Stress signaling boosts interferon-induced gene transcription

Abstract

Promoters of antimicrobial genes function as logic boards, integrating signals of innate immune responses. One such set of genes is stimulated by interferon (IFN) signaling, and the expression of these genes (ISGs) can be further modulated by cell stress-induced pathways. Here, we investigated the global impact of stress-induced p38 mitogen-activated protein kinase (p38 MAPK) signaling in macrophages on their response to IFN. We found that in response to cell stress induced by an inhibitor of protein synthesis, the p38 MAPK-activated transcription factors CREB and c-Jun mediated transcriptional synergism with IFN-activated signal transducers and activators of transcription (STATs). p38 MAPK signaling induced activating histone modification at loci of IFN-stimulated genes (ISGs) and stimulated nuclear translocation of the CREB coactivator CRTC3. Disrupting the synergy between p38 MAPK and IFN reduced the amount of cell death in macrophage cultures after Listeria monocytogenes infection. Our findings uncover a mechanism of transcriptional synergism and highlight the biological consequences of coincident stress-induced p38 MAPK and IFN-stimulated JAK–STAT signal transduction.

Introduction

Macrophages provide innate immunity through recognition, phagocytosis, and killing of pathogens and through the synthesis of regulatory cytokines. The signaling environment of an immunologically activated macrophage is generally determined by a multitude of pathways that arise from the activation of pattern-recognition and cytokine receptors and that may reinforce or inhibit each other. The modular nature of control elements in target gene promoters provides an important means to integrate the output of signaling pathways through cooperativity of transcription factors. Among macrophage-activating cytokines, interferons (IFNs) play a major role (1, 2). Type I IFN (IFN-I, including IFN-α/β) are mainly antiviral, whereas the type II IFN (IFN-γ) increase effector functions against nonviral pathogens. IFN receptors (IFNRs) signal through Janus kinases (JAK) and their tyrosine phosphorylation of signal transducers and activators of transcription (STATs) (3–6). IFN receptor signaling activates the transcription of IFN-induced genes (ISGs) by generating the transcription factors ISGF3 and GAF. ISGF3 consists of a tyrosine-phosphorylated STAT1-STAT2 heterodimer in association with IFN-regulatory factor 9 (IRF9). It is formed predominantly by the type I IFNR and stimulates transcription by associating with IFN-stimulated response elements (ISREs) of ISG promoters. On the other hand, GAF is a homodimer of tyrosine-phosphorylated STAT1 and is generated in larger quantities by IFN-γR signaling. It activates the transcription of ISGs by binding to IFN-γ-activated sequences (GAS) of ISG promoters.

ISG expression is modulated by additional signaling pathways. For example, a large contingent of ISGs is regulated by cooperativity between STAT and nuclear factor-κB (NF-κB) pathways (7–8). Activity of the p38 mitogen-activated protein kinase (p38 MAPK, or p38) pathway has also been linked to the control of ISG expression (9, 10). Pathways mediated by c-Jun N-terminal kinase (JNK) and p38 are activated through a wide variety of stimuli, such as environmental stress, growth factors, inflammatory cytokines or infection. Among the p38 family enzymes (p38α, β, γ, δ), p38α is the only isoform expressed ubiquitously (11, 12). Dual Thr/Tyr phosphorylation by MAP2K activates p38, which creates a complex signaling network by phosphorylation of several downstream targets (13–15). Studies posit a direct involvement of p38 in IFNR signaling and in the transcriptional activation of ISGs through ISRE and GAS promoter elements (10, 16).

In this study, we systemically analyzed the interplay between stress and IFN pathways in macrophages. Our data show that p38 is not directly involved in IFN signaling but rather synergizes with IFN-induced JAK-STAT pathways when cells are simultaneously exposed to IFN and stress. We defined the subset of stress-enhanced ISGs and found that cooperativity between STAT and CREB or c-Jun transcription factors was the basis for the synergy between p38 and IFN signaling. Analysis of macrophages exposed to Listeria monocytogenes underpinned the importance of p38 and IFN pathway cooperation for the course of infection. Our data highlight the role of ISG promoters as convergence points for the coordinated response to IFN and stress.

Results

p38 MAPK selectively enhances IFN-γ- and IFN-β-induced ISG expression

To study the interaction of stress-induced MAPK and IFN pathways without the confounding effects of NF-κB (8), we used the drug anisomycin in combination with IFN-β or IFN-γ, with or without one of two inhibitors of p38 MAPK activity, PH-797804 (PH) or LY-222820 (LY), in bone marrow-derived macrophages (BMDMs) (Fig. 1A) (9, 17). Anisomycin is a bacterial pyrrolidine antibiotic that has the ability to potently and reversibly inhibit eukaryotic protein synthesis. At the very low concentrations used here, anisomycin is known to cause ribotoxic stress in cells, leading to the activation of JNKs and p38 MAPKs, through a mechanism not yet fully understood (18). As expected, anisomycin caused phosphorylation of both p38 and JNK but did not stimulate degradation of inhibitor of κB (IκB). PH blocked p38 activity without consequences for JNK activation or IκB degradation (fig. S1, A to C). Furthermore, neither IFN-γ nor IFN-β promoted p38 phosphorylation, and anisomycin did not stimulate STAT1 Tyr⁷⁰¹ phosphorylation (fig. S1D) (19).

Fig. 1. Selective ISG enhancement by p38 MAPK.

(A) Experimental overview listing the genotypes used and their analysis. (B to E) Quantitative real-time PCR (q-PCR) showing pre-mRNA expression of the indicated genes in WT BMDMs upon stimulation with anisomycin (An; 100 ng/ml), IFN-γ (10 ng/ml for 30 min; B and C) or IFN-β (250 IU/ml for 1.5 hours; D and E) either alone or in after pre-treatments with anisomycin (20 min) alone or with PH-797804 (PH; 1 μM; 1 hour) or LY-2228820 (LY; 200 nM; 1 hour). (F and G) q-PCR-based mRNA expression of the indicated genes in p38α^ΔM BMDMs and their WT counterparts (p38α^fl/fl) stimulated with anisomycin (100 ng/ml) alone, or (F) IFN-β (250 IU/ml for 2 hours) or (G) IFN-γ (10 ng/ml for 2 hours) alone or after a 20-min pre-treatment with anisomycin. (H and I) q-PCR-based measurement of mRNA expression of the indicated genes in WT BMDMs stimulated as described in (B to C) and (D to E), respectively, minus LY. (J and K) Time-course of Ifit2 pre-mRNA and mRNA expression, analyzed by q-PCR, in WT BMDMs treated as described in (H and I), respectively, minus the inhibitor. Data are mean + SD of three individual experiments, except (F and G) which are from triplicate samples from each of two mice. For (B) to (E) and (H) to (K), one-way analysis of variance (ANOVA) corrected for multiple testing with Dunnett’s post-hoc test was used. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; and ns, not significant.

Treatment with IFN-γ increased the abundance of the STAT1 homodimer-regulated pre-mRNAs Irf1, Socs3 and Gbp5 (20) independently of p38. The combined action of stress and IFN-γ further enhanced Irf1 and Socs3 expression through p38-dependent signals. In contrast, Gbp5 expression remained unaffected by either stress or both p38 inhibitors (Fig. 1, B and C, and fig. S1E). Likewise, IFN-β-induced expression of ISGs was unperturbed in the presence of the either p38 inhibitors, but anisomycin-induced enhancement of the Ifit2 and Ifit3 genes was dependent on p38. Expression of the Mx1 gene was not enhanced by anisomycin. (Fig. 1, D and E, and fig. S1F). The results demonstrate these ISGs represented the following categories: ISRE-controlled, stress-enhanced (Ifit2, Ifit3); ISRE-controlled, enhancement-resistant (Mx1); GAS-controlled, stress-enhanced (Irf1, Socs3); and GAS-controlled, enhancement-resistant (Gbp5). Deficiency in p38α (p38α^ΔM) abolished both the anisomycin-dependent enhancement of Ifit2 stimulation by IFN-β and the enhanced expression of the Irf1 gene by IFN-γ. As expected, p38α deficiency did not reduce expression of the enhancement-resistant Mx1 and Gbp5 genes. (Fig. 1, F and G).

These data show that anisomycin enhances the expression of a subset of IFN-γ-induced genes controlled by STAT1 homodimers as well as a subset of IFN-β-induced genes controlled by ISGF3. Thus we next turned to study enhancement of STAT1 homodimer-controlled genes by IFN-β and that of ISGF3-controlled genes by IFN-γ. We found IFN-β-stimulated expression of the Irf1 and Socs3 genes lacked enhancement by anisomycin. In contrast, induction of Ifit2 and Ifit3 genes with IFN-γ was enhanced by anisomycin (Fig. 1, H and I). Thus, we conclude that induction via GAS or ISRE promoter elements determines whether ISG responses to both IFN types are enhanced by stress.

Kinetic studies showed that anisomycin+IFN-γ stimulation for 4 hours resulted in a larger enhancement of the Ifit2 and Ifit3 mRNA compared to the pre-mRNA (Fig. 1, J and K, and fig. S1, G and H). Tristetraprolin (TTP) is an RNA-binding protein that regulates mRNA decay of inflammatory genes, and its destabilizing activity is inhibited by p38 MAPK (21, 22). TTP-deficient BMDMs (TTP^ΔM) produced the same degree of Ifit3 mRNA enhancement and PH still abolished the anisomycin effect as in WT cells (fig. S1I). The data rule out an involvement of TTP-dependent mRNA decay in stress enhancement of ISG expression.

p38 MAPK increases RNA polymerase II binding to stress-enhanced ISGs

The transcription cycle requires dynamic phosphorylation and dephosphorylation of the RNA polymerase II (Pol II) C-terminal domain (CTD) (23). Whereas Ser⁵ phosphorylation of the Pol II CTD is associated with promoter clearance and promoter-proximal pausing, Ser² phosphorylation is indicative of progressive transcript elongation. We considered the possibility that p38 signaling might selectively increase Pol II phosphorylation rather than promoting its recruitment to ISGs. In line with pre-mRNA expression, chromatin immunoprecipitation (ChIP) analysis showed that anisomycin alone failed to recruit Pol II, but anisomycin and p38 enhanced the binding of both Pol II and its Ser⁵-phosphorylated form (Pol II S5P) to the Ifit3 and Ifit2 promoters compared to IFN-β alone. Relative increases in total Pol II and Pol II S5P were similar. Anisomycin had no impact on IFN-β-dependent Pol II recruitment to the enhancement-resistant Mx1 promoter (Fig. 2, A and B, and fig. S2A). The increases of Pol II and Pol II S5P binding to the Irf1 and Socs3 promoters stimulated by IFN-γ or anisomycin+IFN-γ were comparably small (Fig. 2C and fig. S2B). This confirms findings that the Irf1 transcription start is preloaded with Pol II, in line with the gene’s primary response to IFN-γ (24, 25). Anisomycin did not further stimulate Pol II recruitment at the enhancement-resistant Gbp5 gene (Fig. 2D).

