001

S.W. Lukowski et al. / Genomics xxx (2015) xxx–xxx [4] IL6R Genetics Consortium Emerging Risk Factors Collaboration, Interleukin-6 receptor pathways in coronary heart disease: a collaborative meta-analysis of 82 studies, Lancet 379 (2012) 1205–1213. [5] S.A. Huber, P. Sakkinen, D. Conze, N. Hardin, R. Tracy, Interleukin-6 exacerbates early atherosclerosis in mice, Arterioscler. Thromb. Vasc. Biol. 19 (1999) 2364–2367. [6] A.D. Pradhan, J.E. Manson, N. Rifai, J.E. Buring, P.M. Ridker, C-reactive protein, interleukin 6, and risk of developing type 2 diabetes mellitus, JAMA 286 (2001) 327–334. [7] J. Spranger, A. Kroke, M. Möhlig, K. Hoffmann, M.M. Bergmann, M. Ristow, H. Boeing, A.F.H. Pfeiffer, Inflammatory cytokines and the risk to develop type 2 diabetes: results of the prospective population-based European Prospective Investigation into Cancer and Nutrition (EPIC)—Potsdam Study, Diabetes 52 (2003) 812–817. [8] F.A. Houssiau, J.P. Devogelaer, J. Van Damme, C.N. de Deuxchaisnes, J. Van Snick, Interleukin-6 in synovial fluid and serum of patients with rheumatoid arthritis and other inflammatory arthritides, Arthritis Rheum. 31 (1988) 784–788. [9] J. Usón, A. Balsa, D. Pascual-Salcedo, J.A. Cabezas, J.M. Gonzalez-Tarrio, E. MartínMola, G. Fontan, Soluble interleukin 6 (IL-6) receptor and IL-6 levels in serum and synovial fluid of patients with different arthropathies, J. Rheumatol. 24 (1997) 2069–2075. [10] A. Waage, C. Kaufmann, T. Espevik, G. Husby, Interleukin-6 in synovial fluid from patients with arthritis, Clin. Immunol. Immunopathol. 50 (1989) 394–398. [11] M. Humbert, G. Monti, F. Brenot, O. Sitbon, A. Portier, L. Grangeot-Keros, P. Duroux, P. Galanaud, G. Simonneau, D. Emilie, Increased interleukin-1 and interleukin-6 serum concentrations in severe primary pulmonary hypertension, Am. J. Respir. Crit. Care Med. 151 (1995) 1628–1631. [12] E. Soon, A.M. Holmes, C.M. Treacy, N.J. Doughty, L. Southgate, R.D. Machado, R.C. Trembath, S. Jennings, L. Barker, P. Nicklin, et al., Elevated levels of inflammatory cytokines predict survival in idiopathic and familial pulmonary arterial hypertension, Circulation 122 (2010) 920–927. [13] A.M. Chinnaiyan, M. Huber-Lang, C. Kumar-Sinha, T.R. Barrette, S. Shankar-Sinha, V.J. Sarma, V.A. Padgaonkar, P.A. Ward, Molecular signatures of sepsis: multiorgan gene expression profiles of systemic inflammation, Am. J. Pathol. 159 (2001) 1199–1209. [14] C. Coulouarn, G. Lefebvre, R. Daveau, F. Letellier, M. Hiron, L. Drouot, M. Daveau, J.-P. Salier, Genome-wide response of the human Hep3B hepatoma cell to proinflammatory cytokines, from transcription to translation, Hepatology 42 (2005) 946–955. [15] C. Coulouarn, G. Lefebvre, C. Derambure, T. Lequerre, M. Scotte, A. Francois, D. Cellier, M. Daveau, J.