Irf8-regulated genomic responses drive pathological inflammation during cerebral malaria

Abstract

Interferon Regulatory Factor 8 (IRF8) is required for development, maturation and expression of anti-microbial defenses of myeloid cells. BXH2 mice harbor a severely hypomorphic allele at Irf8 (Irf8(R294C)) that causes susceptibility to infection with intracellular pathogens including Mycobacterium tuberculosis. We report that BXH2 are completely resistant to the development of cerebral malaria (ECM) following Plasmodium berghei ANKA infection. Comparative transcriptional profiling of brain RNA as well as chromatin immunoprecipitation and high-throughput sequencing (ChIP-seq) was used to identify IRF8-regulated genes whose expression is associated with pathological acute neuroinflammation. Genes increased by infection were strongly enriched for IRF8 binding sites, suggesting that IRF8 acts as a transcriptional activator in inflammatory programs. These lists were enriched for myeloid-specific pathways, including interferon responses, antigen presentation and Th1 polarizing cytokines. We show that inactivation of several of these downstream target genes (including the Irf8 transcription partner Irf1) confers protection against ECM. ECM-resistance in Irf8 and Irf1 mutants is associated with impaired myeloid and lymphoid cells function, including production of IL12p40 and IFNγ. We note strong overlap between genes bound and regulated by IRF8 during ECM and genes regulated in the lungs of M. tuberculosis infected mice. This IRF8-dependent network contains several genes recently identified as risk factors in acute and chronic human inflammatory conditions. We report a common core of IRF8-bound genes forming a critical inflammatory host-response network.

Figure 1. BXH2 mice do not develop cerebral malaria following Plasmodium berghei infection.

(A) Survival plots of Plasmodium berghei (PbA) infected BXH2 mice (n = 18), heterozygous [BXH2×B6]F1 offspring (n = 15), and ECM-susceptible B6 (n = 19) parental controls. Shown are the combined results of four experiments. (B) Blood parasitemia levels during infection for mice in Panel A following infection with PbA. (C) Irf8 genotype-specific survival curves for 24 [BXH2×B6]F2 mice along with parental controls. (D) Qualitative comparison of representative Evans blue dyed brains from uninfected and infected B6 and BXH2 mice indicating breakdown of the blood-brain barrier in infected B6 (d7 PbA), but not BXH2 (d7 PbA or d16 PbA) mice. (E) Perfused brains from ECM-susceptible B6 and ECM-resistant BXH2 and [BXH2×B6]F1 mice (n = 2 to 4) were collected six days following infection with PbA. Infiltrating leukocytes were enriched by Percoll gradient and stained with CD45, Ly6C and CD11b, or CD45, TCRβ, CD4 and CD8 antibodies. The presence of myeloid and lymphoid infiltrates is observed in the brain of B6 mice compared to BXH2 or F1. Gate R3 denotes infiltrating cells gated by side scatter (SSC-A) and forward scatter (FSC-A).

Figure 2. Transcript profiling of PbA infected B6 and BXH2 brains reveals strain and infection specific differences.

(A) Unsupervised principal components analysis clusters samples according to mouse strain and infection status. (B) Intersection of gene lists generated by pairwise comparisons between infected and uninfected B6 and BXH2 transcript profiles. (C) Euclidean clustered heat map of transcripts regulated in both a strain and infection specific manner (two-factor ANOVA, p_adj-interaction<0.05) illustrated as infection-induced fold change in each strain (d7/d0). Each row represents a unique gene, and in cases where two or more transcript probes for a gene were significant, the average fold change was used. Differential expression patterns clustered into three groups with Group 1 genes being up-regulated by infection in both strains, Group 2 genes up-regulated by infection in B6 mice and unresponsive in BXH2 and Group 3 genes down-regulated by infection, typically more so in B6 than BXH2. See Table S1 for details. Red shaded heat map indicates the presence of one or more IRF8 binding sites within 20 kb of the gene transcription start site as determined by ChIP-seq. (D) Gene ontology for transcripts differentially regulated by infection during ECM pathology in B6 mice (as determined by B6 d7/d0 pairwise analysis, p_adj<0.05). Up-regulated genes are predominantly involved in innate and adaptive immunity processes, while down-regulated genes are not, and rather include a variety of homeostatic biological and metabolic processes.

Figure 3. Genes up-regulated during ECM pathology in B6 mice are significantly enriched for IRF8 binding sites.