Fig. 2. p38 MAPK dependence of RNA polymerase II binding and histone modification at ISG promoters.

(A to H) BMDMs were stimulated with anisomycin (An; 100 ng/ml), (A,B,E,F) IFN-β (250 IU/ml for 2 hours) or (C,D,G,H) IFN-γ (10 ng/ml for 2 hours) alone or in combination with 20 min pre-treatment in anisomycin with or without PH-797804 (PH; 1 μM) for 1 hour, and processed for site-directed ChIP with control IgG or antibodies to phospho-Ser⁵ or phsopho-Ser² Pol II, and total RNA-Pol II, analyzing promoter and gene body association of (A, B, E, and F) Ifit3 and Mx1 and (C, D, G, and H) Irf1 andGbp5 as indicated. (I and J) BMDMs were stimulated as described in (A and C), respectively and processed for site-directed ChIP with ph-H3S28 antibody or control IgG, analyzing binding to (I) Ifit3 and Mx1 and (J) Irf1 and Gbp5 promoters. (K and L) BMDMs were stimulated as described in (I and J), respectively and processed for site-directed ChIP with H2A.Z antibody or control IgG, analyzing binding to the indicated promoters. Data mean + SD of at least three independent experiments. Two-way ANOVA corrected for multiple testing with Dunnett’s post-hoc test was used. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; and ns, not significant.

In analyzing p38-dependent effects on elongation, we observed similar increases of stress-induced, p38-dependent increase of total Pol II and Ser²-phosphorylated Pol II (Pol II S2P) at the Ifit3 and Ifit2 gene bodies compared to IFN-β alone and this effect was not seen at the Mx1 gene body (Fig. 2, E and F, and fig. S2C). Contrasting the recruitment of Pol II and Pol II S5P at the Irf1 and Socs3 promoters, anisomycin enhanced the IFN-γ-induced presence of Pol II and Pol II S2P at the bodies of these genes, but not at the Gbp5 gene (Fig. 2, G and H, and fig. S2D), confirming that IFN-γ increases Irf1 and Socs3 transcription by stimulating Pol II elongation. Consistent with the pre-mRNA data (Fig. 1H), anisomycin did not enhance the IFN-β-dependent recruitment of phosphorylated Pol II or total Pol II to the Irf1 promoter and gene body (fig. S2E).

In conclusion, the data show that binding of Pol II reflects the control of ISGs pre-mRNA synthesis by IFN and stress, in line with stress enhancement through augmented transcriptional initiation. However, the data do not support the notion of selective enhancement of CTD phosphorylation through the p38 pathway, because anisomycin-induced increases in recruitment were highly similar for total Pol II and its phosphorylated forms.

p38 MAPK signaling introduces activating histone modifications at ISG promoters

Stress-signaling introduces activating histone modifications, particularly histone 3 (H3) phosphorylation of at Ser¹⁰ and Ser²⁸ and its acetylation at Lys⁹ and Lys¹⁴, thereby facilitating the transcription of stress-responsive promoters (26, 27). The amount of Ser²⁸ phosphorylation after anisomycin or IFN-β treatment alone was slightly increased, whereas their combination strongly enhanced the p38-dependent deposition of this mark. Anisomycin synergy with IFN was not selective for enhanced ISG expression, because it was observed at both the enhanced Ifit3 and the enhancement-resistant Mx1 promoters. Similar data were obtained for anisomycin and IFN-γ treatment (Fig. 2, I and J). Two further stress-associated H3 marks, Ser¹⁰ phosphorylation and Lys⁹/Lys¹⁴ acetylation also occurred at all examined ISG promoters (fig. S2, F and G). Contrasting the stress-induced histone marks, H3 Lys²⁷ trimethylation (H3K27me3) erasure and H3 Lys²⁷ acetylation (H3K27ac), events which are more generally associated with gene activation (28), occurred in response to IFN without further input by anisomycin (fig. S2, H and I).

Deposition of the histone variant H2A.Z through p38 signaling is associated with some active promoters, but its removal occurs during transcriptional activation of ISGs by IFN-β (29, 30). The decrease of H2A.Z at ISG promoters occurred in response to both IFN-β and IFN-γ, without an effect of cotreatment with anisomycin or PH (Fig. 2, K and L).

Considered together, these results indicate that p38 signaling increases typical stress-induced H3 modifications independently of enhancing ISG expression and that it does not affect IFN-dependent H2A.Z removal, H3K27ac deposition, and H3K27me3 erasure.

Global analyses of stress-enhanced ISGs and their promoter configuration

The data above generated with representative gene loci did not reveal a selectivity factor deciding between enhancement and enhancement resistance of ISGs. We initiated our search for this factor by globally identifying ISGs affected by stress signaling using bulk RNA sequencing (RNA-seq). Anisomycin treatment alone increased the expression of 2208 mRNAs, including those associated with inflammation (Il1a, Il1b, Nlrp3, Il6), TNF signaling (Inhba, Nfkbiz, Tnfsf18), and stress-response pathways (Ptgs2, Nr4a3, Jun, Fosl1) (fig. S3, A and B, and data file S1). 1467 IFN-γ-induced genes and 1757 IFN-β-induced genes included gene sets either repressed or stimulated by anisomycin alone (Fig. 3, A and B, and data files S2 and S3). Among these, 182 IFN-γ-induced ISGs and 116 IFN-β-induced ISGs were enhanced by anisomycin but did not respond to the drug alone, whereas 103 genes showed enhancement after induction by either IFN type (magenta, royal blue, and aquamarine data sets in Fig. 3, A to C, respectively). Further analysis revealed that, in line with our qPCR results, the group of GAS-controlled ISGs (Irf1, Socs3, Cxcl9) was enhanced only by anisomycin combined with IFN-γ, whereas the group of ISRE-controlled ISGs (Ifit1, Ifit2, Ifit3, Mx2) was enhanced by anisomycin combined with either IFN type. Conversely, Gbp5 and Mx1 were not enhanced by the combination of anisomycin and either IFN type (Fig. 3, C to E, and data file S, 2 and 3). Representative genes enhanced by anisomycin combined with IFN-γ (Cxcl9), IFN-β (Usp18) or either (Mx2) were validated by qPCR (Fig. 3F).

Fig. 3. Genome-wide analysis of p38 MAPK-dependent ISG expression.

(A to E) RNA-seq was performed from three individual replicates of WT BMDMs stimulated with anisomycin (An; 100 ng/ml) alone, or with IFN-γ (10 ng/ml; A, C and D) or IFN-β (250 IU/ml; B, C and E) for 2 hours alone or after pre-treatment with anisomycin for 20 min. Differentially expressed genes were then categorized by response into pie charts (A and B), as indicated in the inset legends. For clarity between the two “enhanced” categories, the magenta slices (A) and royal blue slices (B) represent ISGs that were enhanced by anisomycin only when combined with IFN (annotated with “An<IFN<An+IFN”); the dark/navy blue slices represent ISGs that also responded to anisomycin alone (annotated with “An≥IFN<An+IFN”). IFN Log2FC≥ 1, adjusted p-value ≤ 0.05; An+IFN vs IFN or An ≥ 1. Details provided in data files S2 and S3. (C) Scatterplot comparing the Log2-fold-changes in mRNA abundances from BMDMs described in (A and B). Grey dots represent unsignificantly changed genes (“N.S.”), and aquamarine and orange dots represent ISGs whose expression changed in the same direction (enhanced or not enhanced, respectively) by anisomycin in the presence of either IFN type (“common”; IFN Log2FC≥ 1, adjusted p-value ≤ 0.05; An<IFN<An+IFN; An+IFN vs IFN Log2FC≥ 1). Magenta, royal blue, pistachio-green, and pale blue dots represent genes from the same categories described/labeled in (A and B). (D and E) RNA-seq heatmap of the top 40 enhanced ISGs in BMDMs stimulated with anisomycin alone or IFN-γ (D) or IFN-β (E) alone or in combination with anisomycin. Color Key represents row Z-score and genes are clustered by rows. (F) q-PCR-based measurement of Cxcl9, Usp18 and Mx2 mRNA expression after treatment with anisomycin (100 ng/ml), IFN-β (250 IU/ml for 2 hours) or IFN-γ (10 ng/ml for 2 hours) alone or after pre-treatment with anisomycin (20 min) or with anisomycin and PH (1 μM; for 1 hour). Inset graphs display the IFN-γ-specific data on a smaller y-axis scale. Data are mean + SD of at least three independent replicates. For (F), one-way ANOVA corrected for multiple testing with Dunnett’s post-hoc test was used. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; and ns, not significant.

Principal component analysis (PCA) of the top 40 anisomycin-enhanced ISGs revealed a clear separation between IFN-induced and enhanced ISGs, whereas anisomycin samples showed a similar expression profile to the untreated ones (fig. S3, C and D). To identify the differentially expressed genes (DEGs) associated with the different sample traits and phenotypes, we performed weighted gene co-expression network analysis (WGCNA) (31). A total of 26 WGCNA modules (clusters of highly co-expressed genes) were identified. Among these, 4 modules (designated green, dark olive green, grey, and midnight blue) were positively correlated with the IFN and anisomycin with IFN conditions (fig. S3E). The co-expression network in the green and the dark olive green modules was associated with, respectively, IFN-β- and IFN-γ-induced genes that were either enhanced or not enhanced and significantly (FDR ≤ 0.05) associated with gene ontology (GOs) terms related to general regulation of immune responses. The majority of the genes associated with the grey module consisted of ISGs enhanced by anisomycin combined with IFN-γ that were enriched for antiviral responses, whereas the midnight blue module consisted of enhancement-resistant genes enriched for metabolic processes, such as oxidative phosphorylation (OXPHOS) and mitochondrial respiration (fig. S3F).