-P. Salier, Altered gene expression in acute systemic inflammation detected by complete coverage of the human liver transcriptome, Hepatology 39 (2004) 353–364. [16] J. Cubedo, T. Padró, L. Badimon, Coordinated proteomic signature changes in immune response and complement proteins in acute myocardial infarction: the implication of serum amyloid P-component, Int. J. Cardiol. 168 (2013) 5196–5204. [17] K. Nakata, R. Saitoh, J. Amano, T. Ichibangase, M. Ishigai, K. Imai, Comprehensive and temporal analysis of secreted proteins in the medium from IL-6 exposed human hepatocyte, Biomed. Chromatogr. 28 (2014) 742–750. [18] K. Nakata, R. Saitoh, J. Amano, A. Koshiyama, T. Ichibangase, N. Murao, K. Ohta, Y. Aso, M. Ishigai, K. Imai, Alteration of intracellular secretory acute phase response proteins expressed in human hepatocyte induced by exposure with interleukin-6, Cytokine 59 (2012) 317–323. [19] S.A. Thacker, P. Robinson, A. Abel, D.J. Tweardy, Modulation of the unfolded protein response during hepatocyte and cardiomyocyte apoptosis in trauma/hemorrhagic shock, Sci. Rep. 3 (2013) 1187. [20] S. Pochareddy, H.J. Edenberg, Chronic alcohol exposure alters gene expression in HepG2 cells, Alcohol. Clin. Exp. Res. 36 (2012) 1021–1033. [21] K.C. Van Dycke, R.M. Nijman, P.F. Wackers, M.J. Jonker, W. Rodenburg, C.T. van Oostrom, D.C. Salvatori, T.M. Breit, H. van Steeg, M. Luijten, et al., A day and night difference in the response of the hepatic transcriptome to cyclophosphamide treatment, Arch. Toxicol. 89 (2015) 221–231. [22] S. Matsumoto, Y. Sakata, S. Suna, D. Nakatani, M. Usami, M. Hara, T. Kitamura, T. Hamasaki, S. Nanto, Y. Kawahara, et al., Circulating p53-responsive microRNAs are predictive indicators of heart failure after acute myocardial infarction, Circ. Res. 113 (2013) 322–326. [23] A. Courboulin, R. Paulin, N.J. Giguère, N. Saksouk, T. Perreault, J. Meloche, E.R. Paquet, S. Biardel, S. Provencher, J. Côté, et al., Role for miR-204 in human pulmonary arterial hypertension, J. Exp. Med. 208 (2011) 535–548. [24] G. Wang, L.S. Tam, E.K.M. Li, B.C.H. Kwan, K.M. Chow, C.C.W. Luk, P.K.T. Li, C.C. Szeto, Serum and urinary free microRNA level in patients with systemic lupus erythematosus, Lupus 20 (2011) 493–500. [25] L. Gailhouste, L. Gomez-Santos, K. Hagiwara, I. Hatada, N. Kitagawa, K. Kawaharada, M. Thirion, N. Kosaka, R.-u. Takahashi, T. Shibata, et al., miR-148a plays a pivotal role in the liver by promoting the hepatospecific phenotype and suppressing the invasiveness of transformed cells, Hepatology 58 (2013) 1153–1165. [26] C. Camps, H.K. Saini, D.R. Mole, H. Choudhry, M. Reczko, J. Guerra-Assunção, Y.-M. Tian, F.M. Buffa, A.L. Harris, A.G. Hatzigeorgiou, et al., Integrated analysis of microRNA and mRNA expression and association with HIF binding reveals the