(A) Quantitative PCR was used to validate ChIP results using known binding targets of IRF8. Targets were highly enriched in IRF8-immunoprecipitated DNA when compared to control IgG preparations. Representative data from one of five independent experiments is shown. (B) The list of genes regulated by infection in B6 mice (d7/d0 pairwise) was interrogated for IRF8 binding sites within 225 kb of their transcription start site. The graph represents the abundance of IRF8 binding sites in each 25 kb segment. (C) IRF8 and control IgG ChIP-seq sequence reads were mapped to the mouse genome and significant IRF8 binding sites were identified. Light blue (top) track indicates non-specific (IgG) sequencing profile and dark blue track (below) displays IRF8 binding sites. Genes were considered to have an IRF8 binding site if a peak was found within 20 kb of the transcription start site.

Figure 4. IRF8-regulated pro-inflammatory networks commonly activated during cerebral malaria and pulmonary tuberculosis.

Networks were generated for the 53 genes up-regulated in B6 mice by both infections which also possess an IRF8 binding site within 20 kb of the TSS (see Table S2 for details). (A) Top scoring network highlighting IRF signaling and MHC class I antigen presentation with gene circles colored according to fold change during PbA infection in B6 d7/d0 (left), BXH2 d7/d0 (center) and M. tuberculosis infection in B6 d30/d0 (right). (B) Second top scoring network highlights Ifng and Stat1 signaling. (C) Third top scoring network highlights MHC Class II and Fc receptors. Genes included in the list of 53 are represented by black circles, while networked genes added by the software are outlined in gray. IRF8-bound genes are outlined in light blue. Direct (black) and indirect (gray) connections between genes are shown by arrows.

Figure 5. Effect of deletion of Ifng, Stat1, Jak3, Irf1, Irgm1, Il12p40, Ifit1, Isg15 and Nlrc4 on susceptibility to PbA induced cerebral malaria.

Control and mutant mice were infected with 10⁶ P. berghei parasites and survival was monitored. Cerebral malaria susceptible mice succumbed between d5 and d10 post-infection with neurological symptoms, while no mice that survived longer than 13 days developed signs of ECM and these were categorized as resistant. Infection specific B6 and BXH2 controls (n>5 per infection) are plotted alongside each mutant strain.

Figure 6. Characterization of the myeloid compartment in PbA-infected B6, BXH2, [BXH2×B6]F1 hybrids and Irf1^−/−.

Spleens from B6, BXH2, [BXH2×B6]F1 and IRF1^−/− mice were harvested prior to or six days following PbA infection, and processed by flow cytometry. 2×10⁶ splenocytes were stained with CD45, CD11b, Ly6C, Ly6G, F4/80, CD11c and MHCII, and representative cellular profiles are shown for each strain (A). The numbers within contour plots refer to gates R1 (monocytes) or R2 (granulocytes) and are reported as mean ± SD (gated as percentages of CD45⁺ cells). Absolute numbers of day 6 PbA-infected mice are shown for (B) CD11b⁺Ly6G⁺ and (C) CD11b⁺F4/80⁺ populations, indicating an expansion of the myeloid lineage in the BXH2 strain, compared to B6, [BXH2×B6]F1 and IRF1^−/− animals. Reduced numbers of myeloid dendritic cells (D) along with lower serum IL12p40 levels (E) are noted in ECM-resistant BXH2 and [BXH2×B6]F1 compared to ECM-susceptible B6 mice. (F) Levels of secreted IL12p40 were determined in culture supernatants from splenocytes of infected mice. Dashed gray line represents IL12p40 detection limit. Differences were considered significant when p<0.05 and calculated compared to the B6 strain (Student's t-test: *p<0.05, **p<0.01, ***p<0.001).

Figure 7. Characterization of the lymphoid compartment in PbA-infected B6, BXH2, [BXH2×B6]F1 and Irf1^−/− mice.

Characterization of the lymphoid compartment was carried out as described in the legend of Figure 6. Splenocytes were stained with CD45, TCRβ, CD4 and CD8 antibodies and representative cellular profiles are shown for each strain (A), where numbers indicate mean ± SD (gated as percentages of CD45⁺ cells). Absolute numbers are indicated in dot plots for total spleen CD4⁺ cells (B) and CD8⁺ cells (C). (D) Serum IFNγ levels were significantly lower in ECM-resistant BXH2, [BXH2×B6]F1 and Irf1^−/− mice. (E) IFNγ production was assayed in vitro in culture supernatants from infected mice with or without stimulation with PMA/Ionomycin or with IL12p70. The p-values were calculated relative to B6 controls with Student's t-test (**p<0.01, ***p<0.001).