Intersection of the RNA-seq data with available ChIP-seq data sets (32), allowed us to relate IFN-β- and IFN-γ-induced gene expression to promoter binding by STAT1, STAT2, and IRF9 (fig. S3F, and data file S, 2 and 3). Genes bound by STAT1 homodimers and/or ISGF3 were overlayed with the set of anisomycin-enhanced ISGs. In accordance with the pre-mRNA data (Fig. 1), gene expression enhancement by anisomycin with IFN-γ occurred at promoters associating with either STAT1 homodimers or the ISGF3 complex (21 and 78 genes, respectively), whereas genes enhanced by anisomycin with IFN-β or either IFN type were bound only by ISGF3 (57 and 66 genes, respectively). The data confirm our conclusions above (from Fig. 1) by clearly defining the subset of enhancement-responsive ISGs and by demonstrating the relevance of GAS and ISRE elements for enhancement response to both IFN types.

Stress signaling does not alter ISGs promoter accessibility

Further addressing the mechanism of p38-dependent enhancement of ISG expression, dynamic changes of chromatin accessibility were explored by ATAC-seq (Assay for Transposase-Accessible Chromatin with high-throughput sequencing). We hypothesized that increased Pol II binding and transcription after anisomycin and IFN cotreatment might result from increased chromatin opening. For data analysis, we divided ISGs according to RNA-seq into enhanced and not-enhanced sets (data file S, 2 and 3). As expected (25), both enhanced and enhancement-resistant gene sets showed chromatin opening upon stimulation with IFN-γ. Anisomycin alone slightly decreased the chromatin accessibility observed in untreated BMDMs (Fig. 4, A and B; first 3 panels) and, unexpectedly, the drug failed to increase accessible chromatin at either enhanced or enhancement-resistant loci when combined with either IFN-γ or IFN-β (Fig. 4, A to D). Inspection of individual gene loci confirmed chromatin opening by IFN and the absence of effects from anisomycin or PH on enhanced (Irf1, Ifit3) and not-enhanced gene loci (Gbp5, Mx1) (Fig. 4, E and F). Chromatin accessibility at anisomycin-induced, stress response genes revealed by RNA-seq (data file S1) remained similarly unchanged (Fig. 4G), suggesting that such genes are primary response genes with pre-configured promoter chromatin (33). In conclusion, nucleosome remodeling is not a driving force behind stress-enhanced ISG or non-ISG expression, suggesting a mechanism of action at subsequent steps of transcriptional activation.

Fig. 4. IFN, but not anisomycin, increases chromatin accessibility at ISG loci.

(A to D) Summary profile plots and heatmap of chromatin accessibility generated by using normalized read coverages in BMDMs treated with anisomycin (An; 100 ng/ml) alone, (A and B) IFN-γ (10ng/ml for 2 hours) or (C and D) IFN-β (250 IU/ml for 2 hours) alone or after pre-treatment with anisomycin (20 min) or anisomycin and PH-797804 (PH; 1 μM; 1 hour). The colored profile represents enhanced ISGs (A and C) and enhancement-resistant ISGs (B and D) according to RNA-seq. Profiles and heatmaps represent regions from between -2 and +1 kb with regard to the TSS as indicated. Merged data from three individual replicates are shown. (E and F) ATAC-seq genome browser tracks at respective gene loci in WT BMDMs treated with anisomycin alone, (E) IFN-γ or (F) IFN-β alone or in combination with anisomycin with or without PH as indicated. One representative replicate is shown. (G) Summary profile plots and heatmap of chromatin accessibility similar to (A to D) showing An-induced genes according to the RNA-seq (Log2FC≥1) in BMDMs that were either untreated (black; UT) or treated with anisomycin (grey; An). Data in (A to G) are derived from three individual replicates.

Enhanced ISG loci are highly enriched in AP-1 and CREB binding motifs: essential role for CREB in IFN and stress cooperativity

To test the hypothesis that p38 enhances transcriptional initiation by recruitment and/or activation of transcription factors, we first examined promoter binding of STAT complexes. Anisomycin caused a slight increment of IFN-β-induced STAT1 binding to the ISRE at the Ifit3 and Mx1 promoters, but this was unaffected by PH. When combined with IFN-γ, anisomycin slightly augmented STAT1 homodimer binding to the GAS motifs in both the enhanced Irf1 and not-enhanced Gbp5 promoters (fig. S4, A and B). These data do not support increased STAT binding as a cause for ISG enhancement by p38. To narrow down enhancement candidates, we performed enrichment analysis for binding motifs based on RNA-seq data. As expected, both enhanced and not-enhanced gene loci were highly enriched in ISRE. Notably, activator protein-1 (AP-1) binding motifs were enriched specifically in enhanced genes (Fig. 5A). The AP-1 subunit c-Jun is involved in stress and immune responses (34, 35, 36), is a constituent of the type I IFN enhanceosome, and is required for induction of some ISGs by IFN-γ (37). c-Jun phosphorylation occurred in response to anisomycin treatment and was partially inhibited by the JNK inhibitor SP600125 (JNKi-II). PH also decreased c-Jun phosphorylation, albeit to a lesser extent, and the combination of the two inhibitors was slightly more effective (fig. S4, C and D). To assess the input of the c-Jun pathway into the enhancement of ISG expression, we pre-treated BMDMs with one of two JNK inhibitors, JNKi-II and JNK-IN-8, which inhibit JNK activity and JNK activation, respectively (Fig. 5B) (38). Both inhibitors slightly decreased anisomycin-induced enhancement of IFN-β-stimulated Ifit2 and Ifit3 expression, and of IFN-γ-stimulated Irf1 and Socs3 expression, reaching statistical significance only for Ifit3. The decrease in anisomycin-induced Ifit3 enhancement caused by JNK inhibitors (Fig. 5C) appeared to be less than that caused by p38 inhibitors (Fig.1D). Furthemore, the decrease of anisomycin enhanced ISG expression, did not reach statistical significance in c-Jun-deficient BMDMs (cJun fl/fl; Vav-iCre^+/-) (Fig. 5D). IFN did not increase c-Jun activation when provided with or without anisomycin pre-treatment (fig. S4E). These results suggest that c-Jun alone cannot account for the entire ISG enhancement by p38 signaling, although it may be a contributing factor.

Fig. 5. ISG enhancement requires CRE/AP-1 binding sites and CREB activity.

(A) Transcription factor prediction for ISGs enhanced by anisomycin, representing the most enriched motif (p ≤0.05). (B) Experimental overview listing the genotypes used and the timing of their treatment and analysis. (C) q-PCR analysis of pre-mRNA expression in BMDMs that were untreated (UT) or stimulated with anisomycin (An; 100 ng/ml) alone, IFN-β (Ifit3, Ifit2; 250 IU/ml for 2 hours) or IFN-γ (Irf1, Socs3; 10 ng/ml for 2 hours) alone or after a pre-treatment with anisomycin, or with anisomycin and SP600125 (JNKi-II; 20 μM; 1 hour pre-treatment) or JNK Inhibitor XVI (JNK-IN-8; 10 μM; 1 hour pre-treatment). (D) q-PCR showing mRNA expression in cJun^fl/fl;Vav-iCre^+/- compared to their WT counterparts (cJun^fl/fl) upon stimulation with anisomycin (100 ng/ml) alone, IFN-β (Ifit3, Ifit2; 250 IU/ml for 2 hours) or with IFN-γ (Irf1, Socs3; 10 ng/ml for 2 hours) alone or after a 20 min treatment with anisomycin. (E) Site-directed ChIP in BMDMs stimulated with anisomycin (100ng/ml) alone, IFN-β (left panel; 250 IU/ml for 2 hours) or IFN-γ (right panel; 10 ng/ml for 2 hours) alone or in combination with anisomycin (20 min pre-treatment) or with anisomycin and PH (1 μM; 1 hour pre-treatment) and processed for ChIP antibody targeting pSer¹³³ CREB1 on the CRE binding site of Ifit3 and Irf1. IgG isotype was used as negative control. (F and G) q-PCR showing mRNA expression in BMDMs transduced with sgRNA targeting CREB1 or a non-targeting control (NTC) upon stimulation with anisomycin (100 ng/ml; 2 hours and 20 min) alone, IFN-β (Ifit2, Mx1; 250 IU/ml for 2 hours) or with IFN-γ (Irf1, Gbp5; 10ng/ml for 2 hours) alone or in combination with anisomycin (20 min pre-treatment). Data are mean + SD from at least three independent replicates. For (C), one-way ANOVA corrected for multiple testing with Dunnett’s post-hoc test was used. For (D) to (G), two-way ANOVA corrected for multiple testing with Dunnett’s post-hoc test was used. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; ns, not significant.

In addition to the AP-1 consensus element, enhanced ISGs were enriched in 3’,5’-cyclic adenosine monophosphate (cAMP) response element (CRE) consensus sequences, which share core nucleotides with the AP-1 element and represent binding sites for ATF/CREB family members (Fig. 5A). CRE-binding protein 1 (CREB1), is activated by p38 (39, 40) and is able to heterodimerize with c-Jun (41, 42). Anisomycin, but not IFN, activated CREB1 by phosphorylation at Ser¹³³ and this was abolished by PH, but not by JNKi-II (fig. S4, F and G). CREB activation occurs in its DNA-bound state (43, 44). Consistently, pSer¹³³ CREB1 activation at the CRE of enhanced ISG promoters was increased by anisomycin pre-treatment of IFN-stimulated BMDMs, and this was blocked by PH (Fig. 5E and fig. S4H). Following deletion of the Creb1 gene in BMDMs by CRISPR/Cas9 editing (fig. S4I) (45), stress enhancement of Irf1 and Ifit2 mRNAs after stimulation with, respectively, IFN-γ and IFN-β was abolished (Fig. 5F), but expression of the enhancement-resistant Gbp5 and Mx1 genes was unabated (Fig. 5G). Unexpectedly, the lack of CREB1 phosphorylation at Ser¹³³ was without consequences for the enhancement of IFN-γ-induced Irf1 expression and IFN-β-induced Ifit2 mRNA expression when assayed in BMDMs derived from CREB-S133A knock-in mice (fig. S4J) (46). Together, these results demonstrate that CREB1 is essential for the enhancement of ISG expression through p38 signaling without strictly requiring its activation by phosphorylation at Ser¹³³.