[27]

[28]

[29]

[30] [31]

[32] [33]

[34] [35]

[36]

[37] [38]

[39]

[40] [41]

[42]

[43]

[44] [45]

[46]

[47]

[48]

[49]

[50] [51]

9

complexity of microRNA expression regulation under hypoxia, Mol. Cancer 13 (2014) 28. M. Brock, M. Trenkmann, R.E. Gay, S. Gay, R. Speich, L.C. Huber, MicroRNA-18a enhances the interleukin-6-mediated production of the acute-phase proteins fibrinogen and haptoglobin in human hepatocytes, J. Biol. Chem. 286 (2011) 40142–40150. Y. Zhang, B. Zhang, A. Zhang, X. Li, J. Liu, J. Zhao, Y. Zhao, J. Gao, D. Fang, Z. Rao, IL-6 upregulation contributes to the reduction of miR-26a expression in hepatocellular carcinoma cells, Braz. J. Med. Biol. Res. 46 (2013) 32–38. T.H. Nguyen, J. Oberholzer, J. Birraux, P. Majno, P. Morel, D. Trono, Highly efficient lentiviral vector-mediated transduction of nondividing, fully reimplantable primary hepatocytes, Mol. Ther. 6 (2002) 199–209. J.E. Klaunig, P.J. Goldblatt, D.E. Hinton, M.M. Lipsky, B.F. Trump, Mouse liver cell culture. II. Primary culture, In Vitro 17 (1981) 926–934. D. Kim, G. Pertea, C. Trapnell, H. Pimentel, R. Kelley, S.L. Salzberg, TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions, Genome Biol. 14 (2013) R36. S. Anders, P.T. Pyl, W. Huber, HTSeq—a Python framework to work with highthroughput sequencing data, Bioinformatics 31 (2015) 166–169. M.D. Robinson, D.J. McCarthy, G.K. Smyth, edgeR: a Bioconductor package for differential expression analysis of digital gene expression data, Bioinformatics 26 (2010) 139–140. A. Kamburov, U. Stelzl, H. Lehrach, R. Herwig, The ConsensusPathDB interaction database: 2013 update, Nucleic Acids Res. 41 (2013) D793–D800. A. Kamburov, C. Wierling, H. Lehrach, R. Herwig, ConsensusPathDB—a database for integrating human functional interaction networks, Nucleic Acids Res. 37 (2009) D623–D628. C.L. Plaisier, J.C. Bare, N.S. Baliga, miRvestigator: web application to identify miRNAs responsible for co-regulated gene expression patterns discovered through transcriptome profiling, Nucleic Acids Res. 39 (2011) W125–W131. J.G. Bode, P.C. Heinrich, Interleukin-6 signaling during the acute-phase response of the liver, Liver Biol. Pathobiol. 4 (2001) 565–580. A.H. Gordon, A. Koj, Changes in the rates of synthesis of certain plasma proteins following tissue damage due to talc injection. A study using the perfused rat liver, Br. J. Exp. Pathol. 49 (1968) 436–447. E. Olivier, E. Soury, J.L. Risler, F. Smih, K. Schneider, K. Lochner, J.Y. Jouzeau, G.H. Fey, J.P. Salier, A novel set of hepatic mRNAs preferentially expressed during an acute inflammation in rat represents mostly intracellular proteins, Genomics 57 (1999) 352–364. M. Warwas, J. Osada, Changes of the level of proteinase inhibitors in rat plasma during turpentine-induced inflammation, Experientia 41 (1985) 633–634. S.W. Lukowski, R.J. Fish, J. Martin-Levilain, C. Gonelle-Gispert, L.H. Bühler, P. Maechler, E.T. Dermitzakis, M. Neerman-Arbez, Integrated analysis of mRNA and miRNA expression in response to interleukin-6 in hepatocytes, Data in Brief, 2015. (submitted). B.P. Lewis, C.B. Burge, D.P. Bartel, Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets, Cell 120 (2005) 15–20. D. Baumjohann, R. Kageyama, J.M. Clingan, M.M. Morar, S. Patel, D. de Kouchkovsky, O. Bannard, J.A. Bluestone, M. Matloubian, K.M. Ansel, et al., The microRNA cluster miR-17~92 promotes TFH cell differentiation and represses subset-inappropriate gene expression, Nat. Immunol. 14 (2013) 840–848. W. Hu, T. Wang, E. Yue, S. Zheng, J.H. Xu, Flexible microRNA arm selection in rice, Biochem. Biophys. Res. Commun. 447 (2014) 526–530. H. Tilg, E. Trehu, M.B. Atkins, C.A. Dinarello, J.W. Mier, Interleukin-6 (IL-6) as an antiinflammatory cytokine: induction of circulating IL-1 receptor antagonist and soluble tumor necrosis factor receptor p55, Blood 83 (1994) 113–118. M. Brock, M. Trenkmann, R.E. Gay, B.A. Michel, S. Gay, M. Fischler, S. Ulrich, R. Speich, L.C. Huber, Interleukin-6 modulates the expression of the bone morphogenic protein receptor type II through a novel STAT3-microRNA cluster 17/92 pathway, Circ. Res. 104 (2009) 1184–1191. J. Brockhausen, S.S. Tay, C.A. Grzelak, P. Bertolino, D.G. Bowen, W.M. d'Avigdor, N. Teoh, S. Pok, N. Shackel, J.R. Gamble, et al., miR-181a mediates TGF-beta-induced hepatocyte EMT and is dysregulated in cirrhosis and hepatocellular cancer, Liver Int. 35 (2015) 240–253. A.S. Collins, C.E. McCoy, A.T. Lloyd, C. O'Farrelly, N.J. Stevenson, miR-19a: an effective regulator of SOCS3 and enhancer of JAK–STAT signalling, PLoS ONE 8 (2013) e69090. M.P. Gantier, H.J. Stunden, C.E. McCoy, M.A. Behlke, D. Wang, M. Kaparakis-Liaskos, S.T. Sarvestani, Y.H. Yang, D. Xu, S.C. Corr, et al., A miR-19 regulon that controls NFκB signaling, Nucleic Acids Res. 40 (2012) 8048–8058. J.P.J. Skinner, A.A. Keown, M.M.W. Chong, The miR-17~92a cluster of microRNAs is required for the fitness of Foxp3+ regulatory T cells, PLoS ONE 9 (2014) e88997. W. Xie, M. Li, N. Xu, Q. Lv, N. Huang, J. He, Y. Zhang, miR-181a regulates inflammation responses in monocytes and macrophages, PLoS ONE 8 (2013) e58639.

Please cite this article as: S.W. Lukowski, et al., Genomics (2015), http://dx.doi.org/10.1016/j.ygeno.2015.05.001