Combined ChIP-Seq and transcriptome analysis identifies AP-1/JunD as a primary regulator of oxidative stress and IL-1β synthesis in macrophages

Abstract

Background: The oxidative burst is one of the major antimicrobial mechanisms adopted by macrophages. The WKY rat strain is uniquely susceptible to experimentally induced macrophage-dependent crescentic glomerulonephritis (Crgn). We previously identified the AP-1 transcription factor JunD as a determinant of macrophage activation in WKY bone marrow-derived macrophages (BMDMs). JunD is over-expressed in WKY BMDMs and its silencing reduces Fc receptor-mediated oxidative burst in these cells.

Results: Here we combined Jund RNA interference with microarray analyses alongside ChIP-sequencing (ChIP-Seq) analyses in WKY BMDMs to investigate JunD-mediated control of macrophage activation in basal and lipopolysaccharide (LPS) stimulated cells. Microarray analysis following Jund silencing showed that Jund activates and represses gene expression with marked differential expression (>3 fold) for genes linked with oxidative stress and IL-1β expression. These results were complemented by comparing whole genome expression in WKY BMDMs with Jund congenic strain (WKY.LCrgn2) BMDMs which express lower levels of JunD. ChIP-Seq analyses demonstrated that the increased expression of JunD resulted in an increased number of binding events in WKY BMDMs compared to WKY.LCrgn2 BMDMs. Combined ChIP-Seq and microarray analysis revealed a set of primary JunD-targets through which JunD exerts its effect on oxidative stress and IL-1β synthesis in basal and LPS-stimulated macrophages.

Conclusions: These findings demonstrate how genetically determined levels of a transcription factor affect its binding sites in primary cells and identify JunD as a key regulator of oxidative stress and IL-1β synthesis in primary macrophages, which may play a role in susceptibility to Crgn.

Figure 1

Strategy employed to identify primary JunD targets in bone marrow derived macrophages. We performed Jund RNAi and compared whole genome expression profiling in macrophages (basal and stimulated with LPS, 100ng/ml, 8h) transfected with scrambled siRNA to those transfected with Jund siRNA. ChIP-Seq analysis was performed in WKY and WKY.LCrgn2 BMDMs (basal and stimulated with LPS, 100ng/ml, 2h). Transcripts showing a fold change > 3 in the RNAi dataset were examined for JunD/AP1 peaks.

Figure 2

siRNA mediated knockdown of Jund. Assessment of the efficiency of siRNA knockdown of Jund in the unstimulated state (A, C) and following eight hours LPS stimulation (B) using qRT-PCR and Western blotting. siRNA experiments were performed in 4 different WKY rats in triplicate. ***P<0.001 using two tailed unpaired t-test to compare BMDMs transfected for 48 hours with either scrambled control siRNA or Jund siRNA. The Western blot (C) for JunD is representative of four different Jund silencing experiments in WKY BMDMs and is demonstrated alongside β-actin loading control.

Figure 3

Validation of microarray results confirms that JunD has both activatory and repressive roles in controlling gene expression linked with oxidative stress. Validation of differentially expressed genes were carried out using four biological replicates with three technical amplification replicates per siRNA type for the unstimulated (A) and eight hour LPS stimulated (B) data sets. Relative gene expression was normalised to Hprt and used to generate fold change values. *P<0.05; **P<0.01;***P<0.001 using a two-tailed unpaired t-test to detect statistically significant differences between the siRNA groups. Error bars represent standard error of the mean. Confirmation of the differential expression of Jund between WKY and WKY.LCrgn2 BMDMs (C) and key JunD targets; Arg1 (D), Cdo1 (E) and Mt2a (F), influenced by Jund siRNA knockdown with >2 fold change in expression that were also differentially expressed between WKY and WKY.LCrgn2 BMDMs over a timecourse of LPS stimulation. Samples from the WKY and WKY.LCrgn2 strains were amplified using a set of four biological replicates with three technical replicates per sample. ***P<0.001 statistically significantly different to WKY using a two way ANOVA to compare the overall timecourse with Bonferonni’s post-tests to compare individual time points.

Figure 4

Genetically determined differences in Jund expression alter the JunD cistrome and identifies primary JunD targets. (A) Distribution of JunD-peaks relative to transcriptional start sites (TSS) of Ensembl genes. Promoter region defined as 20 kilobases (kb) upstream of the TSS, upstream region between 10kb and 50kb upstream from TSS. (B) Occurrence of peaks within 100 kilobases of the TSS. (C) Twelve base pair AP-1 motif identified by de novo motif analysis present in 63% peaks. De novo motif analysis using HOMER identified two de novo motifs in basal WKY BMDMs (D) and four motifs in LPS stimulated WKY BMDMs (E). The de novo motif identified is displayed on the bottom of each the pair of motifs, the matched consensus motif for a transcription factor on the top. (F) Il1b and Prkca confirmed as primary JunD targets by qPCR validation. The aligned reads comprising peak passing the posterior probability threshold of 0.9 for each JunD-bound gene in the WKY strain in the LPS stimulated state for l1b and the basal state for Prkca are shown in genome browser views along with the peak in the WKY.LCrgn2 strain. Samples from WKY and WKY.LCrgn2 strains were amplified using three biological replicates with three technical replicates per sample. Results expressed as mean fold change over IgG. *P<0.05; **P<0.01; using a one-tailed unpaired t-test to detect statistically significant differences between the strain and condition pairs. Error bars represent standard error of the mean.

Figure 5

Integrative analysis identifies primary JunD targets in basal and LPS stimulated BMDMs.Jund gene expression patterns in WKY and WKY.LCrgn2 BMDMs over the LPS time course were used for Spearman correlation analysis with the rest of the transcripts on the microarrays (P < 0.001 cut off). The correlated genes were used for building the JunD target gene networks and selected based on i. significant differential expression following Jund siRNA knockdown (both basal and after LPS stimulation FDR< 5%), ii. presence of the JunD ChIP-Seq peak in WKY BMDMs (both basal and after LPS stimulation). The outer ring represents all the transcripts correlating with Jund expression levels (P<0.001) in WKY and WKY.LCrgn2 BMDMs (1445 transcripts, 75%, indicated in grey circles) over the LPS time course. Transcripts associated with a JunD ChIP-Seq peak (within 20kb of TSS or within the gene body) are shown as black circles (232 transcripts, 12%). The transcripts correlating with Jund expression levels and down-regulated following Jund siRNA (basal, upper panel) are shown in blue (125 transcripts, 6.5%); and those up-regulated (basal, upper panel) are shown as red circles (116 transcripts, 6%). Primary JunD targets in basal BMDMs are given with the gene names and show a JunD ChIP-Seq peak (24 transcripts, basal state, upper panel). This analysis was repeated for the LPS-stimulated macrophages taking into account JunD siRNA and ChIP-Seq datasets in LPS stimulated macrophages and identified 36 primary Jund targets (lower panel).

Figure 6

JunD expression levels control the production of mature IL-1β in BMDMs and nephritic glomeruli. (A) Western blot of mature IL-1β expression in WKY, WKY.LCrgn2, LEW and LEW.WCrgn2 BMDMs primed with LPS and stimulated with ATP. The result of 3 independent experiments is demonstrated alongside β-actin loading control. ELISAs for IL-1β in LPS primed and ATP activated BMDMs (B) and in nephritic glomeruli (C). **P<0.01 statistically significantly different to WKY using a one way-ANOVA with Bonferonni’s post-tests.