CREB1 acquires Ser¹³³ phosphorylation-independent transcriptional competence by using the CREB-regulated transcription coactivator family (CRTC) (47–49). In particular, CRTC3 is associated with immunoregulatory genes in macrophages (50). Nuclear access of CRTC is under control of kinase-dependent signaling pathways, prompting us to examine a potential role of p38. Indeed, nuclear translocation of CRTC3 was stimulated by anisomycin, but not by IFN treatment and a larger CRTC3 fraction remained in the cytoplasm in the presence of PH (fig. S4, K and L). Thus, CRTC3 may contribute to p38-dependent CREB activity.

Genome-wide identification of CREB and c-Jun binding at ISG loci

We then performed ChIP coupled with sequencing (ChIP-seq) to globally identify ISGs associating with CREB and/or c-Jun, either constitutively or after treatment with anisomycin. Because c-Jun binding to an ISG subset is controlled by IFN-γ (37), treatment with the cytokine was included. 8430 constitutive CREB1 binding sites and 5614 c-Jun binding sites were found in regions -2/+1 kb relative to TSS, with an overlap of 54.3% (Fig. 6A). By integrating these ChIP-seq data with our RNA-seq data identifying IFN-γ- and IFN-β-induced genes, we discovered 575 ISG loci among the c-Jun and CREB1 co-bound genes, and 240 or 132 ISG loci bound by, respectively, CREB1 or c-Jun alone (Fig. 6, B and C, and data file S4). Whereas CREB1 binding was generally constitutive, c-Jun binding was induced by IFN-γ on 43 genes, including 34 known ISGs (Fig. 6D). Gene set enrichment analysis (GSEA) of the ChIP-seq data indicated that genes bound by both CREB1 and c-Jun or by only c-Jun were enriched for MAPK and JAK-STAT signaling pathways, whereas these annotations were not found for genes bound by only CREB1 (fig. S5, A to C). The transcription factor binding site prediction confirmed the GSEA by revealing that genes bound by CREB and c-Jun or c-Jun only were enriched in STAT, IRF1, IRF7 and ISRE binding sites (fig. S5, D and E), whereas genes bound by only CREB1 were devoid of ISRE or IRF binding motifs but enriched for GAS sequences annotated as STAT3 binding motifs (fig. S5F). The data suggest a predominance of promoters with CREB1 and c-Jun binding sites among enhanced ISGs and, thus, associate c-Jun with IFN-γ signaling (37).

Fig. 6. ChIP-seq analysis of CREB1 and c-Jun binding at ISG loci.

(A to D) Venn diagrams of c-Jun- and CREB1-bound genes detected by ChIP-seq, wherein (A) shows genome-wide c-Jun- and/or CREB1-bound genes, (B) shows the CREB1/c-Jun co-bound ISGs derived from RNA-seq data, (C) shows the ISGs that are bound only by CREB1 or only by c-Jun, and (D) shows the ISGs that are bound by c-Jun and which is induced by IFN-γ (see data file S4). (E and F) Scatterplot comparing the Log2-fold-changes in mRNA abundances from BMDMs treated with IFN-γ or IFN-β. Color code of the dots (genes) defined in the inset keys. “N.S.” denotes not significantly altered genes (IFN Log2FC≤1, adjusted p-value ≥ 0.05),”common” denotes genes for which both IFNs had a similar effect (E) or similar lack of effect (F).(G and H) Genome browser tracks from ChIP-seq and RNA-seq data. For ChIP-seq WT BMDMs were treated with An (100 ng/ml) alone, a combination of JNKi-II (JNKi) and PH-797804 (PH) alone (20 μM and 1 μM, respectively), (G) IFN-γ (10 ng/ml for 2 hours) or (H) IFN-β (250 IU/ml for 2 hours) alone or in combination with An (20 min pre-treatment) or with An and the combination of the two inhibitors (1 hour pre-treatment) as indicated. Tracks are shown for c-Jun (purple), CREB1 (green), and IgG (negative control; black). Merged data from three replicates are shown. Previously published ChIP-seq (29) data were used to show binding sites for STAT1, STAT2 and IRF9, and RNA-seq tracks show IFN and An+IFN samples, each one representative replicate.

CREB1 and c-Jun co-bind enhanced ISGs

To corroborate the GSEA and motif discovery data correlating ISG promoter configuration with ISG enhancement, RNA-seq and ChIP-seq data were combined for a further integrated analysis. Among the 182 genes enhanced by anisomycin in combination with IFN-γ, 71 were co-bound by CREB1 and c-Jun; these included Irf1 and Socs3. 33 of the 116 anisomycin-enhanced, IFN-β–induced ISGs and 49 of the 103 commonly enhanced genes-including Ifit2, Ifit3, Mx2, Rsad2 and Isg20-were bound by both transcription factors (Fig. 6E and data file S4). Among the enhancement-resistant, IFN-γ-induced ISGs, a smaller proportion (74 out of 280, which did not include Mx1 and Gbp5) showed co-binding of the transcription factors as did a smaller proportion of both the anisomycin-resistant, IFN-β-induced ISGs and the commonly not-enhanced ISGs (163 out of 440 and 121 out of 300, respectively, which also did not include Mx1 and Gbp5) (Fig. 6F and data file S4). The data confirm that ISG promoters with binding sites for both CREB1 and c-Jun are enriched among p38-enhanced genes, but that binding of both transcription factors is not strictly necessary.

The genetic data (Fig. 5, D and F) suggest a dominant role of CREB1 and a minor role of c-Jun in Irf1 and Ifit2 enhancement, whose promoters were co-bound by both transcription factors. However, a small number of enhanced genes showed binding of either CREB1 or c-Jun alone (fig. S5, G and H). The latter were investigated in c-Jun-deficient BMDMs. Enhancement of Isg15 was unaffected and that of the IFN-γ-induced Batf2 and Slamf7 genes was slightly decreased by c-Jun deficiency. In contrast, the loss of c-Jun completely abolished IFN-γ-induced expression and enhancement of the Il2ra gene and abrogated the enhancement of IFN-β-induced Pdcd1 mRNA expression. In addition, Pdcd1 mRNA enhancement was similarly reversed by pre-treatment with JNKi-II and JNK-IN-8 in WT BMDMs (fig. S5, I to L). These results confirm the impact of c-Jun on IFN-γ signaling (37) and show that the expression of a small subset of ISGs is enhanced by c-Jun.

Browser tracking of the anisomycin-enhanced Irf1 and Ifit loci indicated binding sites for c-Jun and CREB1 close to STAT1 homodimers and ISGF3 bound to GAS elements and ISRE sequences, respectively. In addition, the RNA-seq tracks visually demonstrated stress-enhanced expression of the Irf1 gene and all clustered Ifit genes (Fig. 6, G and H). In contrast, browser tracks of the enhancement-resistant Gbp5 and Mx1 promoters indicated binding of c-Jun but not of CREB1 (fig. S5, M and N).

p38 MAPK enhances ISG expression and nitric oxide-mediated cell death in macrophages infected with Listeria monocytogenes

To validate the relevance of our data for innate immunity to microbes, we infected macrophages with the intracellular bacterium Listeria monocytogenes (Fig. 7A). Pattern recognition receptor signaling in response to L. monocytogenes infection activates stress-responsive MAPK as well as IRF transcription factors, inducing IFN-β synthesis and subsequent ISG expression (52, 53). Ifnb1 pre-mRNA expression was inhibited by PH, indicating AP-1 family constituents of the IFN-β enhanceosome (54) (Fig. 7B). ISGs associated with activated macrophages, such as Nos2, Il1a, Il1rn and Il6, were similarly induced by L. monocytogenes and inhibited by PH. In contrast, PH did not reduce ISG expression in response to IFN-β treatment (Fig. 7, C and D). Type I IFNR-deficient macrophages (Ifnar1^-/-) lacking the background ISG transcription through endogenous IFN-β synthesis, responded with moderate increase of antimicrobial ISG expression to L. monocytogenes infection or IFN-γ treatment alone, and demonstrated a pronounced synergism in the transcriptional induction by the combined treatment (Fig. 7E). This IFN-γ and infection synergism was eliminated by PH, confirming enhancement of IFN-γ-induced ISGs by the infection-induced p38 pathway. Similar results were obtained in WT BMDMs, where Nos2 pre-mRNA expression resulted from L. monocytogenes infection-induced IFN-I synthesis (fig. S6A).

Fig. 7. p38 MAPK enhancement of Listeria monocytogenes-induced ISG expression and its blockade effect of macrophage viability.

(A) Overview of Listeria monocytogenes (Lm) infection and subsequent analysis. (B and C) q-PCR of pre-mRNA expression of the indicated genes in WT BMDMs infected with Lm (MOI 10; 4 hours) alone or after PH treatment (PH-797804; 1 μM; 1 hour pre-treatment). A PH-alone (5 hours) condition is also shown. (D) q-PCR showing Nos2 pre-mRNA expression in WT BMDMs stimulated with IFN-β (250 IU/ml) for 4 hours alone or after 1-hour pretreatment with PH (1 μM). (E) q-PCR assessing the expression of Nos2, Il1a, Il1m, and Il6 pre-mRNA in Ifnar1^-/- BMDMs cultured in the indicated conditions: stimulated with IFN-γ (10 ng/ml, 4 hours), infected with Lm (MOI 10; 4 hours) and treated with PH (1 μM; 5 hours where alone, 1 hour pre-treatment where combined). UT, untreated. (F) Genome browser tracks assessing c-Jun and CREB1 binding to the Nos2 promoter in BMDMs treated with IFN-γ for 2 hours. Tracks are representative of one of three individual replicates. (G and H) Site-directed ChIP with (G) pSer¹³³ CREB1 or (H) c-Jun antibodies (or control IgG) analyzing binding to the CRE/AP-1 binding site of Nos2 in WT BMDMs infected with Lm (MOI 10) alone or in combination with PH (1 μM; 1 hour pre-treatment) for 4 hours. UT, untreated. (I and J) Colony forming unit (CFU) assay in WT BMDMs infected with Lm (MOI 10) alone or in combination with PH (1 μM; 1 hour pre-treatment) for the indicated time. (K) Griess assay analyzing the concentration of nitrite in the medium of WT BMDMs 24 hours after infection with Lm (MOI 10) alone or in combination PH (1μM; 1 hour pre-treatment). PH alone condition as a control. (L) Cell death assessed by propidium iodide (PI; 1 μg/ml) staining and measured by FACS in WT BMDMs 24 hours after infection and treatment as in (K). Data is representative of one of four biological replicates. FSC-A, forward scatter; Pe-Texas-Red-A, PI fluorescence. Data are mean + SD of at least three independent replicates. For (B) to (E), (G), (H), (K) and (L), one-way ANOVA corrected for multiple testing with Dunnett’s post-hoc test was used. For (I) and (J), two-tailed unpaired t-test was used. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; and ns, not significant.

To verify the relevance of ISG enhancement for L. monocytogenes infection, we explored expression of the Nos2 gene, nitric oxide (NO) production, and NO-dependent macrophage death (37, 55, 56). In agreement with Nos2 pre-mRNA enhancement by p38 signals, its promoter contains binding sites for CREB1 and c-Jun (Fig. 7F). L. monocytogenes infection increased the binding of both pSer¹³³ CREB1 and c-Jun at the CRE/AP-1-bound DNA, and PH completely eliminated promoter-bound Ser¹³³CREB1 (Fig. 7, G and H).

Intra-macrophage growth of L. monocytogenes, determined by colony forming unit (CFU) assay, was slightly reduced by PH between 1 and 8 hours, but strongly diminished between 12 and 24 hours after infection (Fig. 7, I and J). Pre-treatment with IFN-γ prior to infection confirmed the anti-microbial effect of this cytokine by decreasing L. monocytogenes growth at 12 and 24 hours and blockade of p38 together with IFN-γ resulted in even stronger bacterial clearance (fig. S6B) (57). To narrow down the beneficial effect of p38 blockade in the context of L. monocytogenes infection, we sought to assess a potential role of p38 in restricting autophagy (58). L. monocytogenes-stimulated autophagy enhances phagosomal escape early after infection while restricting cytoplasmic growth at later stages (56, 59). During autophagosome formation, the cytosolic form of the microtubule-associated protein 1A/1B-light chain 3 (LC3-I) is converted into LC3-II and recruited to autophagosomal membranes and the ratio between the two forms represents the autophagic rate (60). We observed an increased LC3BII/LC3BI ratio in L. monocytogenes-infected BMDMs at both early and late stages of infection. However, neither IFN-γ pre-treatment nor p38 inhibition affected the autophagic rate, thus excluding autophagy as a mechanism for the observed effect of p38 inhibition on bacterial growth (fig. S6C). We and others have previously shown a linear relationship between NO production and cell death by PANoptosis (a form of cell death involving key features of pyroptosis, apoptosis and necroptosis) in WT macrophages at these late stages of infection (56, 61, 62). The data suggest that by sustaining their viability, PH increases the ability of macrophages to kill the invading bacteria. Consistent with this assumption, both NO production and macrophage death, assessed by propidium iodide (PI) staining, were significantly reduced by PH 24 hours after infection (Fig. 7, K and L). Together, these findings demonstrate that p38 enhancement of ISG expression exacerbates consequences of L. monocytogenes infection, whereas p38 blockade is beneficial for the viability of host macrophages.

Stress and IFN synergism during VSV infection

To explore the IFN/stress synergism in the context of viral infection, we infected BMDMs with GFP⁺ vesicular stomatitis virus (VSV-GFP) (63). VSV-GFP infection resulted in increased anti-viral gene expression, similar to IFN-β stimulation. Pre-treatment with IFN-β followed by VSV infection led to enhancement of gene expression, which was decreased by blockade of JNK and p38 (fig. S6D). IFN-β treatment significantly decreased the levels of VSV viral RNA as well as the percentage of infected cells, as assessed by % of GFP⁺ cells. Pre-treatment with a combination of JNK and p38 inhibitors reverted the IFN-β-dependent effect on VSV mRNA expression and increased the percentage of infected cells (fig. S6, E and F).

Together, these data demonstrate that enhancement of anti-viral genes during VSV infection might contribute to a stronger IFN-I response leading to a decrease in VSV infectivity.

Discussion

The co-occurrence of signaling by sensors of stress and IFN receptors is a hallmark of innate responses to many viral and bacterial pathogens. Here, we show that the enhancement of macrophage ISG expression is mediated by cooperative action of the IFN and p38 signaling pathways in the process of transcriptional initiation (Fig. 8). In contrast to what has been found in mouse embryonic fibroblasts (10), p38 signaling in primary macrophages had no effect on transcriptional induction through ISRE or GAS sequences by the IFN receptors alone. Global definition of stress-enhanced ISGs for both IFN types and compilation of their specific attributes compared to stress-resistant ISGs revealed CRE/AP-1 promoter elements as the cardinal elements of distinction. Reportedly, c-Jun and CREB are transcription factors that are activated by the JNK and p38 pathways, respectively (39, 40, 64), with a minor contribution of p38 to c-Jun phosphorylation in some situations (65), as also seen here (fig. S4E). Whereas CREB1 was essential for the enhanced expression of ISGs associating with both CREB1 and c-Jun, the latter was largely dispensable. At the resolution of ChIP-seq many binding sites for c-Jun and CREB were found to overlap, but the experimental set-up did not enable us to assess whether the two transcription factors bind the same site and whether they interact physically in their promoter-associated state. Generating this knowledge is required to understand the lack of c-Jun requirement for the stress-enhancement of co-bound ISGs. It might result either from redundancy with other members of the AP-1 family that act as surrogates in a knock-out situation (66), or from a lack of essential mechanistic input into transcriptional activation during stress enhancement of ISG. The latter possibility is suggested by the result that a small number of genes associating only with c-Jun require the transcription factor for pathway synergy, demonstrating its inherent potential to cooperate with STATs. Surprisingly, promoters induced exclusively by STAT1 homodimers failed to mediate stress input into gene control by IFN-β, while stress enhancement of IFN-γ inducibility took place. This difference between the IFN types may result from the increased presence of STAT1 homodimers in IFN-γ-treated cells, or from a regulatory capacity of yet undefined factors.

Fig. 8. Working model of stress kinase-mediated control of ISG expression.

Stress-induced-p38-mediated control of ISG expression is triggered on a specific subset of ISGs. In the “no enhancement” status, stress kinases do not alter the IFN-γ- and IFN-β-induced transcription of genes controlled by STAT1 homodimers, such as Gbp5, or the ISGF3 (STAT1-STAT2-IRF9) complex, such as Mx1. The presence of CRE/AP-1 binding sites at ISG promoters is characteristic of the “enhancement” status, allowing stress kinases to enhance transcription. Stress-induced-enhancement occurs at both STAT1 homodimer and ISGF3-controlled ISGs when stimulated with IFN-γ, but only at ISGF3-controlled ISGs when stimulated with IFN-β. L. Monocytogenes infection of macrophages causes the production of IFN-β and activation of the p38 MAPK pathway. Together, they synergistically enhance ISG expression, thus amplifying the consequences of infection such as NO production and NO-dependent cell death.

Given the profound transcriptome changes caused by anisomycin treatment alone we were surprised to find a complete lack of p38-dependent chromatin remodeling. Instead, our findings support the assumption that the transcriptional regulation by anisomycin alone occurs largely at constitutively open promoters with poised polymerases, a characteristic of primary response genes (33). This notion is further consistent with the unresponsiveness of ISGs to treatment with anisomycin alone, which may in part be explained by an ability of STATs to increase accessibility of most ISG loci (25) that cannot be executed by stress-activated transcription factors. IFN-dependent promoter opening may be an important prerequisite for CREB and c-Jun to unfold their activity in transcriptional initiation as also suggested by their constitutive or anisomycin-induced association with ISG promoters.

While our study demonstrates the indispensable role of p38 signaling for the enhanced transcriptional induction of ISG, several possibilities remain how it feeds into the mechanism of transcriptional initiation. The most attractive of these, the phosphorylation of CREB at Ser¹³³, was ruled out, as ISG enhancement was observed in macrophages expressing a CREB-S133A mutant. p38 signaling produced robust phosphorylation of promoter-bound CREB1 without an apparent contribution of the modification to ISG enhancement. The partial block of c-Jun phosphorylation by PH may indicate a role for the kinase in the activation of AP-1 for the small group of ISG showing enhancement via c-Jun. A potentially more general mechanism for the acquisition of phospho-Ser¹³³-independent transcriptional CREB activity is to harness the CRTC3 coactivator (47–49). In support of a role for CRTC3 the protein translocated to the nucleus after anisomycin treatment and translocation was largely blocked by PH. One potential target for the p38 effect in this scenario is salt-inducible kinase 2 (SIK2). SIK2 inhibits nuclear translocation of CRTC3, but inhibition can be overcome by its phosphorylation (50, 67). A more likely possibility is the direct phosphorylation of CRTC3 by p38, as the protein contains several MAPK phosphorylation sites and was shown to be a substrate of ERK. Phosphorylation of these sites increases recruitment of protein phosphatase 2A (PP2A) and CRTC3 dephosphorylation-dependent nuclear translocation (68). Finally, stress-induced ISG enhancement may involve histone modification (26, 27, 29). Anisomycin-induced p38 had a profound effect on H3 modification at ISG promoters, increasing phosphorylation at Ser¹⁰ and Ser²⁸. These effects were not specific for enhanced ISG, but may yet be relevant for the transcriptional activities of CREB and c-Jun in enhancing ISG transcription. Stress-induced H3 modification may create a transcription-permissive histone code at all ISG promoters, but the realization of this potential requires stress-activated transcription factors such as CREB and c-Jun in addition to STAT1 homodimers and/or ISGF3.

The pleiotropic effects of the p38 pathway make it difficult to assess the specific contribution of ISG enhancement to immune responses with genetic or pharmacological tools. To circumvent this problem, we made use of the fact that the ISG Nos2 is subject to stress enhancement and that IFN- and stress-dependent production of NO is a contributing factor to a complex form of L. monocytogenes-induced macrophage death referred to as PANoptosis (56, 61). Inhibitors of p38 had little effect on the intracellular growth of L. monocytogenes at early stages of infection, but they enhanced bacterial killing while also protecting macrophages from PANoptotic death at late stages of infection. The data suggest a role of stress-enhanced NO production in controlling the survival of L. monocytogenes-infected macrophages. The reduced efficacy of IFN-β to curtail VSV replication in presence of p38 inhibitors is also in agreement with a blockade of ISG enhancement during viral infection; however, p38 signaling is likely to influence innate immunity to VSV in various ways, including an impact on IFN-β synthesis (69). Thus, the results reported here do not allow for comprehensive conclusions about the role of p38 MAPK during VSV infection.

Studies on murine SARS-CoV-2 infection identified iNOS-dependent cell death as contributing factor to severe disease (70). Studies in human patients indicate an important role for IFN in immunity against COVID-19 (71) and attribute a role also in the cytokine storm associated with severe disease (70). Likewise, p38 activity has been associated with severe COVID-19 and its inhibition is proposed as a therapeutic strategy (72). Data corroborating a role for IFN and the p38 MAPK pathway is shown in a multi-omics study with blood leukocytes from COVID-19 patients, wherein the authors identified STATs as associated factors and the p38 and AP-1 pathway as a key feature distinguishing severe COVID-19 from influenza virus infection (73). Based on these and our studies, we propose that p38-enhanced ISG expression is a means to efficiently harness antimicrobial effector mechanisms while also carrying the risk of exaggerated, disease-associated inflammation.

Materials and Methods

Animal experiments

C57BL/6N mice (wildtype, WT) were purchased from Janvier Labs. p38α^ΔM mice were previously described (74). cJun fl/fl; Vav-iCre ^+/- mice were housed and bred at the University of Veterinary Medicine of Vienna, Vienna. Knock-in CREB-S133A mice were generated as previously described (46). Rosa26-Cas9 knock-in mice were purchased from The Jackson Laboratory. WT, Ifnar1-/- and Ttp^fl/fl;LysM-Cre mice in C57BL6/N background were housed and bred at the Max Perutz Labs.

Mice were housed under identical conditions in a specific-pathogen-free (SPF) facility according to the Federation of European Laboratory Animal Science Association (FELASA) guidelines and additionally monitored for being norovirus negative. Mice were bred under the approval of the institutional ethics and animal welfare committee of the University of Veterinary Medicine of Vienna and the national authority Federal Ministry Republic of Austria Education Science and Research section 8ff of the Animal Science and Experiments Act [Tierversuchsgesetz (TVG), BMWF-68.205/0068-WF/V/3b/2015 and GZ 2020-0.200.397]. The study did not involve animal experiments as defined in the TVG and did not require ethical approval according to the local and national guidelines. Mice were used at an age of 8-to 12-weeks for the isolation of bone marrow. The study did not involve animal experiments as defined in the TVG and did not require ethical approval according to the local and national guidelines.

Cell cultures

All cell lines used were grown at 37 °C, 5% CO₂. Bone marrow-derived macrophages (BMDMs) were differentiated from bone marrow isolated from femurs and tibias of 8- to 12-week-old mice from both sexes, that were pooled for each experiment. Femur and tibia were flushed with Dulbecco’s modified Eagle’s medium (DMEM; Sigma-Aldrich) and cells were cultured and differentiated for 9 to 10 days in DMEM supplemented with 10% of fetal bovine serum (FBS; ;Sigma-Aldrich), 500ng/ml human recombinant M-CSF (a kind gift from L. Ziegler-Heitbrock, Helmholtz Center, Munich, Germany), 100 units/ml penicillin, and 100 ng/ml streptomycin (both Sigma-Aldrich).

Lenti-X 293T cells (Takara Bio) were cultured in DMEM supplemented with 10% FBS, 100 units/ml penicillin, and 100 ng/ml streptomycin (each Sigma-Aldrich). Transfection of cells was performed using Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer’s instructions.

Reagents

Anisomycin (Sigma-Aldrich, catalog # 176880) was used at the concentration of 100 ng/ml. Murine IFN-γ (a kind gift from G. Adolf, Boehringer Ingelheim, Vienna) was used at the concentration of 10 ng/ml, and murine IFN-β (PBL Assay Science, catalog # 12400-1) was used at the concentration of 250 IU/ml. PH-797804 (Selleckchem, catalog # S2726) was used at the concentration of 1 μM, LY2228820 (Selleckchem, catalog # S1494) at the concentration of 200 nM, JNK inhibitor II (SP600125; Sigma-Aldrich, catalog # 420119) at the concentration of 20 μM, and JNK-IN-8 (JNK Inhibitor XVI; Selleckchem, catalog # S4901) at the concentration of 10 μM. LPS (Sigma-Aldrich, catalog # L2637) was used at the concentration of 100 ng/ml. VSV-GFP was grown on Vero cells for 42 hours, after which supernatants were clarified and titrated on MDCK cells by plaque assay, as previously described (75).

Immunoblotting

Cell lysis and immunoblotting was performed as previously described (32). Primary antibody to α-tubulin (catalog # T9026, used at 1:5000) and vinculin (V9131, 1:5000) were purchased from Sigma-Aldrich. Primary antibody to GAPDH was purchased from Millipore (catalog # ABS16, 1:3000). The following primary antibodies were from Cell Signaling Technology: p38 phospho-Thr¹⁸⁰/Tyr¹⁸² (catalog #9211, used at 1:1000); total p38 (catalog #9212, 1:1000); SAPK/JNK phospho-Thr¹⁸³/Tyr¹⁸⁵ (catalog #9251, 1:1000); total SAPK/JNK (catalog #9252, 1:1000); MSK1 phospho-Thr⁵⁸¹ (catalog #9595, 1:1000); total MSK1 (catalog #3489, 1:1000); CREB1 phospho-Ser¹³³ (catalog #9198, 1:1000); total CREB (catalog #4820, 1:1000); c-Jun phospho-Ser⁷³ (catalog #9164, 1:1000); total c-Jun (catalog #9165, 1:1000); total IκBα (catalog #9242, 1:1000); and LC3B (catalog #2775, 1:1000). The HRP-coupled secondary antibodies used were purchased from Jackson ImmunoResearch Inc, catalog numbers 111-035-003 and 115-035-144, each used at 1:6000. For development of protein signals, SuperSignal West Pico PLUS (Thermo Scientific) was used. For signal detection, the BioRad ChemiDoc imaging system was used. For LC3BII/LC3BI ratio, the intensities of the different lanes were analyzed using Image J (76), normalized on the housekeeping gene control and expressed as a ratio of the two proteins of interest.

Immunofluorescence

70෗10⁵ BMDMs were seeded on Millicell EZ SLIDE 8-well glass chamber slides (Thermo Fisher Scientific) with 500 ml of medium per well. The next day, the cells were stimulated for 2 hours with 100 ng/ml anisomycin alone, 10 ng/ml IFN-γ or 250 IU/ml IFN-β alone or pre-treated with anisomycin for 20 min or with anisomycin and 1 μM PH-797804 (1 hour pre-treatment). Cells were fixed with 3% paraformaldehyde for 15 min at room temperature and then permeabilized with 0.1% Triton-X in PBS for 10 min. Blocking was carried out in 0.1% Triton-X and 3% BSA in PBS for 1 hour. The primary antibody (TORC3/CRTC3; Cell Signaling Technology, catalog #2720, 1:100) was diluted in blocking buffer and incubated for 1 hour at RT. Secondary goat anti-rabbit Alexa Fluor® 488 IgG (H + L; TermoFisher Scientific; Catalog#A-11008, 1:500) was incubated for 30 min in blocking buffer. Samples were then stained with 1 μg/ml DAPI in PBS for 10 min. Samples were mounted in ProLong™ Gold Antifade Mountant (TermoFisher Scientific) and left overnight at 4°C before image acquisition. Images were acquired using Zeiss Axio Imager Z2 with ×40 oil objective and 1.6x magnification. Images were processed using the ImageJ software (76). The background range in all images for DAPI was adjusted to 584-7834 and for GFP 943-16383. The color depth was changed to 16 bit. A composite picture was made and DAPI was set to magenta and GFP to green. The color space of the picture was converted to RGB and a scale bar displaying 5 μm was inserted. For statistics, at least 20 cells per condition were analyzed in four individual replicates. The ratio between CRTC3 abundance in the nucleus and in the cytoplasm was obtained by dividing the mean intensity values (integrated intensity) of the GFP signal in the nucleus (DAPI channel) by the mean intensity values (integrated intensity) of the GFP signal in the surrounding cytoplasm (GFP signal-DAPI signal) using the CellProfiler software v. 4.2.1 (77).

RNA preparation and RT-qPCR

RNA was isolated using the NucleoSpin RNA II kit (Macherey-Nagel, Catalog # 740955). The cDNA synthesis was performed by reverse transcription of 400 ng of total RNA using Oligo(dT)18 primers for mRNA, Random Hexamer Primer for pre-mRNA and the RevertAid Reverse Transcriptase (ThermoFisher Scientific) according to the manufacturer’s instructions. When low amount of RNA was obtained the cDNA synthesis was carried out with the LunaScript RT SuperMix Kit (NEB, Catalog#E3010). Real-time q-PCR were run on the Mastercycler (Eppendorf) using the Luna Universal qPCR Master Mix (NEB, Catalog#M3003). Primers for RT-qPCR are listed in Table S1.

Generation of knock-out BMDMs using CRISPR/Cas9 gene editing

Constructs for gRNA expression

Per target gene one gRNA was designed using the VBC-score (https://www.vbc-score.org/) for each two complementary oligos ordered (table S1). Constructs for gRNA expression were cloned as previously described (78) with minor modifications. Non-targeting gRNA sequences were previously published (79). In short, 1 μg CROPseq-Guide-Puro-mCherry2 plasmid (will be shared via Addgene upon publication) was digested in a mixture with 1 μl BsmBI-v2, 3 μl NEBuffer r3.1 (NEB) and ddH20 to 30 μl total volume. The reaction was incubated at 55°C for 60 min after which 2 μl rSAP (NEB) enzyme was added and the reaction further incubated at 37°C for 60 min and heat-inactive at 80°C for 20 min. Meanwhile, corresponding forward and reverse gRNA oligos were mixed and resuspended to 100 μM in nuclease-free water (ThermoFisher Scientific). To anneal and phosphorylate the oligos 1 μl of the mixture was added to 1 μl 10x T4 Ligation Buffer (NEB), 7.5 μl water and 0.5 μl T4 PNK enzyme (NEB). The reaction was incubated at 37°C for 20 min, 95°C for 5 min and then ramped down to 25°C at 5°C/min. 1 μl of the annealed oligo mixture was mixed with 199 μl water.

For the ligation, 1.6 μl digested and rSAP-treated plasmid are mixed with 1 μl diluted oligo duplex, 5 μl 2x Quick ligase buffer, 2.4 μl water and 1 μl Quick ligase (NEB). This 11 μl reaction was incubated for 15 min at 25°C and 2 μl heat-transformed into 10 μl NEB Stable Competent E. coli (NEB) according to the manufacturer’s protocol. Colonies were grown over night on LB (Merck Millipore) plates with 100 μg/ml carbenicillin (Carl Roth) at 32°C and thereafter incubated in 2 μl shaking liquid LB culture with carbenicillin at 32°C over night. Plasmids were isolated using the Qiagen Plasmid Mini prep kit (Qiagen) and correct ligation was confirmed via Sanger sequencing using the primer: GAGGGCCTATTTCCCATGATTCC (5’-3’).

Lentiviral production

Lenti-X 293T cells (Takara Bio, Catalog#632180) were seeded the day before transfection at 8×10⁶ in 8 ml of medium in 10cm dishes. They day after they were transfected with 10148.04 ng of sgRNA plasmid and the packaging plasmid pMDLg/pRRE, pRSV-Rev, pMD2.G (Addgene) in a 1:1:1 ratio using Lipofectamine 3000 (TermoFischer Scientific) according to the manufacturer’s instructions. 1 day after transfection the supernatant was discarded and replace with 7.5 ml of fresh medium. The transfected cells were cultured for other 24 hours. The lentiviral supernatant was collected 48 and 72 hours after transfection and sterile-filtered with a 0.45 mm sterile syringe filter to remove cell debris.

BMDMs transduction

Rosa26 Cas9 knock-in BMDMs (45) (Cas9 BMDMs) were differentiated for a total of 11 days. On day 2, 5×10⁶ cells were seeded into 10-cm non-treated tissue culture plates with 5ml of medium supplemented with 500 ng/ml of recombinant M-CSF. After 4 hours, Cas9 BMDMs were transduced with 5 ml of lentiviral supernatant supplemented with 8 μg/ml of polybrene (Sigma-Aldrich). The next day the cells were harvested, centrifuged for 5 min at 500g, seeded onto 10-cm non-treated tissue culture plates with 5 ml of medium, and transduced for the second time with 5 ml of lentiviral supernatant supplemented with 8 μg/ml of polybrene. Two days after the first transduction the medium was discarded and replace with fresh medium supplemented with 500 ng/ml of recombinant M-CSF and transduced Cas9 BMDMs were differentiated for 5 more days. At day 9 of differentiation, the mCherry⁺ (sgRNA) GFP⁺ (Cas9) cells were FACS sorted using a FACSAria (BD Biosciences). Sorted cells were cultured for two additional days in medium supplemented with 500 ng/ml of recombinant M-CSF. Transduced BMDMs were analyzes by FACS. For RNA isolation, the RNeasy Micro Kit (Qiagen, Catalog#74004) was used according to the manufacturer’s instructions. Proteins were precipitated with acetone from cell lysates prepared using buffer RTL according to the kit supplementary protocol and boiled at 95°C for 7 min prior to be loaded onto a 10% SDS-polyacrylamide gel.

RNA-sequencing (RNA-seq)

10×10⁶ BMDMs were seeded after 10 days of differentiation in 15cm dishes with 20 ml of medium. The day after, the cells were stimulated for 2 hours with 10 ng/ml IFN-γ or 250 IU/ml IFN-β alone or after pre-treatment with 100 ng/ml anisomycin for 20 min. Total RNA was isolated using the Allprep DNA/RNA mini kit (Qiagen, Catalog#80004) according to the manufacturer’s instructions. The quality controls, RNA sequencing, and preliminary analysis were performed by the Biomedical Sequencing Facility (CeMM, Vienna, Austria; https://www.biomedical-sequencing.org/) The experiment was carried out as three biological replicates per condition.

RNA-seq analysis

RNA-seq data were processed and quality-controlled using established bioinformatics software. Raw reads were trimmed using trimmomatic (v.0.32) (80) and aligned to the mouse reference genome (mm10) using STAR (v.2.7.1) (81). Gene expression was quantified by counting uniquely aligned reads in exons using the function summarizeOverlaps from the GenomicAlignments package (v.1.6.3) in R. Gene annotations were based on the Ensembl GENCODE Basic set (genome build GRCm38 release 93. Differential expression analysis was carried out using DESeq2 (v.1.24.0) (82), with an FDR threshold of 0.05. For comparison between treatments and for establishment of enhanced and not enhanced genes a threshold of Log2FC ≥ 1 expression and an adjusted p-value (padj) ≤ 0.05 were considered. GSEA analysis was performed with GSEA (v.4.0.3) using Log2FC values (83). Volcano plots were generated with the ggplot2 package of the R software (v.4.0.2) using the log2-transformed fold change and -log10-transformed adjusted pvalues for gene expression. Scatterplots were generated with the ggplot2 package of the R software (v.4.0.2) comparing the log2-fold-changes in mRNA abundances from IFN-β and IFN-γ treated BMDMs. Venn diagrams were generated using a Venn tool () with the gene lists obtained from the DESeq2 analysis. Heatmaps were generated using pheatmap tool (v.1.0.12) with the normalized counts from the gene lists obtained from the DESeq2 analysis and clustered by row. PCA plot was generated using the function prcomp() from the base package stats in R (v.4.2.0). Weighted gene co-expression network analysis (WGCNA; v.1.71) was performed in R (v.4.2.0) using as input DESeq2 normalized count data (31). Hierarchical clustering with the function hclust() was performed on the 1-TOM (Topological Overlap Matrix) dissimilarity using average as the agglomeration method. Branch pruning of the dendrogram has been carried out using the cutreeDynamic() function. The module clustered genes of cutreeDynamic() and the expression data were then used to compute the principal components of the modules using the moduleEigengenes(). Spearman correlations of the moduleEigengenes and the traits/conditions are color coded in the heatmap. Correlations below 0 are in blue tones and above 0 are in red tones (fig. S3C). GO enrichment analysis of the WGCNA modules was performed with ShinyGO v. 0.76 (http://bioinformatics.sdstate.edu/go/). For defining the groups of enhanced and not enhanced ISG in Fig. 3, A to C, the significantly (padj ≤ 0.05) genes (An+IFN vs IFN sorted by Log2FC≥1 and An vs UT sorted by Log2FC≤2) and the significantly (padj ≤ 0.05) genes (An+IFN vs IFN sorted by 1<Log2FC>-1 and An vs UT sorted by Log2FC≤2) were filtered, respectively. For identification of STAT1- or ISGF3-bound enhanced genes, previously published ChIP-seq data were used (29). PygenomeTracks (84, 85) was used for RNA-seq tracks visualization. The list of differentially expressed genes upon anisomycin stimulation (volcano plot in fig. S3B) is provided in data file S1; the list of differentially expressed genes upon IFN−γ stimulation and STAT1/ISGF3-bound ISG (Fig. 3A and fig. S3F) is provided in data file S2; and the list of differentially expressed genes upon IFN-β stimulation and ISGF3-bound ISG (Fig. 3B and fig. S3F) is provided in data file S3.

ChIP and ChIP-seq

In total, 1.5 × 10⁷ bone marrow-derived macrophages were seeded on a 15cm dish with 20 ml of medium supplement with 500 ng/ml M-CSF. The next day, cells were stimulated for 2 hours with 10 ng/ml IFN-γ or 250 IU/ml IFN-β alone or after pre-treatment with 100 ng/ml anisomycin for 20 min. In order to block JNK and p38 MAPK activity, 1 μM PH-797884 together with 20 μM JNK inhibitor II were added 1 hour prior to anisomycin stimulation. Cells were processed with the ChIP protocol as previously described (32), using the following antibodies: RNA polymerase II CTD repeat YSPTSPS phospho-Ser⁵ (Abcam, catalog #ab5408, 5 μl); RNA Polymerase II phospho-Ser² (Bethyl, catalog #A300-654A, 7 μl); RNA polymerase II CTD repeat YSPTSPS (Abcam, catalog #ab817, 10 μl); histone H3 acetyl-Lys²⁷(Cell Signaling Technology, catalog #8173, 5 μl); histone H3K27me3 (Active Motif, catalog #39155, 5 μl); histone H3 acetyl-Lys⁹/Lys¹⁴ (Cell Signaling Technology, catalog #9677, 5 μl); H2A.Z (Cell Signaling Technology, catalog #50722, 2 μl); histone H3 phospho-Ser²⁸ (Abcam, catalog #ab32388, 5 μl); histone H3 phospho-Ser¹⁰ (Abcam, catalog #ab5176, 10 μl); histone H3 (Abcam, catalog #ab1791, 2 μl); STAT1 (Cell Signaling Technology, catalog #14995, 10 μl); CREB phospho-Ser¹³³ (Cell Signaling Technology, catalog #9198, 5 μl); CREB (Cell Signaling Technology, catalog #4820, 5 μl); c-Jun (Cell Signaling Technology, catalog #9165, 10 μl); and rabbit monoclonal IgG isotype control (Clone DA1E; Cell Signaling Technology, catalog #3900, 1 μl). Real-time qPCR assays were run on the Mastercycler (Eppendorf). Primers for ChIP-qPCR are listed in table S1.

For ChIP-seq, a minimum of 3 IPs per condition were pooled to obtain enough starting material. Library preparation, quality check and sequencing were performed by the Vienna Biocenter Core Facilities NGS Unit (https://www.viennabiocenter.org/vbcf/next-generation-sequencing/). Libraries were sequenced on NovaSeq S1 PE100.

ChIP-seq analysis

ChIP-seq data were processed using the ChIP-seq pipeline from the nf-core framework (nfcore/ChIP-seq v.1.2.2; https://zenodo.org/record/3529400#.YA8YGmRKjZk). Reads were aligned against the Illumina iGenome Mus musculus GRCm38 reference genome.

Downstream analysis of ChIP-seq data was performed using MACS2 narrow-peak calling (included in nf-core/atacseq pipeline). Volcano plots were obtained using ggplot2. PygenomeTracks (84, 85) was used for visualization of ChIP-seq tracks and previously published ChIP-seq tracks (29). Biological replicates were merged when an overlap was found in at least 2 replicates using BEDOPS (v2.4.40), using BED files after narrow-peak calling. All different treatments were merged with the same tool and each peak was annotated to a gene by using the HOMER script annotatePeaks.pl. (86). Venn diagrams were generated using a Venn tool (https://bioinformatics.psb.ugent.be/webtools/Venn/) with the peak lists obtained from the annotated merged BED files and from the gene lists obtained from the DESeq2 analysis of the RNA-seq data merging data from both IFN types (IFN vs UT; Log2FC≥1; padj ≤ 0.05). The genes from the RNA-seq were filtered from the IFN list for padj ≤ 0.05 and Log2FC≥1. GSEA was performed using ChIP-Enrich (48) (Bioconductor) R tool with the hybridenrich method. All enriched pathways or transcription factor binding sites were considered significant with an FDR p-value≤0.05. The built-in gene sets used in this analysis are Kegg_pathway (Kyoto Encyclopedia of Genes and Genomes v.3.2.3. (genome.jp/kegg)) and transcription_factors (Transcription Factors (MSigDB) v.6.0; software.broadinstitute.org/gsea/msigdb/collections.jsp). List of ISG costitutively bound by both c-Jun and CREB1, only by c-Jun or only by CREB1 is provided in data file S4.

ATAC-seq

3×10⁶ BMDMs were seeded onto 6-cm non-treated tissue culture plates on day 7 of differentiation. The next day, cells were stimulated for 2 hours with 10 ng/ml IFN-γ or 250 IU/ml IFN-β alone or pre-treated with 100 ng/ml anisomycin for 20 min. In order to block p38 MAPK activity 1 μM PH-797884 was added for 1 hour prior to anisomycin stimulation. Cells were processed as previously described (25). The quality of the libraries was confirmed on a bioanalyzer to further determine the size distribution. Libraries were sequenced on a NovaSeq SP PE50.

ATAC-seq analysis

ATAC-seq data were processed using the ATAC-seq pipeline from the nf-core framework (nfcore/atacseq v1.2.1, https://zenodo.org/record/3965985#.YA8a8GRKjZk). Reads were aligned against the Illumina iGenome Mus musculus GRCm38 reference genome. Downstream analysis of ATAC-seq data was performed using MACS2 narrow-peak calling and differential chromatin accessibility analysis (included in nf-core/atacseq pipeline).

For the generation of summary profile plots and heatmaps, density information (bigwig) for gene regions and surrounding regions (2 Kb upstream of TSS and 1 Kb downstream of TES) was plotted using deeptools (v3.4.3) (https://zenodo.org/record/3965985#.Yh-DPejMJPZ). The list of genes used for summary profile plots and heatmaps were obtained accordingly to the RNA-seq data. The computeMatrix command was used in the scale-regions mode with the option missingDataAsZero. Subsequently, plotHeatmap was used for the generation of heatmaps and summary profiles. PygenomeTracks was used for track visualization (84, 85).

BMDMs infection with Listeria monocytogenes

For BMDMs infection the LO28 strain of Listeria monocytogenes (L. monocytogenes) was grown overnight in brain heart infusion (BHI) medium at 37°C with continuous shaking. After reaching stationary phase, bacterial concentration was measured with a spectrophotometer and the correct volume of bacterial culture medium (1OD600nm = 1×10⁹ viable bacteria) was transferred into a tube and pelleted by centrifugation. Bacteria were washed twice with PBS and resuspended in DMEM supplemented with 10% FBS. BMDMs were seeded in DMEM supplemented with 10% FBS and 500 ng/ml M-CSF and infected with MOI 10 for the indicated time. After 1 hour the medium was changed to DMEM supplemented with 10% FBS and 50 μg/ml gentamycin (MP Biomedicals). One hour later, the medium was replaced with DMEM supplemented with 10% FBS and 10 μg/ml gentamycin, which was kept until the end of the experiment. Primers for RT-qPCR are listed in table S1.

Colony-Forming-Unit assay

For in vitro colony forming unit (CFU) assays, 5×10⁵ BMDMs were seeded in 200 μl of DMEM supplemented with 10% FBS and 500 ng/ml M-CSF in 96-well plates and infected with L. monocytogenes at different time-points. After each time-point, cells were washed twice with PBS and lysed in 50 μl sterile nuclease-free water twice for 5 min at 37°C. For quantifying bacterial loads, four 1:10 serial dilutions of these lysates were plated on plates containing BHI medium and incubated for 1 day at 37°C.

Measurement of NO production

For measurement of NO production, 1×10⁶ BMDMs were seeded in 500 μl of DMEM supplemented with 10% FBS and 500 ng/ml M-CSF in 24-well plates. After 24 hours of L. monocytogenes infection (MOI 10), NO production was measured indirectly by assaying the concentration of nitrite in the cell culture medium with the Griess method. 100 μl of cell supernatant was transferred into a 96-well plate in triplicates to which it was added 100 μl of sulfanilic acid solution (1% sulfanilamide in 5% phosphoric acid) and 100 μl of NEDD solution [0.1% N-(1-naphthyl)ethylenediamine in double-distilled water] and the absorbance was immediately measured at 545 nm with a plate reader. Serial dilutions of sodium nitrite (NaNO2) were prepared and used as standard curve to assess the NO concentration in the samples.

Propidium iodide staining and flow cytometry

For assessment of cell death, 1×10⁶ BMDMs were seeded in 2 ml of DMEM supplemented with 10% FBS and 500ng/ml M-CSF in 6-well plates. Cells were infected with L. monocytogenes for 24 hours. For flow cytometry, BMDMs were harvested by incubation for 5 min in citric saline (0.135 M potassium chloride, 0.015 M sodium citrate), collected by centrifugation together with the culture medium to include cells in suspension and then washed twice in PBS. Cell pellets were resuspended in FACS buffer (1x PBS + 2% BSA) and stained with propidium iodide (PI; Thermo Fischer Scientific) at a final concentration of 1 μg/ml immediately prior to FACS analysis. Unstained cells and untreated cells were used as negative control, and cells killed by 95°C incubation were used as positive control. FACS analysis was carried out with BD LSRFortessa™ Cell analyzer at the Max Perutz Labs FACS facility (https://www.maxperutzlabs.ac.at/research/facilities/biooptics-facs). Analysis of the percentage of of PI-positive cells was performed with FlowJo (https://www.flowjo.com/). A minimum of 10000 cells were acquired for each dot plot. The percentage of positive cells in the FACS analysis represents the percentage of PI positive cells over the live-cell population. This was calculated using a forward and side-scatter gate to exclude dead cells and debris from the analysis.

VSV-GFP infection and flow cytometry analysis

WT BMDMs were infected with VSV-GFP at MOI 5 for 4 hours for RT-qPCR and for 24 hours for analysis of GFP expression by FACS. For flow cytometry, BMDMs were harvested by incubation for 5 min in citric saline (0.135 M potassium chloride, 0.015 M sodium citrate), collected by centrifugation together with the culture medium to include cells in suspension and then washed twice in PBS. Cell pellets were resuspended in FACS buffer (1x PBS + 2% BSA) and GFP expression was analyzed using BD LSRFortessa™ Cell analyzer at the Max Perutz Labs FACS facility. Analysis of the percentage of GFP-positive cells was performed with FlowJo. A minimum of 8000 cells were acquired for each dot plot. The percentage of positive cells in the FACS analysis represents the percentage of VSV-positive cells over the live-cell population. This was calculated using a forward and side-scatter gate to exclude dead cells and debris from the analysis.

Quantification and Statistical Analysis

Experiments were repeated at least three times with the following exceptions: RT-qPCR of BMDMs derived from p38α^ΔM mice was performed on two biological replicates per condition, and RT-qPCR of BMBM derived from knock-in CREB-S133A mice was performed with cells from 2 mice differentiated into 3 different plates. All data are presented as the mean values with the standard deviation (SD). Differences in mRNA/pre-mRNA expression data, NO measurement, and PI staining were compared using one-way ANOVA and corrected for multiple testing with Dunnett’s post-hoc test. Differences in ChIP data and in mRNA expression data between wild-type and knock-out cells were compared using two-way ANOVA and corrected for multiple testing with Dunnett’s post-hoc test. CFU assay significance was assessed with two-tailed unpaired t-test. All statistical analysis was performed using GraphPad Prism V7 (GraphPad) software. Statistical significance is noted in the figures as follows: ns: p > 0.05, ∗: p ≤ 0.05, ∗∗: p ≤ 0.01, ∗∗∗: p ≤ 0.001, ∗∗∗∗: p ≤ 0.0001. Each dot represents one biological replicate. The number of independent biological replicates (n) for all experiments is indicated in the figure legends. For immunoblots and immunofluorescence, the reported images are representative of at least three independent experiments.

One-sentence summary.

Our findings demonstrate how JAK-STAT and stress kinase signaling results in synergistic transcriptional activation of interferon-induced genes.

Data and materials availability

All data needed to evaluate the conclusions in the paper are present in the main text or the Supplementary Materials. Publicly available raw data used in this paper are available under the accession number GEO: GSE115435 (STAT1, STAT2, IRF9 ChIP-seq; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE115435). All the sequencing data generated for this publication have been deposited in NCBI’s Gene Expression Omnibus and are available under the accession number GEO: GSE199166 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE199166). The paper does not report original code.